Concerns about Nvidia’s high-power 16-pin connector just won’t go away. Even with the updated 12V-2×6 standard, fresh reports of melting connectors on the flagship GeForce RTX 5090 are popping up, making people question whether the design is really up to the job for these power-hungry GPUs.

In the latest cases, hardware site Club386 shared that their RTX 5090 Founders Edition got badly damaged after the 12V-2×6 power connector melted during testing. The photos they posted show badly charred and deformed pins on both the graphics card and the cable, plus damage to the power supply. Importantly, they were using a solid 1000W Be Quiet Dark Power 13 PSU with a single native 12V-2×6 cable, properly installed, no user mistakes.

YouTuber Daniel Owen also ran into the same problem on his RTX 5090 Founders Edition. The connector melted even though he used a 4×8-pin to 12V-2×6 adapter. These failures add to a growing list of issues with the RTX 50-series since launch, happening even when people lower power limits and use the right cables.

The 12V-2×6 connector was supposed to fix the melting problems from the older 12VHPWR design. It has shorter sense pins to better detect if the cable isn’t fully seated and can handle more power. But real-world use, especially with the RTX 5090’s massive spikes that can go over 500-600W, keeps exposing weaknesses. Things like uneven current across the pins, slight cable bending, and heat buildup seem to still be causing trouble.

Club386 didn’t hold back, calling the connector a “flawed” design that’s “not fit for purpose” and pushing Nvidia to finally admit there’s still a problem. These incidents show how tough it is to push hundreds of watts through a small connector during heavy gaming, AI work, or benchmarking.

Nvidia has stood by the connector, saying it meets ATX 3.1 specs and recommending native cables from good PSUs. Some third-party companies like MSI have started offering reinforced cables with better monitoring. Still, these high-profile failures mean that if you’re dropping big money on an RTX 5090, it’s smart to double-check your cabling, make sure everything’s fully seated, and maybe consider undervolting.

If you’re affected, make sure to document everything for RMA purposes. It remains to be seen if or when Nvidia comes up with a proper hardware fix or other solutions to finally end this connector drama.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: Club386)

The post Another RTX 5090 faces power connector meltdown during testing appeared first on Gizmochina.

NVIDIA’s new RTX Spark platform was one of the biggest highlights at Computex Taipei 2026, with major PC brands including ASUS, Dell, Lenovo, and MSI unveiling compact desktop systems powered by the new AI-focused architecture. The launch marks NVIDIA’s first major step into the PC processor market and could reshape the future of AI PCs, creator workstations, and compact desktops.

NVIDIA RTX Spark Platform Takes Center Stage

The RTX Spark platform is an Arm-based system-on-chip (SoC) built around NVIDIA’s Grace and Blackwell technologies. It combines a 20-core CPU with a Blackwell GPU featuring 48 SM units and 6,144 CUDA cores. The platform also supports up to 128GB of LPDDR5X unified memory.

NVIDIA says the chip is designed for local AI development, content creation, gaming, and AI inference workloads. The company also claims graphics performance comparable to an RTX 5070-class GPU in certain scenarios.

ASUS, Dell, Lenovo, and MSI Showcase New Mini PCs

Several manufacturers introduced RTX Spark-powered mini PCs during the event. While each system features a unique chassis design, they share the same core platform and connectivity options.

ASUS ProArt GA10

The ASUS ProArt GA10 features a compact 150 × 150 × 51 mm design and supports up to 140W of cooling capacity. It includes four USB-C ports, HDMI output, and 10 Gigabit Ethernet connectivity.

Dell XPS RTX Spark Desktop

Dell’s new compact desktop also adopts the RTX Spark platform. The system includes four USB-C ports, HDMI output, and 10GbE networking while maintaining a small footprint.

Lenovo RTX Spark Mini PC

Lenovo introduced a newly designed mini PC chassis built around RTX Spark. It offers the same set of connectivity options, including four USB-C ports, HDMI, and 10GbE Ethernet.

MSI EdgeMesa N AI+

MSI’s EdgeMesa N AI+ stands out with its white chassis design. The system targets content creators and AI developers while offering the same RTX Spark hardware foundation.

Shared Features Across All Models

Every RTX Spark mini PC showcased at Computex includes 10GbE networking, 20Gbps USB-C connectivity, PCIe Gen 5 M.2 SSD support, and up to 128GB unified memory.

Why RTX Spark Could Be a Game-Changer for PC Buyers

For consumers, RTX Spark could bring desktop-class AI performance, powerful graphics, and massive memory capacities into much smaller and more efficient PCs. Because the CPU, GPU, and up to 128GB of unified memory work together on a single platform, users may be able to run advanced AI models, edit high-resolution videos, and handle demanding creative workloads locally without relying heavily on cloud services.

If NVIDIA’s performance claims hold up, RTX Spark systems could also offer a compelling alternative to traditional Intel and AMD-powered PCs, increasing competition and giving buyers more choice. 

The trade-off is that some older Windows applications and specialized software may face compatibility or optimization challenges, although Windows on Arm support has improved significantly in recent years.

Read More:

• US- China Tech War: China Just Beat NVIDIA in the Next Big AI Race
• Why Nvidia’s H2+ Humanoid Robot Has the Tech World Talking
• US Expands AI Chip Restrictions Worldwide: What It Means for China

(via)

The post NVIDIA RTX Spark Mini PCs Arrive at Computex 2026 — Could They Replace Traditional Desktops? appeared first on Gizmochina.

For the past two years, Qualcomm has been the only real player to offer an ARM-based chip for Windows PCs. It started with the Snapdragon X Elite in 2024 which Qualcomm followed with the more powerful Snapdragon X2 series in late 2025. 

Qualcomm’s sole reign on the Windows-on-ARM (WoA) market is now officially over, all thanks to Nvidia. The latter announced its new ARM-based RTX Spark PC chips at Computex 2026, and Nvidia wants it to be the Apple Silicon moment for Windows.

The news sent Qualcomm’s stock tumbling more than 10% in premarket trading, wiping out over $10 billion in market value overnight. That reaction tells you a lot about how big a deal this really is.

So what exactly are these two chips, how do they compare, and who should you care about? Let’s break it down.

What Is the Nvidia RTX Spark?

RTX Spark is Nvidia’s answer to the growing wave of ARM-based, all-in-one chips, similar to what Apple’s been doing brilliantly with its M-series and Qualcomm has been working hard with the Snapdragon X lineup.

At its core, RTX Spark is based on the same GB10 Grace Blackwell Superchip that powers Nvidia’s DGX Spark developer workstation. The company just repackaged it for everyday Windows laptops. It combines a custom ARM-based CPU (developed with MediaTek), a Blackwell RTX GPU, and an AI accelerator all on one chip, built on TSMC’s 3nm process with 70 billion transistors.

And the headline numbers are impressive. RTX Spark can pack up to 20 CPU cores, a 6,144-core Blackwell GPU, and up to 128 GB of unified memory. Nvidia claims it delivers over 1 petaflop of FP4 AI compute, which in simple terms is enough to run massive AI and Large Language Model (LLM) models locally. 

The CPU and GPU are connected together via Nvidia’s NVLink-C2C interconnect, which enables up to 600 GB/s of bandwidth between the two.

According to team green, devices powered by RTX Spark are set to arrive in autumn 2026 from major brands including ASUS, Dell, HP, Lenovo, and Microsoft Surface.

What Is the Snapdragon X2 Elite?

Qualcomm launched the Snapdragon X2 series at its annual Snapdragon Summit in September 2025. It is the direct follow-up to the original Snapdragon X Elite from 2024, and it brings a meaningful generational step forward, especially in CPU performance.

The X2 family consists of three chips: the X2 Plus, the X2 Elite, and the top-of-the-line X2 Elite Extreme. All three are built on TSMC’s 3nm process and feature Qualcomm’s third-generation Oryon CPU cores, but the Extreme version is the one that gets the major attention.

The X2 Elite Extreme: Qualcomm’s Best Shot

The Snapdragon X2 Elite Extreme (model number X2E-96-100) is Qualcomm’s most powerful laptop chip ever made. It packs 18 Oryon cores versus 12 cores in the first-generation Snapdragon X Elite. 

Those 18 cores are split into two groups: 12 fast “Prime” cores that can boost up to 5 GHz on two cores (and run at 4.4 GHz across all cores), and 6 “Performance” cores running at up to 3.6 GHz. Together, they share 53 MB of cache.

On the AI side, the X2 Elite Extreme includes an 80 TOPS NPU (Neural Processing Unit), which is Qualcomm’s dedicated AI engine. All three X2 Elite chips share this same NPU rating.

Benchmark numbers from Qualcomm’s own Snapdragon Summit testing were genuinely strong on the CPU side. In Cinebench 2024 and Geekbench 6.3 multi-core tests, the X2 Elite Extreme was able to match or beat Apple’s M4 Pro, scoring 1,964 and 23,693 points respectively.

Compared to the original Snapdragon X Elite, the X2 Elite Extreme offers roughly a 39% improvement in single-core performance and a 50% boost in multi-core performance. In gaming benchmarks, Qualcomm claims a 2.3x improvement over the previous generation.

The chip also supports up to 48 GB of LPDDR5x memory on a 192-bit bus (in the Extreme version), offers PCIe 5.0 storage, and handles up to three USB 4.0 ports. Laptops running the X2 Elite Extreme began arriving in the first half of 2026, including the ASUS Zenbook A16 which features the chip prominently.

GPU and Gaming: Qualcomm’s Weak Spot

Here is where things get complicated for Qualcomm. While the X2 Elite Extreme impresses on CPU tasks, the GPU side tells a different story.

In the demanding 3DMark Steel Nomad benchmark, the X2 Elite Extreme managed just 1,306 points at 13 frames per second. Apple M4 Pro, in comparison, manages around 1620 points in the same test.

The Solar Bay and Steel Nomad Light scores were more respectable, but the GPU improvement is not nearly as dramatic as the CPU gains. One reviewer from Tom’s Hardware found that professional tools like AutoCAD were unsupported, and some games crashed or showed graphical errors when running under emulation on Windows ARM.

RTX Spark vs Snapdragon X2: Head to Head

This is where Nvidia has a structural advantage that Qualcomm simply cannot match right now.

AI Performance: RTX Spark delivers over 100 TOPS of on-device AI computing power. The Snapdragon X2 Elite’s NPU sits at 80 TOPS.

GPU Power: The RTX Spark’s Blackwell GPU supports DLSS 4.5, Nvidia’s AI-powered graphics upscaling technology that makes games look better and run faster. No Snapdragon chip has anything comparable. For gaming and creative work like video editing or 3D rendering, RTX Spark is in a completely different league.

Memory: RTX Spark supports up to 128 GB of unified memory. The X2 Elite Extreme tops out at 48 GB. For heavy AI workloads and large language models running locally, this is a massive difference.

CPU: This is more even. The X2 Elite Extreme is a very strong CPU performer, matching Apple’s M4 Pro in many tests. RTX Spark’s 20-core CPU will be tested properly when devices launch in autumn 2026, but based on Nvidia’s track record, expectations are high.

Software Ecosystem: This may be the most important difference of all. Nvidia brings CUDA — its developer software platform — to RTX Spark. For 15-plus years, virtually every major AI model, AI framework, and GPU-accelerated app has been built on CUDA. Meanwhile, Qualcomm has no equivalent. Snapdragon X laptops have also faced ongoing issues with app compatibility, where x86 apps running through Microsoft’s Prism emulation layer sometimes behave poorly.

Why Qualcomm Is Still Relevant

Despite all of this, Qualcomm is not out of the race. The Snapdragon X2 series chips are available right now, while RTX Spark laptops won’t arrive until autumn 2026. For buyers who need a capable ARM-based Windows laptop today, the X2 Elite Extreme is genuinely impressive, especially for CPU-heavy work, everyday productivity, and battery life.

Moreover, Qualcomm is targeting a broader market just like its smartphone chips. In fact, the company recently unveiled a Snapdragon C platform for affordable AI powered laptops. 

