Graphics Cards - Page 66

The latest NVIDIA GeForce Game Ready 552.12 driver is here. It optimizes Season 3 of Call of Duty: Modern Warfare III and Call of Duty: Warzone, which features DLSS 3 and Reflex. The new driver also supports Diablo IV's new ray-tracing update, and NVIDIA's internal benchmarks showcase that you'll need to enable Frame Generation for playable performance.

On the plus side, at 1080p, it's playable with max settings and ray-tracing on a GeForce RTX 4050 or RTX 4060 Laptop GPU, with the latter hitting close to 100 FPS. Looking at Call of Duty: Warzone, with DLSS 3, the flagship GeForce RTX 4090 can hit an impressive 235.9 FPS at 4K, followed by the GeForce RTX 4080 SUPER with 184.3 FPS.

In addition, the new driver adds more one-click optimal settings in GeForce Experience and the NVIDIA App for nine new titles, including Dragon's Dogma 2 and Horizon Forbidden West Complete Edition.

MSI's Duke series of graphics cards was last seen during the Turing generation, with the flagship GeForce RTX 2080 Ti and GeForce RTX 2070. The Duke series slots between MSI's affordable Ventus line-up and the more OC-ready Gaming series. However, it was a no-show during the Ampere generation of GeForce RTX 30 Series cards, leading many to believe that MSI has simply retired the Duke in favor of the popular Ventus and gaming models.

Well, the Duke is back - at least for those in the Chinese retail market, with MSI launching a GeForce RTX 4060 Gaming Duke 3X in the region. As per the name, this new GPU features a triple-fan cooling setup with a physical design that is all-black and closer to the Gaming series than Ventus.

It's a long one, measuring 319 mm, but thankfully, it's a slim design with a two-slot thickness that includes a black brushed aluminum backplate for additional support. Three fans for a GeForce RTX 4060 is overkill, but that should mean that the Gaming Duke runs quiet and coolly.

NVIDIA has released a new version of its Streamline tool for developers (SDK 2.4.0), which includes DLSS 3.7.0 - the latest version of its upscaling and Frame Generation tech. Streamline is a tool developed by NVIDIA to make it easier for game developers to implement DLSS technologies in their projects.

It's also similar to Microsoft's upcoming DirectSR in that it's open-source and supports third-party plugins and upscalers like Intel XeSS and AMD FSR.

The updates were spotted by the creator of DLSSTweaks (on Reddit), a popular tool that lets users update a game's version of DLSS and add support for DLAA. So, the arrival of DLSS 3.7.0 is the more exciting update for gamers, and people are already testing it to see what difference it makes.

The US government is clamping down even harder on high-performance GPUs and AI GPUs being imported to China, with beefed-up US trade regulations stopping the cut-down GeForce RTX 4090 D graphics card, and H20 AI GPU.

We've seen NVIDIA and AMD restricted on the number of silicon it can send to China, depending on specific performance metrics by the Biden administration. Starting on April 4, GPUs with rated compute performance that is over 70 TFLOPs -- including the cut-down GeForce RTX 4090 D and H20 AI GPUs -- are now restricted, as they offer over 70 TFLOPs of compute performance.

The updated SEC document explains: "License is required (NLR) for computers with an "Adjusted Peak Performance" ("APP") not exceeding 70 Weighted TeraFLOPS (WT) and for "electronic assemblies" described in 4A003.c that are not capable of exceeding an "Adjusted Peak Performance" ("APP") exceeding 70 Weighted TeraFLOPS (WT) in aggregation".

The Steam Hardware & Software Survey results for March 2024 are now available, and the data once again shows that the GeForce RTX 3060 is still the most popular GPU among Steam gamers. As reported earlier today, its share rose dramatically in March, with 6.71% (+0.71%) of PC gamers rocking the mainstream Ampere card.

What about the current generation of GPUs? The GeForce RTX 40 Series, the Radeon RX 7000 series, and Intel's first generation of Arc graphics cards. Although we've seen new models and variants arrive at a steady clip, most recently with the SUPER refresh from NVIDIA and the Radeon RX 7900 GRE global relaunch from AMD, we're now well over a year into the generation.

Now's as good a time as any to take stock and see what's hot and what's not. So, let's examine some trends, starting with the current generation of GPUs - where the GeForce RTX 4060 and GeForce RTX 4070 are the most popular new cards.

