Graphics Cards - Page 153

We have got our first proper look at Intel's upcoming Xe-HP GPU for data centers, with Igor Wallosek and an exclusive look at Intel's multi-tiled GPU. The pictures were shared to Igor from internal Intel slides, and were upscaled using AI software.

The new Intel Xe-HP architecture that we have more details on here packs more Execution Units (EUs) than previously rumors have stated. We have our first proper look at the Intel Arctic Sound 2T and Intel Arctic Sound 1T GPUs with the Arctic Sound 2T being a 2-tile GPU and the Arctic Sound 1T being a 1-tile GPU... pretty obvious, but let's break it down.

Above, we have the Intel Arctic Sound 2T which is a dual-tile Intel Xe-HP with 480 EUs per tile for a total of 7680 shader units (but this still isn't the full 8192 shader units we could expect). Intel is using 32GB of super-fast HBM2e memory and has a 300W TDP on the Arctic Sound 2T GPU. You will need a single 8-pin EPS connector here.

ASUS and GIGABYTE have had their CMP 30HX mining cards revealed, and now it's time for MSI to reveal its CMP 30HX MINER series for GPU cryptocurrency mining.

MSI has 2 different models of CMP 30HX MINER series cards with the MSI CMP 30HX MINER XS and CMP 30HX MINER, with both of the cards having the same TU116-100 GPU and 6GB of GDDR6 memory at 14Gbps on a 192-bit memory bus and even dual fans.

You can see that both of these cards look like other GeForce RTX series graphics cards in MSI's stack, with the MINER XS the mining version of the VENTUS XS while the MINER is identical to the ARMOR series from MSI.

The world of hardware mods on graphics cards hasn't been anywhere near as good as it was a couple of decades ago, but it still happens -- with news of modder VIK-on breaking down a physically dead and no-longer-booting GeForce RTX 2080 Ti and seeing that it supports up to 22GB of GDDR6 memory (up from its stock 11GB GDDR6 memory).

VIK-on purchased a dead GeForce RTX 2080 Ti graphics card that wasn't booting up anymore, with the modder noting that the graphics card was taken to 3 other workshops who didn't fix it but damaged it further. There were missing components from people who obviously didn't know what they were doing but VIK-on worked his magic.

The GPU was resoldered and each of the memory modules were replaced, and the PCB traces that were cut were fixed. VIK-on took his time adding in all of the missing pieces and components from the damaged GeForce RTX 2080 Ti graphics card and then started to verify if a simple strap mod (tricks the GPU BIOS into supporting higher memory capacities) would work.

We had heard that "Lite Hash Rate" (LHR) cards were coming with gimped Ethereum mining performance, but now it looks like they're going to be called "Lower Hash Rate" and GALAX has spilled the beans.

The new GALAX GeForce RTX 3080 LHR and GeForce RTX 3070 LHR are exactly what you think: GeForce RTX 3080 and GeForce RTX 3070 graphics cards with Lower Hash Rate performance: gimped for Ethereum miners. GALAX's new GeForce RTX 3080 LHR packs 10GB of GDDR6 at 19Gbps on a 320-bit memory bus, while the GALAX GeForce RTX 3070 rocks 8GB of GDDR6 at 14Gbps on a 256-bit memory bus.

You should be seeing around 43MH/s of Ethereum mining power out of the RTX 3080 LHR while we can expect around 25MH/s out of the RTX 3070 LHR.

AMD is currently preparing its new Radeon PRO W6800 workstation graphics card, packing its Navi 21 GPU and a huge 32GB of GDDR6 memory.

This is the first Navi 21-based graphics card with 32GB of memory, with the unreleased GPU found on UserBenchmark and noticed as the "AMD Radeon Pro W6800 32GB (1002 73A3, 1002 0E1E)". It might not seem like much, but this is seemingly the same card we saw just recently but with 16GB of GDDR6 at the time.

The new AMD Radeon PRO W6800 workstation graphics card packs the Navi 21 GLXL GPU which in the UserBenchmark scores sits somewhere between NVIDIA's GeForce RTX 2080 SUPER and GeForce RTX 3070 graphic cards.

