Nvidia's RTX Pro 6000 has 96GB of VRAM and 600W of power
(theverge.com)
Nvidia is announcing its RTX Pro Blackwell series of GPUs today, designed to meet the needs of professional designers, developers, data scientists, and creatives.
Nvidia is announcing its RTX Pro Blackwell series of GPUs today, designed to meet the needs of professional designers, developers, data scientists, and creatives.
Decrypting Encrypted files from Akira Ransomware using a bunch of GPUs
(tinyhack.com)
I recently helped a company recover their data from the Akira ransomware without paying the ransom. I’m sharing how I did it, along with the full source code.
I recently helped a company recover their data from the Akira ransomware without paying the ransom. I’m sharing how I did it, along with the full source code.
Speeding up computational lithography with the power and parallelism of GPUs
(semiengineering.com)
A new lithography library brings mask optimization operations to GPUs.
A new lithography library brings mask optimization operations to GPUs.
SVDQuant+NVFP4: 4× Smaller, 3× Faster FLUX with 16-bit Quality on Blackwell GPUs
(hanlab.mit.edu)
With Moore's law slowing down, hardware vendors are shifting toward low-precision inference. NVIDIA's latest Blackwell architecture introduces a new 4-bit floating point format (NVFP4), improving upon the previous MXFP4 format. NVFP4 features more precise scaling factors and a smaller microscaling group size (16 v.s. 32), enabling it to maintain 16-bit model accuracy even at 4-bit precision while delivering 4× higher peak performance.
With Moore's law slowing down, hardware vendors are shifting toward low-precision inference. NVIDIA's latest Blackwell architecture introduces a new 4-bit floating point format (NVFP4), improving upon the previous MXFP4 format. NVFP4 features more precise scaling factors and a smaller microscaling group size (16 v.s. 32), enabling it to maintain 16-bit model accuracy even at 4-bit precision while delivering 4× higher peak performance.
Linux Introducing a Standardized Way of Informing User-Space over Hung GPUs
(phoronix.com)
The upcoming Linux 6.15 kernel is set to finally introduce a standardized way of informing user-space of GPUs becoming hung or otherwise unresponsive.
The upcoming Linux 6.15 kernel is set to finally introduce a standardized way of informing user-space of GPUs becoming hung or otherwise unresponsive.
We were wrong about GPUs
(fly.io)
We’re building a public cloud, on hardware we own. We raised money to do that, and to place some bets; one of them: GPU-enabling our customers. A progress report: GPUs aren’t going anywhere, but: GPUs aren’t going anywhere.
We’re building a public cloud, on hardware we own. We raised money to do that, and to place some bets; one of them: GPU-enabling our customers. A progress report: GPUs aren’t going anywhere, but: GPUs aren’t going anywhere.
800x Speed Boost on Nvidia GPUs
(scmp.com)
A high-performance algorithm that could solve complicated material design problems on consumer GPUs has been developed by Chinese researchers, achieving a groundbreaking 800-fold increase in speed over traditional methods.
A high-performance algorithm that could solve complicated material design problems on consumer GPUs has been developed by Chinese researchers, achieving a groundbreaking 800-fold increase in speed over traditional methods.
Tiny Corp Nearing "Completely Sovereign" Compute Stack for AMD GPUs with Tinygr
(phoronix.com)
George Hotz' Tiny Corp that develops the Tinygrad neural network framework and sells the Tinybox NVIDIA and AMD powered AI workstations is nearing a "completely sovereign" software stack for GPU compute on AMD.
George Hotz' Tiny Corp that develops the Tinygrad neural network framework and sells the Tinybox NVIDIA and AMD powered AI workstations is nearing a "completely sovereign" software stack for GPU compute on AMD.
If GPUs Are So Good, Why Do We Still Use CPUs at All?
(codingstuff.substack.com)
There’s this old video from 2009 that’s been going viral on Twitter recently. Its supposed to give viewers an intuition of the difference between CPUs and GPUs.
There’s this old video from 2009 that’s been going viral on Twitter recently. Its supposed to give viewers an intuition of the difference between CPUs and GPUs.
Nvidia's Christmas Present: GB300 and B300 – Reasoning Inference, Amazon, Memory
(semianalysis.com)
Merry Christmas has come thanks to Santa Huang. Despite Nvidia’s Blackwell GPU’s having multiple delays, discussed here, and numerous times through the Accelerator Model due to silicon, packaging, and backplane issues, that hasn’t stopped Nvidia from continuing their relentless march.
