Graphics Cards - Page 70

Ten years ago, NVIDIA's Maxwell architecture debuted in February 2014 with the GeForce GTX 750 Series - introducing an 'all-new design for the Streaming Multiprocessor (SM)' on a cutting-edge (for the time) 28nm process.

The first GPUs to adopt the new architecture were designed primarily for efficiency, with the GM107 chip seen in the GeForce GTX 750 and GTX 750 Ti bringing massive efficiency gains compared to the previous Kepler architecture.

The Maxwell architecture is also in the NVIDIA Tegra X1 SOC for mobile devices, which pairs Maxwell-based graphics with ARM CPU clusters. In addition to appearing in NVIDIA's Android-based Shield TV hardware, the Tegra X1 (or a custom version) would eventually make its way into the Nintendo Switch.

We've all been scammed or know someone who has been scammed... but this latest one sees a Thai buyer scamming a Vietnamese gamer, sending them a pair of women's shoes... and not the graphics card that they ordered.

I was tagged in a new post on X by "I_Leak_VN," which pointed to a Vietnamese-based dealer that got scammed by a merchant in Thailand. The buyer wanted an NVIDIA GeForce RTX 3060 Ti graphics card, but when they opened the box, they realized they were scammed: a pair of women's shoes was in the box instead.

The Vietnamese dealer was approached by a Thai seller who had the Manli GeForce RTX 3060 Ti graphics card for sale, with the buyer taking the card and paying the asking price. However, once the item was shipped, the box was missing the important graphics card and instead had the pair of shoes inside. The packaging is great, with foam around the GPU box to make it feel like there was some weight to it, but a useless pair of shoes resided inside.

The very latest 12V-2x6 "H++" power connectors have been shipping on NVIDIA's latest GeForce RTX 40 SUPER series graphics cards, offering higher power delivery than what the 12VHPWR "H+" power connector is capable of.

The new 12V-2x6 power connector is shipping on most of the new GeForce RTX 40 SUPER series graphics cards, with the RTX 4080 SUPER, RTX 4070 Ti SUPER, and RTX 4070 SUPER all shipping with the newer Gen5 connectors, but there are a few models of the RTX 4070 SUPER floating around with regular 8-pin PCIe power connectors.

Now we've got @wxnod on X showing us some close-up photos of the differences between the H++ and H+ power connectors, with the new H++ design based on the newer 12V-2x6 power connector revision, which is capable of supporting up to a maximum of an insane 675W power delivery into your graphics card. Both the plugs look virtually identical, where you can see the difference by removing the top shroud as they're labeled "++H" and "+H".

Note: Just to be clear, this new connector is the one NVIDIA is using on its GeForce RTX 40 SUPER series graphics cards -- the updated 12V-2x6 power connector -- which MLID says NVIDIA is mandating on the new GeForce RTX 50 series GPUs.

NVIDIA's next-gen GeForce RTX 50 series graphics cards will reportedly feature new PCIe 6.0 16-pin power connectors, according to the latest rumors.

In a new video, Tom from Moore's Law is Dead explains that his sources have said that NVIDIA will be shifting towards their next-gen GPUs using a next-gen PCIe 6.0 16-pin connector. This move could filter all the way down to the mid-range GeForce RTX 5060, not just the higher-end GeForce RTX 5080 and GeForce RTX 5090 graphics cards.

NVIDIA's next-gen graphics cards are not in danger of sliding to 2025, as some other recent rumors have suggested, according to fresh word from the grapevine.

This comes from Moore's Law is Dead (MLID), and we need to set a quick bit of background here for the context of this comment on the subject of NVIDIA - as it came as a brief, but interesting, aside, as part of the latest chatter from the rumor mill on Intel's Battlemage GPUs.

Basically, MLID was making it clear that there were certain rumors floating around hyping up Battlemage and that the GPUs could be inbound relatively soon, whereas the YouTuber believes this is not the case at all.

AMD's RX 7900 GRE, the alternative spin on the RX 7900 for China (GRE stands for 'Golden Rabbit Edition') has come to another marketplace - the UK - making a previously rumored US launch look more likely.

If you've been following the tale (or should that be tail?) of the RX 7900 GRE, after the graphics card's launch in China, it went on sale in Europe. As VideoCardz pointed out, the GPU has been on shelves in Germany and Italy, and just cropped up in the UK as TechPowerup noticed.

The UK retailer selling the RX 7900 GRE - which sits just under the 7900 XT - is AWD-IT, but it seems that the XFX model being sold went out of stock quickly. (Even though the pricing wasn't appealing at all - ironic as it was highlighted via a popular UK deals website).

We're witnessing a fairly steady flow of rumors around NVIDIA's RTX 5000 GPUs of late - or whatever Blackwell ends up being called - and the latest batch comes from a YouTube leaker.

This is RedGamingTech (RGT) who has been tapping sources to get fresh info on purported specs for Blackwell graphics cards, as well as news on the likely release timeframe.

Obviously as with any rumor, take all this with plenty of seasoning, and the caveat that even if these are NVIDIA's current plans being relayed back accurately, the relevant specs and other details may change as Blackwell is further developed.

