Graphics Cards - Page 116

Just as I had finished off reviewing the big wave of NVIDIA GeForce RTX 4090 graphics cards, with some overclocking, ray tracing, DLSS 3, and 8K benchmarks all on the way... fresh rumors of AMD's next-gen RDNA 3 GPU architecture are here... and boy, they're juicy.

AMD's next-gen RDNA 3 GPU architecture will reportedly be led by the flagship Navi 31 GPU, with leaker "Greymon55" tweeting out that some AIBs have already got their Navi 31 chips and are "testing them" right now. That should make sense, given that AMD is launching its next-gen Radeon RX 7000 series "RDNA 3" GPUs on November 3.

But in some follow-up tweets, Greymon55 was asked if he knew the final rasterization and RT performance on Navi 31, to which Greymon55 replied "2X" but then he was asked again about RT, to which he responded very simply with "more". So we should expect twice the performance of the Radeon RX 6900 XT in raster, and over twice the performance in RT-powered games, lining up with rumors and expectations over the last 12-18 months.

MSI's new custom GeForce RTX 4090 SUPRIM LIQUID X graphics card is one of the best RTX 4090 cards that money (can hopefully soon) buy, with some fake news hitting the internet about the card today.

The news started out on Reddit where a user saw the price of MSI's new GeForce RTX 4090 SUPRIM LIQUID X was $1934.99, a huge $185 premium over its $1750 pricing... but that is completely wrong. The price seems to have spiked up to $1935, but anyone who purchased it at this price is being refunded the price difference.

MSI explained: "MSI is aware of an article from the forum regarding the raised price of RTX 4090 SUPRIM LIQUID X. As a matter of fact, it was our negligence to mismatch the price. The price of RTX 4090 SUPRIM LIQUID X will remain at 1749.99 USD, which is exactly the same price as we launched the product".

We did hear over a week ago now that NVIDIA's beasty TITAN Ada was cancelled after it needed to be mounted TO THE ACTUAL motherboard because it was so big, required dual 16-pin power connectors, and melted PSUs.

But now we have another leaker confirming that NVIDIA has nixed the card, with leaker "kopite7kimi" tweeting out simply that "we won't see TITAN of Ada Lovelace". The previous confirmation that NVIDIA cancelled its TITAN Ada came from leaker Moore's Law is Dead over a week ago now, with his sources saying the card used 600-700W of power, and was tripping breakers, melting power supplies, and "sometimes melting itself lol".

NVIDIA was expected to use the full-fat AD102 GPU with 18176 CUDA cores, and what I would've loved to have seen: 48GB of GDDR6X memory. NVIDIA's current Ada Lovelace flagship GeForce RTX 4090 graphics card ships with 16384 CUDA cores and 24GB of GDDR6X memory.

Intel's new Arc A770 and Arc A750 graphics cards are out in the wild, reviewed by some of the cherry-picked outlets and influencers online, but most people are noticing many issues... one of them being high idle power draw.

But don't worry, Intel has a solution for you since you're effectively a beta tester for Arc GPU solutions, with at least 41W of idle power consumption compared to other GPUs that consume less than 10W of idle power, something has to be done. Intel has a solution, but they want you to work for it.

Intel now has a list of recommendations for "optimal power consumption" for its Arc A7 series GPUs, where you'll need to change some settings in the BIOS for your motherboard, as well as some Windows Power settings, which I find truly ridiculous. Didn't need to do any of this for any AMD Radeon or NVIDIA GeForce graphics card I've owned in decades, but hey.

NVIDIA's new GeForce RTX 4090 graphics card is here and its nifty new 16-pin PCIe power connector (12VHPWR) can be annoying... it sits very high in your case, and can hit the side window... until Cablemod stepped up into the ring.

Cablemod has just announced its new 12VHPWR Right Angle Adapter, which plugs directly into the 12VHPWR port and turns it into a 90-degree angle, better fitting inside of your PC. The new 12VHPWR 90-degree cable requires just an additional 23.2mm of space to get it into your PC.

You can't buy Cablemod's new 12VHPWR Right Angle Adapter just yet, with the company having the 90-degree cable on pre-sale starting October 31... a great present to yourself, or your RTX 4090 that is, for Halloween. It doesn't really help if you're already using a GeForce RTX 4090, but if you haven't pulled the trigger yet, the new 90-degree right-angle cable from Cablemod is going to be your second purchase.

