AMD Instinct MI300X
gfx942
Records held
23
Runs
2,099
Operations
4314 implementations
Records held on this GPU
Operation · workload
History
Record
Implementation
Since
Show all 23 records ›Showing all 23 records ⌄
Coverage by operation family
Family
Operations
Runs
With source
gemm
1
1,764
1,764
moe
1
237
237
mla-attention
1
79
79
trimul
1
19
19
Serving evidence
End-to-end serving results measured on this hardware — a separate corpus with its own resolver, never ranked against the kernel records above.
Model · workload
Scenario
Best reported
Runs
Llama-2-70B (MLPerf reference)llama2-70b-99 · Server
server
24,861 tok/s
4
Llama-2-70B (MLPerf reference)llama2-70b-99 · Offline
offline
27,854 tok/s
4
Llama-2-70B (MLPerf reference)llama2-70b-99 · Offline
offline
27,854 tok/s
4
Llama-2-70B (MLPerf reference)llama2-70b-99 · Server
server
24,861 tok/s
4
Llama-2-70B (MLPerf reference)llama2-70b-99 · Interactive
interactive
8,840 tok/s
2
Mixtral-8x7B (MLPerf reference)mixtral-8x7b · Server
server
47,804 tok/s
2
Mixtral-8x7B (MLPerf reference)mixtral-8x7b · Offline
offline
53,334 tok/s
2
Llama-2-70B (MLPerf reference)llama2-70b-99 · Server
server
25,463 tok/s
2
48 serving runs across 28 comparison groups · Open in the serving resolver →
Record activity
No record transitions on this GPU in the trailing year.
Sources: GPU MODE KernelBot (2025-09-23) · June 9 Researcher Reciprocity License v1.0last observed 2025-09-23