Graphics Cards - Page 116

We did hear rumors about NVIDIA's new GeForce RTX 3090 SUPER graphics card before it was confirmed, and later released in the form of the current flagship Ampere GPU: the GeForce RTX 3090 Ti.

NVIDIA never released the GeForce RTX 3090 SUPER, but we did hear rumors of the RTX 3090 SUPER back in July 2021 with rumors of the GeForce RTX 3090 SUPER, GeForce RTX 3080 SUPER, and GeForce RTX 3070 SUPER. Instead, NVIDIA went with the "Ti" branding over the "SUPER" release with the GeForce RTX 3090 Ti, GeForce RTX 3080 Ti, and GeForce RTX 3070 TI graphics cards.

In a new leak onto the NGA forums, we see the black frame that NVIDIA was eyeing off... eventually going with the silver frame even on the flagship GeForce RTX 3090 Ti. AMD did something similar with their fastest Navi 2X-based graphics cards in the Radeon RX 6x50 card releases. I dig the style, and wish NVIDIA made its Founders Edition graphics cards available in more color options.

A prototype version of NVIDIA's upcoming AD102-based GeForce RTX 4090 has been leaked, with a triple-fan cooler teased that won't be used in the final design.

The new NVIDIA GeForce RTX 4090 prototype was leaked by "Greymon55" but we don't know which AIB partner has made this, although it looks very similar to many different triple-fan coolers on high-end graphics cards that are already on the market. It looks very similar to triple-fan ASUS and MSI graphics card designs, but is also giving me some triple-fan Vega GPU-cooling Radeon VII vibes.

NVIDIA's new GeForce RTX 4090 graphics card has 16384 CUDA cores, 24GB of ultra-fast GDDR6X memory and up to 450W of power according to the latest rumors. We're expecting 600W to be unleashed, but that could be for a beefed-up GeForce RTX 4090 or higher-end GeForce RTX 4090 Ti that will see Ada Lovelace chewing down another 150W+ of power.

NVIDIA's new GeForce RTX 4080 Founders Edition graphics card just turned up in some fresh leaks, with a gander at its newly-minted font and RTX 3090-class chunky cooler. But now... now it's time for some color.

COLORFUL's next-gen iGame GeForce RTX 4080 Ultra OC graphics card is now in the wild, after an anonymous source provided me with this image. This of course could be fake, but if it's real then COLORFUL has a real nice-looking design on its new iGame GeForce RTX 4080 Ultra OC graphics card.

The card itself features a single 16-pin power connector, with provisions for two more on what I'm assured to see as some monster custom GeForce RTX 4080 or GeForce RTX 4090 from COLORFUL. COLORFUL's design is very similar (except its switched white to black) from its iGame GeForce RTX 3050 Ultra W DUO OC 8G, that I reviewed earlier this year. I loved the design then, and even said I'd love to see a higher-end graphics card with that design and here we have the iGame GeForce RTX 4080 Ultra OC. Yep, love that style.

NVIDIA's next-gen GeForce RTX 4080 graphics card is in the headlines in a big way today, where the purported GeForce RTX 4080 Founders Edition smiles for the camera.

The new picture is coming from leaker "KittyYukko" with a single picture to Twitter, teasing the GeForce RTX 4080 Founders Edition with a chunky cooler that looks like the bigger GeForce RTX 3090 series Founders Edition cooler. We could be looking at a dual-slot design since the photo is from the top, and it could be fake, but we're just reporting what we're seeing.

NVIDIA is expected to use the AD103-300 GPU inside of the GeForce RTX 4080 graphics card, with 9728 CUDA cores, or 76 SMs enabled in total (out of the possible 84 SMs). Previously, the GeForce RTX 4080 was expected to have 10,240 CUDA cores, or 80 SMs. Fresh rumors also suggested it will feature 16GB of GDDR6X memory clocked at 23Gbps on a 256-bit memory bus that would feed out up to 768GB/sec of memory bandwidth on a 340W TDP.

Intel is talking up the ray tracing performance of its Arc GPU hardware, where the company has flown Ryan Shrout and Tom Peterson out to Berlin, Germany for IFA 2022 and revealed new details on the RTU (Ray Tracing Unit) and its performance.

Intel's Ryan Shrout and Tom Peterson spoke with PCGamer's Senior Hardware Editor, Jacob Ridley, where they talked about a fair amount but when it came to ray tracing we've got some confident new comments from the company. The Ray Tracing Unit (RTU) on Arc GPUs is capable of delivering "real" ray tracing performance notes Tom Peterson, with the Arc A770 and Arc A750 battling the RTX 3060 and keeping up or beating it.

