Graphics Cards - Page 178

NVIDIA has worked with Bandai Namco on something truly unique to celebrate the 40th anniversary of the release of a game that changed everything: PAC-MAN.

What did NVIDIA do? NVIDIA researchers trained a new AI model called "NVIDIA GameGAN" that 'watched' 50,000 PAC-MAN episodes to re-produce a fully functional version of the game, all without "an underlying game engine".

How did all of this work? Well, GameGAN, in NVIDIA's own words "the first neural network model that mimics a computer game engine by harnessing generative adversarial networks, or GANs. Made up of two competing neural networks, a generator and a discriminator, GAN-based models learn to create new content that's convincing enough to pass for the original. You may have experienced GANs by using NVIDIA GauGAN, a deep learning model developed by NVIDIA Research that turns rough doodles into photorealistic masterpieces with breathtaking ease".

We already know that both AMD and NVIDIA will be launching next generation graphics cards later this year, where we have already heard rumblings of NVIDIA launching its next-gen Ampere-based GeForce RTX 3080 Ti and other RTX 30 series cards at Computex 2020 in September.

Now we have DigiTimes with a report that teases: "With AMD and NVIDIA set to launch their next-generation GPUs in September, graphics cards vendors are expected to cut prices for older-generation products, stimulating demand further in third-quarter 2020".

There are some things to note here, as DigiTimes didn't say anything about gaming-focused graphics cards... but we all know that they're coming. We should expect to see NVIDIA unveil a slew of new Ampere-baed GeForce RTX 30 series graphics cards towards the later half of the year -- with previous rumors pegging September 2020 for a Computex release.

NVIDIA unveiled its new Ampere GPU architecture last week, with the first Ampere-based product out of the gate being NVIDIA's new A100 GPU. Normally we'd see NVIDIA dub it the Tesla A100 -- just like the Volta-based Tesla V100, but nope -- now we might know why.

According to a new report from Heise, NVIDIA is dropping the Tesla branding because it is too confusing with Elon Musk's automotive company, Tesla Motors. NVIDIA has had the TESLA brand around for over a decade now, with teh first member of the Tesla graphics card with the Tesla C870-based on the Tesla G80 GPU back in 2007.

VideoCardz has also picked up on the fact that NVIDIA has removed 'Tesla' from the name of its Turing-based Tesla T4 accelerator, now known simply as NVIDIA T4. You can still see 'Tesla' in the URL, too -- at least for now.

Epic Games blew our minds with their recent unveiling of Unreal Engine 5 and its recent video, which is an utter work of art and one of the examples of the closest we've gotten to photorealistic graphics.

Well, now we know what kind of grunt you'll need inside of your gaming PC to pull it off. Epic used Sony's next-gen PlayStation 5 to run the Unreal Engine 5 demo, but World Today News has compiled some statements from Epic Games that detail just what we need to run the demo on our PCs (if we could).

Kim Libreri, the Chief Technology Officer (CTO) of Epic Games, was paraphrased by World Today News: "Can the demo work on a modern PC with the GeForce RTX 2070 Super? According to Mr. Libreri, yes, and the results will be 'pretty good.' For comparison: the PS5 graphics accelerator, on which the demonstration was run, is capable of providing up to 10.28 teraflops (TFLOPS), and the GeForce RTX 2070 Super has a theoretical performance of 9 TFLOPS".

It looks like the United States just made one of the largest changes in recent years, with the Trump administration paving the way for the future of chip production on American soil.

It first started in 2017 days before Donald Trump was inaugurated as the 45th President of the United States, where I wrote a story that he was pushing both TSMC and Foxconn at the time to consider manufacturing their chips in the US.

Fast forward to 2020 and just days ago the White House was talking with Intel and TSMC to get plants built in the US -- on top of that, throw in the coronavirus pandemic and the tentacle of issues it throws out into the world. Having the world's bleeding edge chips being made in Taiwan which has possible issues with China to face in the coming years, as well as China and global relations post-COVID-19, and we have some glaring issues in worldwide supply of chips that go into virtually all products.

NVIDIA has just made its new Ampere GPU architecture official, unveiling the first Ampere-based GPU in the A100 GPU. You can read more about that here, but alongside the NVIDIA A100 GPU is the new DGX A100 system.

NVIDIA's new DGX A100 3rd Generation integrated AI system packs a huge amount of computing power, centered around 8 x NVIDIA A100 GPUs, as well as 2 x 64-core/128-thread AMD Rome CPUs with 1TB of RAM. There's a huge 15TB of PCIe 4.0 NVMe SSDs for kick ass storage, and 9 x Mellanox ConnectX-6 VPI for 200Gb/s over the network.

The new DGX A100 is built for data science and AI, with up to 5 PFLOPs of peak FP16 performance, and up to 2.5 TFLOPs of TF32 performance, right up to 156 TFLOPs with FP64 performance.

GTC 2020 NVIDIA has officially unveiled its first GPU based on the Ampere GPU architecture, the new NVIDIA A100, which the company has in full production and is already shipping to customers worldwide.

NVIDIA's new A100 GPU packs an absolutely insane 54 billion transistors (that's 54,000,000,000), 3rd Gen Tensor Cores, 3rd Gen NVLink and NVSwitch, and much more. The GPU itself measures 826mm2 (!!!) on TSMC's 7nm node, as well as packing a huge 40GB of HBM2 memory from Samsung, and up to 600GB/sec bandwidth through NVLink.

The new A100 is being compared against the V100, which is based on the Volta GPU architecture. NVIDIA's current-gen Tesla V100 comes with 16GB and 32GB HBM2 options, while the V100 was built on the 12nm node at TSMC it packs just 21 billion transistors in comparison.

On the eve of NVIDIA's next-gen Ampere GPU reveal, AMD has decided to unveil its new Radeon Pro VII graphics card aimed at professionals.

AMD's new Radeon Pro VII is the consumer-focused Radeon VII, but for professionals. We have the same 16GB of HBM2 on a 4096-bit memory bus with up to 1TB/sec of memory bandwidth. The same Vega 20 GPU on the 7nm node is used, with 3840 cores, 240 TMUs, and 64 ROPs.

What is new here with the Radeon Pro VII is that AMD has shifted over to the PCIe 4.0 standard, versus the consumer-focused Radeon VII on PCIe 3.0. The same GPU clocks are here, with peak GPU clocks of up to 1750MHz -- while AMD delivers a new multi-GPU bridge for the Radeon Pro VII which the company calls Infinity Fabric Link with up to 168GB/sec of bandwidth.

NVIDIA will be unveiling its next-gen Ampere GPU architecture tomorrow morning, with the super-insane Ampere A100 GPU to lead the charge as well as a bunch of A100-powered systems.

We are to expect NVIDIA founder and CEO Jensen Huang to burst out onto a (virtual) stage and unveil the new Ampere GPU architecture, talk about the latest innovations that NVIDIA has achieved in AI, high performance computing (HPC), data science, autonomous machines, healthcare, and graphics during the GTC 2020 online keynote.

I was fully booked and I'm sure many hundreds of others were, for the annual GTC 2020 which was meant to kick off on March 23 -- but was canceled due to the coronavirus pandemic. NVIDIA shifted gears to go all-online for GTC 2020 this year, with the full announcements that will be in the keynote to be found here.

We are so very close to NVIDIA announcing its new Ampere GPU during its now online-only GTC 2020 keynote, with NVIDIA CEO and founder Jensen Huang to take the virtual stage.

Before that, Jensen has been up to something in his kitchen -- cooking up something delicious. In a newly-posted video to NVIDIA's official YouTube account titled 'What's Jensen been cooking?' the man himself pulls something out of the oven... which seems to be a new DGX system that I wrote about a few days back as the Ampere-based HGX / DGX A100 system.

NVIDIA's super-fast DGX solutions are some of the fastest pieces of kit you can buy, with previous DGX designs powered by Pascal, Volta, and Turing GPUs. But now, a new system is about to be unveiled which will be powered by the new Ampere GPU architecture.