AMD's FSR 3 or FidelityFX Super Resolution 3 with frame generation launched in September 2023 alongside the Radeon RX 7700 XT and Radeon RX 7800 XT. However, with support in just two games on day one, and only a handful more added by the year's end, adoption for AMD's answer to DLSS 3 has been slow... to say the least.

The good news is that January 2024 saw more games add FSR 3 support than the first four months of the technology. The adoption looks to be picking up steam thanks to FSR 3's improved performance in games like Avatar: Frontiers of Pandora and AMD confirming that its latency reduction tech Anti-Lag + (a key component of frame generation) is finally coming back.

And with that, AMD has updated its list of games with FSR 3 and FSR 2 support, and those 'coming soon.' FSR 3 includes both Super Resolution upscaling and Frame Generation, with FSR 2 including Super Resolution only. Here's a look at the current slate of PC games with FSR.

NVIDIA released its GeForce RTX 2080 Ti graphics card five years ago, launching with 11GB of VRAM, but now modders are doubling that to 22GB of VRAM and selling them on eBay as AI GPUs.

The US sanctions on China have stopped high-powered graphics cards from entering the country, with the likes of the GeForce RTX 4090 banned, which is why we saw the introduction of the modified GeForce RTX 4090 D sent to China. We've seen modders in the country taking apart gaming graphics cards and turning them into AI GPUs, and now we've got modded RTX 2080 Ti cards hitting the market. Oh boy.

The modded consumer-based gaming-ready GeForce RTX 2080 Ti has been modded, rocking 22GB of VRAM, which is far better than 11GB for AI workloads. The VRAM acts as a high-speed buffer for those huge AI datasets, complex models, and calculations that are needed for AI workloads. Interesting move, modders.

AMD is getting ready to reintroduce Anti-Lag+ technology for PC gamers with Team Red's graphics cards.

The news broke on X (formerly Twitter) when somebody (TheRazerMD) asked a question about when AMD might be planning on bringing back Anti-Lag+, which is a tech to combat input lag (much like NVIDIA Reflex, Team Green's rival take).

AMD's Frank Azor, who is Chief Architect of Gaming Solutions & Marketing, responded to say that yes, it was inbound again, and that Anti-Lag+ is "coming soon" in fact (hat tip to VideoCardz on this one).

File this under 'w' for worrying, but a Redditor has reported an invasion of red fire ants inside their PC, with the insects apparently eating the GPU alive, we're told.

The story goes like this: Redditor Thejus_Parol says that they noticed rising GPU temperatures with their Zotac GTX 1060.

When investigating what might be going on, checking if the fans could have malfunctioned, they discovered "ants marching on my GPU and on top of my case" and even coming out of the graphics card's heatsink.

NVIDIA's RTX 4090 GPU seems to be coming down in price (finally), though we're only witnessing some initial rumblings of this - but there should be some bigger price falls to come soon enough.

Those looking for the most powerful desktop GPU on the market have been pretty much out of luck in recent times, what with NVIDIA RTX 4090 pricing spiralling upwards.

That's been happening ever since the ban in China on the RTX 4090, resulting in more stock being shipped to Asia - even now, post-ban - with the RTX 4090 D alternative variant seeming to have failed to capture the attention of Chinese buyers in the way NVIDIA hoped it would.

Ubisoft's open-world co-op pirate game Skull and Bones has been a long time coming; development began in 2013, and now, just over a decade later, the game is finally making its debut on PC and consoles. On day one, PC gamers will be pleased to know that the game will feature Ray-Traced Global Illumination to enhance lighting with DLSS 2 Super Resolution support to boost performance on GeForce RTX-powered rigs.

Unfortunately, there is no DLSS 3 Frame Generation, but with the latest GeForce Game Ready 551.52 WHQL drivers offering optimizations for the game - odds are it will run smoothly. Or, you'd hope so, considering how long it has been in development.

What has arrived with DLSS 3 Frame Generation (and DLSS 2 and DLAA) support this week is the latest action RPG from celebrated studio DON'T NOD and publisher Focus Entertainment - Banishers: Ghosts of New Eden.

NVIDIA has just pushed out its new GeForce Game Ready 551.52 WHQL drivers, which optimize the experience for Ubisoft's upcoming Skull and Bones, which drops on February 16 on the PC. Download the new GeForce Game Ready 551.52 WHQL drivers right here (665MB download)

The new GeForce Game Ready 551.52 WHQL drivers offer support for Skull and Bones, whereas on the PC, it will allow GeForce RTX series GPU owners to enable Ray-Traced Global Illumination lighting to enhance image quality, as well as enhance game performance with DLSS 2 -- not DLSS 3, unfortunately.

There are also GeForce Experience's one-click optimal settings that enable you to have the optimal game configuration for over 1000 titles that are supported. NVIDIA's new drivers introduce support for 3 more games: Helldivers 2, Suicide Squad: Kill the Justice League, and Three Kingdoms Zhao Yun.