Meanwhile, RTX Spark is likely more aimed at creators, AI developers, and serious gamers willing to pay a premium. In fact, the starting price of the laptop could be around $1800. 

The Bottom Line

Nvidia’s RTX Spark is a serious statement of intent. It takes the company’s unmatched software ecosystem, combines it with genuinely next-generation hardware, and packages it into a laptop chip that can go head-to-head with Snapdragon X2 platform and even Apple M5 chips.

Qualcomm’s X2 series is also not a bad chip. On CPU performance, it holds its own against the best in the business. But the GPU weakness, the software compatibility headaches, and the relatively modest NPU performance all become harder to defend once RTX Spark laptops start hitting store shelves.

For buyers looking at Windows laptops today, the Snapdragon X2 Elite Extreme is a strong choice, particularly in machines like the ASUS Zenbook A16. But if you can wait until late 2026, RTX Spark could be worth holding out for.

For more daily updates, please visit our News Section.

Tech enthusiast? Get the latest news first! Follow our Telegram channel and subscribe to our free newsletter for your daily tech fix!

The post Qualcomm Snapdragon X2 Elite vs Nvidia RTX Spark: ARM chip for Windows, but which is better? appeared first on Gizmochina.

Nvidia‘s upcoming RTX Spark laptops may end up costing a lot more than many expected, putting them directly in competition with Apple’s MacBook Pro lineup.

Fresh pricing estimates from a Morgan Stanley report suggest systems powered by the flagship N1x chip could start at around $2,899. Meanwhile, laptops based on the standard N1 platform are expected to begin at roughly $1,799. If those figures prove accurate, RTX Spark won’t be targeting the mainstream market. Instead, Nvidia appears to be aiming squarely at creators, developers, and power users willing to pay a premium.

The hardware itself is certainly ambitious. RTX Spark combines Nvidia’s ARM-based Grace CPU with a Blackwell RTX GPU, offering up to 20 CPU cores, 6,144 CUDA cores, and as much as 128GB of unified memory. Nvidia is also claiming up to 1 petaflop of AI performance, a figure that highlights just how heavily AI workloads are baked into the platform.

Gaming doesn’t seem to be an afterthought either. During Computex, Nvidia showcased Forza Horizon 6 running at over 100 FPS at 1440p, while partners repeatedly emphasized the platform’s ability to deliver high performance without the bulk usually associated with gaming laptops.

Several manufacturers are already on board. Microsoft’s Surface Laptop Ultra, Dell’s refreshed XPS 16 Creator Edition, Asus’ ProArt P16 and P14, HP’s OmniBook X 14 and OmniBook Ultra 16, Lenovo’s Yoga Pro 9n, and MSI’s Prestige N16 Flip AI+ were all shown as part of the initial wave.

What’s perhaps more interesting is the scale Nvidia is aiming for. The company says the ecosystem could grow to around 30 laptop models and 10 desktop systems by the fall, with more devices expected from Acer and Gigabyte.

Of course, impressive specifications only tell part of the story. Windows on Arm still has questions to answer when it comes to software compatibility and real-world efficiency. Battery life will matter too, especially if Nvidia wants these machines to be viewed as genuine MacBook Pro alternatives rather than just powerful Windows laptops.

Still, RTX Spark is shaping up to be one of Nvidia’s most ambitious PC initiatives in years. The pricing may raise some eyebrows, but the company clearly isn’t trying to compete on affordability.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Via)

The post Here’s how much Nvidia’s RTX Spark-powered laptops could cost appeared first on Gizmochina.

NVIDIA has officially pulled the wraps off RTX Spark, a new Arm-based PC platform that could end up being one of the more interesting Windows launches of the year.

Announced by Jensen Huang during his Computex 2026 keynote, RTX Spark combines an NVIDIA Grace CPU, a Blackwell-based RTX GPU, and a large pool of unified memory into a single platform aimed at laptops and compact desktops.

The hardware itself is what makes the announcement stand out.

The CPU uses a 20-core Arm design developed in collaboration with MediaTek. According to NVIDIA, it combines ten Cortex-X925 performance cores with ten Cortex-A725 efficiency cores, a configuration that should look familiar to anyone following the smartphone chipset market. In a way, RTX Spark feels like NVIDIA taking ideas that worked well in flagship phones and scaling them up for PCs.

Then there is the GPU.

RTX Spark integrates a Blackwell graphics processor with 6,144 CUDA cores and fifth-generation Tensor Cores. NVIDIA claims up to 1 PFLOP of AI performance, which is the sort of figure that would have sounded absurd for a thin laptop not all that long ago.

Perhaps the most interesting part is the memory architecture. Instead of treating system memory and graphics memory as separate pools, RTX Spark uses up to 128GB of unified LPDDR5X memory shared across the entire system. For AI workloads especially, that could end up being just as important as the raw processing power.

NVIDIA is also leaning heavily on its existing software ecosystem. CUDA, TensorRT, DLSS, Reflex, ray tracing, and the rest of the RTX stack are all part of the package. That gives the platform an advantage that many Windows-on-Arm efforts have struggled with in the past: software support.

The company showed plenty of AI-focused demos during the keynote, but realistically, many buyers will probably care just as much about battery life and gaming performance. NVIDIA says RTX Spark devices can be as thin as 14mm while still delivering high-end graphics capabilities, although we’ll have to wait for independent testing to see how those claims hold up.

The first systems are expected from ASUS, Dell, HP, Lenovo, MSI, and Microsoft’s Surface division later this year.

There is still a lot we don’t know, including pricing and real-world battery life. But one thing is fairly clear already: NVIDIA is no longer content with supplying GPUs for PCs.

(Via)

The post NVIDIA RTX Spark’s custom 20-core Grace CPU details revealed appeared first on Gizmochina.

At Computex Taipei 2026, NVIDIA‘s CEO took the stage and officially unveiled the RTX Spark, the company’s first-ever processor built specifically for Windows PCs. It’s an Arm-based chip that packs a 20-core Grace CPU (built alongside MediaTek), a Blackwell RTX GPU with 6,144 CUDA cores, an NPU, and up to 128GB of unified LPDDR5X memory, all on a single TSMC 3nm package. 

As a result, NVIDIA says the RTX Spark is capable of 1 petaFLOP of AI performance. That’s a lot of zeroes.

The chip, formerly rumored under the codename “N1X,” is clearly NVIDIA’s answer to Apple silicon, and a direct challenge to Qualcomm’s Snapdragon X series that’s been dominating the Windows on Arm conversation for the past year. 

And NVIDIA isn’t just bringing its hardware here. The full software stack, including CUDA, TensorRT, DLSS 4.5, Reflex, G-SYNC, and RTX ray tracing, is all coming along for the ride.

What is NVIDIA RTX Spark capable of?

So what can it actually do? NVIDIA says the RTX Spark can run 120 billion parameter AI models locally, handle 12K 4:2:2 video editing, render 3D scenes larger than 90GB, and push AAA games at 1440p above 100 frames per second. The latter is powered by DLSS 4.5 and Frame Generation, which NVIDIA has been shipping on its desktop cards for a while now.

Adobe is on board too, and not in a half-hearted way. The company is doing a complete architectural overhaul of both Photoshop and Premiere to support RTX Spark, promising up to 2x improvements in AI and graphics performance. Blackmagic Design, Blender, CapCut, ComfyUI, and OTOY have also signed on.

The laptops powered by the RTX Spark will be slim and trim. NVIDIA says 14mm of thickness, 3 pounds of weight, and availability in 14-inch and 16-inch sizes, with OLED displays and precision-machined aluminum chassis. 

The first PCs with the new chip will launch this fall from brands like ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI. 

One thing that’s missing is the actual benchmark numbers. It’s still a few months until fall, but we can expect leaks revealing some performance figures before that. 

For more daily updates, please visit our News Section.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories! 

The post NVIDIA RTX Spark is here, and it wants to be the Apple Silicon moment for Windows PCs appeared first on Gizmochina.

Nvidia is about to announce its first Arm chips for Windows laptops at Computex tomorrow. However, someone may have spoiled the surprise for Nvidia. 

Just a day before the expected announcement, detailed specifications of Nvidia’s upcoming N1-series Arm processors for PCs and laptops have surfaced online, courtesy of VideoCardz, which cited internal Nvidia documents.

The leak reveals at least four chip variants in the N1 family, covering everything from high-performance workhorses to power-efficient options for thin, mainstream laptops.

Some context first

For quite a while now, the Windows-on-Arm story has mostly been Qualcomm’s story. The company’s Snapdragon X Elite chips, built on a 4nm process with Qualcomm’s own Oryon CPU cores, brought real competition to Intel and AMD for the first time in years on thin-and-light laptops. 

Qualcomm made big claims about beating Intel Core i7 chips on multi-threaded benchmarks, and for the most part, the real-world results backed that up.

Now Nvidia wants in. And Nvidia isn’t just bringing a CPU to the fight; it’s bringing a Blackwell GPU along for the ride.

Nvidia N1X

The top-tier N1X reportedly shares its core design with the GB10 processor inside Nvidia’s DGX Spark desktop AI supercomputer. That’s not a casual comparison. The full-fat N1X is said to pack a 20-core CPU — ten Cortex-X925 performance cores and ten Cortex-A725 efficiency cores — paired with a Blackwell 2.0 GPU featuring 48 Streaming Multiprocessors, which works out to 6,144 CUDA cores.

A slightly trimmed N1X variant is also in the works, dropping to 18 CPU cores (nine performance, nine efficiency) and a 40-SM GPU with 5,120 CUDA cores. Both N1X chips are designed to run at 45W to 80W. Crucially, though, that power figure covers the entire chip package: CPU and GPU together.

Nvidia N1

The standard N1 lineup is for thinner, more affordable devices. Two variants have reportedly been planned. The higher-end one pairs eight Cortex-X925 performance cores and four Cortex-A725 efficiency cores with a 20-SM GPU delivering 2,560 CUDA cores. 

The second, more entry-level option steps down to a 10-core CPU (seven performance, three efficiency) and a 16-SM GPU with 2,048 CUDA cores. The whole N1 family runs within an 18W to 45W power envelope.

Memory and Storage

The two families differ significantly under the hood. The N1X supports up to 128GB of LPDDR5X memory across a 16-channel interface, while the standard N1 caps at 64GB with an 8-channel setup. 

Likewise, N1X supports up to three M.2 SSDs, while the N1 tops out at two.

How Long Has This Been Cooking?

According to VideoCardz, at least one of the leaked slides is dated 2024, meaning Nvidia may have been working on this for two years or more. Not every chip listed in these documents is guaranteed to ship; roadmaps change, and some variants may quietly disappear. But the scale of what’s been leaked suggests Nvidia had serious, long-term ambitions here.

For more daily updates, please visit our News Section.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories! 

(Source)

The post Nvidia’s first ARM chip for PCs, N1 and N1x, leaks and it looks like a big deal appeared first on Gizmochina.

NVIDIA’s DLSS 5 rollout hasn’t exactly been smooth, and now there’s another twist. The company’s own promotional trailer has disappeared from YouTube, but not for the reason you’d expect.

The video, which was first shown at GTC last month, highlighted DLSS 5’s new neural rendering capabilities. The demo clips, including scenes from Resident Evil: Requiem, drew attention for their lighting and overall visual quality. At the same time, though, the tech quickly became a talking point for less positive reasons.

A lot of the criticism has focused on how DLSS 5 affects visuals, especially character faces. Some players feel the AI processing smooths things out too much, giving scenes a slightly artificial look. The reaction online has been fairly divided, and in some cases quite harsh. Reports suggested the official trailer had a low like ratio despite pulling in millions of views.

NVIDIA CEO Jensen Huang has addressed those concerns more than once, saying developers still have control over how the tech is used and that it’s meant to support, not override, a game’s original style. Still, the debate hasn’t really settled.

