med-mastodon.com is one of the many independent Mastodon servers you can use to participate in the fediverse.
Medical community on Mastodon

Administered by:

Server stats:

345
active users

#cuda

2 posts2 participants0 posts today

#NVIDIA Bringing #CUDA To #RISCV
NVIDIA's drivers and CUDA software stack are predominantly supported on x86_64 and AArch64 systems but in the past was supported on IBM POWER. This week at the RISC-V Summit China event, NVIDIA's Frans Sijstermans announced that CUDA will be coming to RISC-V.
#AMD for their part with the upstream #opensource #AMDKFD kernel compute driver can already build on RISC-V and the #ROCm user-space components can also be built on RISC-V.
phoronix.com/news/NVIDIA-CUDA-

www.phoronix.comNVIDIA Bringing CUDA To RISC-VNVIDIA announced this week that they are bringing their CUDA software to RISC-V processors.

#GPUHammer is the first attack to show #Rowhammer bit flips on #GPU memories, specifically on a GDDR6 memory in an #NVIDIA A6000 GPU. Our attacks induce bit flips across all tested DRAM banks, despite in-DRAM defenses like TRR, using user-level #CUDA #code. These bit flips allow a malicious GPU user to tamper with another user’s data on the GPU in shared, time-sliced environments. In a proof-of-concept, we use these bit flips to tamper with a victim’s DNN models and degrade model accuracy from 80% to 0.1%, using a single bit flip. Enabling Error Correction Codes (ECC) can mitigate this risk, but ECC can introduce up to a 10% slowdown for #ML #inference workloads on an #A6000 GPU.

gpuhammer.com/

GPUHammerGPUHammer

New #ZLUDA 5 Preview Released For #CUDA On Non-NVIDIA #GPU
For now this ability to run unmodified CUDA apps on non-#NVIDIA GPUs is focused on #AMD GPUs of the #Radeon RX 5000 series and newer, which is AMD Radeon GPUs with #ROCm. Besides CUDA code samples, GeekBench has been one of the early targets for testing.
phoronix.com/news/ZLUDA-5-prev

www.phoronix.comNew ZLUDA 5 Preview Released For CUDA On Non-NVIDIA GPUsZLUDA Version 5-preview.43 was released today as this open-source CUDA implementation for use on non-NVIDIA GPUs, with one of the current focuses being on enabling CUDA on AMD Radeon GPUs with ROCm.

#ZLUDA Making Progress In 2025 On Bringing #CUDA To Non-NVIDIA #GPU
ZLUDA #opensource effort that started half-decade ago as drop-in CUDA implementation for #Intel GPUs and then for several years was funded by ##AMD as a CUDA implementation for #Radeon GPUs atop #ROCm and then open-sourced but then reverted has been continuing to push along a new path since last year. Current take on ZLUDA is a multi-vendor CUDA implementation for non-NVIDIA GPUs for #AI workloads & more.
phoronix.com/news/ZLUDA-Q2-202

www.phoronix.comZLUDA Making Progress In 2025 On Bringing CUDA To Non-NVIDIA GPUs

China’s few remaining weaknesses:

Manufacturing the most advanced chips is a still an issue because ”China lacks a domestic alternative to the cutting-edge lithography tools produced by ASML, a Dutch company. … chip designers [are] reliant on SMIC, a state-owned foundry. “

China may initially struggle awhile with “… the software used by coders to program chips. Nvidia’s platform, called CUDA, is still by far the best in the world. Nearly all AI developers learn how to use it. And it works only with Nvidia’s chips. Switching to an alternative is costly, because it pulls developers out of an enormous network of fellow users that can help solve problems.”

“Huawei has created a substitute for CUDA, called CANN …But the software is years behind Nvidia’s [It’s buggy and unenthusiastic used by local techies. But,] Huawei has beaten the odds before, and it may well do the same again with CANN.”

#china #technology #CUDA
economist.com/business/2025/05

The Economist · Huawei and other Chinese chip firms are catching up fastBy The Economist