GIGABYTE is prepping to launch a new Radeon RX 7900 GRE graphics card with the company's WindForce 3X cooling. It's a design that we've across its entire Radeon and RDNA 3 line-up, from the Radeon RX 7600 to the flagship Radeon RX 7900 XTX. Adding a new GIGABYTE Radeon RX 7900 GRE into the mix is interesting because of the card's origins.

The Radeon RX 7900 GRE or Golden Rabbit Edition was created specifically for the Asian and Chinese markets, with the SKU and various models also making their way to other regions (like Australia) for new builds.

The Radeon RX 7900 GRE uses the Navi 31 GPU found inside the Radeon RX 7900 XT and Radeon RX 7900 XTX - albeit cut down. With 16GB of VRAM on a 256-bit bus, 80 RDNA 3 Compute Units, 5,120 Stream Processors, 160 AI Accelerators, and a total board power (TBP) rating of 260W, the card's overall performance sits in between the Radeon RX 7800 XT and the Radeon RX 7900 XT.

NVIDIA is forming a new business unit that has the goal for next-gen chips for AI processors, consoles, cloud computing, and more.

The news is coming from 9 different sources of Reuters; NVIDIA wants to help companies make their own custom AI chips... businesses that otherwise go to their competitors like Broadcom and Marvell. Right now, custom silicon design for data centers is "dominated" by the likes of Broadcom and Marvell, so NVIDIA wants a slice of that pie (and many others).

Research firm 650 Group says that the data center custom chip market will grow by up to $10 billion this year, and then double that in 2025. The broader custom chip market was worth around $30 billion in 2023, amounting to around 5% of annual global chip sales, says Needham analyst Charles Shi.

Japanese manufacturer Sycom has just teased a new cooling solution for NVIDIA's just-released GeForce RTX 4080 SUPER graphics card, using dual 120mm fans from Noctua.

Sycom has been releasing interestingly-styled GeForce RTX 40 series graphics cards for a while now, with RTX 4070 and RTX 4060 Ti graphics card releases using Noctua cooling. Now, the new RTX 4080 SUPER graphics card from Sycom replaces the GPU's vapor chamber with a cooling solution in the hopes of thermal dissipation that's much better for these new "power-hungry" GPUs.

The new hybrid GeForce RTX 4080 SUPER Hydro graphics card from Sycom is only packaged with its G-Master Hydro series desktop gaming PCs, but the cooler itself is from Asetek in the form of the "Hybrid GFX 240mm" which features a dedicated 240mm radiator, and dual Noctua NF-A12x15 120mm fans.

NVIDIA has just announced its new RTX 2000 Ada workstation GPU, which offers 1.5x the performance of the previous-gen RTX A2000 12GB workstation GPU (and 4GB more framebuffer, with 16GB GDDR6 ECC memory).

The new NVIDIA RTX 2000 Ada workstation GPU features the AD107 GPU with 2816 CUDA cores, 16GB of GDDR6 ECC memory on a 128-bit memory bus and PCIe 4.0 x8 lanes. Not bad, considering NVIDIA is charging $625 (before taxes) for the workstation GPU.

Still, there is up to 1.7x faster ray-tracing performance, up to 1.8x AI throughput over the previous-gen workstation GPU, up to 2x the performance boost across professional graphics, rendering, AI, and compute workloads, all within the same 70W TDP as the previous-gen workstation GPU.

NVIDIA has just launched the new GeForce RTX 4080 SUPER GPU; a $200 price cut to $999 USD alongside a small increase in performance has made it feel more like a relaunch than a refresh. Replacing the existing GeForce RTX 4080 as a powerful 4K gaming GPU, many retailers in the US are selling out of the flagship SUPER GPU - with MSRP models especially hard to find.

According to a report by Videocardz, retailers like BestBuy and MicroCenter are still offering models at the $999 price point. However, these are 'pick-up' only in stores with available units. It looks like the GeForce RTX 4080 SUPER is very popular with enthusiast PC gamers, especially with rising prices for GeForce RTX 4090 cards.

The report also notes that buying stock directly from vendors like ASUS, MSI, and Zotac is not an option, with many 'out of stock' GeForce RTX 4080 SUPER cards no longer available for purchase.

The very latest on AMD's next-gen RDNA 4 GPU architecture and new Navi 48 and Navi 44 GPUs is here, where we're learning more about the two new GPUs and their rumored specifications.

In his latest video, Tom from Moore's Law is Dead covers the latest he's heard from his industry sources on Navi 48 and Navi 44 GPUs from AMD. Starting with AMD's new flagship Navi 48 GPU, which will be a monolithic die manufactured on TSMC's N4P process, with MLID hearing that Strix APU and RDNA 4 discrete GPU engineering samples have been boosting between 3.0GHz and 3.3GHz.

Not only that, but AMD has added FP8 and Matrix Hardware, with RDNA 4 reportedly "ready" to be "able" to launch in Q4 2024, but MLID notes that there is a "slim chance" for a Q3 2024 release or later Q1 2025 release. MLID adds that he has "no proof" that AMD ever planned to support GDDR7 memory on smaller configurations, with the company reportedly toying with GDDR7 on a multi-chip GPU, but for the monolithic Navi 48 and Navi 44 GPUs... we're looking at GDDR6 memory.