Tom continued, saying that the ray tracing performance Intel offers with its Arc GPUs is competitive or better than NVIDIA's ray tracing hardware. Tom said: "The RTU [ray tracing unit] that we have is particularly well suited for delivering real ray tracing performance. And you'll see that when you do ray tracing on comparisons with an [RTX] 3060 versus A750 or A770, we should fare very, very well. Yeah, we're definitely competitive or better than NVIDIA with ray tracing hardware".

The US government has just issued new export licensing requirements to both NVIDIA and AMD, stopping them from exporting their respective advanced GPU designs for AI-based applications to both Russia and China.

The new export controls aren't banning specific GPUs like NVIDIA's new Hopper H100 or Ampere A100 GPUs, nor is it targeting AMD's new Instinct MI200 or Instinct MI250X GPUs but rather a particular threshold of performance. NVIDIA's current A100 GPUs get close and the H100 GPUs would blow past it, so they've received notification from the US government of the new export limitations.

AMD and its current Instinct MI200 and upcoming MI250X are also slapped with the new licensing requirements, since their next-gen GPUs are super-fast at AI-related tasks and the US government can't have the Chinese government having that type of AI power. The Biden administration along with the US Commerce Department have installed new rules that will block the exporting of chip design software that is necessary to make next-gen chips... these new export restrictions are an extension of that.

Jon Peddie Research's new report shows that GPU shipments passed 84 million units in Q2 2022, which is down 34% year-over-year.

JPR noted in its new report that overall, GPUs will have a compound annual growth rate of 3.8% between 2022-2026, where the firm predicts GPUs will hit an install base of 3.1 billion units. Over the next 5 years, JPR said that the penetration of discrete GPUs (dGPUs) in the PC will grow to reach 30%.

In the last quarter, AMD's overall GPU market share dropped 1.1%, Intel's overall GPU market share actually increased by 2% while NVIDIA's GPU market share dropped 3.15%. When it comes to overall GPU shipments, AMD shipments dropped by 7.6%, Intel shipments dropped 9.8%, but NVIDIA (which also sells the most graphics cards) saw the steepest decline with a drop of a painful 25.7%.

AMD showed off its new Radeon RX 7000 series reference graphics card during its Ryzen 7000 series "Zen 4" processor unveiling, with what looks like the design of the new flagship Radeon RX 7900 XT graphics card.

The next-gen Radeon RX 7900 XT reference design from AMD is continuing with its black color scheme, which is something AMD leaned into a little heavier with its suped-up Navi 21-based Radeon RX 6950 XT graphics card. AMD CEO Dr. Lisa Su said during the live stream that performance on the new RDNA 3-powered Radeon RX 7000 series GPU looks "absolutely wonderful".

We can't see any PCIe power connectors on the sneak peek of the Radeon RX 7000 series graphics card from AMD, but we should expect dual 8-pin PCIe power connectors on the flagship Radeon RX 7900 XT. NVIDIA has been rumored to be tapping the newer 16-pin PCIe power connector for its next-gen Ada Lovelace-powered GeForce RTX 40 series GPUs, while it's not known if AMD will use the newer 16-pin power connector... or continue on with dual 8-pin PCIe power connectors.

AMD officially unveiled its next-gen Ryzen 7000 series "Zen 4" CPUs during its livestream today, with the new flagship Ryzen 9 7950X processor tested with AMD's next-gen RDNA 3-based Radeon RX series GPU.

During the live stream, AMD CEO Dr. Lisa Su showed off what AMD is playing around with internally: their new Zen 4-powered Ryzen 9 7950X processor with 16 cores and 32 threads of CPU power at up to 5.7GHz with their new RDNA 3-based GPU in the world's first public demo. Out of all of the games on the planet, AMD tested its next-gen GPU silicon with Lies of P.

The game was run at 4K on Ultra graphics settings, with the Ryzen 9 7950X processor and a pre-production RDNA 3-based Radeon RX 70000 series GPU. We don't know the exact specs of the new GPU just yet, but the flagship model is expected to feature AMD's next-gen Navi 31 GPU that should pack 12288 Stream Processors, and 24GB of GDDR6 memory and also be made on TSMC's 5nm process node: just like the new Zen 4 processors.

NVIDIA's next-gen GeForce RTX 4060 Ti and GeForce RTX 4060 graphics cards will be consuming more power this time around for an x060-class GPU: rumor has it we're looking at up to 280W.

The new Ada Lovelace GPU architecture will debut with the GeForce RTX 40 series graphics cards, where we should see the flagship GeForce RTX 4090 first... but the GeForce RTX 4060 Ti and GeForce RTX 4060 graphics cards are in the news with some purported 3DMark TimeSpy Extreme benchmark results.