Now the situation has taken a more unusual turn. The DLSS 5 trailer on NVIDIA’s GeForce YouTube channel was removed, but it wasn’t taken down by NVIDIA itself. Instead, it was pulled following a copyright claim.

According to reports, Italian broadcaster La7 used parts of the trailer in a program and then ended up triggering a copyright claim through YouTube’s system. The result was that not only NVIDIA’s original video, reportedly with over 2 million views, but also other uploads using the same footage were blocked.

It’s the kind of issue that points back to YouTube’s automated Content ID system, which doesn’t always get things right. In this case, the trailer is clearly NVIDIA’s own content, so the takedown is likely temporary. Still, it adds another layer to what has already been a somewhat messy rollout for DLSS 5.

For now, the conversation around the technology, and how it’s being used, doesn’t seem to be slowing down anytime soon.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

The post YouTube took down NVIDIA’s own DLSS 5 trailer following a copyright issue appeared first on Gizmochina.

Trying to squeeze every last bit of performance out of high-end GPUs is nothing new. But in the case of the RTX 5090, one ambitious modding attempt ended in a very expensive lesson.

In a recent YouTube video titled “I Killed My 5090…”, the creator behind Frame Chasers walked viewers through what started as a bold experiment on a Gigabyte RTX 5090 Aorus Master, and ended with a completely dead card.

The idea, at least on paper, made sense. Some Gigabyte variants include an unpopulated pad on the PCB for a second 12V-2×6 (16-pin) power connector. It’s there for alternative board designs, including versions that use dual connectors. Seeing that unused pad, Frame Chasers decided to take things further by soldering in a second power connector.

That wasn’t the only modification. He also performed a shunt resistor mod, a well-known (and risky) tweak that effectively bypasses NVidia’s built-in power limits. The goal was simple: let the GPU draw far more power than stock settings allow.

At first, it appeared to work. The system booted, benchmarks ran, and the card seemed stable under load. For a moment, it looked like the mod had paid off.

Then things went very wrong.

The card didn’t fail during a benchmark run or stress test. Instead, it died while idling. According to the video, two large holes were burned straight through the PCB, and at least one capacitor was knocked off the board from extreme heat.

Frame Chasers believes the failure came down to a dangerous combination of factors. The shunt mod reportedly increased baseline power draw, meaning the GPU was pulling more wattage even at idle. On top of that, the fans weren’t spinning despite rising temperatures. The card was also set to “performance mode” in the NVidia Control Panel, which keeps clocks elevated instead of allowing the GPU to downclock at idle.

The GPU die and VRAM chips themselves may still be intact. The catastrophic damage seems localized to the PCB and surrounding power components.

For enthusiasts, extreme overclocking can be thrilling. There’s always the temptation to push just a little further. But this incident is a stark reminder that modern GPUs operate within tightly controlled power and thermal envelopes for a reason.

(Source | Via)

The post Modified RTX 5090 dies at idle due to extreme temperature appeared first on Gizmochina.

Cyberpunk 2077 just hit its five-year anniversary — hard to believe, considering how rocky its launch was and how far it’s come since. Despite everything, it’s still one of the most visually striking games on PC, thanks in no small part to NVIDIA’s early push for ray tracing. To mark the milestone, NVIDIA is dusting off something fans probably thought would never see daylight again: an original GeForce RTX 2080 Ti Cyberpunk 2077 Edition, and yes, it’s signed by Jensen Huang himself.

These cards were already rare when they were new. NVIDIA only made 200 of them back in 2020, and none ever hit retail shelves. Seventy-seven were given away during the original Cyberpunk promotion, and the rest essentially vanished into storage. Now, NVIDIA is offering up two more units — and at least one reportedly carries Jensen’s signature, which pretty much turns it into a collector’s grail piece rather than a GPU anyone would actually install.

The card itself is impossible to mistake: a custom AL5052 aluminum shroud in that unmistakable yellow-and-black Cyberpunk scheme, turquoise accents, and a matching backplate. Under all that flair, though, it’s still a standard RTX 2080 Ti Founders Edition. It launched with support for the game’s original ray-traced effects and DLSS 2.0, but modern features like RT Overdrive, DLSS 3.5 Ray Reconstruction, and DLSS 4 are out of reach for Turing. Not that it matters — whoever wins one of these isn’t going to be gaming on it. This is the kind of hardware you put behind glass.

If you don’t win but really want the look, there are third-party reproduction shroud kits floating around on eBay for people who happen to own a regular RTX 2080 Ti FE PCB. It’s not quite the same, but it’s a way to get close.

As for entering the giveaway, it’s straightforward: just head to NVIDIA’s X post embedded above and comment “GeForce Season.”

That’s it. Two winners will be picked, and they’ll walk away with one of the rarest pieces of GeForce history ever made — a signed relic from the Turing era and a neat little slice of Cyberpunk memorabilia.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source)

The post NVIDIA celebrates Cyberpunk 2077’s fifth anniversary with an ultra-rare giveaway appeared first on Gizmochina.

Key Highlights:

• Jensen Huang warns that China is catching up fast in AI and could “win the race.”
  
• He criticizes U.S. restrictions and overregulation for slowing innovation.
  
• Huang clarifies he wants America to lead by accelerating, not isolating.

Huang’s Warning at the London AI Summit

Nvidia CEO Jensen Huang has warned that China is moving rapidly toward AI dominance and could “win the AI race” if the U.S. continues to slow its innovation through overregulation and export restrictions. Speaking at the Financial Times Future of AI Summit in London, Huang said China is “nanoseconds behind America” in AI progress, highlighting how quickly the global gap is closing.

Huang urged Western countries to speed up AI development rather than hold it back. “It’s vital that America wins by racing ahead and winning developers worldwide,” he later wrote on X (formerly Twitter).

China’s Rapid AI Growth

According to Huang, China’s government-backed energy subsidies and strong state support are helping local companies advance faster. Beijing’s investments in semiconductor manufacturing and AI research are creating an ecosystem that rivals the U.S. in both scale and ambition.

He noted that while China faces technological hurdles, its developer community and policy environment are highly motivated and unified under national goals for AI self-reliance.

U.S. Restrictions and Policy Concerns

The U.S. government recently reaffirmed its ban on Nvidia’s advanced Blackwell AI chips being sold to China, citing national security reasons. Huang criticized these restrictions, arguing they hurt U.S. competitiveness and inadvertently push China to accelerate its own chip development.

He also pointed to state-level AI regulations and high energy costs as factors slowing innovation in the U.S. “Cynicism and overregulation are making it harder for American companies to move fast,” he said.

Market Reaction and Clarification

After his remarks were reported, Nvidia’s stock briefly dipped, prompting Huang to clarify that he wasn’t predicting U.S. defeat. Instead, he emphasized that the U.S. must act faster to maintain its lead. Nvidia remains the world’s most valuable company, valued at around US$4.7 trillion, even after a recent dip from its US$5 trillion milestone.

Balancing Global AI Competition

Huang’s comments come amid rising U.S.–China tensions over technology and trade. While Washington focuses on limiting China’s access to cutting-edge AI hardware, Beijing is boosting domestic innovation.

Huang’s message was clear: the U.S. should lead through innovation and collaboration, not isolation. “We want America to win,” he said, “but winning means building faster, not building walls.”

Read More:

• US AI Tools Under Fire: Are America’s Top Coders Using Chinese Models?
• AI Chip War Heats Up: U.S. Blocks Nvidia, China Responds with Total Ban
• China’s XPENG Shocks the World with Full-Stack Physical AI: Meet the IRON Robot and Flying Car

(via)

The post China Will Win the AI Race: Nvidia CEO Jensen Huang’s Warning to the West appeared first on Gizmochina.

Key Highlights:

• U.S. government has officially blocked Nvidia from selling its advanced Blackwell and B30A AI chips to China.
• China has responded with a new rule banning foreign AI chips from all state-funded data centers.
• Global AI chip war deepens as both nations tighten technology controls

White House Blocks Nvidia’s Advanced Blackwell Chip Sales to China

The Trump administration has decided not to allow Nvidia to sell its latest Blackwell AI chip to China. White House spokeswoman Karoline Leavitt confirmed the decision, stating, “The Blackwell chip is not something we’re interested in selling to China right now.”

This decision follows months of speculation about whether a scaled-down version of the chip might be cleared for sale. The move aligns with President Trump’s strategy to maintain America’s technological edge in artificial intelligence and prevent China from acquiring high-end U.S. hardware.

US Treasury Secretary Calls Blackwell the “Crown Jewel”

U.S. Treasury Secretary Scott Bessent described Nvidia’s Blackwell chip as the “crown jewel” of American AI innovation. He indicated that sales to China might be reconsidered only after the chip becomes outdated in 12 to 24 months. According to Bessent, Nvidia’s rapid pace of innovation means that the Blackwell could soon be several generations behind the company’s latest models.

Nvidia’s Market Collapse in China

Nvidia’s position in China has deteriorated sharply. Once controlling 95% of the Chinese AI data center market in 2022, the company now holds no market share. CEO Jensen Huang has expressed hope that Nvidia will eventually return to the Chinese market, but admitted there are no current plans to do so. Nvidia is reportedly redesigning its B30A chip in hopes of meeting future U.S. export conditions.

China Retaliates with Ban on Foreign AI Chips

In a direct response, Beijing has ordered that all new data centre projects receiving state funding must use only domestically produced AI chips. Projects that are less than 30% complete are required to remove or cancel any foreign chip orders, while those further along will be reviewed individually.

This directive represents one of China’s most aggressive steps toward achieving AI chip self-sufficiency. It is expected to benefit domestic manufacturers such as Huawei, Cambricon, and MetaX. However, industry analysts warn that Chinese AI chips still lag behind Nvidia’s products in performance and software ecosystem support, potentially slowing China’s AI development.

The Global AI Divide Widens

The escalating U.S.–China chip restrictions are reshaping the global AI landscape. The United States aims to secure national security and preserve its technological dominance, while China seeks independence from U.S. technology and to strengthen its domestic semiconductor ecosystem.

While American tech giants like Microsoft, Meta, and OpenAI continue to build massive AI data centres using Nvidia’s most advanced chips, China risks falling behind in computing capacity despite gains in local chip production.

Both the U.S. and China are pursuing self-reliance in AI technology, setting the stage for a deeper divide in global innovation. The U.S. is safeguarding its AI lead, while China is accelerating its domestic semiconductor push. 

Nvidia and other American chipmakers are losing access to one of the world’s largest markets, while Chinese chip firms stand to benefit, though at the cost of slower advancement in high-end AI capabilities.

Read More:

• NVIDIA’s Market Share in China Drops from 95% to Zero Amid US Export Restrictions
• China’s XPENG Shocks the World with Full-Stack Physical AI: Meet the IRON Robot and Flying Car
• Tech War Intensifies: China’s Lithography Breakthrough Challenges U.S. Semiconductor Edge

(via)

The post AI Chip War Heats Up: U.S. Blocks Nvidia, China Responds with Total Ban appeared first on Gizmochina.

Nvidia is betting big on the future of intelligent connectivity, investing $1 billion in Nokia to accelerate work on AI-driven 5G and 6G networks. The move gives Nvidia a 2.9% stake in the Finnish telecom giant and signals a deeper partnership aimed at blending AI and radio access networks (RAN) — what both companies are calling AI-RAN.

Under the deal, Nvidia will buy almost 166.4 million new Nokia shares at $6.01 each, pending regulatory approval. The funding will help Nokia speed up its development of AI-based RAN software, expand its cloud and data center solutions, and enhance its optical and switching technologies — all key to building the next generation of AI-ready infrastructure.

The collaboration will see Nvidia’s CUDA and AI platforms integrated into Nokia’s telecom systems, effectively turning networks into smart, adaptive systems capable of handling AI workloads closer to the user.

T-Mobile US is already signed on as a partner, with field trials planned for 2026 to test performance, power efficiency, and real-world scalability.