AMD is busy handling a million-and-one things between its new Zen CPU architectures, RDNA GPU architectures, semi-custom designs for Microsoft, Sony, and other companies -- but the RDNA 3 architecture is the diamond in all that silicon rough for me.

The new RDNA 3 architecture could be revealed as early as December this year according to Tom from Moore's Law is Dead, who hasn't been right on everything but has definitely been over-the-target on many facts surrounding CPUs and GPUs over the last couple of years.

He started by saying that RDNA 3 "could slip 1 or 2 quarters" but it "looked like they might try to do some reveal at the very end of December, I think it's fair to say that RDNA 3 should launch at the earliest early 2022 and you shouldn't be surprised if it actually slips to mid-to-late 2022 but I certainly hope it doesn't".

PowerColor revealed its Radeon RX 6700 XT HELLHOUND graphics card in March 2021, but now the company has unleashed the new Radeon RX 6700 XT HELLHOUND Spectral White. Check it out:

The new Radeon RX 6700 XT HELLHOUND Spectral White graphics card is pretty self-explanatory, with PowerColor making the whitest graphics card ever. The custom PCB is white which looks fantastic just from the preview pictures, but the fans and entire cooling shroud are also white.

This isn't just some "hey guys, we've made a white-themed graphic card" but more like "let's make the whitest graphics card that has ever been made".

GIGABYTE is working away on their biggest and most badass custom Big Navi graphics card yet, with the AORUS Radeon RX 6900 XT Xtreme WaterForce on the way. Check it out:

We don't know the exact details about the AORUS Radeon RX 6900 XT Xtreme WaterForce, but we do know it'll rock a pre-binned Navi 21 XTXH GPU like the other super-premium Radeon RX 6900 XT graphics cards. GIGABYTE will be using a full custom PCB with 3 x 8-pin PCIe power connectors, and a pre-installed water block.

You will need a pre-existing liquid-cooled setup in your gaming PC if you want to use the AORUS Radeon RX 6900 XT Xtreme WaterForce, with the card looking quite dapper with its custom styling from GIGABYTE.

Intel is still hard at work on its upcoming Xe-HPG based GPU with a new tease of the DG2-128EU GPU, with a proposed draft for a DG2-based laptop board design leaked by Igor's LAB. Check it out:

The board diagram is pretty tasty with a tease of the Intel DG2-128EU (BGA1379 socket) as well as an Intel Tiger Lake-H CPU with 8 cores, which drops in a few days' time. This particular board packs 6 memory modules with 2 attached to the GPU, we should see 4GB of GDDR6 used here while the other 4 memory modules could be soldered system memory.

We saw a leaked board diagram last year for the higher-end Intel DG2-384EU version, with the DG2-384EU model having its 6 memory modules around the GPU versus the lower-end DG2-128EU having just 2 memory modules attached to its GPU. You can see that below:

Intel is getting awfully close to revaling its first gaming-focused GPU ever -- let's forget about the Intel i740 as it wasn't technically a GPU -- with Intel DG2 being "right around the corner" says Intel Senior Game Developer Relations Engineer Pete Brubaker.

Brubaker said it pretty plainly in his tweet, which was linking to a job posting by Intel which is looking for a new Senior Game Developer Relations Engineer. In the tweet, Brubaker said: "Come work with us at Intel! DG2 is right around the corner, it's about to get exciting".

This marks an exciting turn for Intel and its upcoming Xe-HPG gaming GPU, with Intel actively working with game developers on getting their games optimized for Intel's future graphic cards. Intel DG2 is a big step up from the DG1 which was effectively just a developer kit.

I just finished writing an article about AMD and its RDNA 3 GPUs taking the performance crown from NVIDIA in 2022, with the star of that show being the purported Radeon RX 7900 XT graphics card.

AMD's new Radeon RX 7900 XT graphics card is getting some more solid rumors from Tom at Moore's Law is Dead, who says in his new video that the Radeon RX 7900 XT "should at a minimum be 40% better than the RX 6900 XT" and that "Ray Tracing & Geometry Performance should get a decent uplift per DCU".