It looks like Valve is about to kick off its annual Steam Summer Sale, with SteamDB developer Pavel Djundik tweeting out that the Steam Summer Sale will begin on June 25.

Djundik tweeted that "Chinese sources are saying the Steam summer sale starts on June 25", clarifying that these sources are "generally trustworthy, but not fully confirmed". He followed up that tweet with another, tweeting: "Confirmed now, Jun 25 to Jul 9".

But then Avi Thour just came out and gave exact times, tweeting "CONFIRMED! The 2020 Summer #Steam Sale runs from 25th June @ 9:55 A.M. Pacific to 9th July @ 10:00 A.M. Pacific. Hold your money boys!"

I'm sure you've never heard of Zeal-All before, and don't worry because neither did I -- but they are a Chinese motherboard manufacturer and they've just released a new LGA1151 motherboard with an integrated GeForce GTX 1650 Ti.

Zeal-All's new ZA-SK1050 motherboard uses the Intel B150 chipset, supporting both 6th and 7th generation Intel Core processors (Skylake and Kaby Lake). It is a small board for custom enterprise use, with Zeal-All aiming the new motherboard at AI algorithms, according to its website.

The board has HDMI, Ethernet, 4 x USB 3.0 ports, and 3.5mm audio ports. Zeal-All's interesting new motherboard has an integrated GeForce GTX 1050 Ti with 4GB of RAM, while the board itself supports up to 32GB of RAM, and has 4 x SATA ports for storage.

An exciting new rumor has emerged through a leak from ChinaTimes, which is reporting that AMD's next-gen Zen 4 processors, as well as AMD's upcoming RDNA 3 (and not RDNA 2 which is on 7nm) as well as NVIDIA's new Hopper GPUs will be made on the 5nm node by TSMC.

A leaked investment plant at-a-glace shows what TSMC has planned for its 5nm N5 and enhanced N5+ nodes. In the image, TSMC refers to AMD's upcoming Zen 4 architecture CPUs and RDNA 3-based GPUs, NVIDIA's next-gen Hopper GPU architecture, and many others including Qualcomm's next-gen Snapdragon 875 and Snapdragon X60 5G modem.

The list of technologies planned for TSMC's new 5nm node include Broadcom for a new high-speed network processor, other 5G modems, MediaTek's new Dimensity 2000 series 5G chips, Intel's next-gen Xe GPU architecture, Apple's upcoming A15 processor, Huawei HiSilison Kirin 1100, and more.

Alright everyone, here we go. This is an absolutely juicy rumor, and more like a gigantic juicy steak. Even writing those words made me salivate a little, but not like the bucket of drool these new specs on the GeForce RTX 3080 Ti.

YouTube channel Moore's Law is Dead has been teasing some purported specs on NVIDIA's next-gen Ampere GPU architecture, and more specifically the GA102 GPU. GA102 will power the next-gen GeForce RTX 3080 Ti, the successor to the GeForce RTX 2080 Ti which is based on the Turing GPU architecture and is the flagship GeForce RTX graphics card right now.

These new rumored specs on GA102 have it packing 5376 CUDA cores on the Ampere architecture, 10% more IPC than Turing, and on the 7nm node that lets GPU clocks scale much higher to 2.2GHz and beyond. The lower-end Ampere GPUs will reach the dizzying heights of 2.5GHz.

It was just yesterday I wrote about NVCache, a new Ampere GPU architecture feature that will use system RAM and SSDs to speed up game load times and optimize VRAM usage. We're in throwing distance of the new Ampere-based GeForce RTX 3000 series cards with some exciting news.

In a new video from YouTube channel Moore's Law is Dead, according to "exclusive insider info" secured by Tom -- NVIDIA's new Ampere cards are not just a die shrink of Turing with more RT Cores. It is not just the "Pascal version of Turing" and instead Ampere is a "multipurpose architecture".

One of the more exciting parts of the new Ampere rumors is that ray tracing performance is significantly better than Turing -- so much so that it reportedly offers 4x better performance per tier. This means a GeForce RTX 3060 will offer the same ray tracing performance as the flagship GeForce RTX 2080 Ti -- if not better.

