Graphics Cards - Page 37

Geekbench is a synthetic benchmarking tool that often does not indicate actual in-game performance, so take the following with a grain of 'let's wait and see' salt. Over the weekend, results for NVIDIA's upcoming GeForce RTX 5070 GPU appeared online. It is set to launch on March 5, with pricing starting from $549.

The tests cover both OpenCL and Vulkan APIs, and the GeForce RTX 5070's scores point to the GPU being around 20% faster than the GeForce RTX 4070 - a modest upgrade in line with other GeForce RTX 50 Series releases. However, compared to the GeForce RTX 4070 SUPER refresh from 2024, the performance improvement drops to around 5%. The GeForce RTX 5070 is launching with a lower CUDA Core count than the RTX 4070 SUPER - 6144 versus 7168, a 14% reduction. This could cause it to fall behind in some gaming workloads.

In many ways, the GeForce RTX 5070 sounds like a GeForce RTX 4070 SUPER replacement, with the Geekbench results hinting that it will be around 20% slower than the GeForce RTX 5070 Ti. However, it's worth noting that early Geekbench entries for the Radeon RX 9070 XT show the GeForce RTX 5070 outperforming RDNA's upcoming flagship card.

AMD will formally unveil the Radeon RX 9070 and Radeon RX 9070 XT at the end of the week, on February 28. The company's new flagship RDNA 4 GPUs target the mid-range PC gaming market. During the event, we expect an architectural breakdown of RDNA 4, the new AI-powered FSR 4 upscaling, and how these GPUs perform when playing the latest games.

Over the weekend, Videocardz posted what it has confirmed to be the official gaming performance figures for the Radeon RX 9000 Series. The figures show AMD's performance compared to the Radeon RX 7900 GRE. The Radeon RX 9070 is said to be 21% faster for 4K gaming, and the Radeon RX 9070 XT is 42% faster. These results, which cover 30+ games, include titles with ray-tracing, and performance in Cyberpunk 2077 is said to be 66% faster on the Radeon RX 9070 XT compared to the Radeon RX 7900 GRE.

Based on our review of the ASRock Radeon RX 7900 GRE Steel Legend, this would put ray-tracing performance in the realm of the GeForce RTX 4070 Ti SUPER. This is a massive improvement for Radeon. However, as the Radeon RX 7900 GRE is a lesser-known GPU compared to the Radeon RX 7900 XT, Radeon RX 7900 XTX, and RTX 4070 Ti, we're wondering if the comparison is being made due to price.

NVIDIA has now admitted that its GeForce RTX 5080 graphics cards are also affecting by missing ROPs, joining the RTX 5090 and RTX 5070 Ti with the same issue.

The company released a statement to The Verge, with NVIDIA GeForce Global PR director Ben Berraondo explaining: "Upon further investigation, we've identified that an early production build of GeForce RTX 5080 GPUs were also affected by the same issue. Affected consumers can contact the board manufacturer for a replacement".

NVIDIA has also amended its previous statement released when it was discovered that its new ultra-enthusiast flagship GeForce RTX 5090 (and RTX 5090D) and latest GeForce RTX 5070 Ti were missing ROPs in some cases.

AMD is busy working on its next iteration of AFMF (AMD Fluid Motion Frames) and it should arrive alongside the RX 9070 graphics cards.

That's the contention of a report from VideoCardz, which observes that AFMF 2.1 is inbound (taking the baton from version 2.0), and it promises to improve the image quality of the tech.

As a quick refresher, AFMF is Team Red's driver-level frame generation, which has the major advantage of not requiring the game developer to code in support. It works with everything (well, most games) via the driver, but at the same time, the weakness is that it upscales the full frame (interface and all the trimmings, which can make UI elements look shoddy).

We've just caught a hint about how AMD might be pricing its incoming RDNA 4 graphics cards, and it's supposedly from the horse's mouth.

This info comes from VideoCardz, and it should be regarded with caution as ever, in the form of a slide which is purportedly from a recent AMD press briefing.

