Graphics Cards - Page 19

AMD's RX 9070 GRE has recently dropped considerably in price over in China, which might be looked upon as an interesting development with the graphics card.

Why? Because the price reduction of the equivalent of $100 (or just under) could conceivably back up the theory that AMD has another variant of the RX 9070 GRE with more video RAM ready and waiting in the wings.

The price drop was highlighted by Uniko's Hardware on X (as flagged by VideoCardz), with the RX 9070 GRE having sunk from 4199 RMB (approximately $590) at launch to 3600 RMB ($505) now. That's for an entry-level model, and as ever, fancier third-party graphics cards can command a fair premium over that (but they've also gone down in price).

AMD's next-gen RDNA 5 gaming GPUs could feature up to 128 Compute Units, and over 12,000 RDNA 5 cores on its top die for the flagship RDNA 5 card according to new rumors.

In a new post by Chiphell member ZhanZhanHao, AMD's flagship RDNA 5 gaming GPUs would feature up to 128 cores per compute unit, which is a huge upgrade (double) over the current flagship RDNA 4 GPU inside of the Radeon RX 9700 XTX which has 64 CUs.

AMD is expected to launch at least four different GPU die configurations for RDNA 5 according to leaks, in both the gaming-focused Radeon RX series and workstation-based Radeon PRO series GPUs. We should expect the higher-end RDNA 5 die with 96 CUs (12,288 cores), while the mid-tier card would sport 40 CUs, low-end die with 24 CUs, and the entry-level die with 12 CUs. Here's how the RDNA 5 stack would look:

When we reviewed the MSI GeForce RTX 5070 Ti Gaming Trio graphics card earlier this year, we were not only impressed by its performance but also by MSI's compact and stylish redesign of its popular Gaming Trio model for the RTX Blackwell generation. With the combination of lighting and translucent, frosted materials on the front and a vented metal backplate featuring a holographic MSI dragon logo on the rear, we dubbed it the "best-looking GPU in MSI's arsenal."

And it's this design that serves as the basis for the new custom MSI GeForce RTX 5070 Ti 16G MLG EDITION graphics card from the company. The translucent touches are still present, but now feature a darker purple tone, accompanied by the GPU shroud and backplate arriving in a vibrant red color that is not typically seen in custom GPU designs.

If you're wondering what MLG is, it refers to the MLG (Mo-Loong-Gi, or Loong: Nia, meaning "Dragon Princess" in Chinese) character, an original creation from MSI. Her likeness is etched on the metal backplate of the MSI GeForce RTX 5070 Ti 16G MLG EDITION GPU next to the holographic MSI dragon.

GALAX has just launched its new custom GeForce RTX 5090 D V2 HOF OC LAB Plus-X Edition graphics card in China, its new flagship RTX 5090 D V2 with a knocked-down 24GB GDDR7 memory capacity.

The ongoing tensions between the US government and China saw the birth of the RTX 5090 D earlier this year, and now the cut-down RTX 5090 D V2 which has less VRAM (24GB versus 32GB on the RTX 5090 and RTX 5090 D).

GALAX's new RTX 5090 D V2 HOF OC LAB Plus-X graphics card has 24GB of GDDR7 as well as lower GPU clocks and power limits than its RTX 5090 D. The regular RTX 5090 D has a 2580MHz GPU boost clock, where its new RTX 5090 D V2 has a 2525MHz GPU boost, with a TDP of 575W (from 600W).

NVIDIA posted its latest financial results for fiscal Q2 2026, with $46.7 billion in revenue, a 56% year-over-year increase in Data Center revenue driven by Blackwell hardware. However, it's also good news for NVIDIA's Gaming and AI PC segment that includes the company's latest RTX Blackwell generation of GeForce RTX hardware.

According to the report, fiscal Q2 2026 Gaming revenue reached $4.3 billion, representing a 14% increase from Q1 2026 and a 49% rise from the same period last year. During this time, NVIDIA launched the mainstream-focused GeForce RTX 5060 GPU on May 19, which NVIDIA says has become the "fastest-ramping x60-class GPU ever" for the company.

Although we're not sure if "fastest-ramping" translates to "fastest-selling," it's clear that the GeForce RTX 5060 has been a massive success in recent months. This also means that we wouldn't be surprised if it starts to make big moves up the Steam Hardware & Software Survey results for August 2025.

