Skip to content

Half of models fails to load due to lack of TensileLibrary.dat for gfx906 in hipBLAS image #7869

@Expro

Description

@Expro

LocalAI version:
v3.9.0

Environment, CPU architecture, OS, and Version:
hipBLAS container image

Describe the bug
Many models (especially MoE one) fails to load on gfx906 GPUs due to lack of TensileLibrary.dat. Relevant line from log:

DEBUG GRPC stderr id="Nemotron-3-Nano-30B-A3B-127.0.0.1:43781" line="rocBLAS error: Cannot read /opt/rocm-6.4.3/lib/rocblas/library/TensileLibrary.dat: No such file or directory for GPU arch : gfx906" caller={caller.file="/build/pkg/model/process.go" caller.L=146 }

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions