Graphics Cards - Page 103

ASUS is rolling out a GPU trade-in program in the UK where you can hand in your old NVIDIA or AMD graphics card for a discount towards a brand-new GeForce RTX 4070 Ti or RTX 4080 GPU. The promotion is live now, will end on March 17, 2023, and isn't something we've seen before. At least, I haven't.

Typically, second-hand GPUs are on-sold to friends of a friend or a workmate. Or put up on Facebook Marketplace or another auction platform. And with that, even though the amounts and cash-back offered by ASUS might not be up there with second-hand pricing, it covers pretty much every major GPU release for the past several years - and is an excellent little all-around promotion. So yeah, we'd love this to make its way to the US or Australia.

ASUS has listed the participating retailers where the cash-back offer is available for the following GPUs.

NVIDIA's AD106 chip from its new Ada Lovelace GPU range is rumored to power the GeForce RTX 4060 graphics card, which has yet to be formally revealed. According to a new listing and 3DMark benchmark results (via Chiphell), the AD106 GPU from NVIDIA boasts 4608 CUDA Cores, a 2535 MHz boost clock, a 128-bit memory bus, and a power rating of only 180W.

The benchmark results compare the chip to GA104 (the GeForce RTX 3070 Ti) and TU104 (the older GeForce RTX 20 Series). And based on the results, it looks either faster or slower than the GeForce RTX 3070 Ti, depending on the test. DirectX 12 and raytracing look to be the clear winner. Still, the big takeaway is that the generational jump means that the level of performance for the NVIDIA GeForce RTX 4060's AD106 will be RTX 3070 Ti at around 27% less power.

Harukaze5719 on Twitter has posted the results in a handy chart you can see here, where apparent wins in 1080p and 1440p tests like FireStrike and TimeSpy showcase that NVIDIA will be positioning AD106 GPU products as mainstream. As x060 class GPUs from NVIDIA are always the most popular, it will be interesting to see what price point the RTX 4060 eventually hits.

Back in the day, graphics cards from 3dfx Interactive - namely those with the Voodoo name - were what you craved to play cutting-edge games like Quake and Unreal. And although the rise of GeForce and Radeon slowly overtook Voodoo as the PC gaming GPUs of choice, those that have a fondness for classic late-'90s PC games and tech will no doubt have fond memories of 3dfx hardware.

This makes this eBay listing for a Voodoo 5 6000 interesting because it was the last 'Voodoo' card from 3dfx and a product that never actually made it to market. One of the reasons it never did come out was its use of multiple chips and four 166 MHz VSA-100 processors, each with 32 MB of 166 MHz SDRAM. The Voodoo 5 6000 was going to be the first 128 MB graphics card, and according to reports, it went through various prototype stages with only test cards produced. This is why this for-real Voodoo 5 6000 (Rev 3700A) on eBay currently has a bid of USD 9,500.

As per the seller, gtastuntcrew302.

As per our report last week, it has been over two months since AMD released a driver update covering RDNA 2-based Radeon RX 6000 GPUs and older. The past four driver updates have been exclusive to the new RDNA 3-based Radeon RX 7900 XT and Radeon RX 7900 XTX GPUs, which launched in December 2022. Even though it makes sense for AMD to focus on the new product launch and fix any issues that might have been discovered in the following weeks, it's still strange.

Namely, because day-one drivers are critical for new game releases, ensuring performance and compatibility is where it needs to be. On this front, high-profile PC games like Forspoken (an AMD-sponsored title) and the impressive Dead Space remake have launched without non-RDNA 3 Radeon driver support.

Naturally, the community has been wondering what's happening - especially when unified drivers are supposed to be the norm. On the NVIDIA front, even with the arrival of the GeForce RTX 40 Series, the current 528.24 WHQL GeForce Game Ready Driver still supports GeForce GTX 10 Series GPUs.

The Steam Hardware Survey is a pretty good gauge to see what's popular in the PC gaming space; for example, most Steam users (63.46%) are running Windows 10 64-bit and are gaming in 1080p.

The most popular GPU is the NVIDIA GeForce GTX 1650 (6.04%), which is closely followed by the GeForce GTX 1060, GeForce RTX 3060, GeForce RTX 2060, and GeForce GTX 1050 Ti. These GPUs represent around 30% of all Steam users, and with that, most PC gamers use the GPUs to play games.

It gets interesting with the arrival of new GPUs and how the numbers change from month to month to reflect growth and change.

The NVIDIA RTX 6000 Ada workstation GPU isn't designed for gaming, but that hasn't stopped someone from putting its traditional 2-slot blower-fan cooling design to the test using the 3DMark Time Spy performance benchmark.