Nokia CEO Justin Hotard described the partnership as a “fundamental redesign” of how networks operate — bringing computing intelligence to the edge of connectivity. Nvidia’s Jensen Huang added that telecommunications form the “digital nervous system” of modern economies and that AI-RAN could be a generational shift for national infrastructure.

The deal also pulls in Dell Technologies as part of the broader ecosystem, hinting at new opportunities in data center and edge computing integration.

If successful, this could mark one of the most significant shifts in telecom since the rollout of 4G — paving the way for 6G networks that think, learn, and adapt in real time.

The post Nvidia will invest $1 billion in Nokia to push AI-powered telecom networks appeared first on Gizmochina.

NVIDIA has lost its entire share of the advanced chip market in China after once holding a 95% dominance. This sharp decline follows strict US export restrictions that prevent the company from selling its most advanced chips to Chinese companies. “At the moment, we’re 100 percent out of China,” confirmed NVIDIA CEO Jensen Huang during a recent event.

Sudden Market Collapse

The US government has banned the export of NVIDIA’s high-end AI chips to China since 2022. This includes the NVIDIA A100, NVIDIA H100, and NVIDIA H200. Although NVIDIA was allowed to sell a downgraded chip, the NVIDIA H20, the Chinese cyberspace regulator launched a security investigation, discouraging local companies from buying it. As a result, the company has been forced out of the Chinese market entirely.

Huang’s Warning

Jensen Huang criticized the export restrictions, calling them a “mistake.” He warned that cutting off China would push the market toward domestic competitors like Huawei Technologies. “What harms China could oftentimes also harm America, and even worse,” he said. Huang emphasized the importance of rethinking policies to maintain global leadership in technology.

China’s Rapid Self-Sufficiency Drive

China is moving fast to reduce its dependence on foreign chips. Domestic companies are investing heavily to develop their own AI chips and manufacturing technologies. Huawei has already introduced an advanced AI chip roadmap and new clustering methods to replace NVIDIA’s products. Other tech giants like Alibaba Group Holding, Tencent Holdings, ByteDance, and Baidu are also pouring resources into semiconductor research and development.

A Powerful AI Talent Base

Huang highlighted that China has nearly 50% of the world’s AI researchers. He warned that excluding the country from US technology could slow down global AI progress. He also noted China’s strong work culture, provincial competition, and deep talent pool that enable it to close the gap with the US quickly. According to Huang, China is only “nanoseconds behind” in chipmaking.

Strategic Impact on the US

Analysts warn that the restrictions may hurt US companies more than expected. By pushing NVIDIA out of the Chinese market, the US may be allowing Chinese firms to strengthen their supply chains and become more independent. This could reduce American companies’ global competitiveness and limit their access to one of the world’s largest chip markets.

Event and Industry Reaction

Huang made these remarks on October 6 at a Citadel Securities event in New York, where he spoke with Konstantine Buhler of Sequoia Capital. A video of the interview was released the following week and widely reported by the South China Morning Post. His comments reflect growing concern in the tech industry about the long-term consequences of the US-China tech war.

The collapse of NVIDIA’s market share in China highlights the impact of geopolitical tensions on the global semiconductor industry. While the US aims to protect its national security, the strategy may also be boosting China’s technological independence. As the chip race accelerates, the world is watching closely to see whether policy shifts will ease tensions or deepen the divide between the world’s two largest economies.

Read More:

• Dell’s Latest Mini PC Brings Powerful AI Performance with Nvidia GB10 Superchip Inside
• Apple’s Big Move: Vietnam Replaces China for New Smart Home Devices
• China’s Smart Robotics industry is booming with more than 450,000 companies

The post NVIDIA’s Market Share in China Drops from 95% to Zero Amid US Export Restrictions appeared first on Gizmochina.

NVIDIA’s latest Blackwell-based GB300 GPUs are starting to show what they can do, and early results point to a massive jump in efficiency compared to the company’s previous generation. A recent CoreWeave benchmark shows just how far the new chips push AI performance when running one of today’s heaviest workloads.

In tests using the DeepSeek R1 model, just four GB300 GPUs matched the performance of 16 H100s. That works out to roughly six times the throughput per card, thanks to both raw hardware upgrades and a more efficient architecture.

The GB300 NVL72 platform can scale up to 37TB of memory (40TB max) with 130TB/s of memory bandwidth, allowing it to tackle extremely large AI models without the bottlenecks seen in older hardware. It also makes use of 4-way tensor parallelism (TP4), while the H100 required 16-way splits (TP16) to reach similar scale. Fewer splits mean less communication overhead, and NVIDIA’s fifth-gen NVLink and NVSwitch interconnects help keep latency low.

For businesses running large AI services, that means faster token generation, lower costs per inference, and a more straightforward path to scaling. CoreWeave noted that the efficiency gains translate directly into practical benefits for workloads such as reasoning models, where performance-per-watt and latency are critical.

Of course, the hardware won’t come cheap. Systems built on the GB300 NVL72 are expected to start at around $300,000, putting them firmly in enterprise territory. But for companies already investing heavily in AI infrastructure, the performance leap could justify the cost.

If these early numbers hold true as deployments expand, the GB300 could mark one of the biggest generational jumps NVIDIA has delivered yet — and a new baseline for high-end AI computing.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

The post NVIDIA GB300 GPUs deliver huge AI efficiency gains in Deepseek R1 inference test appeared first on Gizmochina.

Nvidia has just announced that it will be expanding its GeForce Now service to India soon. The company’s cloud gaming will be powered by up to an RTX 5080 GPU, bringing high end gaming performance for a subscription.

Nvidia GeForce Now India Launch Confirmed

Nvidia has confirmed its partnership with Brothers Picture to help bring GeForce Now to the region. To recall, this service is already available in certain other markets and is finally expanding to India as well. The tech giant has revealed some of the upcoming lineup of AAA games that will be powered by GeForce Now, which includes Call of Duty: Black Ops 7. Apart from this, it has confirmed that more than 4,500 other titles will also be supported.

For those unaware, cloud gaming basically eliminates the need for a powerful gaming PC to enjoy high fidelity games. So even gamers on a budget laptop or weaker PCs can enjoy next gen titles with the help of an internet connection. After getting the subscription, Nvidia uses Indian data centers and the latest RTX powered systems for high resolution gaming, up to 4K resolution and 240fps. There will be multiple tiers for this subscription, which includes a free model with ad support and the performance and Ultimate plans.

The brand recommends a minimum connection speed of 25Mbps to enjoy the Ultimate tier of GeForce Now. Having a fiber optic connection and over 100Mbps broadband speeds will also help lower latency and make the experience more seamless by reducing latency. Mobile gaming is still dominant in India owing to its accessibility. But the subscription service could make PC gaming more affordable. Nvidia aims on rolling out the cloud gaming subscription in November 2025.

For more daily updates, please visit our News Section.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

The post Nvidia GeForce Now with RTX 5080 Cloud Gaming is Finally Coming to India appeared first on Gizmochina.

The AI boom has shifted into overdrive in 2025, and consumer GPUs are no longer just for gamers. Nvidia and AMD have turned their latest graphics cards into miniature AI workhorses, cramming them with faster memory, specialized tensor hardware, and lower-precision compute modes that cater to generative AI, LLM inference, and deep learning training.

If you’re running Stable Diffusion, fine-tuning LLaMA, or spinning up transformer-based workflows locally, these are the GPUs worth your attention right now.

Nvidia RTX 5090

The GeForce RTX 5090 leads the current generation with unmatched AI performance. Nvidia built this Blackwell-based card with 32GB of GDDR7 memory, delivering 1.79TB/s of bandwidth. It runs on 5th-gen Tensor Cores and supports new data formats like FP4 and FP8, which help speed up inference and training. The card delivers up to 838 TOPS of INT8 performance and outpaces the 80GB A100 in LLM benchmarks, hitting over 5,800 tokens per second with optimized models.

Stable Diffusion runs significantly faster on the 5090 compared to the 4090. Early tests show up to 2x speed increases when using FP4. With a TDP of 575W, the card requires high-end cooling and power delivery, but for AI developers working locally, the performance gains justify the size and heat.

Preview	Product
	Nvidia GeForce RTX 5090 Founders Edition	Check Price on Amazon

Please Note: When you buy something using the links in our articles, we may earn a small commission at no cost to you.

Nvidia RTX 5080

The RTX 5080 delivers many of the same AI features as the 5090 at a lower price. It uses 16GB of GDDR7 memory and pushes 960GB/s of memory bandwidth. Its 5th-gen Tensor Cores support FP4 and FP8 operations, and the card delivers around 450 TOPS for INT8 inference. It runs at 360W and has a smaller CUDA core count than the 5090 but retains strong generative AI performance.

In practice, the 5080 performs 10-20% better than the RTX 4080 Super in AI benchmarks. It also outperforms the RTX 4090 in certain inference tasks where faster memory and newer tensor features come into play. This makes the 5080 a solid option for creators running LLMs or diffusion models that fit within 16GB of VRAM.

Preview	Product
	NVIDIA GeForce RTX 5080 Founders Edition	Check Price on Amazon

Please Note: When you buy something using the links in our articles, we may earn a small commission at no cost to you.

Nvidia RTX 4090

The RTX 4090 continues to serve as the gold standard for AI workloads among mainstream users. It ships with 24GB of GDDR6X memory, delivering roughly 1TB/s of bandwidth. The card features 4th-gen Tensor Cores and supports FP16 and BF16 operations. It delivers over 330 FP16 TFLOPS and performs well in both training and inference.

LLMs with up to 30 billion parameters can run on the 4090 with 8-bit quantization. Stable Diffusion and other image generation models also benefit from the card’s high compute performance. Despite newer cards on the market, the RTX 4090 remains a reliable option for AI professionals and researchers.

Preview	Product
	PNY NVIDIA GeForce RTX 4090 24GB Verto Triple Fan Graphics Card DLSS 3 (384-bit PCIe 4.0, GDDR6X,…	Check Price on Amazon

Please Note: When you buy something using the links in our articles, we may earn a small commission at no cost to you.

Nvidia RTX 4080 Super & 4070 Ti Super

Nvidia launched the 4080 Super and 4070 Ti Super in early 2024. These Ada Lovelace refreshes improved on memory bandwidth and AI performance. The 4080 Super uses 16GB of GDDR6X and offers around 736GB/s bandwidth. It includes 4th-gen Tensor Cores and delivers up to 418 INT8 TOPS. The card consumes 320W and remains efficient for training and inference at mid-range batch sizes.

The 4070 Ti Super also received 16GB of memory and an upgraded bus. It delivers about 353 INT8 TOPS and consumes 285W. While it cannot match the 4090 or 5080 in compute throughput, it performs well in local LLM inference and image generation tasks, making it a strong choice for budget-conscious developers.

AMD Radeon RX 9070 XT

AMD’s RDNA 4-based RX 9070 XT introduces significant AI upgrades to the Radeon family. The card includes second-generation AI accelerators, support for FP8, and improved ray tracing. It ships with 16GB of GDDR6 memory and reaches 640GB/s of memory bandwidth. Its estimated FP32 compute performance is around 48.7 TFLOPS.

The RX 9070 XT provides around 389 INT8 TOPS. It runs at 300W and supports ROCm on Linux, allowing compatibility with PyTorch and TensorFlow. The card is best suited for AI-enhanced gaming, FSR4 upscaling, and small-scale inference tasks.

Preview	Product
	RX 9070 XT OC Edition 16GB GDDR6 (PCI Express 5.0, 20 Gbps​, 16GB GDDR6) Graphic Card" rel="nofollow noopener sponsored" target="_blank">ASUS Prime Radeon RX 9070 XT OC Edition 16GB GDDR6 (PCI Express 5.0, 20 Gbps​, 16GB GDDR6)…	Check Price on Amazon

Please Note: When you buy something using the links in our articles, we may earn a small commission at no cost to you.

AMD Radeon AI Pro R9700

AMD launched the Radeon AI Pro R9700 as a workstation-grade GPU aimed at AI developers and creative professionals. It comes with 32GB of GDDR6 memory and uses the same RDNA 4 architecture as the 9070 XT but with double the compute units. The card provides around 383 INT8 TOPs, supports FP8 operations, and runs at 300W.

The R9700 supports ROCm on both Linux and Windows, making it AMD’s most developer-friendly card yet. Its large VRAM buffer allows for fine-tuning and inference of LLMs beyond the capacity of the RX series. It performs well in multi-GPU setups and is positioned as a cost-effective alternative to Nvidia’s workstation-class cards.

For more daily updates, please visit our News Section.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

The post Best Consumer-Grade AI GPUs of 2025 appeared first on Gizmochina.

Nvidia and AMD have struck a rare arrangement with the US government, agreeing to hand over 15% of revenue from sales of certain AI chips to China in exchange for export licenses. The deal covers Nvidia’s H20 and AMD’s MI308 accelerators, both tailored for China after Washington’s 2023 export restrictions limited access to top-tier GPUs.

The agreement comes amid an intensifying US-China contest over advanced computing hardware, which powers AI models used in everything from language processing to military simulations. The US has argued that high-performance GPUs can give Beijing an edge in strategic projects, including autonomous weapons and mass-surveillance platforms, and has sought to slow China’s access to them through strict export controls.

Deal keeps China market open but sparks security concerns

By accepting the revenue-sharing terms, Nvidia and AMD retain a foothold in one of their most important markets while remaining formally compliant with US rules. Critics, however, say the compromise undermines the point of those rules. Christopher Padilla, a former US export control official, called the arrangement “bribery or blackmail, or both,” warning that it risks weakening the policy’s deterrent value.

The White House has not publicly commented, but the deal was approved under the Trump administration. Its structure is unusual: rather than banning specific products outright, it monetizes their sale to a strategic competitor, raising questions about whether security priorities are being traded for fiscal gain.

Nvidia, for its part, insists it is operating within government guidelines. “We follow the rules the US government sets for our participation in worldwide markets,” the company said in a statement. CEO Jensen Huang has framed the issue in terms of global competitiveness, arguing that “America cannot repeat 5G and lose telecommunication leadership” and that US-developed AI could set the world standard if it moves fast enough.

National security analysts remain divided on whether chips like the H20, engineered with performance caps to meet US requirements, can still advance Chinese defense capabilities. Some argue even downgraded accelerators are powerful enough for large-scale AI training, narrowing the intended gap between US and Chinese computing capacity.

For now, the revenue-sharing deal buys both companies time in a market they cannot afford to lose, while giving Washington a measure of control over what gets shipped. But it also exposes the fragility of export control regimes when commercial interests collide with national security aims. If this revenue-sharing model spreads to other sectors, the line between protecting critical technologies and monetizing their sale to strategic competitors could become increasingly blurred.

In related news, we recently covered a story on China quietly achieving 5nm production without EUV technology, exploring how SMIC bypassed chip sanctions through engineering ingenuity.

For more daily updates, please visit our News Section.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

(Via 1,2)

The post Unbelievable: Nvidia and AMD agree to pay 15% of China chip revenue to the US! appeared first on Gizmochina.

At Computex 2025, the tech industry’s biggest players put forward their boldest visions for the future of artificial intelligence, both at the cloud and at the edge.

But it wasn’t just the usual suspects like Nvidia and Intel grabbing attention. From MediaTek’s AI ambitions to a neuromorphic chip developed by German researchers, this year’s event revealed a diverse and evolving AI silicon landscape.

Here’s a closer look at the most compelling AI chips and architectures introduced at the show.

Nvidia Unveils Grace Blackwell and NVLink Fusion

Nvidia introduced its Grace Blackwell NVL72 architecture at Computex 2025, positioning it as a new standard for high-performance AI computing. The hybrid architecture combines GPUs and CPUs in a tightly fused package and is capable of running trillion-parameter models, real-time inference, and large-scale scientific workloads.

The company also launched NVLink Fusion, a next-generation interconnect designed to enable high-speed chip-to-chip communication. Nvidia confirmed that companies like MediaTek, Qualcomm, and Marvell will adopt NVLink Fusion for building semi-custom AI compute solutions. The move opens Nvidia’s ecosystem to external partners and expands the reach of its AI infrastructure technology.

MediaTek’s Custom AI Ambitions and 2nm Leap

MediaTek announced its plan to tape out its first 2nm chip with TSMC in September 2025. According to the company, the new chip will offer up to 15% better performance and 25% greater power efficiency than current 3nm chips. MediaTek aims to use the new node for both mobile devices and high-performance AI applications.

The company also highlighted its collaboration with Nvidia on the DGX Spark, a compact AI supercomputer for developers. MediaTek contributed a custom 20-core CPU architecture to the system and is using its ASIC expertise to design chips tailored for AI model training and inference. MediaTek’s IP in high-speed I/O, SerDes, and optical interfaces supports this initiative.

Intel Expands AI and GPU Lineup with Arc Pro and Gaudi 3

Intel introduced the Arc Pro B50 and B60 GPUs, designed for AI inference and workstation-grade computing. The B60 GPU includes 24GB of memory, while the B50 version ships with 16GB. Both models support multi-GPU scaling and are optimized for applications in architecture, engineering, and design.

Intel also showcased Project Battlematrix, a workstation-class platform supporting up to eight Arc Pro B60 GPUs and capable of running AI models with up to 150 billion parameters. The platform is based on Intel Xeon processors and targets AI developers and researchers.

In addition, Intel rolled out new deployment options for its Gaudi 3 AI accelerators. The Gaudi 3 PCIe cards support scalable inference in existing data centers, while rack-scale configurations support up to 64 accelerators per rack and 8.2TB of high-bandwidth memory.

TUM Introduces Cloud-Free AI Chip Inspired by the Human Brain

Researchers from the Technical University of Munich (TUM) revealed the AI Pro chip, a neuromorphic processor that does not require cloud connectivity or internet access. The chip performs real-time, on-device inference using a memory-compute integrated architecture. It employs hyperdimensional computing, allowing it to recognize patterns using limited training data.

Professor Hussam Amrouch led the development of the chip, which consumes only 24 microjoules per task, significantly less than conventional AI processors. The AI Pro chip targets specific applications in healthcare, robotics, and autonomous navigation where local processing and data security are essential.

Qualcomm Previews Snapdragon X2 and Teases September Launch

Qualcomm did not launch new chips at Computex but used the stage to confirm the dates for its next major event. The company announced that the annual Snapdragon Summit will take place from September 23 to 25 in Hawaii. Qualcomm is expected to launch the Snapdragon 8 Elite 2 for smartphones and possibly introduce the Snapdragon X2 Elite for Windows laptops.

CEO Cristiano Amon said Qualcomm plans to expand its presence in the PC market, aiming for more than 100 Snapdragon-powered laptop models by next year. He also highlighted the growth of native Arm-compatible apps, claiming that over 1,400 games now run on Snapdragon devices.

The AI Chip Race Isn’t Slowing Down

Between Nvidia’s rack-scale ambition, MediaTek’s 2nm plan, Intel’s workhorse GPUs, and TUM’s sci-fi-like brain chip, Computex 2025 made one thing abundantly clear: the AI chip market isn’t just booming — it’s evolving. Fast.

And the conversation is no longer about which chip is faster. It’s about where those chips live, how they connect, and what new kinds of intelligence they can unlock.

For more daily updates, please visit our News Section.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

The post The Most Interesting AI Chips Unveiled at Computex 2025 appeared first on Gizmochina.

Huawei might be stepping in to fill the void left by Nvidia in the Chinese market. Following the export ban from the US, Huawei officially announced the Ascend 920, which is its fastest AI chipset. The timing of the announcement is oddly coincidental and could be arriving at a big cost to Nvidia. So here’s what we know.

How the Huawei Ascend 920 AI Chip is Changing the Game

For those unaware, the United States announced stricter restrictions on the import of AI chips, which affects chipmaking giant, Nvidia. The company managed to circumvent previous regulations to sell its Nvidia H20 processors in the Chinese market. However, the latest export ban could be cracking down on this. But with AI growing exponentially in the region, Huawei’s new announcement arrives at the perfect time to fill in the gap.

As per the official announcement, the new Ascend 920 will hit mass production later this year and is based on SMIC’s 6nm process. It promises to offer up to 900 TFLOPS of BF16 compute and 4000GB/s of memory bandwidth. This chip is supported by HBM3 memory modules and will also support PCI 5.0, along with next gen interconnect protocols to help large model training. With this chip, the company could replace the Nvidia H20 by being an effective alternative.

To recall, the H20 is being used by other Chinese tech giants like Tencent and ByteDance. But after these new US restrictions, the Ascend 920 might be the next in line. The current gen Ascend 910C offers around 60 percent of the Nvidia H100’s performance, while the Ascend 920 possibly offering training efficiency by 30 to 40 percent.

For more daily updates, please visit our News Section.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

The post Nvidia May Lose Billions as Huawei Deploys Its Fastest AI Chip Under US Export Ban appeared first on Gizmochina.

The 8GB version of Nvidia’s RTX 5060 Ti 8GB turns out to lag significantly behind the 16GB version in certain scenarios. In VRAM-intensive games such as the is Horizon Zero Dawn Remastered, or in 4x Frame Generation, the 8 GB version falls significantly behind the 16 GB version.

Recent benchmarks, shared by a Chinese reviewer on Bilibili, highlighted significant performance gaps between the 8GB and 16GB models of the RTX 5060 Ti. In Cyberpunk 2077 at 1440p with max settings and DLSS 4 Quality (no Frame Generation), the 8GB card limps along at 9.1 FPS compared to the 16GB’s 18.5 FPS—a 200% difference.

Enabling 2x Frame Generation narrows the gap, with 61 FPS for the 8GB and 62 FPS for the 16GB. But crank it to 4x Frame Generation, and the 16GB pulls ahead by 22%, hitting 113 FPS versus the 8GB’s 92.4 FPS. 4x Frame Generation requires more VRAM, and the 8GB model just can’t keep up.

In Horizon Zero Dawn Remastered, the 16GB card shines, delivering 29% higher average FPS at 1080p and 16% at 1440p. More critically, the 8GB version’s 1% lows at 1080p are 30% worse, making gameplay feel choppier. And this is not even surprising. As per official system requirements, HZD Remastered ideally requires 16GB of VRAM. In less VRAM-hungry games like Assassin’s Creed: Shadows, both models performed neck-and-neck, as 8GB is enough for 1080p/60 FPS in this game.

Nvidia has been quiet, not sending 8GB models for review, which hints that Nvidia is well aware of the shortcomings. With AAA games only getting more demanding, the $50 jump to the $429 16GB model looks like the smarter bet for future-proofing.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source)

The post RTX 5060 Ti 8GB struggles in VRAM-heavy games and 4x Frame Generation appeared first on Gizmochina.

Nvidia has recently released new drivers for its GeForce series of graphics cards to address the annoying game crashing issue. The company has released the 576.02 GPU drivers that finally offer a fix to the various issues being faced by gamers worldwide. So here are all the details.

What does Nvidia’s latest GPU drivers bring?

Over the last couple of days, Nvidia received many reports from GeForce GPU users regarding game crashes, stability issues, and black screen problems. But the latest 576.02 GPU drivers have been released to fix the many issues caused by the 572.83 drivers that was released back in March 2025.

In the release notes of the Nvidia 576.02 drivers, Nvidia states that the stability issues have been fixed for systems running on Windows 11 version 24H2, which was facing BSOD (Blue screen of death) when playing certain titles with DLSS 4 games with Multi Frame Generation Enabled.

Playing with these settings on RTX 50 series GPUs caused stability issues and random black screen problems as well. The latest drivers help fix crashes for various games like Fortnite, while addressing the Star Wars Outlaws hanging issue when the game is idle. Even Overwatch 2 had a stuttering issue with V-Sync enabled.