NVIDIA's new GeForce RTX 4060 Ti graphics card reportedly scores around 8600 points in TimeSPy Extreme, with an average GPU clock of 2600MHz+ and power draw of 270-280W or so says leaker "QbitLeaks". The GeForce RTX 4060 on the other hand will score around 6000 points, with an average GPU clock of 2700MHz+ and power draw of 230-240W.

If you thought the recent price drops on AMD Radeon RX series and NVIDIA GeForce RTX series graphics cards were good, we might not have seen the best of it according to the latest reports.

A new report on MyDrivers said that according to industry sources in Taiwan that the recent price cuts hasn't shifted as many graphics cards as they wanted, with warehouses and retailer still filled. A new second wave of price cuts is now on the way, which will be a bigger effort (hint: even bigger price cuts) than the first price cuts to graphics cards from AMD and NVIDIA.

AMD and NVIDIA further slashing prices won't automatically shift hundreds of thousands or millions of graphics cards needed to prepare for the next-gen GPUs around the corner, so expect some tactic changes. First, we'll see these price cuts get more AMD Radeon RX 6000 series and NVIDIA GeForce RTX 30 series graphics cards into gamers' hands at an even better price than today.

Intel's new Arc A380 graphics card has been tested in 50 games across a spectrum of titles released between 1999 and 2022, including DirectX 9, DX10, DX11, DX12, OpenGL, and Vulkan titles.

The folks over at PC Games Hardware has tested the Intel Arc A380 graphics card, which is great to see after hearing Intel Arc GPUs will be using a DirectX 9 to DirectX 12 emulator... as the Arc A-series desktop GPUs have no native DX9 API capabilities. This means we get Arc hardware + Arc software + GPU compatibility in 50 games as a great early look at the Arc GPUs (but not up against AMD and NVIDIA in these results).

As for the games tested, the new Intel Arc A380 graphics card ran games flawlessly including Crysis, Control, Cyberpunk 2077, Elden Ring, F1 2022, FEAR (one of my favorites), Red Dead Redemption 2, Stray, The Witcher 2 and 3, and many, many more. As for the games that didn't work, PC Games Hardware notes that Max Payne 2 -- which runs on the older DirectX 8 API -- "doesn't start at all".

Intel's new Arc A380 graphics card has been put through its paces in multiple ways, with an interesting new one here: Ethereum mining on Intel's new GPU.

The developer of Nanominer has added support recently for Intel's new Arc GPU, where it's been benchmarked with some Ethereum mining... performance wise the new Intel Arc A380 graphics card spits out just 10.2MH/s without any optimizations enabled. We're looking at 75W of power which isn't bad for power consumption, but performance wise the new Intel Arc A380 is worse than NVIDIA's lower-end GeForce GTX 1660 series GPUs.

Ethereum miners are preparing for Proof of Stake in the coming weeks, making Ethereum mining on the Intel Arc A380 kinda useless. It's good to see the performance and whatnot of Intel's new GPU, but when the DAG size for Ethereum is 5127MB, then the VRAM on the Intel Arc A380 becomes an issue.

Intel has just detailed its XeSS technology that will be launching alongside its upcoming Arc GPU desktop graphics card family, which drops next month. Ahead of that, we've got some juicy new details on Intel's upcoming XeSS upscaling tech.

In a new demo, Intel's Ryan Shrout and Tom Peterson use an Intel Arc A770 graphics card running Shadow of the Tomb Raider at 1440p and at maximum graphics settings, with ray tracing enabled. Tom ran his PC with XeSS on the "Balanced" preset, while Ryan was running Shadow of the Tomb Raider at the native 2560 x 1440 resolution with XeSS disabled.

XeSS on "Balanced" was running Shadow of the Tomb Raider at around 60-80FPS for the most part, while natively, the Intel Arc A770 was pushing out less than 60FPS, and spent most of its time at 40-50FPS or so.

NVIDIA has confirmed that it pulled in $6.7 billion in revenue during its earnings results for Q2 FY2023, with the company explaining the reason: lower sales in GeForce RTX series graphics cards because of a "tough macroeconomic environment".

During the call, NVIDIA's chief financial officer Colette Kress explained that the 33% drop in Gaming revenue for NVIDIA from this time last year, 44% sequentially. Kress explained that the decreases in GeForce GPUs were "primarily attributable to lower sell-in of Gaming products, reflecting reduced channel partner sales due to macroeconomic headwinds".

