Video Cards & GPUs News - Page 141

All the latest graphics cards and GPU news, with everything related to Intel Arc, NVIDIA GeForce, AMD Radeon & plenty more - Page 141.

Follow TweakTown on Google News

SK Hynix teases HBM3 with 12-Hi 24GB stack layout, 6400Mbps speeds

Anthony Garreffa | Nov 12, 2021 8:43 PM CST

The world of consumer graphics cards shifted away from HBM to GDDR6 pretty quickly, with flagship NVIDIA GeForce RTX 30 series GPUs rocking ultra-fast Micron GDDR6X memory. But the world of HBM didn't stop, and now SK Hynix is showing off their 24GB HBM3 memory clocked at 6.4Gbps.

SK Hynix teases HBM3 with 12-Hi 24GB stack layout, 6400Mbps speeds

During the recent OCP Summit 2021, ServeTheHome spotted some of that delicious new HBM3 memory from SK Hynix and snapped a photo of it. It was pretty fast of SK Hynix to display it since the company only announced its new HBM3 memory less than a month ago now -- and it's as thin as a piece of A4 paper, too.

We don't have the final specification of HBM3 from JEDEC just yet, but SK Hynix has already bumped the initial spec from 5.2Gbps to 6.4Gbps -- and that's a good thing. We should see next-gen accelerators and other HBM3-powered products, with some absolutely brutally-fast memory bandwidth. Hell, the 12-Hi stack on a 1024-bit memory interface can drive 819.2GB/sec per stack... with multiple stacks, we're looking at memory bandwidth numbers never before seen.

Continue reading: SK Hynix teases HBM3 with 12-Hi 24GB stack layout, 6400Mbps speeds (full post)

ZOTAC is giving away GeForce RTX 3080 Ti, 3070 Ti Resident Evil GPUs

Anthony Garreffa | Nov 12, 2021 7:41 PM CST

ZOTAC is running a new campaign that starts today and runs through until December 17, 2021 where you can win a wicked ZOTAC GAMING Resident Evil: Welcome to Raccoon City Limited Edition Graphics Card or Mini PC hardware.

ZOTAC is giving away GeForce RTX 3080 Ti, 3070 Ti Resident Evil GPUs

The new ZOTAC "Escape to Win" Mini Game isn't hard to enter, and you can win some pretty damn good prizes -- the grand prize of course, being huge. In what I think is the longest name ever given to a graphics card, you can win one of the ZOTAC GAMING GeForce RTX 3080 Ti and 3070 Ti AMP Extreme Holo Resident Evil: Welcome to Raccoon City Edition.

Not only can you win one of the ZOTAC GAMING GeForce RTX 3080 Ti and 3070 Ti AMP Extreme Holo Resident Evil: Welcome to Raccoon City Edition graphics cards, but there's also the ZOTAC ZBOX MAGNUS ONE Resident Evil: Welcome to Raccoon City Edition. This is a full Mini PC that crams a powerful Intel Core i7 processor and ZOTAC GAMING GeForce RTX 3070 graphics card into a Mini PC that is just 8.3 litres.

Continue reading: ZOTAC is giving away GeForce RTX 3080 Ti, 3070 Ti Resident Evil GPUs (full post)

EVGA unveils Precision X1 OC software: new VF Curve Editor, and more

Anthony Garreffa | Nov 9, 2021 10:28 PM CST

EVGA's new Precision X1 OC software is here, and it packs a new VF Curve Editor and frame rate synchronization which syncs your frame rate limit with the NVIDIA Control Panel.

EVGA unveils Precision X1 OC software: new VF Curve Editor, and more

Not only do we have the synchronization of the built-in frame rate limit between Precision and NVIDIA Control Panel, but you can tweak that sucker all the way to 1000FPS -- while there's also some fixed issues with the LED temperature mode fails and resets to rainbow mode and VF curve value is cleared when Apply is used on another page both fixed.

1000FPS + EVGA doesn't feel right, not after the EVGA GeForce RTX 30 series GPUs that were being killed from super-high FPS inside of Amazon's new game New World.