The update also brings fixes for Control, The Last of US Part 1, Uncharted Legacy of Thieves Collection, and even Hellblade 2. The 576.02 GPU driver update is currently live and can be found on the Nvidia app. So if you’ve been facing any of these aforementioned problems, make sure your GPU drivers are up to date.

For more daily updates, please visit our News Section.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

The post Nvidia finally addressed the frustrating game crashing issue with the latest 576.02 GPU drivers appeared first on Gizmochina.

Nvidia’s Blackwell GPU series is facing renewed scrutiny after an RTX 5070 owner reported a melted power cable. While these incidents remain rare, they highlight potential design flaws that could impact the RTX 50 series, including less power-hungry cards.

The latest case, documented by X user ere9w, involved a Zotac RTX 5070 paired with a 750-watt Seasonic PSU in a new gaming rig. Moments after powering on, smoke emerged, revealing a melted cable rather than the GPU’s 12V-2×6 connector.

The user noticed a damaged pin on the GPU’s power connector

In response to the similar issues with the RTX 4090 series, Nvidia initially blamed improper seating before endorsing the revised 12V-2×6 configuration. While inspecting the GPU, Ere9w noticed a damaged pin on the GPU’s power connector. Judging from the difficulty of misaligning the two parts, many users suggested it was a manufacturing defect. The user also insisted on proper installation.

This incident echoes findings from overclocker Der8auer, who analyzed an RTX 5090 with a melted connector and cable. Using thermal imaging, he identified uneven current distribution, with some wires carrying excess load. The 12V-2×6 design lacks a mechanism to monitor pin-specific power draw, leading to poor load balancing and overheating. This flaw now appears to affect the RTX 5070 and 5080, not just the power-hungry 5090. It raises questions about the series’ reliability.

Nvidia’s silence (as of writing) leaves gamers wary, especially with the high cost of these GPUs. While the RTX 40 series prompted a connector fix, the RTX 50 series’ issues suggest unresolved challenges in Blackwell’s architecture. Users investing in these cards—priced upwards of $500—face risks of damage, potentially voiding warranties if deemed user error. Until Nvidia addresses this, the melted cable reports could dampen enthusiasm for the otherwise promising RTX 50 lineup.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source)

The post After RTX 5090, RTX 5070 is also facing melted power cable issues appeared first on Gizmochina.

Chinese researchers have achieved a groundbreaking performance boost in supercomputing, using domestically developed GPUs. Their new computing model delivered a nearly tenfold speed improvement compared to Nvidia-powered US supercomputers. This achievement challenges America’s dominance in high-performance computing and shows China’s growing self-reliance in advanced technology.

China’s response to US Tech sanctions

The United States has placed strict sanctions on China, restricting access to Nvidia’s advanced GPUs like the A100 and H100. Additionally, Nvidia’s CUDA software is not compatible with non-Nvidia hardware, making independent algorithm development difficult. In response, China has focused on developing its own hardware and software solutions to bypass these technological barriers.

What is the innovation in Parallel Computing

Led by Professor Nan Tongchao from Hohai University in Nanjing, the research team created an innovative “multi-node, multi-GPU” parallel computing model. This approach optimizes data transfer and task coordination, reducing performance losses and improving efficiency. Unlike the US TRITON system, which required 64 nodes to achieve a 6x speedup, the Chinese model accomplished the same with only 7 nodes. This represents an 89% reduction in node usage, making the system more efficient and cost-effective.

The supercomputing system was built using a domestic x86 computing platform. It used Hygon 7185 CPUs, which feature 32 cores and 64 threads, running at 2.5 GHz. The GPUs, developed in China, were equipped with 128GB of memory. The system also utilized a 200 Gb/s network bandwidth to enable efficient communication between nodes.

Flood simulation success

The research team tested their supercomputer by simulating a flood at the Zhuangli Reservoir in Shandong, China. Using 200 nodes and 800 GPUs, the system completed the simulation in just three minutes. This marked a 160x speed improvement over traditional methods. Such advancements make real-time flood simulations possible, improving disaster response, flood control, and reservoir management.

What are the future applications?

The study, published in the Chinese Journal of Hydraulic Engineering on January 3, 2025, has made its research code open-source. This allows other researchers to build upon the findings. Potential applications include hydrometeorology, sedimentation modeling, and surface-water interactions. Future work will focus on expanding the model’s applications and testing its stability in real-world engineering projects.

US-China rivalry in supercomputing

China’s success in supercomputing highlights the unintended effects of US sanctions. By restricting access to Nvidia technology, the US aimed to slow down China’s progress. However, the sanctions have instead pushed China to develop its own powerful computing solutions. 

With this breakthrough, China is not only catching up but also surpassing US supercomputers in certain areas. This marks a significant shift in the global tech race, as China continues to strengthen its position in high-performance computing.

Read More:

• Nvidia RTX 5050 and RTX 5060 registered at EEC: new budget GPUs on the horizon
• Global shortage of Nvidia GeForce RTX 5090 graphics card – wait time extends to 4 months
• China’s Smart Robotics industry is booming with more than 450,000 companies
• China’s Top AI Chatbot ERNIE Bot will be free for all users

(via)

The post Chinese GPUs outperform Nvidia in supercomputing study: Challenge to US tech dominance appeared first on Gizmochina.

The RTX 5080 and RTX 5090 have been in the conversation for a while. But let’s be real—not every gamer can afford those cards. Here comes the RTX 5050 and RTX 5060, which just popped up in the Eurasian Economic Commission. Could these be the affordable GPUs we’ve been waiting for?

RTX 5050 and RTX 5060 spotted on EEC

Actually, the RTX 5050 is a bit of a surprise. Nvidia skipped the RTX 4050 for desktops, so seeing a desktop version of the RTX 5050 is unexpected but kinda exciting. Zotac’s filing suggests they’re planning to make one, which could be a game-changer for budget gamers.

The RTX 5050 is expected to sit in the $250 range, which is way more palatable than the $749+ price tag of the RTX 5070 Ti. Given it supports all the RTX 50 series features like DLSS 4, it will be a solid option for gamers who don’t need top-tier performance.

Now the question is whether it’ll use DDR6 or DDR7 memory. We will have to wait and see as the leaks we have come across so far are kinda conflicting. If Nvidia manages to impress gamers with the affordable 50 series cards, it will mean a serious competition to AMD and Intel.

About the RTX 5060 and 5060 Ti?

The RTX 5060 and 5060 Ti are also in the filings. They reportedly could hit the market sooner—maybe within the next few months. These cards will probably sit in the middle, offering a balance between performance and price.

But here’s a catch: Nvidia’s been dealing with production delays. So don’t be surprised if these cards get pushed back to late spring or even summer. And if the RTX 5050 does make it to desktops, it might not arrive until later this year.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

For more daily updates, please visit our News Section.

The post Nvidia RTX 5050 and RTX 5060 registered at EEC: new budget GPUs on the horizon appeared first on Gizmochina.

Nvidia launched the RTX 5080 and 5090 on January 30. Now a major French retailer has revealed the launch date of the RTX 5070 Ti to be February 20.

TopAchat, one of the biggest PC hardware retailers in France, has listed the cards that boast 16GB GDDR7 memory. Users can enter their email address to get notified as soon as the cards become available.

RTX 5070 Ti listed on TopAchat

The retailer has listed 6 custom GeForce RTX 5070 Ti graphics cards, all of which are set to become available at the same time. However, there isn’t any mention of the non-Ti cards yet, which are expected to arrive soon after this release.

According to the product description, the chips are based on the company’s Blackwell architecture and “offers spectacular performance”. The cards support DLSS 4 and real-time rendering with Ray-Traycing.

On the inside, the 5070 Ti cards equip the GB203-300 GPU packing 8960 CUDA cores. The GDDR7 memory provides up to 896GB/sec of memory bandwidth. In comparison, the RTX 4070 Ti packs 7,680 CUDA cores and 12GB of GDDR6X memory, and the RTX 5070 packs 6,144 CUDA cores – only marginally improved on the RTX 4070 – with the same 12GB of VRAM but upgraded to GDDR7. Both graphics cards rely on Multi Frame Generation to compete with pricier options.

In the US, the lowest price of the 5070 Ti models is reportedly $750. Given the short supply of the recently released RTX 50 series cards, I don’t expect widespread availability soon after launch. This could also lead to prices soaring well above MSRP, much like we saw with the RTX 5090.

Although the opinions about whether the RTX 5070 or 5070 Ti cards are worth the money are mixed, they seem to be the most economical option to experience DLSS 4 until the RTX 5060/5060 Ti cards arrive.

Tech enthusiast? Get the latest news first! Follow our Telegram channel and subscribe to our free newsletter for your daily tech fix!

For more daily updates, please visit our News Section.

(TopAchat)

The post RTX 5070 Ti to launch on February 20: details here appeared first on Gizmochina.

If you were hoping to get your hands on Nvidia’s shiny new RTX 5080 or 5090 graphics cards that were launched last week, you’re out of luck. Both GPUs sold out across the U.S. in less than an hour after they went on sale, leaving retailers empty-handed. It all happened so fast that people are calling it a “paper launch,” wondering if Nvidia even had enough cards to sell in the first place.

One major US system integrator had just one RTX 5090 on launch day

Honestly, we probably should have seen this coming. Before the launch, Nvidia’s head of GeForce community, Tim Adams, warned everyone that the RTX 5080 (Geekbench) and 5090 (3D Mark) would be in high demand and tough to find. But “tough to find” turned into “impossible to find” in no time. Newegg (Electronics Store) said it sold out “within minutes” and called the response “overwhelming.”

Meanwhile, PC hardware YouTuber Gamers Nexus reported that one major US system integrator had just one RTX 5090 on launch day, with a small shipment arriving later. Another company sold out of its 20 cards in three minutes—all through pre-built PCs. Yikes.

Now, retailers are scrambling to restock. Micro Center has a banner on its website telling customers to check back for updates, and U.K. retailer Overclockers UK says it could take two to six weeks for the RTX 5080 and three to 16 weeks for the RTX 5090 to come back in stock. Even the lucky few who managed to snag an RTX 5090 at Best Buy are facing delays. One Reddit user revealed they won’t get theirs until February 21. So, yeah, it’s going to be a while before these cards are easy to find.

According to the January 2025 Steam hardware survey, most people are still rocking RTX 3060s—a GPU that’s two generations old. Sure, current games like Dragon Age: The Veilguard still run fine on older cards, but upcoming titles like Avowed and Monster Hunter: Wilds are starting to recommend 40-series GPUs for features like ray tracing. For a lot of people, this launch felt like the perfect chance to finally upgrade after years of dealing with shortages and sky-high prices during the COVID-19 pandemic. But, well, here we are again.

The prices have also skyrocketed amid production delays, where the RTX 5090 was previously reported to hit a $6000 price tag.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

For more daily updates, please visit our News Section.

(Via)

The post Nvidia’s RTX 5080 and 5090 sold out in minutes—here’s what happened appeared first on Gizmochina.

The launch of Nvidia’s RTX 50 series has left gamers frustrated and empty-handed, with scalpers reselling the RTX 5090 for as much as $6000—three times the MSRP. Not to mention, the situation has left many questioning if they’ll ever be able to get their hands on these next-gen GPUs at a reasonable price any time soon.

Notably, retailers like Best Buy and Newegg listed the RTX 5080 and RTX 5090 at 9 AM EST, but stock vanished almost instantly. Only a lucky few managed to grab a card, but most were met with disappointment as checkout errors and sold-out notifications became the norm. Nvidia’s own store didn’t even list the Founders Edition cards, which added to the disappointment.

Meanwhile, eBay scalpers are having a field day. The RTX 5090 is listed on the platform for jaw-dropping prices, with some sellers asking for over $6000. Even the RTX 5080, with an MSRP of 1999, is being resold for 1,800 to 2,200. Some sketchy listings have even been removed for violating eBay policies. But sellers with verified stock are raking in massive profits. Check out the benchmark listings of the RTX 5080 here.

