run into dead loop when tuning the tma persistent kernel #4332
Description
I'm try running matmul_kernel_tma_persistent this kernel with various configs, but it often get into dead loop with GPU 100% utilization.
Is that a bug?
{{ message }}
I'm try running matmul_kernel_tma_persistent this kernel with various configs, but it often get into dead loop with GPU 100% utilization.
Is that a bug?