[Feature] Enable AMX FP16 on next generation CPU

intel / xFasterTransformer

Apache License 2.0

355 stars 61 forks source link

[Feature] Enable AMX FP16 on next generation CPU #456

Closed wenhuanh closed 3 months ago

wenhuanh commented 3 months ago

need export env DNNL_MAX_CPU_ISA=AVX512_CORE_AMX_FP16 when running

changqi1 commented 3 months ago

除了compute和compute_resident，其他的kernel也需要支持这个amx fp16的支持

wenhuanh commented 3 months ago

revised and updated.

changqi1 commented 3 months ago

@wenhuanh We will need to check ISA to select AMX_FP16 or AVX512_FP16. But currently this is OK. And you need to give its perf improvement in GNR vs. EMR.

wenhuanh commented 3 months ago

Perf status was updated in related JIRA.