VESA has just announced the updated DisplayPort 2.1 specification today, also announcing that any DisplayPort 2.0-ready product is now a DisplayPort 2.1-certified product.... and those products do NOT include NVIDIA's new GeForce RTX 40 series GPUs (unfortunately, and very silly I might add).

The updated DisplayPort 2.1 specification means that any and all DisplayPort 2.0-certified products (this includes UHBR aka Ultra-High Bit Rate products) like GPUs, monitor scalar chips, docking station chips, DP40 + DP80 cables, and anything else that's DP2.0-ready is now DP2.1-ready.

AMD has its new Ryzen 6000 series with DisplayPort 2.0 compatibility, as well as Intel's did-they-even-release-it Arc A770 and Arc A750 graphics cards have DP2.0 ports, which have now all been silently upgraded to DP2.1 overnight. NVIDIA's new GeForce RTX 4090 does NOT have DisplayPort 2.0, so there's no upgrade to DisplayPort 2.1, which is something that I've covered extensively before, during, and after the RTX 4090 launch.

Acer's new Predator BiFrost Intel Arc A770 OC graphics card is for sale in Taiwan, with a listing on Acer's official store in Taiwan having the card in stock for NT$12,900 which works out to a shave over $400 USD.

If you can't buy Intel's slick Arc A770 Limited Edition (which there were so few made) then you might want to try your chances for a custom AIB model, with the Acer Predator BiFrost Intel Arc A770 OC graphics card featuring factory overclocking, as well as 16GB of GDDR6 memory.

16GB of GDDR6 memory is important because remember Intel offers their new Arc A770 graphics card with either 16GB GDDR6 or 8GB of GDDR6, depending on the model. Intel's own in-house Arc A770 Limited Edition graphics card features 16GB of GDDR6 memory, while the custom GUNNIR Intel Arc A770 Flux graphics card has just 8GB of GDDR6 memory.

GUNNIR's new custom Intel Arc A770 Flux graphics card has been picked up by Expreview, with the beautiful-looking card put through its paces. Check it out:

The new custom GUNNIR Intel Arc A770 Flux graphics card features a fully custom design, with an 8+4 VRM phase design, dual 8-pin PCIe power connectors, a triple-fan cooling system, and a dual-slot design. The design of the card looks badass, with GUNNIR shifting away from traditional-looking graphics cards with its custom Intel Arc A770 Flux.

Inside, GUNNIR has 8GB of GDDR6 memory (remember, Intel's new Arc A770 comes in 16GB and 8GB variants) with the GPU factory overclocked from 2100MHz up to 2400MHz, the team at Expreview was able to push the GUNNIR Intel Arc A770 Flux graphics card and its GPU up to a much higher 2.65GHz with overclocking.

Leadtek has just introduced its new custom GeForce RTX 4090 Winfast Hurricane graphics card, the first custom GeForce RTX graphics card since the introduction of the Leadtek GeForce RTX 4090 Winfast Hurricane graphics card in 2020.

You might not remember the brand Leadtek, but they're one of the oldest names in the graphics card business. I still remember the day I purchased the Leadtek GeForce Ti 4400... man, those were the days.

The new Leadtek GeForce RTX 4090 Winfast Hurricane graphics card, which is a new 3.5-slot beast that is thicker than its predecessor (obviously) and is 3cm longer than its predecessor. Leadtek uses a triple-fan cooling solution with an X-shaped shroud with 3 x 100mm fans keeping Ada nice and cool.

NVIDIA has officially "unlaunched" its GeForce RTX 4080 12GB graphics cards, but what about all of the AIB partners that had announced their custom GeForce RTX 4080 12GB cards... and started making retail packaging for them? Well, some of those costs will be recouped and covered by Uncle Jensen.

We can't just see these boxes have stickers slapped over them, replacing the "GeForce RTX 4080 12GB" to whatever NVIDIA calls it: GeForce RTX 4070 12GB or GeForce RTX 4070 Ti 12GB. It will be tacky, and not look good for the biggest GPU launch in NVIDIA's history. Nope. NVIDIA is working with AIB partners that had already made retail boxes for the now-unlaunched GeForce RTX 4080 12GB.

NVIDIA is reimbursing some AIB partners for the retail packaging on the GeForce RTX 4080 12GB, where the boxes will be collected, and destroyed. Disappointing. We don't know if NVIDIA is repaying the AIB partners for the losses they'll make on the retail packaging, or if it'll include everything (the people who actually have to destroy the packaging, which will take time).

