Graphics Cards - Page 109

AMD has officially unveiled its new Radeon RX 7900 XTX and Radeon RX 7900 XT graphics cards, and now AIB partners can unveil their custom Radeon RX 7900 series GPU designs.

ASUS is here with its new TUF Gaming Radeon RX 7900 XTX and TUF Gaming Radeon RX 7900 XT graphics card, both featuring a triple-slot and triple-fan design. Both of the new cards will require 3 x 8-pin PCIe power connectors, up from the 2 x 8-pin PCIe power connectors on AMD's in-house reference Radeon RX 7900 XTX and Radeon RX 7900 XT graphics cards.

Remember that AMD includes the brand new DisplayPort 2.1 connector on its new Radeon RX 7000 series GPUs, which can drive 8K 165Hz and 4K 480Hz (and an incredible 1440p at 900Hz). The new ASUS TUF Gaming Radeon RX 7000 series GPUs are the first graphics cards from ASUS that rock the new DisplayPort 2.1 connector.

AMD has formally unleashed its new Navi 31-powered Radeon RX 7900 XTX and Radeon RX 7900 XT graphics cards, both powered by the new RDNA 3 GPU architecture, and man... they're 4K and 8K gaming powerhouses.

The new Radeon RX 7900 XTX has some serious power inside of itself, with AMD completely thrashing NVIDIA in one department: including the new DisplayPort 2.1 standard while NVIDIA has stuck almost in the 90s with its DisplayPort 1.4 connectors on its fleet of new GeForce RTX 40 series graphics cards.

DisplayPort 2.1 can drive far, far higher refresh rates at higher resolutions where we're talking about the truly dizzying heights of 8K 165Hz and 4K 480Hz, which should melt most gamers' eyeballs. Hell, DP2.1 can even drive an absolutely bonkers 1440p at 900Hz (up from 1440p 480Hz on DisplayPort 1.4).

AMD invited the tech press, influencers, YouTubers, and its partners to Las Vegas for its huge RDNA 3 GPU reveal, with my invite sadly declined over personal reasons... but it hasn't stopped the unleashing of some real game-changing Radeon RX 7000 series graphics cards.

I've already covered AMD's new Radeon RX 7900 XTX graphics card, but now we have the second-best RDNA 3-powered Radeon RX 7900 XT graphics card that is just slightly cut down from its bigger brother.

AMD's new Radeon RX 7900 XT still rocks the impressive Navi 31 GPU, which is the first consumer GPU chiplet with 84 CUs and a 2GHz game clock. There's 20GB of GDDR6 memory on the Radeon RX 7900 XT spread out on a 320-bit memory bus, with 300W TBP.

AMD has finally unveiled its new Radeon RX 7900 XTX and Radeon RX 7900 XT graphics cards, where we'll be diving into the new flagship Radeon RX 7900 XTX first. Spoiler alert: AMD's new monster Radeon RX 7900 XTX launches December 13 for just $999... laughing in the face of the $1500+ pricing on NVIDIA's also monster GeForce RTX 4090 graphics card.

The new AMD Radeon RX 7900 XTX graphics card is powered by the new RDNA 3 GPU architecture, offering 96 CUs, and a 2.3GHz GPU game clock across its 12288 cores. AMD slaps a large 24GB of GDDR6 memory, matching its competitor in the NVIDIA GeForce RTX 4090 with 24GB of GDDR6X.

AMD's new Radeon RX 7900 XTX and its 24GB of GDDR6 memory is spread out on a 384-bit memory bus, joined by 96MB of Infinity Cache (lower than the 128MB of Infinity Cache on Navi 21-powered GPUs) which offers up to an insane 5.3TB/sec of memory bandwidth (yes, you read that correctly). AMD has a 355W TBP (Total Board Power) for the Radeon RX 7900 XTX, only 55W more than the Radeon RX 7900 XT (300W).

NVIDIA's new monster GeForce RTX 4090 graphics card is now in the wild, with high-end custom designs from the likes of GALAX and its new flagship GeForce RTX 4090 HOF (Hall of Fame) graphics card now teased in all of its PCB nudity.

The new GALAX GeForce RTX 4090 HOF graphics card features power-hungry dual 16-pin power connectors, while inside of the card itself we have the usual AD102-300 GPU with its 16384 CUDA cores and 24GB of GDDR6X memory clocked at 21Gbps offering 1TB/sec+ of memory bandwidth.

