Under GPU-CC, LLM serving losses come from a serialized VM–GPU bridge, not compute; simple scheduling and loader changes recover most of the gap on Blackwell.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The Serialized Bridge: Understanding and Recovering LLM Serving Performance under Blackwell GPU Confidential Computing
Under GPU-CC, LLM serving losses come from a serialized VM–GPU bridge, not compute; simple scheduling and loader changes recover most of the gap on Blackwell.