The new RDNA 3-powered Radeon RX 7900 XT is where AMD could actually take the performance crown from whatever NVIDIA has at the time from Ampere + Samsung to battle the RDNA 3 + TSMC GPU from AMD. NVIDIA will at most have an Ampere refresh, but the Samsung 8nm node is holding them back -- AMD however, will be sliding into the next-gen 5nm node from TSMC on top of the next-gen RDNA 3 architecture.

AMD's next-gen RDNA 3 architecture has been in the headlines quite a lot this past week, with rumors flying hot and fast -- but now we have some solidification on those hot AF rumors.

Our mate Tom from Moore's Law is Dead has published a new video that says that his sources have said the current rumors of RDNA 3 being 3x faster than AMD's current RDNA 2-based flagship Radeon RX 6900 XT are "utter bullshit" and that his sources told him in "harsh terms" that the performance numbers are "horse shit".

We're being told to expect at least a 40% performance uplift from RDNA 3 over RDNA 2, and that it should be 3x faster -- than RDNA, not RDNA 2 -- something AMD itself has said for a number of years now.

We aren't far away from the launch of NVIDIA's new GeForce RTX 3080 Ti graphics card, with another custom card turning up: the new GIGABYTE GeForce RTX 3080 Ti Gaming OC.

GIGABYTE's new GeForce RTX 3080 Ti Gaming OC graphics card was leaked by our friend Tom @ Moore's Law is Dead, who posted the picture he received from someone who had the GIGABYTE RTX 3080 Ti Gaming OC in-hand. The retail box is virtually identical to the non-Ti model, except for where it has 12GB of GDDR6X versus the 10GB of GDDR6X on the GeForce RTX 3080.

It was only just back 4-5 weeks ago now that GIGABYTE had 12 -- yes, 12 new custom GeForce RTX 3080 Ti graphics cards in the pipeline sliding through the EEC (Eurasian Economic Commission). This new GIGABYTE RTX 3080 Ti Gaming OC is just one of them, but now spotted in-person ahead of launch.

We've already seen crazy pricing on NVIDIA's upcoming GeForce RTX 3080 Ti graphics cards that is landing between $1300 through to a huge $2250 -- but that doesn't hold up against the stupidly high $3500 pricing of the RTX 3080 Ti in the UAE.

MSI's custom GeForce RTX 3080 Ti SUPRIM X graphics card is being sold at a UAE-based retailer for $3500, which I cannot even begin to comprehend. $1300 was bad enough, but $2250 was just insane... and now we're talking about $3500. That should be the entire cost of a kick ass new gaming PC **and** a GeForce RTX 3080 Ti.

Reddit user FaisalKhatib saw the new MSI GeForce RTX 3080 Ti SUPRIM X graphics card at a retailer in the UAE, priced at $3500 and available to buy right now a month out from it being released, and available even before NVIDIA itself has confirmed the GeForce RTX 3080 Ti in the first place.

We heard some rumbles about AMD's upcoming Navi 33 GPU just over 24 hours ago now, and now it's time for news on the bigger, next-gen Big Navi GPU -- Navi 31 which is reportedly up to 3x faster than Navi 21 that powers the Radeon RX 6800 XT and Radeon RX 6900 XT graphics cards.

KittyYYuko was behind the recent Navi 33 rumors and is now fueling the new Navi 31 rumors, where in reply to Paul Eccleston from RedGamingTech on Twitter who said Navi 31 and Navi 32 would have MCM designs (multi-chip modules) with 2 chiplets for compute (with 80 compute units each) and 1 x I/O die.

He said that Navi 33 was a monolithic GPU with 80 compute units (the way GPUs have been made for decades) while performance-wise, Navi 31 is "about 2.5x faster than Navi 21". Navi 31 will reportedly have similar ray tracing power as NVIDIA, with Eccleston saying that it would be "a tiny bit slower like for like" while machine learning "still handled as lower precision ops".

Can you believe we're barely a month out from NVIDIA revealing new graphics cards already? Computex 2021 is right around the corner -- if you can believe that we're nearly half way through the year already -- meaning it's new hardware time (again).

