-
Notifications
You must be signed in to change notification settings - Fork 0
All issues
Issue creation is restricted in this repository
- #8 · parda21 opened
on Sep 19, 2026
Issues
is:issue state:open
is:issue state:open
Search results
External fp8 baseline: vLLM (fp8 weight-only) tokens/s on the same GPUs
benchmarkmeasurements from a GPUmeasurements from a GPUhelp wantedwe cannot do this one alonewe cannot do this one aloneroadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.Ada (RTX 4070) needs ~17 SM clocks per warp step where Ampere (A16) needs 11, same SASS
help wantedwe cannot do this one alonewe cannot do this one aloneroadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.Benchmark results wanted: post your GPU here
benchmarkmeasurements from a GPUmeasurements from a GPUhelp wantedwe cannot do this one alonewe cannot do this one aloneStatus: Open.#8 In parda21/NWC;Paper
roadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.llama.cpp port: GGML tensor type, CUDA mmv kernel, converter
help wantedwe cannot do this one alonewe cannot do this one aloneroadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.Colab notebook on a free T4 (sm_75)
help wantedwe cannot do this one alonewe cannot do this one aloneroadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.Second model family with perplexity check (Llama 3.1 8B, Mistral 7B)
roadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.Tensor-core accumulation path (mma.m16n8k16) for A100 / H100 SXM
roadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.Measure H100 PCIe and H100 SXM
benchmarkmeasurements from a GPUmeasurements from a GPUhelp wantedwe cannot do this one alonewe cannot do this one aloneroadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.Measure RTX 4080 Super
roadmapplanned work, shown in the README roadmapplanned work, shown in the README roadmapStatus: Open.