NVIDIA's RTX 3060 is busy cementing its lead as the most popular graphics card with Steam gamers.

If you follow the Steam hardware survey, you'll likely be acquainted with the dominance of the RTX 3060 in the GPU department of late, as the NVIDIA board has been at the top of the rankings since October 2023 (when it dethroned the GTX 1650, in fact).

The RTX 3060 has been up and down a little in its percentages since then. The GPU rose to a mighty 6.27% last October, then dropped back down towards the 5% mark - still remaining in place as the top graphics card, mind - before grabbing a 6.17% share of Steam gamers in February.

Ahead of a new generation of discrete GPU hardware for gaming, especially when a release is expected within months, not years, a steady stream of information arrives from known sources and leakers. On the Radeon front, several credible sources have confirmed that AMD won't be releasing a flagship Radeon RX 7900 XTX killer for its upcoming RDNA 4 generation. Instead, AMD is focusing on the mid-range and competitive markets.

The reasoning for this comes down to AMD's next-gen chiplet design not being ready, so it will instead release monolithic Radeon RX 8800 XT, RX 8700 XT, and Radeon RX 8600 XT GPUs. Today, we've got some performance expectations from known Radeon leaker @All_The_Watts over on X - albeit in cryptic form.

As you can see, we get emojis instead of actual words - a fun presentation that, although posted on April 1, is not an April Fool's joke. The rumor, as deciphered, is straightforward, offering performance info for Navi 48 (Radeon RX 8800 XT) and Navi 44 (Radeon RX 8600 XT).

GIGABYTE is preparing to launch a limited-edition GeForce RTX 4080 SUPER, with only 300 units to be made. The AORUS GeForce RTX 4080 SUPER Xtreme ICE will boast the highest out-of-the-box overclock of all RTX 4080 SUPER cards, with an impressive Boost Clock speed of 2700 MHz. This is roughly 6% higher than the reference Boost Clock speed of 2550 MHz and 30 MHz higher than the ROG Strix model.

Unlike other Xtreme models from GIGABYTE, the AORUS GeForce RTX 4080 SUPER Xtreme ICE is air-cooled via the company's WINDFORCE system that includes a vapor chamber, "Bionic Shark Fans," and plenty of heat pipe action. This info comes from a leaked GIGABYTE presentation for the new GPU via @I_Leak_VN on X., showcasing the GPU's design, goals, and specs.

GIGABYTE also lists the AI performance for the AORUS GeForce RTX 4080 SUPER Xtreme ICE in its slides, with its 885 AI TOPs performance being considerably higher than the reference RTX 4080 SUPER's 836 AI TOPs.

According to rumors, Intel's second generation of discrete graphics cards for desktops, codenamed Battlemage, is on track for release sometime in the second half of 2024. The first generation of Alchemist Arc graphics has come a long way since its 2022 debut, partly thanks to Intel's continued driver updates and software support. Arc GPUs have gone from products you wouldn't recommend to impressive and affordable mainstream gaming options.

With shipping manifest leaks (via @momomo_us on X) confirming the existence of two Battlemage GPUs - Battlemage G10 and Battlemage G21 - it's widely accepted that G10 will be targeting the mainstream and mid-range PC gaming market, with G21 being the mainstream or entry-level replacement for current Arc models like the A750 and A770.

With this latest update and last week's SiSoft Sandra database surprise appearance from Battlemage "Xe2-HPG," it looks like the silicon is now in the hands of Intel partners - pointing to a launch sooner rather than later. Intel has yet to formally announce Battlemage or provide details on what to expect hardware and performance-wise, but we know quite a bit based on the information out there.

NVIDIA launched its modified GeForce RTX 4090 D graphics card last year, reducing CUDA core counts and TDPs to meet US export restrictions on high-performance chips going into China... and now a custom ASUS ROG Strix RTX 4090 D has been unlocked to 600W power with impressive results.

In their official sense, all of the GeForce RTX 4090 D graphics cards ship with compliant specs approved to go into the Chinese (and other restricted markets) with lowered TDPs of 425W, down from 450W on the full-spec RTX 4090. There is no factory overclocking, and no dual-BIOS features allowed that push higher GPU clocks... until now.