The 30518 graphics score represents a 72% boost over its workstation predecessor, the RTX A6000 Ampere-based GPU. The card is slower than the GeForce RTX 4090, but its intended use isn't gaming. According to the source, it reportedly hit closer to 36000 before a Windows update, so its performance could be much closer to the RTX 4090.

Specs-wise, the NVIDIA RTX 6000 Ada's 18,176 CUDA cores is an 11% increase over the RTX 4090's 16,384 CUDA Core count, and it also features more memory in the form of 48GB GDDR6 over a 384-bit interface. The card is also rated at 300W, which makes it an efficient powerhouse for workstation tasks.

As per a new report at The Wall Street Journal, the state-run China Academy of Engineering Physics (CAEP) - reportedly the country's top nuclear weapons research institute - has managed to get a hold of Intel and NVIDIA hardware. Which we assume have been used to do all of that atomic research.

Hardware from both suppliers has been on an export blocklist since 1997, but even so, the chips were obtained from resellers in China. The CAEP is specifically on the blocklist because of its work in nuclear science, with the report claiming that the study of "computational fluid dynamics" is code for nuclear explosions. To circumvent the restrictions, the CAEP has set up multiple entities and over a dozen aliases, and many of these have been added to the blocklist as recently as June 2020.

So then, how did they find out about the chips? A review of research papers published by CAEP over the past decade found references to hardware from Intel and NVIDIA - like an Intel Core i7-7800X processor and an NVIDIA GeForce GTX 1080 Ti graphics card. Multiple research papers feature sections about "maintaining nuclear stockpiles," which is a scary thought in its own right. The report notes that the chips used cover everything from 7nm to 14nm process nodes, though there are no records of Intel and NVIDIA chips launched within the last two years being used.

And not just DirectX 9 gaming, but modern titles also benefit from performance increases. And with a price drop for the Intel Arc A750 to USD 249 - Intel is positioning the GPU as a more affordable mainstream alternative to NVIDIA's most popular GPU, the GeForce RTX 3060. Those who buy an Arc GPU before July 31 will also be eligible for two free games; Nightingale and The Settlers: New Allies.

With the new price and impressive driver-based performance increases, Intel claims that the Intel Arc A750 delivers 52% more performance per dollar compared to the GeForce RTX 3060 across several DX9, DX11, and DX12 titles.

From Metro Exodus to popular competitive jams like Overwatch 2, Fortnite, and League of Legends, Intel is looking to deliver on the promise of Arc with this new driver update.

With AMD RDNA 3 "Navi 31" GPUs now out, specifically the Radeon RX 7900 XT and Radeon RX 7900 XTX, it looks like there could be room for a significant update for the tech via the implementation of 3D V-Cache. Stacking cache in this way has led to AMD's Ryzen 7 5800X3D and the new Zen 4-based Ryzen 7000 X3D CPUs being touted as ideal for gaming. With significant performance boosts found specifically in gaming-related tasks.

And with that, it's fair to expect that a Radeon RX 7000 GPU with 3D V-Cache could lead to similar results. Engineer Tom Wassick noticed the potential for an RDNA 3 upgrade after dissecting a Radeon RX 7900 XT, where he saw "keep-out zones" in the design that are similar to what's found on the Ryzen X3D CPUs.

"A linear array of 'spots' that look remarkably like the keep-out zones on X3D," Tom Wassick writes. "Could they be considering stacked MCD functionality?" With the groundbreaking multi-chip design in RDNA 3, it makes sense that AMD would also be looking into something like a 3D V-Cache to boost performance that isn't all about power and clock speeds.

The first GeForce RTX 4070 testing results (the non-Ti model) have surfaced via the GPU-Z validation database, covering a Colorful RTX 4070 iGame Ultra White OC. Indicating that partners are testing cards in their labs, the listing also confirms a few things related to the card's clock speeds.

Based on the information datamined by Matthew Smith on Twitter, we learn that the GeForce RTX 4070 has a Base Clock speed of 1920 MHz and a Boost Clock speed of 2475 MHz. No official word on the power rating, but previous reports indicate that the RTX 4070 will be an efficient 200W or so card.

Compared to the GeForce RTX 4070 Ti's 2610 MHz, it's a 135 MHz reduction in Boost Clock speed. Based on the same AD104 chip as the RTX 4070 Ti, the baseline RTX 4070 is rumored to feature 5888 CUDA Cores versus the 7680 in the Ti model - another reduction. How the difference in clock speeds and specs relate to performance remains to be seen, but it sounds like we won't have to wait too long to find out.

AMD driver updates for its Radeon GPUs have been limited to the new RDNA 3-based Radeon RX 7000 Series for two months, leaving RDNA 2 GPUs behind. Although listed as a December update, the last RDNA 2 driver for the Radeon RX 6000 Series (which includes the 6700 XT, 6800 XT, and 6900 XT) surfaced at the end of November 2022.

