Metraipolis › archive › dc_ec5364ef18e7
Hand-Written PTX Tensor-Core GEMM Kernels: A Multi-Precision Study on NVIDIA L4 — arXiv:2608.10103
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.10103
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-19T06:08:54.082Z
- SHA-256 of the archived text
2fe94016044d0e41c1ac29a20b0faed914a443e9aca48725eb34cf8c5cd2dcbb- Stable URL of this page
https://metraipolis.com/documento/dc_ec5364ef18e7
How to cite
Hand-Written PTX Tensor-Core GEMM Kernels: A Multi-Precision Study on NVIDIA L4 — arXiv:2608.10103 — arXiv (en), https://arxiv.org/abs/2608.10103 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ec5364ef18e7 (Metraipolis archive dc_ec5364ef18e7, sha256 2fe94016044d0e41c1ac29a20b0faed914a443e9aca48725eb34cf8c5cd2dcbb)
BibTeX
@misc{dcec5364ef18e7,
title = {Hand-Written PTX Tensor-Core GEMM Kernels: A Multi-Precision Study on NVIDIA L4 — arXiv:2608.10103},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.10103},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ec5364ef18e7; sha256 2fe94016044d0e41c1ac29a20b0faed914a443e9aca48725eb34cf8c5cd2dcbb},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.10103. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
High-performance Tensor Core kernels rely on a low-level PTX pipeline built from asynchronous data movement with this http URL , warp-level matrix loads with ldmatrix, and matrix multiply-accumulate operations with this http URL . However, most application code accesses Tensor Cores indirectly through the WMMA C++ API. This paper asks a focused, practical question: when does replacing WMMA with hand-written PTX actually pay off? To answer this question, we conduct a controlled, single-GPU study on an NVIDIA L4 GPU (Ada, SM89), comparing double-buffered WMMA baselines with a family of hand-written PTX GEMM kernels across FP16, INT8, and INT4 arithmetic and square problem sizes from $N=512$ to $N=8192$. Every kernel is profiled with Nsight Compute across the full metric set, and PTX speedups are reported relative to the corresponding same-precision WMMA baseline. Hand-written PTX provides no end-to-end speedup for FP16, because its instruction-level gains are offset by operand-packing overhead. In contrast, the PTX kernels achieve consistent speedups of 1.4x-1.8x for INT8, driven primarily by lower instruction counts and better global-memory coalescing, and 2.9x-4.3x for INT4, where native mma.sync.m16n8k64.s4 execution avoids the software-emulated sequence used by the WMMA path. Relative to the FP16 WMMA baseline, the best quantized kernels reach 34.4x (INT8) and 98.7x (INT4) at $N=8192$. Across these experiments, occupancy is a poor predictor of throughput. For large matrices, performance instead tracks memory-system behavior -- particularly global-load coalescing and DRAM-active cycles -- more closely than Tensor Core utilization. These results identify the precisions and operating regimes in which the additional complexity of hand-written PTX is justified. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-12 in cs.AI). Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.10103 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-19T05:57:28.334Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-12?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).