It's a presentation slide about 'what gamers want' from their graphics cards and unsurprisingly, one of those factors is more accessible GPUs when it comes to pricing.

With the current GeForce RTX 50 Series stock shortages and difficulty finding cards at close to retail prices, ASUS is starting a new ROG Elite Rewards loyalty program that will give its members the opportunity to purchase a GeForce RTX 50 Series GPU at retail price directly from the source.

ASUS started this program in response to scalpers and bots driving up graphics card prices in recent weeks. The goal is to ensure that PC gamers and creators have the chance - and priority access - to obtain a powerful GeForce RTX 50 Series GPU. With the limited availability of GeForce RTX 50 Series GPUs, this is a fantastic move from ASUS.

"No scalping, no inflated prices, just a fair shot at cutting-edge performance," ASUS tells us. ROG Elite Rewards members, a free program you can sign up for, will enter a drawing, and those selected will be given the option to buy a GeForce RTX 50 Series card directly from ASUS.

NVIDIA is reportedly engaging in "scarcity marketing" of its new GeForce RTX 50 series GPUs, which is why the new Blackwell gaming cards are hard to find worldwide... sigh.

In a new post from industry insider @Jukanlosreve on X we're learning that the real reason that NVIDIA's new GeForce RTX 50 series GPUs are hard to find and pretty much out of stock worldwide is because the company is engaging in "scaracity marketing".

GPU inventory is "completely sufficient" but NVIDIA is reportedly controlling the supply and not releasing GeForce RTX 50 series GPUs -- right now the RTX 5090, RTX 5080, and RTX 5070 Ti have been released -- which is "thereby creating the impression of shortage". However, we're to expect that by April or May that GeForce RTX 50 series GPUs will be released in "large quantities".

We've just been treated to another hint that the incoming RX 9070 XT from AMD is going to be a seriously powerful graphics card.

This new leak comes from VideoCardz, which used its sources to get hold of what are purportedly some official benchmarks from an AMD press briefing.

The eyebrow-raising takeaway here is that if they're genuine, the leaked benchmarks - which are real-world gaming tests, thankfully - show the 9070 XT is some 42% faster than the RX 7900 GRE at 4K (Ultra) resolution.

NVIDIA has given us an update on the issue with black screens that has been plaguing a number of folks who've purchased a new Blackwell GPU.

These black screen bug check crashes have been happening on RTX 5090 and 5080 graphics cards since the launch of these GPUs a while back.

NVIDIA said it was going to investigate this bug when reports of problems first emerged earlier this month, and an admin over at NVIDIA's GeForce forums has just provided an update, stating:

If you're a GeForce RTX 50 Series early adopter and have a GeForce RTX 5090 or GeForce RTX 5070 Ti, there's a slight chance that it might not be performing as it should. After some users reported slightly slower-than-expected performance and discrepancies in their GPU specs versus the official specs, NVIDIA has confirmed that an issue affects the first run of RTX Blackwell cards.

In an official statement, NVIDIA has confirmed that 0.5% or 1 in 200 GeForce RTX 5090, RTX 5090D, and RTX 5070 Ti are shipping with fewer ROPs than specified. "The average graphical performance impact is 4%, with no impact on AI and Compute workloads," NVIDIA's statement reads. "Affected consumers can contact the board manufacturer for a replacement," it continues, adding that the "production anomaly" has been fixed.

NVIDIA didn't mention the GeForce RTX 5080 in its statement; however, a new case over the weekend on Reddit points to the ROPs issue potentially affecting RTX 5080 cards, too. With this in mind, let's break down how to check the ROP count on your GeForce RTX 50 Series GPU and the correct values you should see.

NVIDIA's GeForce RTX 50 Series launch has been plagued by issues, from availability to pricing to the return of melting power cables and connectors when trying to game on the new GeForce RTX 5090 flagship. The past few days have seen the RTX 50 Series make headlines for another issue: GPUs being shipped with missing ROPs.