Continue reading: EVGA unveils Precision X1 OC software: new VF Curve Editor, and more (full post)

NVIDIA announces A2 Tensor Core GPU: GA107 GPU + 16GB GDDR6 memory

Anthony Garreffa | Nov 9, 2021 7:47 PM CST

NVIDIA has just announced its new A2 Tensor Core, its latest entry-level Ampere-based accelerator using the GA107 GPU which packs 1280 CUDA cores, with 16GB of GDDR6 memory.

NVIDIA announces A2 Tensor Core GPU: GA107 GPU + 16GB GDDR6 memory

The new NVIDIA A2 Tensor Core is quite a bit cut down compared to the A16 Tensor Core, which has 4 x GA107 GPUs compared to the 1 x GA107 GPU on the new A2 Tensor Core. That huge cut down means the TBP (total board power) drops to just 60W, but you can also support it at 40W, meaning you don't need external power.

NVIDIA is clocking the GA107 GPU at 1440MHz base, 1770MHz boost, offering up to 4.5 TFLOPs of single-precision compute power. This means the low-end GeForce RTX 3050 Ti Mobile GPU beats the A2 Tensor Core, with the RTX 3050 Ti Mobile GPU packing up to 7.1 TFLOPs of compute performance.

Continue reading: NVIDIA announces A2 Tensor Core GPU: GA107 GPU + 16GB GDDR6 memory (full post)

NVIDIA GeForce RTX 3080 Ti Laptop GPU: 7424 CUDA cores, 16GB GDDR6

Anthony Garreffa | Nov 8, 2021 7:43 PM CST

NVIDIA is expected to unveil its new flagship Ampere laptop GPU at CES 2022 in January, with a new entry onto the Geekbench list for an Intel 12th Gen Core "Alder Lake-S" CPU with an NVIDIA GeForce RTX 3080 Ti Laptop GPU.

NVIDIA GeForce RTX 3080 Ti Laptop GPU: 7424 CUDA cores, 16GB GDDR6

The new posting includes some new details, with the laptop SKU rocking the GA103S GPU with 58 Compute Units which will translate to 7424 CUDA cores (256 CUDA cores disabled). NVIDIA should offer the new GeForce RTX 3080 Ti Laptop GPU with 8GB or 16GB of GDDR6 memory, depending on the configuration, on a 256-bit memory bus.

We should expect the GeForce RTX 3080 Ti to beat everything on the market obviously, and will find its way into the next-gen flagship laptops of 2022 powered by CPUs like Intel's new Core i9-12900HK mobile processor under the new Alder Lake CPU architecture.

Continue reading: NVIDIA GeForce RTX 3080 Ti Laptop GPU: 7424 CUDA cores, 16GB GDDR6 (full post)

AMD Instinct MI250X: Aldebaran MCM GPU, 128GB HBM2e memory, 500W power

Anthony Garreffa | Nov 8, 2021 4:49 PM CST

AMD has finally announced its next-generation MI200 HPC GPU codenamed Aldebaran, based on the new CDNA 2 architecture on the 6nm process node.

AMD Instinct MI250X: Aldebaran MCM GPU, 128GB HBM2e memory, 500W power

The first MCM technology is now here in the form of the codename Aldebaran GPU, offering up 2 x GPU chiplets and an enormous 128GB of ultra-fast HBM2e memory. There's an incredible amount of GPU horsepower, memory bandwidth, and so much more going on inside of the AMD Instinct MI200.

Inside, the new Aldebaran GPU has two dies: one secondary, and one primary. Each of the GPU dies has 8 Shader Engines each, for a total of 16 Shader Engines while each of the Shader Engines has 16 Compute Units with full-rate FP64, packed FP32, and 2nd Generation Matrix Engine for FP16 and BF16 operations.

Continue reading: AMD Instinct MI250X: Aldebaran MCM GPU, 128GB HBM2e memory, 500W power (full post)

AMD's next-gen RDNA GPUs could have 3D Infinity Cache technology

Anthony Garreffa | Nov 8, 2021 3:56 AM CST

AMD's next-gen RDNA 3 GPU architecture is really going to open up a can of whoop-ass on NVIDIA in 2022, with the MCM-based (multi-chip module) graphics card now rumored with new 3D Infinity Cache technology.