The driver, AMD Radeon Software 22.11.2, marks the last time a Radeon software update supported the RX 6000 Series - with the four subsequent driver releases being exclusive to the Radeon RX 7900 XT and Radeon RX 7900 XTX. As both cards made their debut in December 2022 and launched with some significant issues relating to power draw, it makes sense that that was the focus for AMD. That said, it doesn't make sense that the bulk of the Radeon GPUs out in the wild have been forgotten, especially when driver updates are crucial to improving day-one performance for many PC games.

With the release of high-profile games like Square Enix's Forspoken (AMD sponsored) and the Dead Space remake from EA - not getting timely Radeon software updates for these games is a little concerning. On the NVIDIA side, Game Ready Drivers supporting the company's entire GPU line-up have continued without pause, even with the arrival of three brand-new GeForce RTX 40 Series cards.

NVIDIA might be planning to release a GeForce RTX 4090 Ti with 24GB of GDDR6X and a 600W TGP in addition to a TITAN-class Ada Lovelace graphics card with a whopping 48GB of GDDR6X and a 800W TGP. This comes as NVIDIA leaker kopite7kimi releases specs for the currently planned GeForce RTX 4090 Ti.

He lists RTX 4090 Ti using PG136/139-SKU310 and featuring 18176 FP32 (CUDA Cores), 96M L2 Cache, 24G GDDR6X memory, and total power of 600W. The CUDA Core count here represents a 10.9% increase over the GeForce RTX 4090.

This sits alongside his previous leak of "the beast" using PG137-SKU0 featuring the same CUDA Core count as this new RTX 4090 Ti leak but with double the memory (48G GDDR6X) and a total power rating of 800W. With images we posted today showcasing a new four-slot cooler for an upcoming Founders Edition card from NVIDIA, we assumed that it was related to a single product.

Intel Arc GPUs launched last year as a new player in the discrete GPU market, with products that would compete directly with offerings from NVIDIA and AMD. As with GPUs, drivers play an essential role in overall performance, stability, and compatibility. Ahead of Intel's Arc debut, the company made it clear that DX9 (DirectX9 API) performance hasdissues.

Last December, Intel released a driver update for Arc that presented a massive bump to DX9 performance. For example, popular online shooter CS:GO saw an 80% increase in 1080p performance on the Intel Arc 770.

The word is that Intel is prepping another driver update for the Arc line-up, which is said to remove a bottleneck or issue that could lead to an across-the-board improvement. As per PC Games Hardware's translation via Videocardz.com, "Intel is currently preparing a major driver update intended to increase performance across the board one of the undoubtedly existing [bottlenecks] has been identified and eliminated."

GPU manufacturers use ECC (Eurasian Economic Commission) listings to register GPU model names for use at a later date, often indicating that a product launch is happening relatively soon. A new ECC filing by Chinese graphics card maker Shangke Group (MaxSun and Soyo) lists A GeForce RTX 4060 and GeForce RTX 4050 alongside the RTX 4070.

The list doesn't mention any upcoming "Ti" models but indicates the expected release cadence of the GeForce RTX 4060 to follow the launch of the RTX 4070. What's interesting is that both of these cards represent the mainstream offerings for the GeForce RTX 40 Series, so even with a price increase over the previous Ampere generation - we should expect them to launch with prices relatively close to the RTX 3050's USD 249 and the RTX 3060's USD 329.

Also, the fact that both have shown up in this listing points to board partners being made aware of the entire Ada line-up and that NVIDIA could release these cards sooner than expected. And with NVIDIA launching its full laptop/mobile RTX 40 Series line-up, we could also see an accelerated timeline for desktops.

Pictures of what we assume will be NVIDIA's GeForce RTX 4090 Ti graphics card have surfaced, showcasing a massive four-slot cooling design that features what looks to be a two-slot exhaust. This could be the reason for the vertical orientation of the I/O ports, which is an interesting design change. The images were posted by the aptly named MEGAsizeGPU.

The RTX 4090 Ti, or next-gen Titan based on Ada Lovelace architecture, is expected to become the flagship GPU in the RTX 40 Series line-up. The image lists 699-1G137 SKU as the card, confirming this. The overall design will make it the largest Founders Edition GPU to date - with a more considerable physical presence than the already chunky RTX 4090 and RTX 4080 GPUs. It also features a gold-colored cooler design, putting it in the Titan-class range for NVIDIA.

It could be a prototype or the real deal; it's hard to say. As per established NVIDIA leaker Kopite7kimi, the RTX 4090 Ti, or "the beast," is the PG137-SKU0 shown here. Granted that info comes from July 2022, where AD102-450-A1 is listed as the chip alongside 18176 FP32 (CUDA Cores), 48G 24Gbps GDDR6X memory, and an eyebrow-raising 800W power limit. And with that, it would also require dual 12VHPWR 16-pin connectors.

