【执行测试错误】：no kernel image is available for execution on the device

[RoadTLife]

执行命令：
python tests/test_flash_mla.py
错误信息：

b=128, s_q=1, mean_sk=4096, h_q=16, h_kv=1, d=576, dv=512, causal=True, varlen=False
CUDA error (csrc/flash_fwd_mla_metadata.cu:76): no kernel image is available for execution on the device

环境情况：

• CUDA版本：12.4
• 显卡：A100

[defei-coder]

only support sm90(Hopper+)

[RoadTLife]

这个库是那些原因，导致只能在H系列上跑了呢？

[defei-coder]

this repo used features only supported by sm90, such as WGMM, and this only built with sm_90 arch.