But when it comes to the PCB of this monster new GeForce RTX 4090 HOF, GALAX is using a hefty 28+4 phase VRM which makes it the most powerful RTX 4090 so far. GALAX used a similar design and power delivery system for its Ampere-based flagship GeForce RTX 3090 Ti HOF (which used a 24+4 phase VRM) but the new GeForce RTX 4090 HOF is another step up from that, big time.

We are now mere hours away from AMD's huge Radeon RX 7000 series "RDNA 3" graphics card reveal, where we will be introduced to the Radeon RX 7900 XTX and Radeon RX 7900 XT... and now, we've got our first real look at the Navi 31-powered cards.

A couple of pictures were shared by leaker @9550pro who said they have pictures of the new Navi 31-based designs from a closed QQ chat group. It looks like we're scoping out some prototype Radeon RX 7900 XTX and Radeon RX 7900 XT graphics cards, but we know that the dual 8-pin PCIe power connectors will be standard on the reference cards from AMD.

Up against the AMD Radeon RX 6900 XT reference graphics card, the new Radeon RX 7900 XTX -- or the Radeon RX 7900 XT, we don't know what we're looking at here exactly -- the new Navi 31-powered flagship graphics card is longer. Still, maintains the dual 8-pin PCIe power connectors versus the monster 16-pin 12VHPWR power connector on NVIDIA's new GeForce RTX 40 series graphics cards.

Intel has just announced that its new Data Center GPU Flex Series cards have been added to their family of PluggableDevices, something that is called Intel Extension for TensorFlow.

Intel Extension for TensorFlow is an open-source solution that runs TensorFlow applications on Intel AI hardware, allowing the high-performance deep learning extension for the TensorFlow PluggableDevice interface. It will allow Intel XPU (GPU, CPU, and more) devices to be readily accessible to TensorFlow developers.

With the new Intel Extension, the company says that developers can train and infer TensorFlow models on Intel AI hardware with "zero code change". Intel Extension for TensorFlow is built on the foundations of the oneAPI software components, with most of the performance-critical graphs and operators being highly optimized by Intel oneAPI Deep Neural Network (oneDNN) which is an open-source, cross-platform performance library for Deep Learning applications.

Intel has its new Arc A770 and Arc A750 graphics cards in the wild now, with the flagship Arc A770 Limited Edition from Intel packing 16GB of GDDR6... except, Intel launched it with some issues... the GDDR6 memory wasn't at the right clock speed.

Intel's new Arc A770 Limited Edition packs 16GB of GDDR6 memory that's meant to be clocked at 17.5Gbps, but gamers were noticing it was only clocked at 16Gbps. This results in lower memory bandwidth, with the 16GB of GDDR6 memory at 17.5Gbps meant to deliver 560GB/sec, but at 16Gbps the memory bandwidth drops to 512GB/sec.

The company said there was a "gap" on their side, but don't worry, the multi-billion-dollar corporation has put "two checks in place in our validation to ensure it doesn't happen again moving forward". I really hope it doesn't, as this is a seriously sloppy move by Intel to ship out its flagship Arc A770 Limited Edition graphics card with 16GB without the "gap" or whatever that means.

GIGABYTE has a bunch of custom GeForce RTX 40 series GPU designs on the way, with a few new GeForce RTX 4080 16GB graphics card designs in the works that have now been leaked. Check them out below:

We have the AORUS GeForce RTX 4080 Xtreme Waterforce graphics card, which is a new water-cooled GeForce RTX 4080 (duh, it's kinda obvious) while we have a new series from the company in the new "AERO" with a white theme.

The company is using a totally new design for its custom AORUS GeForce RTX 4080 Xtreme Waterforce graphics card, with a pre-installed water block... and we'll have to make it clear that this is the first RTX 40 series graphics card with a pre-installed waterblock. Very... very cool to see, pun intended.

NVIDIA has its new flagship GeForce RTX 4090 graphics card on the market and in the wild, lighting up gamers' PCs (and in some ways, far too much) while the new GeForce RTX 4080 16GB graphics card isn't too far away now.

In fact, it's so close that the new NVIDIA GeForce RTX 4080 16GB has turned up in the UK with a retailer listing custom designs from Palit starting from £1450 (which works out to around $1680 USD or so). Palit has its upcoming GeForce RTX 4080 OmniBlack 16GB graphics card for £1450 (which works out to around $1680 USD or so).

From there, there's the Palit GeForce RTX 4080 GameRock 16GB graphics card for £1500 (which works out to around $1740 USD or so) while the flagship Palit GeForce RTX 4080 GameRock OC 16GB graphics card is another £1530 (which works out to around $1776 USD or so).

We all know there are a bunch of AIB partners that sell NVIDIA GeForce graphics cards, AMD Radeon graphics cards, and just a few are now on the market selling Intel Arc graphics cards... but who leads the pack in terms of the number of custom cards made?

That would be GALAX.

GALAX makes an incredible 187 different designs of NVIDIA's previous-gen Ampere-based GeForce RTX 30 series graphics cards, which is a staggering amount of cards. We have the big AIBs like COLORFUL, ASUS, GIGABYTE, MSI, and ZOTAC... while EVGA is still on the list with 63 different GeForce RTX 30 series GPU designs, but zero going forward as EVGA terminated its relationship with NVIDIA.

We are only days away from AMD's huge Radeon RX 7000 series "RDNA 3" graphics card reveal on November 3, with slow trickles on leaks for what to expect.

Another day, another confirmation that AMD won't be leading with the Radeon RX 7900 XT and its 20GB of GDDR6 memory but rather the purported Radeon RX 7900 XTX with 24GB of GDDR6 memory will be the flagship Navi 31-based GPU design that we'll see unleashed on November 3.

I wouldn't be trusting random leakers, but VideoCardz picked up that leaker "chi11eddog" teased the Radeon RX 7900 XT with 20GB of VRAM, and the higher-end Radeon RX 7900 XTX with 24GB of VRAM "confirmed" says the leaker. The leaker shared two product stickers with hidden names, leaving out the "XT" and "XTX" and their respective 20GB and 24GB of frame buffers.

NVIDIA's new GeForce Game Ready 526.47 WHQL drivers are here, offering the best day-0 gaming experience for Sackboy: A Big Adventure.

Sackboy: A Big Adventure features ray-traced reflections, shadows, and ambient occlusion effects as well as support for NVIDIA's new Ada Lovelace-exclusive NVIDIA DLSS Super Resolution technology. NVIDIA's very, very impressive DLSS Super Resolution technology amplifies performance by up to 2.5x on the new GeForce RTX 4090 graphics card.

Not only does Sackboy: A Big Adventure get day-0 support in the new GeForce Game Ready 526.47 WHQL drivers, with the new drivers also supporting Victoria 3, WRC Generations - The FIA WRC Official Game, and it also introduces DLSS 3 technology into F1 22.

COLORFUL has just introduced 8 new custom GeForce RTX 3060 series graphics cards, 4 in the new GeForce RTX 3060 8GB variant, and 4 new GeForce RTX 3060 Ti GDDR6X variants.

NVIDIA's new GeForce RTX 3060 8GB is based on the same GA106 GPU that packs 3584 CUDA cores, but it's had its memory gimped to 8GB on a 128-bit memory bus that reduces overall memory bandwidth down to just 240GB/sec (which is down from the 12GB GDDR6 on the regular RTX 3060, with a much bigger 360GB/sec memory bandwidth).

COLORFUL is also re-releasing previous GeForce RTX 3060 12GB variants, but are adding "BattleAx Deluxe" and boosting GPU clock speeds from 1777MHz (reference clocks) up to 1867MHz. The TDP will change depending on which GeForce RTX 3060 variant you purchase, but it won't go above 220W.

NVIDIA is in some serious hot water over the 16-pin "12VHPWR" power adapter with its new flagship GeForce RTX 4090 graphics card, with multiple reports of cards catching fire and the "12VHPWR" power adapter melting, or worse.

The story gets juicier now with Igor's Lab reporting that NVIDIA reportedly notified all of its AIB partners that all damaged cards that they receive back from consumers, need to be sent directly to NVIDIA HQ for "failure analysis".

This news lines up with NVIDIA CEO Jensen Huang flying out to Taiwan to meet with TSMC (Taiwan Semiconductor Manufacturing Company) to nail 3nm wafers for its next-gen Blackwell GPUs... but maybe Uncle Jensen was calming AIB partners down as they're about to get reamed with returns, bad reviews from customers, and more.