The reason behind this price hike

The root of the problem lies in a combination of high demand and limited supply. Reports suggest a design flaw in Nvidia’s Blackwell GPU chips caused production delays, which has resulted in the limited availability. Third-party manufacturers like MSI received far fewer chips than expected, further reducing the supply of custom cards.

Unfortunately, the situation may not change soon. Analysts suggest not to expect significant stocks any time soon. And even when it finally happens, scalpers are likely to dominate the market. If past GPU launches are any indication, demand will remain higher than the supply for months, if not longer.

For now, gamers can only hope that Nvidia ramps up production as soon as possible and find ways to curb scalping. Until then, the dream of upgrading to an RTX 5080 or RTX 5090 remains out of reach for most.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

For more daily updates, please visit our News Section.

(Source)

The post Nvidia RTX 5090 & 5080 prices skyrocket amid production delays: RTX 5090 hits $6000 appeared first on Gizmochina.

Nvidia‘s GeForce RTX 5080, which is set to launch alongside the flagship RTX 5090 on January 30, has already been spotted in benchmark platforms. For those interested about the 80-class GPU, these benchmarks might help you decide – or leave you scratching your head.

Spotted on Geekbench

The numbers spotted on Geekbench browser (via Benchleaks) gives us a sneak peak at what to expect from the RTX 5080.

First things first, the test environment included an AMD Ryzen 7 9800X3D processor, 32GB of DDR5-6000 memory, and an MSI motherboard.

In the Vulkan test, it scored 261,836 points. This is about 22% better than the RTX 4080. Not bad, right? But then we look at the OpenCL results: 256,138 points. This is just a 6.7% improvement—a bit underwhelming if you ask me.

Here’s the crux of it: Nvidia’s 80-class GPU generations have always outperformed the previous generations’ 90-class models. But this time, the RTX 5080 hasn’t reached the same heights as the RTX 4090. That’s quite surprising, and not a good omen for those who harbor hopes for a sub- or under-performance king.

Also spotted on Blender Open Data platform

If you’re into Blender, the RTX 5080 showed up on the platform too. According to Reddit user TruthPhoenixV, it scored 9,063.77—around 9.4% faster than the RTX 4080 and 8.2% ahead of the RTX 4080 Super. To be fair these are decent gains, but nothing that screams “next-gen revolution.”

The RTX 5080 is solid on the spec sheet though. The card’s based on Nvidia’s Blackwell architecture that leverages 10,752 CUDA cores. This is paired with 16GB of GDDR7 memory and rated AI performance at 1,801 TOPS. And it all starts at $1,000.

Should you be excited? That’s the big question. These leaks offer a decent preview, but they’re not the full story. Nvidia’s 5080 seems like a solid step up from the 4080, but not a game changer—especially if you’re eyeing that RTX 4090-level performance.

The RTX 5080 will debut on January 30, 2025. And for now, it’s best to wait until reviews drop after the launch. Until then, keep your expectations balanced. The RTX 5080 might not be the giant leap forward that we’ve seen from Nvidia in the past, but it still could be a contender for gamers looking for solid upgrade.

In related news, the RTX 5090 leak has revealed a staggering 46% performance improvement. On the other hand, the giant has also responds to concerns about RTX 5090 melting power connectors

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

For more daily updates, please visit our News Section.

Source: Benchleaks, TruthPhoenixV

The post Nvidia RTX 5080 spotted on Geekbench and Blender: slower than the RTX 4090? appeared first on Gizmochina.

Conversations around the Nvidia GeForce RTX 5090 are not only exciting but also fairly positive for the most part. But there’s at least one major concern floating around: the fear of power connectors melting, just like we saw with the RTX 4090. And given the $1,999 price tag, nobody wants to deal with a potential hardware failure again.

But Nvidia’s on it, reassuring fans that they’re doing everything they can to avoid a repeat of the issues from the 4090.

A quick recap: what went down with the RTX 4090

When the RTX 4090 hit the market in October 2022, some users started reporting overheating and even melting power connectors. At first, Nvidia pointed to user error, claiming that people were plugging the connectors in wrong. However, the experienced PC community couldn’t accept it so easily. So Nvidia took a closer look and made some changes.

They ended up redesigning the 12V-2×6 16-pin cable, making it more durable and adding shorter, more conductive pins to help with heat management. This fix definitely reduced the number of overheating incidents, but some 4090s still had issues, especially those with older cables. The original cables, designed for a 600-watt power draw, just weren’t cutting it for the demanding GPU.

But what about the RTX 5090?

Now, the RTX 5090 is about to drop on January 30, 2025, and it’s actually pulling even more power than the 4090—575 watts up from 450 watts of the previous generation. But Nvidia’s confident that the updated 12V-2×6 connector will be able to handle it. At a recent event, the company assured everyone that the new design is built to withstand the added power demand, so the risk of melting connectors should be a lot lower.

Some manufacturers aren’t just relying on Nvidia’s new connector, though. MSI, for example, has added yellow-tipped power adapters to their builds, which lets users see if the connector is securely plugged in—something that could help prevent the user errors that Nvidia has previously pointed to as the cause of connector issues.

While Nvidia’s tweaks and their partners’ innovations suggest they are taking the right steps to keep things safe, users may still consider upgrading their power supply before replacing their existing card with the RTX 5090.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

Source: Quasar Zone

The post Nvidia responds to concerns about RTX 5090 melting power connectors appeared first on Gizmochina.

Nvidia’s upcoming GeForce RTX 5090 has been on the conversation for quite a while, and for good reason. A recent leak has given us a sneak peek at its performance, and it’s looking like a major upgrade over the RTX 4090. With up to 46% more power in synthetic benchmarks, this flagship GPU has already grabbed the attention.

So, what do the benchmarks say?

Some 3D Mark scores for the RTX 5090 have spotted online, with quite impressive numbers. Check out the results:

• 3D Mark Speed Way (1440p): 14,383
• 3D Mark Steel Nomad (4K): 14,133
• 3D Mark Port Royal (1440p): 36,673
• 3D Mark Time Spy (1440p): 48,732
• 3D Mark Time Spy Extreme (4K): 25,485
• 3D Mark Fire Strike (1080p): 101,767
• 3D Mark Fire Strike Extreme (1440p): 60,778
• 3D Mark Fire Strike Ultra (4K): 33,114

These scores show that the RTX 5090 is roughly 30-46% faster than the RTX 4090, depending on the benchmark. Nvidia’s own gaming tests back this up, suggesting a 30% performance boost in raw gaming power. For instance, in Cyberpunk 2077—a notoriously demanding game—the RTX 5090 needs DLSS to hit playable frame rates with Path Tracing enabled. This means we’re looking at some serious muscle under the hood.

As far as these benchmarks are concerned, the RTX 5090 is going to be a beast. Sure, synthetic benchmarks don’t always tell the full story, as performance can vary depending on your CPU, RAM, and game optimizations. But one thing is for sure—this card is set to be the king of desktop GPUs for quite a while.

And don’t expect AMD or Intel to dethrone it anytime soon. AMD’s Radeon RX 9070 XT and even its upcoming RDNA-based GPUs are unlikely to match the sheer power of Nvidia’s Blackwell flagship. Intel’s Battlemage and Celestial series? Same story.

Should you care?

Absolutely. The RTX 5090 will easily be the fastest graphics card money can buy for the foreseeable future. Of course, we’ll have to wait for real-world testing to see how it handles in different games and workloads, but if the leaks are accurate, this GPU is going to set new standards. For users looking for nothing but the best, it’s hard to imagine a better choice. For more daily updates, please visit our News Section.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

(Source)

The post Nvidia GeForce RTX 5090 leak reveals up to 46% performance improvement appeared first on Gizmochina.

Nvidia CEO Jensen Huang recently visited China to celebrate Chinese New Year with employees and reaffirm the company’s commitment to its workforce and technological collaboration. The visit comes amidst ongoing US-China geopolitical tensions and export restrictions on advanced AI chips. Huang aimed to lift employee morale and highlight Nvidia’s dedication to China’s technology sector. The visit highlights the importance of maintaining global cooperation to drive technological progress, even amidst rising geopolitical tensions.

Huang attended events in Nvidia’s Shenzhen and Beijing offices, including a Spring Festival celebration in Beijing. Praising the team, he highlighted Nvidia China’s low turnover rate of 0.9%, compared to a global average of 2%. Over the years, Nvidia’s workforce in China has grown by more than 50%, reaching nearly 4,000 employees. Huang acknowledged the employees’ loyalty and significant contribution to Nvidia’s global success.

China’s Role in Nvidia’s Growth

China is a crucial market for Nvidia, contributing $5.4 billion in revenue during Q3 FY2024, a 34% year-over-year increase. The region accounts for 17% of Nvidia’s total revenue and remains one of the company’s largest markets after the US and Singapore. Nvidia has collaborated with over 3,000 Chinese start-ups and supports 1.5 million local developers using CUDA, its proprietary AI programming platform.

Geopolitical Challenges

The visit coincided with tightened US export controls on AI chips, targeting China. Nvidia has expressed opposition to these restrictions, warning that they could harm innovation and US technological leadership. In response, China announced an antitrust investigation into Nvidia. Despite these challenges, Huang reiterated Nvidia’s commitment to the Chinese market, describing the company’s 25-year journey in China as an honor.

Huang’s visit also fueled speculation about Nvidia’s potential Asia-Pacific headquarters, with Taiwan being a possible location. As China continues to focus on technological self-reliance, Huang emphasized Nvidia’s role in modernizing the sector and reaffirmed the company’s dedication to supporting innovation and collaboration.

The post Nvidia CEO chooses China over Trump’s inauguration: Visits his employees in Beijing and thanks them for loyalty appeared first on Gizmochina.

Nvidia’s latest RTX 5090 series has been making waves, but not everyone is thrilled. Sure, the new GPUs boast AI-driven features like DLSS 4 and Multi-Frame Generation (MFG) that promise to push gaming visuals to the next level.

But here’s the catch: when you strip away the AI wizardry, the raw performance upgrade over the RTX 4090 feels modest. Recent tests of the RTX 5090D—a special version designed for the Chinese market—give us a clearer look at what’s really going on.

What the tests reveal

Thanks to details shared by Chiphell’s forums, we know the RTX 5090D delivers about a 20% boost in raw gaming performance compared to the RTX 4090. No DLSS, no frame generation—just raw, unassisted performance. That’s not bad by any means, but let’s be honest, it’s not the jaw-dropping leap many gamers were hoping for.

Now, when DLSS 4 and MFG are turned on, the RTX 5090D supposedly hits Nvidia’s promised performance levels, and it’s seriously impressive.

Reports also shed light on the hardware itself. The RTX 5090D was overclocked to 2600 MHz during testing. But the reviewer downclocked the memory to achieve this. That dropped its bandwidth to 896 GB/s, which pretty much cancels out the benefits of Nvidia’s shiny new GDDR7 memory. For comparison, the RTX 4090 was pushed to 2610 MHz, and its memory bandwidth clocked in at 912 GB/s. So, not a huge difference there.

What’s more, the RTX 5090D seems to lack configurable power limits, which is a bummer for anyone who loves to tinker. Overclocking potential? Not looking great.

If you’re all about raw performance, the RTX 5090D feels like an incremental step forward rather than a game-changer. On paper, that 20% boost is fine, but it’s not the leap we’ve come to expect. And while DLSS 4 and MFG are great for squeezing out extra performance, not everyone’s a fan of leaning on AI to make up for hardware gains.

It’s still early days, and there’s more to learn about the full RTX 5090 lineup. But for now, if you’re rocking an RTX 4090, you might not feel the need to upgrade just yet.

Tech enthusiast? Get the latest news first! Follow our Telegram channel and subscribe to our free newsletter for your daily tech fix!

The post Nvidia RTX 5090D vs RTX 4090: A 20% raw performance boost without DLSS appeared first on Gizmochina.