There have been numerous cases of GPU cables melting on high-end and flagship GeForce RTX Series GPUs, making it somewhat normalized in 2025. And that's in the sense that we've seen enough cases now that it's surprising, and not in the sense that it's a widespread or standard issue. This has led to a shift from the 16-pin 12VHPWR connector that was introduced alongside the GeForce RTX 40 Series to the newer 12V-2x6 connector for the RTX 50 and Radeon RX 9000 Series.

Today's report of a GeForce RTX 5090 failing, a PNY ARGC OC model, wasn't due to an improperly inserted cable or an overheated and melted connector. Instead, one of the GPU's capacitors exploded, and the noise was so loud that it caused the PC user to jump out of their chair.

This report, accompanied by the photos you see, originates from Reddit and the pcmasterrace Subreddit, with RoboDogRush reporting that their GeForce RTX 5090 "exploded" while they were editing a video, not while gaming. This indicates that the GeForce RTX 5090 wasn't being stressed in the same way it would be running DOOM: The Dark Ages in 4K with Path Tracing and an uncapped frame rate.

If you're like us and obsessed with PC hardware, gaming, and what's on the horizon, you probably know the name Kepler_L2. As one of the most trusted and respected leakers of GPU information, if they're posting anything related to specs or hardware, it's worth taking note of. The latest Kepler_L2 drop over on the Anandtech Forums (via Wccftech) offers images of GPU die configurations or block diagrams for AMD's upcoming RDNA 5 or UDNA generation of Radeon graphics.

According to the image, the top RDNA 5 GPU will be a beast, featuring eight shader arrays and 16 shader engines, which deliver a total of 96 Compute Units. This would be AMD's flagship, a GeForce RTX 5090-like alternative, something it has chosen not to offer with its current RDNA 4-powered Radeon RX 9000 Series. There, the flagship Radeon RX 9070 XT features only 64 Compute Units.

The image also offers some insight into the memory configuration for AMD's next-gen flagship RDNA 5 or UDNA GPU, with 16 Unified Memory Controllers delivering a maximum bus or interface size of 512-bit. Now, will all of this hardware make its way into a commercial flagship Radeon RX 10090 XT desktop graphics card built for PC gaming? Or will it be limited to the workstation and AI space?

If AMD's RX 9070 GRE ever gets shipped outside of Asia - it's currently exclusive to that region - we might just get a 16GB flavor of the graphics card, which would be a definite crowd pleaser.

Of course, the RX 7900 GRE graphics card started in Asia only, and was later distributed elsewhere, so it isn't like it's out of the question for the RX 9070 GRE to follow in those footsteps.

The idea that AMD may be planning a variant of the RX 9070 GRE with 16GB of video RAM - currently it has a 12GB VRAM pool - comes from the Board Channels forum over in China (as flagged by VideoCardz).

ASUS is known for its premium PC gaming products, with the scuttlebutt throughout the industry being that every product the company releases comes with "ROG Tax" in comparison to its competition.

While that company's prices are certainly high compared to the competition, they do make some of the best products gamers can get their hands on, dominating multiple categories with innovative and impressively high-end variations of already established, and sometimes brand new product designs. For example, ASUS managed to squeeze an extra 10% performance out of a GeForce RTX 5090 with its 30th anniversary edition card, which also comes with a unique aesthetic and design language that is unlike any other GPU on the market.

There is also the ROG Azoth Extreme Wireless Gaming Keyboard, which we gave an Editor's Choice Award, and the recently reviewed ROG Strix OLED XG27UCDMG gaming monitor. However, sometimes ASUS decides to create a product solely for style points, and while in most cases, you can find a cheaper alternative that offers the same performance, you can't fault the company for providing a stylistic option. Introducing the ROG Astral RTX 5080 16GB Dhahab CORE.

After the US government granted licences to export AI GPUs to China, the nation has banned the NVIDIA H20 GPU, according to a recent report.

It was only last week we heard rumors that NVIDIA had sent notices to companies on its AI GPU production line to stop production of H20 AI GPUs headed for China. Those two companies were Arizona-based Amkor Technology and Samsung Electronics. The story of NVIDIA selling its highly sought-after AI chips to China has been nothing short of a fiasco, with NVIDIA initially set to lose a staggering $5.5 billion over trade restrictions on China.

However, the US government reversed its stance on NVIDIA's chip exports by opening up a license that the company and others like it can apply for. This license would grant NVIDIA the right to export the H20 to China, provided the company paid the US government 15% of the revenue from these sales. Now it doesn't seem any of that is going to happen, as China has now banned the sale of H20 GPUs across the nation, or at least that is what the recent report is claiming.