The GALAX GeForce RTX 4090 HOF (Hall of Fame) might be the most powerful version of the most powerful GPU on the planet. Rocking dual 16-pin '12VHPWR' power connectors, and a huge 666W power limit, it also features a robust 28+4 phase power design.

A 4.3-inch full-color LCD panel dubbed the "HOF Panel III" can also display system info and play videos, which is cool but probably unnecessary. All of this is wrapped up in a physical design that is more eyebrow-raising and gaudy than something you'd call sleek or stylish.

The triple-fan cooler design includes 112mm fans on both sides, and a 92mm fan in the center, not to mention a chunky heatsink. There's even a "hyper boost button" to increase fan speeds which GALAX notes will reduce GPU temperatures by around 10 degrees Celsius.

A new AMD article titled "Never a Better Time to Upgrade with Radeon Graphics" outlines the reasons to choose either an AMD Radeon RX 7000 or 6000 Series graphics card but ends up making RDNA 3 cards like the Radeon RX 7900 XT look like poor choices.

The pricing surrounding the new flagship GPUs from AMD and NVIDIA has been one of the most talked about parts of the next generation of PC gaming tech. Especially when it comes to the models that sit right below the top - the GeForce RTX 4080 and the Radeon RX 7900 XT.

Starting with the GeForce RTX 4090 (USD 1599) compared to the GeForce RTX 3090 (USD 1499), we see a massive leap forward in 4K gaming performance. And a slight difference in launch pricing. It's a similar story for the flagship AMD Radeon RX 7900 XTX (USD 999) compared to the AMD Radeon RX 6900 XT (USD 999), which also presents a giant leap forward in terms of 4K and ray-tracing performance.

Microsoft's DirectStorage technology brings lightning-fast loading times to PC games, with near-instantaneous gaming similar to what we saw when the PlayStation 5 launched. How it works is that it uses the GPU for asset decompression, which dramatically improves loading. And it requires a modern M.2 SSD to work. Forspoken, from Square Enix, is the first game to use the DirectX technology on PC, and German publication PC Games Hardware has been checking it out to see what effect it might have on in-game performance and frame rates.

For the GeForce RTX 4090, you're looking at a 10% drop in overall frame rates when comparing a standard SATA SSD without DirectStorage enabled to it running on an M.2 SSD. An average FPS of 83.2 compared to 75.3/74.4. Interestingly the 1% lows aren't affected, which means that the overall performance was smooth despite the lower average.

Of course, this is a single system, one game, and unspecified hardware, so it might not point to a situation where a 10% drop in performance would be the norm for DirectStorage-enabled games. The tests are also limited to 4K, so it could be a different story in 1440p. Also, with NVIDIA RTX IO being NVIDIA's implementation of DirectStorage for its GPUs, we're unsure if there were any specific driver accommodations for this feature for Forspoken -an AMD-sponsored title.

It's no secret that high-end CPUs and GPUs perform better in colder temperatures, especially when overclocking enters the picture. It's the reason why LN2 or liquid nitrogen is used to cool hardware pushing insane speeds - as opposed to, say, lava.

Thanks to a Chinese content creator by the name of "苏打baka," we've now got confirmation that a powerful gaming PC rocking an Intel Core i9-13900K CPU and NVIDIA GeForce RTX 4090 will have no problem functioning during the next Ice Age.

Some regions of China are experiencing some of the coldest temperatures they've seen for decades right now, so she decided to take her gaming rig to see how it holds up in a sub-zero -53C climate. And apparently, no hardware was damaged in the creation of the video.

AMD has released Adrenalin 23.1.2 drivers for its Radeon GPUs, except that the update is exclusive to the Radeon RX 7900 Series of graphics cards that launched late last year. Alongside support for the new game Forspoken it expands Vulkan API support, though what's missing is support for older AMD GPUs - specifically the Radeon RX 6000 Series. And with Forspoken being an AMD-sponsored title on PC, having no day-one support for the RX 6000 Series is baffling.

This means no optimized support for those playing Forspoken on a Radeon RX 6950 XT, and it's been 48 days since AMD released a driver covering its full line-up of GPU offerings. That driver update arrived on December 8, 2022, so this is the third Radeon RX 7000 Series exclusive driver from AMD in a row. Perhaps AMD is currently focused on optimizing the new RDNA 3 architecture and fixing bugs related to power-draw and performance, outlined by the "Known Issues" section of the AMD Software: Adrenalin Edition update.

And even though the new drivers are limited to the Radeon RX 7900 XT and RX 7900 XTX GPUs, the download package is still a sizable 590 MB.