According to TrendForce, Nvidia is working on the next generation of B100 and B200 GPUs that are based on the Blackwell architecture. The new GPUs are expected to hit the market in the second half of this year and will be used for CSP cloud customers. For those unfamiliar, “CSP cloud customers” refers to customers who use Cloud Service Providers (CSPs) for their cloud computing needs. The company will also add a streamlined version of the B200A for OEM enterprise customers with edge AI needs.

It is reported that TSMC‘s CoWoS-L packaging capacity (that is used by the B200 series) continues to be tight. It is mentioned that the B200A will switch to the relatively simple CoWoS-S packaging technology. The company is focusing on the B200A to meet the requirements of the Cloud Service Providers.

B200A technical specifications:

Unfortunately, the technical specifications of the B200A are yet to be fully clear. For now, it can only be confirmed that the capacity of HBM3E memory is reduced from 192GB to 144 GB. It is also reported that the number of layers of the memory chip is halved from eight to four. Nonetheless, the capacity of a single chip is increased from 24GB to 36 GB.

The power consumption of the B200A will be lower than the B200 GPUs and it will not require liquid cooling. The air cooling system of the new GPUs will also make them more easy to set up. B200A is expected to be supplied to OEM manufacturers around the second quarter of next year.

Supply chain surveys show that NVIDIA’s main high-end GPU shipments in 2024 will be based on the Hopper platform, with H100 and H200 for the North American market and H20 for the Chinese market. As the B200A will become available around the second quarter of 2025, it is expected to not interfere with the H200, which will arrive on or after the third quarter.

(Source)

The post Nvidia is working on an AI GPU with 144GB HBM3E memory appeared first on Gizmochina.

It looks like the graphics card that introduced many gamers to ray tracing and high-fidelity visuals is nearing its end. According to a new rumor, Nvidia is planning to discontinue production of the ever-popular RTX 3060, which has been a top performer in the Steam Hardware Survey.

The news comes from a post on the Chinese tech forum Bobantang, known for insider scoops within the hardware industry. While the information hasn’t been officially confirmed by Nvidia, Board Channels has a track record of accuracy, having previously predicted the RTX 3060 Ti’s discontinuation.

According to the forum post, Nvidia has notified its manufacturing partners to place final orders for the RTX 3060. This doesn’t mean the card will vanish overnight. Existing stock will continue to trickle through retailer channels for the next few months, allowing gamers a final window to snag this workhorse GPU.

The RTX 3060 was launched in February 2021 at a time when crypto-mining and chip shortages drove GPU prices to absurd heights. Initially, it was available only in a 12GB GDDR6 version, with an 8GB variant introduced later. 

Both versions struck a good balance between price and performance. Although early buyers faced inflated costs, the 3060 eventually became a budget-friendly option for gamers wanting a taste of Nvidia’s Ampere architecture.

The card currently reigns at the top of the Steam Hardware Survey with a 5.71% market share. It offered a significant jump in performance over its predecessor, the GTX 1660 Ti, enabling smooth gameplay at 1080p and even respectable performance at 1440p with some graphical adjustments. The inclusion of ray tracing capabilities, albeit at a limited capacity, further added to its appeal.

But with Nvidia’s new RTX 40 series arriving, the landscape is changing. The RTX 4060, as the successor to the RTX 3060, offers better performance at a similar price. While the 3060 will likely still be available in the used market, the 4060 is a more attractive option for new buyers.

The post Nvidia reportedly discontinues its ever-popular RTX 3060 GPU appeared first on Gizmochina.

We have previously come across numerous reports claiming that OpenAI used YouTube content to train its text-to-video model Sora. Now it’s reported that companies like Apple, Nvidia, Anthropic, and more are also using the ‘publicly available data’ generated by users to train their AI models. Apparently, Apple used tens of thousands of YouTube videos with subtitles to train Apple Intelligence, which is against the platform’s content policy.

The news comes from an investigation by Proof News that was co-published with Wired.

According to the investigation, Apple and the other companies were using a dataset called YouTube Subtitles that included transcripts of 173,536 YouTube videos from over 48,000 channels. Videos in the dataset span from educational channels such as Khan Academy and MIT to news sites including The Wall Street Journal, to some of the platform’s top creators like MrBeast and Marques Brownlee.

According to Marques Brownlee, Apple technically avoids “fault” as they sourced their AI from companies that used the transcripts from YouTube videos instead of using the data directly. Nonetheless, the data/transcripts still contribute to the AI models, in which the creators invested their time and money. Brownlee concluded by saying that this is going to be an evolving problem for a long time.

Proof News also created a tool for creators to search for their content in the dataset. The YouTube Subtitles dataset does not include imagery from videos but does include some translated subtitles in languages. The dataset was reportedly created by a non-profit research lab called Eleuther AI, which focuses on promoting open science norms.

None of the companies mentioned above immediately commented on the matter. YouTube chief executive Neal Mohan has already made it clear in an interview that companies using YouTube videos to train their AI models is a “clear violation” of the platform’s policies.

(Source)

The post Apple reportedly used YouTube videos to train Apple Intelligence appeared first on Gizmochina.

Sharp-eyed gamers have spotted a curious development on the Xbox store. Foreign media outlet Gamerant has discovered a “Play with GeForce Now” button appearing for certain games, including “Resident Evil 8: Village” and “Street Fighter 6.”

Microsoft and Nvidia might have a bigger collaboration on the way

This has fueled speculation that Microsoft may be preparing to integrate cloud gaming services directly into the Xbox ecosystem. GeForce Now, a cloud gaming platform by NVIDIA, could potentially allow Xbox players to access these titles without needing downloads.

As previously reported, ten new games were added to the GeForce Now library this month, including the aforementioned “Resident Evil 8: Village” and “Street Fighter 6.” This alignment with the “Play with GeForce Now” button sightings further strengthens the possibility of an official announcement from Microsoft.

While the exact timeline remains unclear, it’s expected that Microsoft will formally unveil Xbox’s support for cloud gaming services within the next few months. Whether this integration will extend to offering GeForce Now within the Xbox Game Pass Ultimate (XGPU) subscription remains to be seen.

(Via)

The post “Play with GeForce Now” Button Now Visible on Some Games on the Xbox Store appeared first on Gizmochina.

In celebration of their 25th anniversary, hardware brand ASL (founded in 1999) released a special edition RTX 4080 SUPER graphics card. This marks ASL’s first foray into the RTX 4080 SUPER category.

The card has a white design with gold accents for the special edition

The card boasts a sleek white design with gold accents highlighting ASL’s branding and the anniversary theme. A full-size metal backplate reinforces the card and continues the celebratory motif.

Despite the commemorative design, the core frequency and power consumption adhere to NVIDIA’s official specifications. Measuring 347mm long and occupying three PCI-Express slots (61.8mm thick), the card offers the expected performance of an RTX 4080 SUPER. For connectivity, ASL includes 3 DisplayPort interfaces and a single HDMI port. To ensure sufficient power delivery, ASL recommends using a 750W power supply with this graphics card.

(Via)

The post ASL has Launched a 25th Anniversary Special RTX 4080 SUPER Graphics Card, and it Looks Sick appeared first on Gizmochina.

Nvidia (NVDA) has dethroned Microsoft (MSFT) to become the world’s most valuable company by market capitalization. This surge is fueled by Nvidia’s dominance in the market for artificial intelligence (AI) hardware. Nvidia’s stock price climbed 3.5% to $135.58, pushing its market cap to a whopping $3.335 trillion. This surpassed Microsoft’s $3.317 trillion market cap and Apple’s $3.286 trillion valuation. Nvidia’s stock has nearly tripled in value this year, reflecting the high demand for its AI processors.

The rapid growth of AI in the past few years has fueled the company’s growth

This surge is a reflection of widespread investor optimism about the future of artificial intelligence technology. However, some analysts warn that this optimism could cool down if there are signs of a slowdown in spending on AI technology.

Nvidia’s dominance in the AI hardware market is a key factor in its success. Their processors are considered superior to competitors’ offerings, leading to a situation where demand heavily outpaces supply. This has made Nvidia the most traded company on Wall Street, with daily turnover averaging a staggering $50 billion.

Nvidia’s stock price recently hit a record high, adding over $110 billion to its market cap. This increase is equivalent to the entire value of Lockheed Martin. The company’s market value has grown explosively, expanding from $1 trillion to $2 trillion in just nine months, and reaching $3 trillion in just over three months.

Nvidia has consistently exceeded Wall Street’s expectations for revenue and profit, driven by the surging demand for its graphics processors as companies race to incorporate AI applications. The company has also recently split its stock 10-for-one, potentially making it more attractive to individual investors. However, despite the stock’s gains, the company’s earnings valuation has actually fallen due to increasing analyst expectations for future earnings.

(Via)

The post Nvidia Becomes the World’s Most Valuable Company, Leaving Microsoft Behind appeared first on Gizmochina.

The competitive landscape of China’s AI chip market is undergoing a significant shift as Huawei gains ground against Nvidia. This development follows US sanctions on Huawei and export curbs on Nvidia’s advanced processors, altering the dynamics of the market.

Huawei’s self-developed 910B Ascend chips are rapidly gaining traction across various industries, facilitated by the increasing popularity of “AI boxes.” These AI boxes are all-in-one machines that integrate AI chips, industry-specific algorithms, and pre-trained large AI models, providing ready-to-use solutions for small companies and government entities. They offer a cost-effective way for these organizations to fine-tune large model-based services without substantial investments.

Previously dominated by Nvidia, the mainland market is now seeing a potential shift, with Huawei poised to overtake its American rival. In response to US export restrictions on its A800, H800, and L40S processors, Nvidia has introduced lower-grade chips, such as the H20, tailored for China. However, these chips are being sold at a discount compared to Huawei’s Ascend 910B chips, reflecting the intensified competition and market challenges Nvidia faces.

Huawei’s Ascend ecosystem has grown rapidly, boasting 40 hardware partners, 1,600 software partners, and 2,900 AI application solutions. Prominent Chinese companies like Tianjin Sitonholy and SenseTime are adopting Huawei chips for their AI boxes, with SenseTime anticipating a doubling of its generative AI sales due to these AI boxes.

Analysts predict that Huawei’s expanding AI chip business will soon dominate the market. AI boxes, deemed essential for businesses, are expected to predominantly feature local players led by Huawei. Insights from Pacific Securities and Counterpoint analysts underscore Huawei’s technological capabilities and substantial market potential. However, the pace of Huawei’s ascendancy hinges on its ability to ramp up production, as manufacturing high-performance AI accelerators and server processors is more complex than mobile system-on-a-chip products.

Despite these advancements, Nvidia remains a significant player, having accounted for 90% of AI chips sold in China in the first half of the previous year, compared to Huawei’s 6%. Nevertheless, the market tide is turning, with more Chinese companies launching AI boxes using Huawei chips. Nvidia’s tailored-for-China chips, including the H20, are currently selling at a discount compared to Huawei’s offerings.

Looking ahead, Huawei is expected to prioritize AI chips over mobile chips to secure a leading market position in China. Collaborations with companies like iFlyTek aim to address computing-power bottlenecks, with products like the Spark Number 1 AI box benchmarked against Nvidia’s A100 processor.

Nvidia’s strategic response to this competitive pressure includes price cuts and the introduction of tailored chips for the Chinese market. Despite these efforts, analysts highlight the uncertainty surrounding Nvidia’s future in China amid US-China tensions and a fiercely competitive landscape.

As China’s share of the global AI industry is projected to exceed 30% by 2035, the ongoing battle between Huawei and Nvidia in the AI chip market will have far-reaching implications for both companies and the broader industry.

Related:

• HUAWEI MateBook 14 2024 Review：120Hz OLED Display
• Nvidia and MediaTek are Teaming Up to Develop an ARM-based SoC for AI Apps, Rumor States
• HMD is working on three new mid-range phones with Snapdragon chips after Unisoc T606-powered Pulse and Aura

(via)

The post Huawei becomes a strong competitor to Nvidia in China’s AI chip market appeared first on Gizmochina.