GPU Pipeline Hazards: Scoreboard Stalls and Warp Divergence
A fine kernel can still profile poorly: warp divergence taxes throughput ~1/N; alone on a long scoreboard wait, SM issue util can hit 0%. Interactive demos.
5 min readConcept
Explore machine learning concepts related to scoreboard. Clear explanations and practical insights.
A fine kernel can still profile poorly: warp divergence taxes throughput ~1/N; alone on a long scoreboard wait, SM issue util can hit 0%. Interactive demos.