Jagadish Krishnamoorthy

icon indicating a website link https://jaglinux.github.io

icon company AMD icon location San Ramon, CA icon location AI frameworks, AI models engineer. More projects at https://github.com/jaglinux

Results 2 issues of Jagadish Krishnamoorthy

[Bug Fix] Support threads_per_head < 64 for wavefront size of 64

1 comment

When launching apply_rotary_pos_half kernel, only threads_per_head of 64 is supported for wavefront size of 64. This change adds support for threads_per_head < 64 such as 4, 8, 16. Remove the...

[Do NOT MERGE] [release/2.5] Enable tf32 testing on test_nn

43 comment

1. In Context.cpp, for the variable allow_tf32 remove static const variable type to non const variable. This allows us to capture the env variable HIPBLASLT_ALLOW_TF32 changes. 2. Add ROCm arch...