AMD's next-gen RDNA GPUs could have 3D Infinity Cache technology

In a new tweet, leaker Greymon55 simply teases "3D IFC" which would imply "3D Infinity Cache". The next-gen RDNA 3 architecture from AMD is expected to have multiple GPU chiplets (MCM technology, or multi-chip module) and should -- at least from this new rumor -- split the Infinity Cache across the GPU chiplets in its MCM design, to 256MB per chip.

AMD's next-gen Navi 31 GPU is expected to have 512MB of Infinity Cache, but now this could be split into a next-gen 3D stacked design through 3D Infinity Cache technology. 3D IFC hasn't been formally announced just yet, so this is rumor territory as usual. AMD is using its 3D V-Cache technology on its next-gen EPYC Milan-X processors, with a huge 768MB of L3 cache expected on the flagship EPYC 7773X (64 cores, 128 threads) processor.

Continue reading: AMD's next-gen RDNA GPUs could have 3D Infinity Cache technology (full post)

GeForce RTX 4090 rumor: 2x RTX 3090 perf, power: 'can you accept it?'

Anthony Garreffa | Nov 8, 2021 12:57 AM CST

NVIDIA is having to compete with AMD in a totally different way in 2022 and beyond, with AMD moving into the world of MCM (multi-chip module) GPUs with multiple GPU chiplets making for a monster flagship Radeon RX 7900 XT. But NVIDIA won't take the fight laying down.

GeForce RTX 4090 rumor: 2x RTX 3090 perf, power: 'can you accept it?'

We've heard rumors that NVIDIA's flagship GeForce RTX 4090 under the new Ada Lovelace GPU architecture will offer twice the performance of the Ampere-based GeForce RTX 3090, but a new rumor from Greymon55 teases: "Double performance, double power consumption, can you accept it?"

Yes, Greymon55... I can accept this.

Continue reading: GeForce RTX 4090 rumor: 2x RTX 3090 perf, power: 'can you accept it?' (full post)

EVGA loses truck full of GeForce RTX 30 series GPUs, someone stole it

Anthony Garreffa | Nov 7, 2021 7:23 PM CST

EVGA has lost an entire truckload of GeForce RTX 30 series graphics cards, where a thief has stolen a truck carrying EVGA graphics cards being taken to its distribution center in California.

EVGA loses truck full of GeForce RTX 30 series GPUs, someone stole it

The truck was filled with EVGA GeForce RTX 30 series graphics cards including GeForce RTX 3060 through to the GeForce RTX 3090, with MSRPs of those cards ranging between $329 and $1960. We're talking about a gigantic amount of the most sought-after graphics cards -- an entire truckload of them -- out in the wild.

EVGA was quick to action, so if you purchase a stolen EVGA GeForce RTX 30 series graphics card you will want to ask for proof of purchase and a receipt -- because if you purchase that card and register for warranty, it won't happen. EVGA will also then investigate that, and want to know how you came to own a stolen graphics card.

Continue reading: EVGA loses truck full of GeForce RTX 30 series GPUs, someone stole it (full post)

AMD's next-gen MCM-based GPU, Instinct MI200 accelerator, teased again

Anthony Garreffa | Nov 6, 2021 7:41 PM CDT

AMD CEO and President Dr. Lisa Su herself teased the next-gen AMD Instinct accelerator, as well as next-gen AMD EPYC CPUs that will be unveiled during their Accelerated Data Center Premiere on November 8 at 11AM ET.

AMD's next-gen MCM-based GPU, Instinct MI200 accelerator, teased again

In a new tweet, AMD CEO Dr. Lisa Su said that she was "looking forward to showcasing" AMD's upcoming EPYC and Instinct accelerators and that the world should join her and AMD on November 8 at 11 AM ET for the big unveiling event. We should see the new AMD Instinct MI250X accelerator, which will be the world's first MCM-based (multi-chip module) GPU with up to 128GB of super-fast HBM2e memory.

AMD's purported Instinct MI250X accelerator will be the first MCM-based GPU, with the codename Aldebaran GPU featuring 2 x Compute Tiles with 110 Compute Units and 128GB of HBM2e memory that should spit out a considerable amount of memory bandwidth per second. We should see a gigantic 383 TFLOPs of FP16 compute performance, which blows away the Instinct MI100 with its 185 TFLOPs of compute performance.

Continue reading: AMD's next-gen MCM-based GPU, Instinct MI200 accelerator, teased again (full post)