ASRock has released a new custom L-Type or right-angle 12V-2x6 cable for graphics cards that includes a notable safety feature when you pair it with an ASRock Taichi or Phantom Gaming Series PSU. First spotted by Videocardz, ASRock's new 12V-2x6 cable has a 'Cable Over-Temperature Protection' feature that includes an NTC sensor at the connectors that monitors cable temperatures.

Now, if you've been following the whole GPU cables melting story, you'll know that it began with the introduction of 12VHPWR power with the GeForce RTX 40 Series. Basically, high-end GPUs like the GeForce RTX 4090 became prone to cables burning and melting due to incorrect insertion or other stress. The shift to the newer 12V-2x6 connector standard ahead of the arrival of the GeForce RTX 50 Series was seen as a way to mitigate this issue, but cases of GPU cables melting are still cropping up.

This includes GPUs for AMD's new Radeon RX 9000 Series, which recently saw the first case of a cable connected to a Radeon RX 9070 XT burning out. To minimize the risk of a GPU cable melting, ASRock is releasing a custom cable with a temperature sensor that "sends a signal to the PSU to ensure safe operation."

With the rise of PC cases with panoramic views, alongside the increased size of enthusiast graphics cards in recent years, the idea of vertically mounting a GPU is becoming increasingly common. And when it comes to brackets and riser cables to support the latest PCIe 5.0 cards like the GeForce RTX 50 and Radeon RX 9000 Series, Lian Li's new 'Multi-Directional Vertical GPU Bracket Kit' might just be the ultimate solution.

This all-in-one solution not only supports vertical orientation for GPUs, but it also allows forward and backward tilting, as well as a three-level height adjustment (with 10.16mm increments) for improved airflow for bottom-mounted fans or radiators.

The Multi-Directional Vertical GPU Bracket Kit mounts directly to the motherboard standoffs for secure installation and is built to support all modern GPUs, with its reinforced support arm rated up to 10kg. Lian Li notes that this will prevent GPU sag, while also supporting the largest quad-slot GPUs.

A couple of weeks ago, ASUS and Noctua's latest GPU collaboration launched with the release of the ASUS GeForce RTX 5080 Noctua Edition. Sporting three of the company's second-generation Noctua NF-A12x25 G2 120mm fans, an advanced vapor chamber, and other high-quality thermal bits and pieces, it's reportedly one of the quietest high-end GeForce RTX 50 Series graphics cards on the market.

It's also one of the largest, as putting three Noctua fans on a GPU that you'd typically associate with PC and radiator cooling means that the GeForce RTX 5080 Noctua Edition is a big boy. With dimensions of 385x151x80 mm, it's a rare quad-slot GPU. However, it also ships with a generous +115 MHz increase to the Boost Clock, so you're getting overclocked performance.

As a premium and custom GPU with cooling designed in collaboration with Noctua, if you're wondering why ASUS and Noctua didn't team up to create a custom (and near-silent) GeForce RTX 5090, the answer is simple. Getting access to the RTX 5090's GB202 chip is increasingly difficult due to shortages and the insatiable demand for high-end NVIDIA hardware for AI.

A US retailer is selling a blower-style RTX 5090 graphics card which is a dual-slot effort, but if you want this spin on the Blackwell flagship, it's seriously pricey - and the GPU comes with other caveats, too.

This is the AFOX GeForce RTX 5090 32GB Professional graphics card and it is being sold by HydraCluster Tech Builds for a towering $5,999 - which is supposedly $500 off the retail price (add seasoning now, and lots of it).

Tom's Hardware picked up on VideoCardz originally noticing the new RTX 5090, which as noted is a two-slot affair - but it's anything but compact in its other dimensions, certainly in terms of its beefy width.

NVIDIA has reportedly halted production of its H20 AI chip that is designed specifically for the Chinese market, with NVIDIA sending notifications to Arizona-based Amkor Technology and Samsung Electronics to stop manufacturing.

NVIDIA's AI chips, specifically the H20, are yet again up in the air. The sale of AI chips to China has been nothing short of a fiasco, with NVIDIA initially set to lose a staggering $5.5 billion over trade restrictions implemented on China for high-powered AI chips. These trade restrictions were set to ensure the US remains at the top of the global AI race, and at the time cited national security risks as China would use NVIDIA's GPUs to develop a sophisticated AI that the Chinese military could then utilize.

Only a month later, the H20, which is a heavily modified version of the H100 GPU, was banned from entering China. However, by July, the US government had changed its stance on NVIDIA chips sold to China. NVIDIA was able to secure a license, which would approve exporting the H20 to China, provided that 15% of the revenue from these sales would be directly allocated to the US government.