The last that we heard, NVIDIA had pulled the reveal of its GeForce RTX 3080 Ti graphics card up to May 18, from a May 25 announcement, but now VideoCardz says it can confirm NVIDIA is planning to announce the new GPU on May 31 alongside the GeForce RTX 3070 Ti.

NVIDIA is reportedly setting product review NDAs separately for the two cards, with the GeForce RTX 3070 Ti review NDA lifting on June 2 while the higher-end GeForce RTX 3080 Ti review NDA lifts on June 9. As for when the new GPUs will be released... well we don't know that exactly, but it shouldn't be long after the reviews -- will you be able to find one, that's an entirely different question altogether.

Call of Duty: Warzone was recently updated and included with it NVIDIA's magic DLSS (Deep Learning Super Sampling) technology which provides a gigantic injection of performance for GeForce RTX gamers. COD: Warzone actually hits 8K 60FPS on the GeForce RTX 3090 with DLSS on Ultra Performance mode.

The new Warzone update with DLSS enabled and turned onto "Performance Mode" sees the GeForce RTX 3090 going from 100FPS average at 4K to a huge 151FPS average... around 50% free performance turning DLSS on. But what about 8K? That's where the fun really begins, and something I just personally did a few days ago on my Dell UP3218K monitor, the Core i9-10900K processor and MSI GeForce RTX 3090 SUPRIM X graphics card.

Call of Duty: Warzone at 8K simply requires the GeForce RTX 3090, it's unplayable at anything less than 60FPS or so at 7680 x 4320 as its driving those 33 million pixels in front of your eyeballs and melting your graphics card in the process.

NVIDIA is ramping into the announcement and release of its new GeForce RTX 3080 Ti which will pack 12GB of GDDR6X memory, with some online retailers teasing custom GIGABYTE and MSI models early.

The new listings for MSI's upcoming custom GeForce RTX 3080 Ti Ventus 3X OC have appeared on Polish retailer X-Kom, as well as Australian and New Zealand outlets listing GIGABYTE and MSI custom GeForce RTX 3080 Ti graphics cards.

The pricing of the custom GeForce RTX 3080 Ti graphics cards from these early listings has it pegged at between $1300 or so through to an insane $2250. These are most likely placeholder prices so I wouldn't be putting away the Benjamins or Dogecoin just yet -- and yeah I'm serious about buying your GeForce RTX 3080 Ti with Dogecoin because Newegg will do that for you.

We are still quite far away from the release of AMD's next-gen RDNA 3 architecture, but early rumbles of the Navi 33 are now making their way out of the cracks of the internet.

KittyYyuko posted on Twitter that Navi 33 (or something close to that) is like the Navi 21 (a flagship GPU) but with "Next Gen IP Core" (which would be RDNA 3). The new Navi 33 GPU is reportedly based on a new GPU architecture, and has 80 compute units (CUs) and 5120 cores.

With this much raw power inside of the Navi 33 we would expect the Radeon RX 7600 XT, or whatever it is named, should have similar power to the current near flagship Radeon RX 6800 XT and Radeon RX 6900 XT graphics cards. That would be a big deal for AMD, and great for gamers -- if there's enough of these new Navi 3X cards to go around whenever they make it to market.

NVIDIA is getting into the spirit of Overwatch... at least in China, with NVIDIA GeForce China approving a slick custom GeForce RTX 3080 Founders Edition graphics card inspired by Overwatch.

This isn't the first time NVIDIA has customized a GeForce RTX Founders Edition graphics card, with the GeForce RTX 2080 Ti Cyberpunk 2077 graphics card looking just as dapper -- but now we have the GeForce RTX 3080 Founders Edition themed with Overwatch and it looks pretty damn good, if you could find it that is.

NVIDIA has painted the card in gold and black, with Tracer's face on the back of the card and an Overwatch logo sticker on the fan. I doubt there will be many of these Overwatch-themed GeForce RTX 3080 graphics card made or available, but if you can get it you're probably desperate enough at this point you don't care whatever the hell artwork is on your graphics card.