Merry Christmas has come thanks to Santa Huang. Despite Nvidia’s Blackwell GPU’s having multiple delays, discussed here, and numerous times through the Accelerator Model due to silicon, packaging, and backplane issues, that hasn’t stopped Nvidia from continuing their relentless march.
GPU Glossary
(modal.com)
We wrote this glossary to solve a problem we ran into working with GPUs here at Modal : the documentation is fragmented, making it difficult to connect concepts at different levels of the stack, like Streaming Multiprocessor Architecture , Compute Capability , and nvcc compiler flags .
We wrote this glossary to solve a problem we ran into working with GPUs here at Modal : the documentation is fragmented, making it difficult to connect concepts at different levels of the stack, like Streaming Multiprocessor Architecture , Compute Capability , and nvcc compiler flags .
Ask HN: How Are AMD GPUs/ROCm for Machine Learning in 2024?
(ycombinator.com)
Ask HN: How Are AMD GPUs/ROCm for Machine Learning in 2024?
Ask HN: How Are AMD GPUs/ROCm for Machine Learning in 2024?
How AMD Is Taking Standard C/C++ Code to Run Directly on GPUs
(phoronix.com)
Back at the 2024 LLVM Developers' Meeting was an interesting presentation by AMD engineer Joseph Huber for how they have been exploring running common, standard C/C++ code directly on GPUs without having to be adapted for any GPU language / programming dialects or other adaptations.
Back at the 2024 LLVM Developers' Meeting was an interesting presentation by AMD engineer Joseph Huber for how they have been exploring running common, standard C/C++ code directly on GPUs without having to be adapted for any GPU language / programming dialects or other adaptations.
GPU Accelerated Object Storage
(acceleratedcloudstorage.com)
GPU-Optimized Cloud Object Storage for Real-Time Data Access
GPU-Optimized Cloud Object Storage for Real-Time Data Access
AMD Developing Next-Gen Fortran Compiler Based on Flang, Optimized for AMD GPUs
(phoronix.com)
AMD today went public with details on the "AMD Next-Gen Fortran Compiler" as a new Fortran compiler they are working on based on LLVM's Flang.
AMD today went public with details on the "AMD Next-Gen Fortran Compiler" as a new Fortran compiler they are working on based on LLVM's Flang.
Fujitsu, AMD Plan to Pair Monaka CPUs with Instinct GPUs
(theregister.com)
Fujitsu and AMD announced plans on Friday to develop a new, more energy-efficient AI and HPC compute platform that will pair the Japanese tech vendor's next-gen CPUs with the House of Zen's Instinct accelerators.
Fujitsu and AMD announced plans on Friday to develop a new, more energy-efficient AI and HPC compute platform that will pair the Japanese tech vendor's next-gen CPUs with the House of Zen's Instinct accelerators.
Security flaws found in Nvidia GeForce GPUs
(pcworld.com)
Graphics card manufacturer Nvidia is currently issuing a warning to all owners of GeForce GPUs. According to an Nvidia security bulletin, several security vulnerabilities requiring urgent attention have been discovered in the company’s own display drivers and other software.
Graphics card manufacturer Nvidia is currently issuing a warning to all owners of GeForce GPUs. According to an Nvidia security bulletin, several security vulnerabilities requiring urgent attention have been discovered in the company’s own display drivers and other software.
Beelink Open-Sourced Multi-Functional GPU Ex Docking Station for Mini PCs
(gamingonlinux.com)
Hardware vendor Beelink have "open-sourced" their Multi-Functional EX Docking Station that you can use to extend their GTi Ultra Series mini PCs with a full GPU.
Hardware vendor Beelink have "open-sourced" their Multi-Functional EX Docking Station that you can use to extend their GTi Ultra Series mini PCs with a full GPU.
Ask HN: What is the cheapest ($/hour) H100 renting service with quick spin-up?
(ycombinator.com)
Something that can be programmatically spun up and turned off in seconds would be ideal.
Something that can be programmatically spun up and turned off in seconds would be ideal.
Benchmarking Llama 3.1 405B on 8x AMD MI300X GPUs
(dstack.ai)
At dstack, we've been adding support for AMD GPUs with SSH fleets, so we saw this as a great chance to test our integration by benchmarking AMD GPUs. Our friends at Hot Aisle, who build top-tier bare metal compute for AMD GPUs, kindly provided the hardware for the benchmark.
At dstack, we've been adding support for AMD GPUs with SSH fleets, so we saw this as a great chance to test our integration by benchmarking AMD GPUs. Our friends at Hot Aisle, who build top-tier bare metal compute for AMD GPUs, kindly provided the hardware for the benchmark.