NVIDIA had had a shaky start to its new Ada Lovelace GPU architecture, with the launch of the GeForce RTX 4090 graphics card being successful -- selling quicker than most expected, but it's been hit with controversy as well -- RTX 4090 cards catching fire, and the 16-pin "12VHPWR" power connector melting.

Not only that, but NVIDIA also "unlaunched" the GeForce RTX 4080 12GB graphics card that surely pissed AIB partners off who were gearing up their custom GeForce RTX 4080 12GB cards (and retail packaging).

Now there are reports that NVIDIA CEO Jensen Huang has flown to Taiwan, to meet with TSMC to make sure that the company secures 3nm wafers for its next-gen GPUs that won't see the light of day until 2024. Jensen also reportedly met with multiple AIB partners, giving them a personal update on their existing GeForce RTX 30 series "Ampere" GPUs.

AMD will be unveiling their next-gen RDNA 3 GPU architecture and at least a couple of members of the Radeon RX 7000 series graphics card family on November 3, but ahead of that the RDNA 3 GPUs get official Vulkan 1.3 support.

The Vulkan 1.3 specification was unleashed earlier this year in January 2022, supported by all three major GPU vendors: AMD, Intel, and NVIDIA. NVIDIA's just-launched GeForce RTX 4090 powered by the Ada Lovelace GPU architecture has support for the latest Vulkan 1.3.3.1 specification, with AMD's new Radeon RX 7000 series and the new RDNA 3 GPU architecture supporting Vulkan 1.3, too.

The conformance list for Vulkan 1.3 has the Intel Arc A-series GPUs, NVIDIA GeForce RTX 30 and RTX 40 series GPUs, as well as AMD and its upcoming "RDNA 3 sGPU" product family. AMD themselves have confirmed that some of their "undisclosed products" based on the RDNA 3 GPU family have successfully completed Vulkan 1.3.3.1 conformity test suite.

Manli has released its new custom GeForce RTX 3060 graphics card with 8GB of GDDR6 memory, with virtually identical specs to the GeForce RTX 3060 with 12GB of GDDR6 memory.

The new GeForce RTX 3060 has 8GB of GDDR6 clocked at 15Gbps but it's on a smaller 128-bit memory bus with up to 240GB/sec of memory bandwidth. Manli makes a GeForce RTX 3060 LHR (Lite Hash Rate) graphics card with 12GB of GDDR6, with this card being identical to the card except for the less VRAM, and slower memory bus.

The higher-end GeForce RTX 3060 with 12GB GDDR6 memory is on a wider 192-bit memory bus, with far higher memory bandwidth of up to 360GB/sec, so the cull down to 8GB GDDR6 + 240GB/sec memory bandwidth is significant.

AMD released its CDNA-based Instinct MI100 accelerator nearly two years ago now, back in November 2020, and while the third-gen CDNA architecture is nearly here, the first-gen MI100 accelerator has hit eBay... in the nude.

The AMD Instinct MI100 accelerator is based on the CDNA architecture and made on TSMC's 7nm process node, with its GPU codenamed "Arcturus" packing 7680 FP32 cores and 32GB of HBM2 memory. The card has a 300W TDP, chewing down its power from dual 8-pin power connectors -- but not regular PCIe power connectors, but rather the EPS ones that go into your motherboard.

The card was posted on eBay in non-functional order, with a price of 500 EUR (about $500 USD or so) which is a rather huge price drop from its MSRP at the time of $6400. The card doesn't work, just to be sure you know that, so this would be a collector's item. But man, in the nude the AMD Instinct MI100 accelerator is a beautiful card under that cooler.

AMD has confirmed that its next-gen Radeon RX 7000 series "RDNA 3" graphics card will NOT be using the new PCIe Gen5 "12VHPWR" power connector... which is something we kinda knew already, but now 100% confirmed from AMD themselves.

Scott Herkelman, General Manager for AMD Radeon (and ex-NVIDIA GeForce exec) confirmed on Twitter with Kyle "HardOCP" Bennett, where he said that the Radeon RX 6000 series and upcoming RDNA 3 GPUs will "not use this power connector".

Given that there are multiple reports of GeForce RTX 4090 graphics cards MELTING, it makes sense that AMD won't use the power connector. We've known for a while now that AMD's upcoming Radeon RX 7000 series flagship GPU will be using 2 x 8-pin PCIe power connectors, but now we have official confirmation from AMD themselves.