Intel's upcoming DG1 graphics card has been spotted hanging out in the SiSoft Sandra database, with its 768 cores at up to 1.5GHz (expect this to be higher at launch) as well as 3GB of VRAM.

This isn't going to be some GeForce killer, but we should instead expect this particular DG1 graphics card from Intel to start knocking around the enterprise market. There will be another SKU that will be for gamers, and I'm sure it'll also have much more VRAM (at least 4GB, more likely 6-8GB and hopefully many more GPU cores and clock speed).

However, in this form the Intel DG1 has 1.9TFLOPs of compute power -- meaning it is faster than the original PlayStation 4. It's also good enough for minimum spec for VR gaming, in case you wanted to do that.

We are beginning to learn much more about NVIDIA's next-gen Ampere GPU architecture, with some super-juicy rumors that -- as always -- require a considerable amount of salt.

The latest on Ampere is from YouTube channel Moore's Law is Dead, with Tom explaining that has has some "exclusive indsider info" on Ampere. What is talked about in the video above is that Ampere is not just a die shrink (down from 14nm to 7nm) with additional RT cores -- but instead it is a full evolution of the Turing architecture.

One of the new nuggets of information is that NVIDIA will introduce something called NVCache with Ampere, which will act similar to AMD's own technology that debuted with Vega in HBCC (High Bandwidth Cache Controller). Like HBCC, NVCache will use some of the system RAM and SSD of your system, and super-speed game load times and optimize VRAM usage.

We are hearing more and more about NVIDIA's next-gen Ampere and Hopper GPUs, with more rumors flying in the last week than the last few months. Now? We have news that NVIDIA is bursting right into 5nm with Hopper.

According to a new report from DigiTimes, NVIDIA has "already pre-booked TSMC's 5nm production capacity in 2021" while "also in discussion with Samsung for smaller volume orders". The reason? To "prevent AMD from getting any bigger, NVIDIA has decided to catch up, even leapfrog AMD, in adopting TSMC's 7nm and 5nm EUV nodes".

Now remember, there's no one going on the record here so take this all with a grain of salt. There are some big power moves going on, I'm sure we can all see that by now -- but this is all just so very interesting.

If the dizzying heights of 8K at 7680 x 4320 weren't high enough, then 15,360 x 8640 (16K) might tickle your pickle -- and USB 4.0 is going to be dropping in 2021 and it should support both displays... all on a single cable.

CNET is reporting that USB 4.0 Type-C ports will pack DisplayPort 2.0 wiring, which will enable single-cable 8K displays and even more insane single-cable 16K displays. USB 4.0 will feature up to 40Gbps of bandwidth, and up to 80Gbps in x2 mode, and will pack tunneling technology that Thunderbolt currently enjoys meaning you can daisy chain devices all day long.

The 8K and 16K resolutions will both run at 60Hz over a single-cable USB 4.0 Type-C connector, as well as next-gen VR headsets. USB 4 is getting some major injections of bandwidth and features thanks to some of the Thunderbolt technology from Intel, where Thunderbolt 3 adopted the USB-C physical connector and now runs in USB alt mode.

The last few days have been pretty crazy in the GPU world, with purported specs on NVIDIA's upcoming Ampere-based GeForce RTX 3080 Ti blowing us away -- and then Intel with its "Father of All" GPUs in the Xe-HP. In the middle of all of this, the diamond-in-the-rough was lost: RTX 3060.

But don't worry... we're going to dive into that right now. In the leaked specs from a few days ago on the upcoming GA1xx range of Ampere GPUs, there was the lowest end card: the GeForce RTX 3060. The GeForce RTX 3060, at least according to these leaked specs, will be based on the GA104 GPU.

Now, just how powerful is the GA104 GPU (again, in these leaked specs)? Pretty damn powerful. We're looking at more power than the current flagship GeForce RTX 2080 Ti which has a price of over $1200. NVIDIA typically charges far less than $500 for a mid-range GeForce RTX 2060 (or RTX 2060 SUPER) in their Gx104 variant.