Pull requests / #1104
#1104 docs: visual community RTX PRO 6000 concurrency results (64K/128K)
closed · @CC-David-CC · 0 comentários · No GitHub
BenchmarksNVIDIA / CUDADocumentation
Descrição
Adds a visual community report for seven Flash-Next variants on RTX PRO 6000 Blackwell 96 GB, Ryzen 9 7950X and 128 GB RAM; FP16 KV, 64K/128K actual input + 512 output  118 completed cohorts / 800 requests. Reports aggregate decode and effective throughput including prefill, placement/memory, failures and unmeasured points. Experimental fork cd9fcca, mostly single observations. Includes the overview PNG/SVG, reproducible plot script, compact CSV and community index entry. Full historical evidence stays linked in the fork: https://github.com/CC-David-CC/Strata-a5500/tree/453310bce60466301e25cb69e77a142a17880605/bench/concurrency-20261005
No site
Links install, modelos, releases.