HKEPC tested an ASUS ROG Strix GeForce RTX 4090 D graphics card after ASUS introduced a new 600W overclock power setting for its ROG Strix GPUs. The ASUS ROG Strix RTX 4090 D is identical to the ROG Strix RTX 4090 non-D, with a new AD102-205 GPU that packs fewer CUDA cores than the full RTX 4090.

COLORFUL has just announced its new GeForce RTX 4070 Ti and GeForce RTX 4070 SUPER graphics cards in its iGame Vulcan series, with a new Fog Hill of Five Elements Edition inspired by the anime series.

The new COLORFUL iGame Vulcan GeForce RTX 4070 Ti and RTX 4070 SUPER graphics cards are both inspired by Samsara Animations' "Fog Hill of Five Elements" anime series. The Chinese AIB beautifully uses the gold and black theme on the cards.

Underneath, the design of COLORFUL's iGame Vulcan GeForce RTX 4070 Ti SUPER and RTX 4070 SUPER packs a gigantic triple-slot cooling design with a triple-slot fan keeping the Ada Lovelace GPU cool. The heatsink features a massive aluminum-finned array that spans the entire length of the cooler, with a cut-out on the back that assists the third fan in venting hot air out of the shroud.

VASTARMOR has just launched their new Radeon RX 7900 XTX Super Alloy graphics card, around 4 months after the Chinese company first unveiled the new RX 7900 XTX design. Check it out:

The new VASTARMOR Radeon RX 7900 XTX Super Alloy features the highest GPU boost clock of any RX 7900 XTX on the market, with 2680MHz out of the box. This means VASTARMOR's new RX 7900 XTX Super Alloy graphics card has a higher boost clock than Navi 31 is capable of. This is the same 2680MHz that we find on the ASRock Taichi and Aqua cards, PowerColor's Liquid Devil, and SAPPHIRE's NITRO+ variants.

VASTARMOR's new card is a chunky boi, measuring in at a large 33cm long and 6.9cm thick, which makes it one of the biggest graphics cards on the market. The company includes a triple-fan cooler, with ARGB lighting, too. Pricing kicks off at 6799 RMB (around $940 USD), which means it's considerably cheaper than the MSRP of the AMD Radeon RX 7900 XTX, which was priced at 7999 RMB.

Every so often a weird and wonderful, highly surprising story pops up around a DIY tech project, and they don't come any more astonishing than FuryGPU, frankly.

As Tom's Hardware picked up, FuryGPU is the accomplishment of games developer Dylan Barrie (who seems to work at Respawn, or someone of the same name does on LinkedIn, at any rate).

Barrie had the idea to build his own GPU - completely from scratch - and actually managed to pull this off.

Astera Labs was demoing its working PCIe 6.0 connectivity with next-gen Aries PCIe 6.0 retimers, spotted on the NVIDIA GTC 2024 showfloor by our friends at Tom's Hardware. Check it out:

We don't even have PCIe 5.0 across the board with all PC users and gamers, but PCIe 6.0 is coming with NVIDIA's new Blackwell B100, B200, and GB200 AI GPUs all ready for Gen6. AI workloads require ridiculous amount of throughput through the GPUs to process everything at full speed, with NVIDIA using its in-house custom NVLink and NVSwitch to offer massive GPU-to-GPU communication, offering up to 1.8TB/sec of bandwidth on GB200-powered systems.

Not every customer wants the entire NVIDIA solution, which is where PCIe 5.0 x16 connections step in for the current fleet of AI GPUs, but Blackwell introduces PCIe 6.0, doubling the 128GB/sec of bandwidth of an x16 connection on PCIe 5.0 to a huge 256GB/sec bandwidth on PCIe 6.0 x16.

Introspect Technology is a member of JEDEC, the body that covers memory -- including the next-gen GDDR7 memory standard -- with Introspect shipping the world's first GDDR7 memory test system.

The just-launched Introspect M5512 GDDR7 Memory Test System is the world's first commercial solution for testing JEDEC's fresh new JESD239 Graphics Double Data Rate (GDDR7) SGRAM specification. GDDR7 is a category-creating memory solution that enables graphics memory engineers, GPU design engineers, product engineers in both memory and GPU spaces, and system integrators to quickly design new GDDR7 memory devices, debug protocol errors, characterize signal integrity, and perform detailed memory read/write functional stress testing without requiring any other tool.