AMD could change the game with its next-gen Radeon GPU release, renaming the new RDNA 5-based graphics cards to the Radeon PTX series... and there's a (pretty) good reason for that.

Moore's Law is Dead covers the topic in his new video, pondering whether AMD should redo its branding with the new Radeon PTX naming scheme, given the ray tracing and path tracing performance inside of RDNA 5. In his video, MLID says that we could see AMD change up the naming scheme for the next-gen RDNA 5-based Radeon GPUs, with the AT3 GPU called the Radeon PTX 1060 XT, and the AT4 GPU called the Radeon PTX 1050 XT.

You might be thinking what the hell "PTX" is, well... NVIDIA rebranded from GeForce GTX to GeForce RTX for the rise of Ray Tracing (RT + X = RTX) so AMD could do something similar with PTX (PT = Path Tracing + X).

AMD's next-generation RDNA 5-based discrete GPUs have had some very interesting new leaks surface, with more details on the "Alpha Trion 3" and "Alpha Trion 4" otherwise known as AT3 and AT4 in recent leaks.

In his latest video, Moore's Law is Dead discusses some new information he has on AMD's next-gen RDNA 5-based Alpha Trion 3 and Alpha Trion 4 GPUs. Starting with the Alpha Trion 3 "AT3" which is said to have 48 CUs of RDNA 5 with 20MB of L2 cache, either a 384-bit LPDDR6 or 256-bit LPDDR5X memory controller, 8 PCIe Gen5 lanes, and performance somewhere between the desktop GeForce RTX 4070 and Radeon RX 9070 graphics cards.

The Alpha Trion 4 "AT4" GPU is more cut down than AT3, with 24 CUs of RDNA 5 and 10MB of L2 cache, a 128-bit LPDDR5X memory controller, the same 8 PCIe 5.0 lanes, and performance between the desktop RTX 3060 and RTX 4060. Both AT3 and AT4 will have much higher performance in ray tracing and path tracing, with releases in 2027.

We have heard about 12VHPWR connectors melting on NVIDIA's RTX 40 series graphics cards, but never on AMD cards.

That discrepancy was due to AMD never adopting the newer power connector standard with the company's latest generation of graphics cards, as AMD explicitly stated it was sticking with the 2x 8-pin connectors for their reliability. After the fiasco of melting connectors with NVIDIA's high-end RTX 40-series GPUs, this even became a selling point for AMD graphics cards.

However, with the RX 9000 series, some partners, such as ASRock, decided to adopt the 12VHPWR connector, and now we have the first report of a melting 12VHPWR connector on an AMD Radeon graphics card. The user reporting the melting was rocking an ASRock Radeon RX 9070 XT Taichi OC, and judging by the details from the user, there were a few reasons as to why it melted.

With the release of AMD's FidelityFX SDK 2.0, FSR 4 is finally available for all game developers, with integration available via signed DLLs. And to make things super easy for developers creating titles with Unreal Engine 5, this update also includes an Unreal Engine 5 plugin for developers to add FSR 4 support in games.

AMD's FSR 4 is the latest iteration of the company's FidelityFX Super Resolution technology that moves to an AI solution similar to NVIDIA DLSS. As such, it's currently exclusive to AMD's RDNA 4 desktop GPUs, like the Radeon RX 9070 XT, which include advanced AI hardware. Per our reviews and hands-on time with FSR 4, it's safe to say that it presents a massive leap forward in image quality compared to FSR 3, and finally closes the gap between FSR and DLSS.

However, unlike previous versions of FSR, FSR 4 is not officially 'open source' yet, with code available to all developers. Well, with the release of FidelityFX SDK 2.0, it was for a moment, as AMD briefly released the FSR 4 source code before taking it down.

MaxSun is prepping its new Intel Arc Pro B60 Dual 48GB Turbo graphics card for release, featuring two Arc Pro B60 GPUs each packing 24GB of VRAM for a total of 48GB GDDR6. Check it out:

We've been keeping a close eye on this graphics card, as I'm personally a long-time (30 years+) fan of all things GPU, and especially dual-GPU offerings. MaxSun's upcoming Arc Pro B60 Dual 48GB Turbo graphics card fits into a regular PCIe slot that splits into two 8-lane connections, with a GPU assigned to each of those 8 lanes.

The PC detects the card as two GPUs, even though it's physically connected to a single PCIe slot, and unlike other dual-GPU graphics cards, this isn't for gaming. MaxSun's new Arc Pro B60 Dual 48GB Turbo graphics card is for professionals, and more so, AI workloads.