ZLUDA + ROCm runs CUDA-targeted Windows apps on AMD GPUs at ~3% slower than native
_underlines_ · reddit · 2026-09-14
A Reddit post details running CUDA-targeted Windows applications on AMD GPUs via ZLUDA with ROCm/HIP—notable since AMD officially had to stop open-sourcing work in this direction.
Benchmark result: only about 3% slower than native CUDA, making it a solid drop-in for CUDA-restricted workloads. The author hopes it finds its way into many projects and optimizations locked to CUDA.
More from Infra
- Would LITE accepting MRVL's merger offer prove its moat was overstated? — BenBajarin · 2026-10-30
- Claymation body horror on 8GB: three wrong settings, not the GPU, made local renders slow — roro9- · 2026-10-30
- durable-object-migrator: zero-downtime exports for Cloudflare Durable Objects — samgoodwin89 · 2026-10-30
- An unnamed firm already runs a fairly scaled dedicated vehicle for compute lending, says analyst — matt_slotnick · 2026-10-30
- GPU break-even throughput vs open-weight token prices up 25-83% since May — bookwormengr · 2026-10-30
- Sam Altman says the chip supply chain's invisible hand is failing — and DRAM is why — SnoozeDoggyDog · 2026-10-30