The new M5512 GDDR7 Memory Test System is the perfect solution for these engineers, AMD and NVIDIA, and AIB partners to get ready to launch GDDR7-powered graphics cards, the first of which will be NVIDIA's next-gen GeForce RTX 50 series GPUs.

NVIDIA's RTX 4060 has a recommended price of $299, though it has sunk a bit below this before - although a fresh low has now been reached with its price tag in the US.

VideoCardz noticed that - at least at the time of writing - there's a Zotac RTX 4060 Twin Edge OC White Edition on sale for $279 at Newegg.

Now, while that pricing may have been seen as part of local offers (in-store pick up) in the past, this is the first time that we're aware of that the vanilla RTX 4060 has sunk to $279 via an online listing you can just click and buy, there and then.

Smuggling video cards into China is understandable, considering the current trade restrictions and the U.S. flat-out banning the sale of the GeForce RTX 4090 to the region. However, smuggling a batch of second-hand seven-year-old GPUs is more of a stretch because you wouldn't think it would be worth trying to smuggle 44 Radeon RX 580 graphics cards through Chinese customs.

According to a new report on MyDrivers (via Tom's Hardware), Shanghai Pudong International Airport customs officers caught a passenger trying to sneak in a batch of old Radeon GPUs into the country. Based on the seizure image all 44 Radeon RX 580 GPUs look to be the same make and model, presumably used for crypto mining.

Naturally, this is a case of someone trying to avoid paying import fees by not declaring that they have a bunch of old graphics cards they will wipe down and sell for a quick profit.

A new NAS motherboard has appeared on AliExpress, powered by an Intel Kaby Lake-G processor and integrated AMD Vega graphics. Yeah, Vega is back, baby.

The new Topton N9 NAS motherboard features an Intel Core i7-8705G "Kaby Lake-G" processor, with 4 cores and 8 cores at up to 4.1GHz. The CPU is designed on the 14nm node process, with a 65W TGP. Topton's new N9 NAS motherboard features an AMD Radeon RX Vega M GL GPU as a separate tile, with 20 Compute Units and 4GB of HBM2 memory.

Topton's new N9 NAS motherboard also features an integrated Intel HD Graphics 630 GPU with 24 Execution Units that uses shared memory. But the bigger deal here is not one or two, but 8 x 2.5GbE ethernet ports (this is a NAS motherboard, remember), so plenty of storage + networking options are required.

Samsung teased its next-gen GDDR7 memory last week at NVIDIA's recent GPU Technology Conference (GTC) event, but now it's officially listed GDDR7 at both 32Gbps and 28Gbps speeds on its website.

Samsung's next-gen GDDR7 memory comes in 16Gb density, allowing for 2GB of GDDR7 memory per module, which is the same capacity as we see on both GDDR6 and GDDR6X memory. Each GDDR7 memory module from Samsung will feature a 32-bit wide memory interface with a refresh rate of 16K/32ms. Samsung says its new GDDR7 memory modules come in 266 FBGA packaging, with two speeds listed: 32Gbps and 28Gbps.

The company will offer its standard 32Gbps GDDR7 memory die and a low-voltage optimized 28Gbps GDDR7 die. We should expect to see next-gen high-end flagship GPUs from NVIDIA like the GeForce RTX 5090 and GeForce RTX 5080 graphics cards debut with 32Gbps GDDR7, while more mid-range and higher-end models (but not flagship) cards like the RTX 5070 and RTX 5060 use 28Gbps GDDR7 memory.

AMD recently re-launched the Radeon RX 7900 GRE, a GPU that outperforms the Radeon RX 7800 XT and NVIDIA's GeForce RTX 4070 while being priced the same as the latter. However, a hard lock (which AMD has said was unintended) on memory speeds at launch limited the GPU's overclocking potential.

This meant that an overclocked Radeon RX 7800 XT could be pushed to reach Radeon RX 7900 GRE performance (while drawing more power) with no option for the RX 7900 GRE to do the same. Still, it's an impressive GPU, with a price that brings some more competition to the stacked mid-range graphics card market that now has four viable choices - the Radeon RX 7800 XT ($499), GeForce RTX 4070 ($549), Radeon RX 7900 GRE ($549), and GeForce RTX 4070 SUPER ($599).

There's some good news for those eyeing a Radeon RX 7900 GRE. The memory overclocking limits have been removed courtesy of the latest Radeon drivers, Adrenalin Edition 24.3.1.