Raster Operators (ROPs) are a key part of the render pipeline and are dedicated bits of hardware that make up the specs of a modern GeForce RTX graphics card. The new flagship GeForce RTX 5090 ($1999) ships with 176 ROPs; however, some early adopters noticed that their cards only reported 168 ROPs - and that performance was not where it should be.

NVIDIA released a statement to confirm that this issue has been corrected and that those affected can contact their board manufacturers for a replacement. According to Team Green, 0.5% or 1 in 200 GeForce RTX 5090, RTX 5090 D, and RTX 5070 Ti cards are affected. The company didn't mention the GeForce RTX 5080 in its statement, but over the weekend, Redditor "gingeraffe90" reported that GPU-Z was listing their RTX 5080 as having 104 ROPs instead of 112.

A new report from financial analyst and reporter Dan Nystedt and Taiwanese news outlet Commercial Times suggests that NVIDIA's GeForce RTX 5070 and RTX 5060 Series GPUs are experiencing production delays.

Although the GeForce RTX 5060 has yet to be formally announced, the GeForce RTX 5070 is still on track for a March 5, 2025, release with a starting price of $549 USD. With full production for the GeForce RTX 5070 pushed back until mid-March, this could lead to another GeForce RTX 50 Series launch (like the RTX 5090, RTX 5080, and RTX 5070 Ti) with limited stock and availability.

The GeForce RTX 5070 Ti, which launched last week, is currently sold out on Newegg, with all models listed as "Out of Stock. "According to this report, NVIDIA discovered "performance issues in the chips" for the GeForce RTX 5060 and RTX 5070, which required a fix.

Thailand's DIY market is totally exhausted of NVIDIA GeForce RTX 50 series GPUs, with a retailer telling customers that they'll be waiting until at least July for new Blackwell GPUs.

In a post on Weibo, "UX_thoughtful" said that the DIY section of a retailer in Thailand says that there no NVIDIA GeForce RTX 50 series GPUs whatsoever in the country, reporting that "come to Thailand to see the market, there is no 50 series graphics cards. In the DIY area, and the clerk said to wait until July".

NVIDIA launched its new GeForce RTX 5090 and RTX 5080 a few weeks ago now, with the RTX 5070 Ti launching days ago... we're not even into March yet, but Thailand has zero stock of the new RTX 50 series GPUs and it'll be a few more months to wait at the very least. Saddening day for PC gamers in Thailand.

AMD is just weeks away from unleashing its next-gen Radeon RX 9070 series "RDNA 4" graphics cards, with news breaking that we'll see NO reference model, leaving the future of RDNA 4 in the hands of AIB partners and custom Radeon RX 9070 series graphics cards.

In new rumors, we're learning that we won't see a "Made-By-AMD" (MBA) variant of the new RDNA 4-powered Radeon RX 9070 series GPUs, which means it'll be the first time in Radeon history that AMD won't have a reference model. There have been renders of the reference model so far, and now we know why official shots of the Radeon RX 9070 series reference cards haven't been released.

We are learning this news from a new share on X from AMD, where there is a render of the Radeon RX 9070 series GPU but the company states in the bottom left: "Artistic Render. Not Available for Purchase". Damn... would've been nice to see a new (possibly even more curvier) Radeon RX 9070 series GPU... but it appears we won't.

NVIDIA has confirmed issues of GeForce RTX 5090, RTX 5090D, and RTX 5070 Ti graphics cards shipping with missing ROPs, which reduce gaming performance.

The company has issued a statement, explaining that less than 0.5% of its cards are affected, and that consumers can contact their AIB board partners to get replacements sent out. Recently, TechPowerUp reported that they found missing ROPs on their GeForce RTX 5090 graphics card sample, with 188 ROPs found versus 176 ROPs on the RTX 5090.

NVIDIA explains: "We have identified a rare issue affecting less than 0.5% (half a percent) of GeForce RTX 5090 / 5090D and 5070 Ti GPUs which have fewer ROPs than specified. The average graphical performance impact is 4%, with no impact on AI and Compute workloads. Affected consumers can contact the board manufacturer for a replacement. The production anamoly has been corrected".