I don't think NVIDIA has ever done this before, but the company has just announced that it is launching its recently announced GeForce RTX 4080 12GB graphics card.

NVIDIA founder and CEO Jensen Huang unveiled the new Ada Lovelace-powered GeForce RTX 4090, GeForce RTX 4080 16GB, and GeForce RTX 4080 12GB graphics cards during GTC 2022 last month and the "GeForce Beyond" event. But now, now that GeForce RTX 4080 12GB has been "unlaunched" says NVIDIA.

In a new post on their website, the company explains that the GeForce RTX 4080 12GB is a "fantastic graphics card" but it's "not named right". NVIDIA must have been (forcefully) listening to its fans, adding that "having two GPUs with the 4080 designation is confusing". None of their internal market research, hundreds of staff that were probably like "WTF, Jensen", could point them towards "this is a bad idea (launching the RTX 4080 12GB, which is really the RTX 4070/4070 Ti).

51RSIC is a company that you've never heard of, but they're making their name known today with the introduction of the new GeForce RTX 4090 24GB Gamer graphics card, which has a built-in spirit level.

No, you won't be using this GeForce RTX 4090 to rid your house of spirits, but rather the Chinese GPU maker has crafted the 51RISC GeForce RTX 4090 24GB Gamer graphics card with a thinner cooler than the rest of the chunky bois that have been released so far.

Virtually all high-end graphics cards made in the last couple of years, but especially the chunky GeForce RTX 3090, GeForce RTX 3090 Ti, and GeForce RTX 4090 graphics cards are triple-slot and thicker. The new 51RISC GeForce RTX 4090 24GB Gamer graphic card is a proper 3-slot card, with the company not providing users with a GPU support bracket, the built-in spirit level is what you've got. Niche, but cool.

We are on the eve of NVIDIA's next-gen GeForce RTX 4090 graphics card being unleashed, but that hasn't stopped the weekend leak game from continuing to tease GPUs that we, unfortunately, won't see this year.

NVIDIA's beasty new GeForce RTX 4090 is a powerhouse you'll find out about shortly, but earlier this year we heard crazy rumors of the full Ada Lovelace GPU and its next-gen power with an 800W or even 900W+ insane card that would be known as the new TITAN... and hopefully, TITAN Ada.

Well, Moore's Law is Dead pumped another video out over the weekend with one of his trusted sources giving us an update on TITAN Ada... and yeah... what an update. Tom at MLID explains: "this comes from somebody that over about 6 months they have proven to me they are more than legit, based on the pictures and information they've fed me before things have been made official. This person tells me that as of now, TITAN Ada is cancelled. It's just too much for now".

Samsung has just hosted its recent Samsung Tech Day 2022 event, where the company updated the world on its work on next-gen GDDR7 memory.

The company hosted its Samsung Tech Day 2022 event in-person for the first time in 3 years, at the Signia by Hilton San Jose, California, where the South Korean giant talked about everything it is working on: 5G models, 200-megapixel ISOCELL HP3 camera sensors, DRAM and NAND flash memory, next-gen GDDR7 memory, and so much more.

Samsung talked about its next-gen DRAM solutions during the Samsung Tech Day 2022 event, where it has 32Gb DDR5 DRAM, and 8.5Gbps LPDDR5X memory, but this is the doozy: totally next-gen 36Gbps GDDR7 DRAM for the data center, HPC, mobile, gaming and automotive market segments.

Intel has officially launched its new Arc A770 and Arc A750 graphics cards, with Arc A770 Limited Edition and Arc A750 Limited Edition graphics cards being reviewed by a few outlets and YouTubers today.

Well, perfect timing would have it that leaker Moore's Law is Dead is back with some fresh Arc launch leaks, where his sources have told him it truly is a dire situation out there in the odyssey that is Arc. MLID's first source said that Intel has been ghosting their GPU purchaser, but fully responding to their CPU guy.

The second source said that they were simply never getting Intel Arc A770 and Arc A750 graphics cards, while the third source said that they simply have "no info" on Arc A770 shipments and that they don't think that they'll get any Arc A-series desktop GPUs whatsoever.

Now that Intel's new Arc A770 and Arc A750 graphics cards have been sent to (cherry-picked) reviewers, UL Benchmarks have unleashed a new 3DMark feature test.