Translation: the neutron bomb that went off in the GPU crypto mining boom -- and the massive drops in cryptocurrency pricing, and power costs and power concerns in multiple countries in the future -- and the massive prices of GPUs in the last 12-18 months haven't helped, oh and the now back-to-Earth pricing of GPUs isn't enough to spur huge sales of our previously confident sales of GeForce GPUs.

Alright... and so we begin. NVIDIA has just teased their next-gen Ada Lovelace GPU architecture, with CEO Jensen Huang saying during NVIDIA's recent Q2 FY2023 results that their "new next-generation coming".

NVIDIA had to say something to help start the hype train, considering their Q2 results weren't good at all... gaming revenues plummeted 33% annually. Jensen addressed this, saying that NVIDIA is off the highs and that the macro condition turned "sharply worse".

Jensen said that NVIDIA is planning to reduce the sell-in over the next few quarters, which will help get rid of the GeForce RTX 30 series graphics cards in the channel. Jensen highlighted that NVIDIA's current-gen Ampere GPU is the most popular GPU that the company has ever made, sitting in the top most 15 popular gaming GPUs on Steam "and it remains the best GPUs in the world, and it will be very successful for some time" he added.

Intel has unveiled its new Data Center GPU Flex Series GPUs, which were previously codenamed "Arctic Sound-M" with the introduction of the Flex Series 170 and Flex Series 140.

It seems that Intel is stripping itself of the Arc GPU naming with the rebranding of Arctic Sound-M, flexing into a new brand name: Flex Series. The consumer Arc GPU side isn't doing so well, with rumors that the entire division is in trouble and analysts saying Intel should package up the GPU business and sell it off... now a rebranding of the workstation Arc cards to Flex.

Intel says that its new GPU is built to handle the various workloads that users will have, without compromising on performance or quality. The company adds that the new Flex GPU series help lower and optimize the "total cost of ownership for diverse cloud workloads like media delivery, cloud gaming, AI, metaverse, and other emerging visual cloud use cases".

It looks like NVIDIA's next-gen GeForce RTX 4090 graphics card has entered production, with leaked OEM documents teasing the GeForce RTX 4090 with 24GB of VRAM is in production.

A post on the Baidu forums spotted by "harukaze5719" on Twitter, teases that the AD102 GPU + 24GB memory graphics card aka the RTX 4090 on this paperwork. We don't know which AIB partner is getting their custom GeForce RTX 4090 graphics card made, as those bits are obviously blurred out. We do know that production kicked off on August 16.

The listing shows the RTX 4090 with the AD102 GPU and 24GB VRAM, there's 3 DP + 1 x HDMI display output: which is the regular display output on graphics cards. Here's hoping those DP connections are the ultra-fast new DisplayPort 2.0 standard.

Birentech has rolled out some new details on its new Biren BR100 GPU, with the Chinese company revealing more on its powerful new GPU at Hot Chips 34.

The new Birentech BR100 GPU uses an in-house GPU architecture, baked onto the 7nm process node and packing 77 billion transistors in total. Quite the hefty monolithic GPU from Birentech, competing directly against AMD and NVIDIA.

The BR100 GPU inside is fabricated on TSMC's 2.5D CoWoW design with 300MB of on-chip cache, with 64GB of HBM2e memory with 2.3TB/sec of memory bandwidth on the PCIe 5.0 x16 (CXL interconnect protocol). Birentech is using two chiplets on its BR100 GPU, with each of the GPU chiplets packing 16 SPCs (Streaming Processing Clusters). BR100 has a 550W TDP, and comes in an OAM module.

NVIDIA's next-gen GeForce RTX 4080 graphics card has some fresh new specs, with leaker "kopite7kimi" providing some updates on the second-fastest Ada Lovelace GPU.

The new GeForce RTX 4080 graphics card was first rumored with a rather large 420W, but the last purported power consumption on the RTX 4080 was scaled down to 320W. But now, kopite7kimi is saying that the GeForce RTX 4080 will have total card power of 340W, but more interestingly: 16GB GDDR6X at a now faster 23Gbps bandwidth.

NVIDIA's current-gen Ampere-based GeForce RTX 3080 has 10GB of GDDR6X, which isn't enough in its class against AMD's fastest Radeon RX 6900 XT (at the time) with 16GB GDDR6, and the replacement of the RX 6900 XT in the Radeon RX 6950 XT continued with 16GB GDDR6X. NVIDIA's faster GeForce RTX 3080 Ti only has 12GB of GDDR6X, while the real dominance happens with the GeForce RTX 3090 and GeForce RTX 3090 Ti both with 24GB of GDDR6X memory.