intel/compiler: Restrict cs_threads to 64

Our current GPGPU_WALKER code only supports up to 64 threads.

On HSW we could use up to 70 and TGL up to 112, but only if the walker
is adjusted so the width does not exceed 64. Work to support this is
in progress.

Previous to this change, we might try to downgrade to SIMD8 if the
SIMD16 shader spilled. Since HSW and TGL have the max number of
threads above 64, we would then try to emit an invalid GPGPU walker

Fixes: 93204506 ("i965/cs: Emit compute shader code and upload programs")
Signed-off-by: Jordan Justen's avatarJordan Justen <>
Reviewed-by: Jason Ekstrand's avatarJason Ekstrand <>
Reviewed-by: Paulo Zanoni's avatarPaulo Zanoni <>
Tested-by: Paulo Zanoni's avatarPaulo Zanoni <>
parent 09323634
......@@ -8841,8 +8841,10 @@ brw_compile_cs(const struct brw_compiler *compiler, void *log_data,
src_shader->info.cs.local_size[0] * src_shader->info.cs.local_size[1] *
/* Limit max_threads to 64 for the GPGPU_WALKER command */
const uint32_t max_threads = MIN2(64, compiler->devinfo->max_cs_threads);
unsigned min_dispatch_width =
DIV_ROUND_UP(local_workgroup_size, compiler->devinfo->max_cs_threads);
DIV_ROUND_UP(local_workgroup_size, max_threads);
min_dispatch_width = MAX2(8, min_dispatch_width);
min_dispatch_width = util_next_power_of_two(min_dispatch_width);
assert(min_dispatch_width <= 32);
Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment