Alphane MoonM to HardwareEnglish • 6 months agoAMD signs $4.9 billion deal to challenge Nvidia’s AI infrastructure leadarstechnica.comexternal-linkmessage-square3arrow-up119arrow-down11cross-posted to: business
arrow-up118arrow-down1external-linkAMD signs $4.9 billion deal to challenge Nvidia’s AI infrastructure leadarstechnica.comAlphane MoonM to HardwareEnglish • 6 months agomessage-square3cross-posted to: business
minus-squareAgent Karyolink2•6 months agoCurious how well this works with some open source projects that have separate implementationa for Nvidia and AMD/Intel GPUs. Although it sounds like the devs would have to be aware of AdaptiveCpp.
minus-squareVerdant BananalinkEnglish2•edit-26 months agoPortProton handles it just fine https://linux-gaming.ru/t/ustanovka-portproton-ispolzovanie-wine-proton-bez-steam/24 even has the setting for it to make use of it the GPU packages that are dependencies for AdaptiveCPP is as follows: hip-runtime-amd rocm-llvm nvidia-utils cuda open source nouveau will not work for it
Curious how well this works with some open source projects that have separate implementationa for Nvidia and AMD/Intel GPUs.
Although it sounds like the devs would have to be aware of AdaptiveCpp.
PortProton handles it just fine
https://linux-gaming.ru/t/ustanovka-portproton-ispolzovanie-wine-proton-bez-steam/24
even has the setting for it to make use of it
the GPU packages that are dependencies for AdaptiveCPP is as follows:
hip-runtime-amd
rocm-llvm
nvidia-utils
cuda
open source nouveau will not work for it