NVIDIA's new GeForce RTX 5090 and GeForce RTX 5070 Ti graphics cards are shipping with missing ROPs, seeing the flagship RTX 5090 experiencing 4.5% reduced performance.

The news started with TechPowerUp user "Wuxi Gamer" discovering that some review samples and even retail graphics cards with missing ROPs, with TPU noting that its ZOTAC GeForce RTX 5090 SOLID graphics card sample was underperforming by a few percentage points, falling behind NVIDIA's in-house GeForce RTX 5090 Founders Edition.

The missing ROPs are a major issue, with ROPs being responsible for pixel data, taking care of tasks like blending and anti-aliasing. NVIDIA usually reduces ROPs count to keep separation between SKUs, but this is a far bigger issue.

AMD's new RDNA 4-based Navi 48 GPU inside of the flagship Radeon RX 9070 XT graphics card has 53.9 billion transistors, meaning it has more transistors than NVIDIA's new GB203 GPU inside of the GeForce RTX 5080.

In a new post on Tieba Baidu, we're learning that AMD's new Navi 48 GPU should be 350mm2, which is smaller than the 390mm2 in previous leaks. AMD recently confirmed its new RDNA 4-based Navi 48 GPU has 53.9 billion transitors in a press briefing earlier this week.

If the new Baidu figures are correct, we're looking at AMD's new RDNA 4-based Navi 48 GPU being smaller than both NVIDIA's previous-gen Ada Lovelace-based AD103 GPU, and its new Blackwell-based GB203 GPU that is inside of the new GeForce RTX 5080 graphics card.

ASUS's new flagship ROG Astral GeForce RTX 5090 has caught fire with smoke billowing out of it, leaving burn marks on both the GPU and motherboard of a Redditor. Check it out:

In a post on Reddit, user "Impossible-Weight485" was browsing the internet when his PC shut down all of the sudden. After restarting his system, the graphics card caught fire and smoke started coming out of it. Not a good thing to see for one of the most expensive graphics cards on the planet.

The Redditor explained: "I was playing PC games this afternoon, and when I was done with the games, my PC suddenly shut down while I was browsing websites. When I restarted the PC, the GPU caught fire, and smoke started coming out. When I took out the GPU, I saw burn marks on both the GPU and the motherboard".

ASUS's in-house overclocking team has pushed NVIDIA's new GeForce RTX 5090 graphics cards to new limits: overclocking the GB202 GPU to a record 3.5GHz, and the 32GB of GDDR7 to 34Gbps, breaking multiple OC world records in the process.

SAFEDISK is ASUS's in-house OC secret weapon, overclocking a custom ASUS GeForce RTX 5090 graphics card under LN2 cooling and various tweaks. The GB202 GPU was pushed to an impressive 3480MHz while the 32GB of GDDR7 memory was pushed up to 34Gbps on its wide 512-bit memory bus.

The overclocker used AMD's Zen 5-based Ryzen 9 9950X processor with 16 cores and 32 threads of processing power, 48GB of DDR5-8000 memory from G.SKILL (dual 24GB DIMMs) and an ASUS ROG Crosshair X870E APEX motherboard. It will be interesting to see these results re-run with AMD's new Ryzen 7 9800X3D processor, too.

AMD's RX 9070 XT graphics card has seemingly been spotted in more leaked benchmarks and this time the claim is that this is going to be a seriously fast piece of hardware.

As you can see in the above post from Tomasz Gawroński (a regular contributor of leaks on X), these are apparently FurMark results for the RX 9070 XT.

That's a little muddied by the fact that the benchmarks (for 4K resolution) in question are attributed to a 'Radeon RX 7800 XT' card, but as Gawroński points out, the GPU carries the same PCI-ID as that previously attached to the RX 9070 in the likes of Geekbench leaks. So, it appears someone is trying to fudge things and cover up the identity of the RDNA 4 GPU.