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:

373
active users

#Nvidia

30 posts28 participants0 posts today
Continued thread

Eine taktische Allianz mit Amerikas neuer Kleptokratie in Verbindung mit seinen eigenen Interessenkonflikten ist Trumps Markenzeichen. Es ist eine Sache, wenn dies lediglich korrupt ist, aber es ist weitaus schlimmer, wenn dadurch die militärische Sicherheit Amerikas verkauft wird.

#Nvidia #China #USA #Trump

Nicht Chinas fürchte ich.

Apple-KI-Framework MLX: Künftig Support für Nvidias CUDA

Zwar laufen in Macs keine Nvidia-GPUs mehr, dennoch soll Apples MLX nun bald auch dort laufen. Das macht interessante Portierungen möglich.

heise.de/news/Apple-KI-Framewo

heise online · Apple-KI-Framework MLX: Künftig Support für Nvidias CUDABy Ben Schwan
#Apple#CUDA#IT

Filed #FreeBSD Bug288300 and opened corresponding review D51407 to update #nvidia drivers to latest Production Branch 570.172.08.

bugs.freebsd.org/bugzilla/show

reviews.freebsd.org/D51407

Details:
nvidia.com/en-us/drivers/detai

Linux counterparts for x11/linux-nvidia-libs:
nvidia.com/en-us/drivers/detai

Note that x11/nvidia-settings and x11/nvidia-xconfig would be updated separately, as upstream (nvidia GitHub repo) is not yet in sync.

bugs.freebsd.orgMaking sure you're not a bot!

Provided patch to fix building x11/nvidia-driver* on #FreeBSD src main at commit 4dd828c80828637452a8a4e07a64e294c82e5d8b and later on Bug 288236.
bugs.freebsd.org/bugzilla/show
Also opened review for the patch as D51340, which is already accepted by 2 reviewers.
reviews.freebsd.org/D51340
As discussed with Konstantin on Bug 288236, turned out that formerly constant macros are now really "variables", the fix is applied conditionally.
As the above-mentioned commit does NOT have corresponding version bump, there are broken time window on main branch only before above-mentioned commit around 10 days.

bugs.freebsd.orgMaking sure you're not a bot!

#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