The new 3DMark feature test will benchmark Intel's new XeSS, its new AI-boosted upscaling technology that works similar to AMD FSR (FidelityFX Super Resolution) and NVIDIA DLSS (Deep Learning Super Sampling). The company provides a nifty 3DMark Intel XeSS frame inspector tool, something that lets you check out the comparison between image quality in an awesome interactive side-by-side comparison between native rendering, and Intel XeSS AI-boosted rendering.

3DMark's new XeSS feature test is based on the scene from the 3DMark Port Royal benchmark, which runs two times to show the effect of XeSS on both performance, and image quality. The first run of the XeSS feature test will measure your baseline performance, where it will render the scene with temporal anti-aliasing (TAA) at the native display output resolution.

NVIDIA announced its next-gen GeForce RTX 4090 graphics card not too long ago now, with the company saying that there would be a "starting price" of $1599... and ZOTAC is sticking to that.

ZOTAC's upcoming fleet of custom GeForce RTX 4090 graphics cards will be priced from $1599 through to $1699, whereas the new ZOTAC GAMING GeForce RTX 4090 Trinity graphics card will cost $1599, the overclocked model -- the ZOTAC GAMING GeForce RTX 4090 Trinity OC -- will cost $1649.

The new flagship ZOTAC GAMING GeForce RTX 4090 AMP Extreme AIRO will cost $1699 -- a full Benjamin more than the RTX 4090 Trinity non-OC card -- with GPU boost clocks of up to 2580MHz (more with OC) compared to 2535MHz and 2520MHz on the RTX 4090 Trinity OC and RTX 4090 Trinity, respectively.

Intel's new Arc A770 and Arc A750 graphics cards have now reached cherry-picked reviewers, but there's one custom Arc A770 graphics card that has my attention: Acer's interesting new Predator Arc A770 BiFrost graphics card.

The new Acer Predator Arc A770 BiFrost graphics card has been picked up, and reviewed by Israeli tech site Plonter.info, with a full review you can check out here. Plonter has a few awesome things here: they're the first media outlet to review the Acer Predator Arc A770 BiFrost graphics card, as well as the first outlet to review a custom Arc graphics card. Congrats, guys!

Acer has built a very interesting card here, with a full custom design with a dual-fan cooler (turbine extract + axial throughput) while it packs the higher chunk of 16GB GDDR6 memory, versus 8GB GDDR6 memory that will come on some custom Arc A770 graphics cards. Intel's own in-house Arc A770 Limited Edition graphics card has 16GB GDDR6, but AIB partners can choose between 8GB and 16GB GDDR6 on their custom Arc A770 designs.

Intel reached out to me a few weeks ago now asking for my address so they could send me Arc A770 and Arc A750 graphics cards, but after covering Arc negatively -- not just opinion pieces, but sourcing leaks and rumors -- got me put on the back burner. Don't worry, Intel promises that I might get a card, one day.

Well, after the tiniest graphics card sampling round in my career, and now Intel's new Arc A770 and Arc A750 graphics cards are in the hands of... well... people who obviously don't push Intel back too much. I think 95% or more of the reviewers who were cherry-picked, were even alive when Intel launched the i740 on AGP... but I digress.

Intel's new Arc Alchemist GPU architecture was initially unveiled close to FOUR YEARS ago back in January 2019, when Intel posted an inspirational video on Twitter that featured ex-RTG boss and now AXG boss Raja Koduri, as well as former Intel CEO Bob Swan. In the video, Intel was recruiting staff, while Raja was saying that Intel was effectively starting from zero in discrete graphics.

Intel has confirmed that Arc A770 graphics card reviews will go live tomorrow for the cherry-picked reviewers -- we won't be getting one, but I'm sure that's no surprise given that I've been reporting everything about Arc, positive or negative -- but reviews will go live tomorrow.

Leading into the launch of Arc A-series GPUs on the desktop, Intel AXG boss Raja Koduri sat down with Ryan Shrout and Tom Peterson to talk all things Arc. Koduri talked about building the Arc Alchemist GPU architecture and bringing it to market -- something the company teased nearly FOUR YEARS ago if you can believe it.

Koduri also talked about the issues of working on the Arc GPU architecture during the pandemic, with teams working from home, some in the labs, some debugging, some everywhere. The ex-RTG boss and father of Vega discussed optimizing software for the Arc GPUs, especially APIs like DirectX 11 -- an API that Arc GPUs don't perform well with whatsoever.