79f26e9e12
* vulkan: Add bfloat16 support This adds bfloat16 matrix multiply support based on VK_KHR_shader_bfloat16. The extension is required for coopmat multiply support, but matrix-vector multiply trivially promotes bf16 to fp32 and doesn't require the extension. The copy/get_rows shaders also don't require the extension. It's probably possible to fall back to non-coopmat and promote to fp32 when the extension isn't supported, but this change doesn't do that. The coopmat support also requires a glslc that supports the extension, which currently requires a custom build. * vulkan: Support bf16 tensors without the bf16 extension or coopmat support Compile a variant of the scalar mul_mm shader that will promote the bf16 values to float, and use that when either the bf16 extension or the coopmat extensions aren't available. * vulkan: bfloat16 fixes (really works without bfloat16 support now) * vulkan: fix spirv-val failure and reenable -O
50 lines
1.5 KiB
Plaintext
50 lines
1.5 KiB
Plaintext
#version 450
|
|
|
|
#include "types.comp"
|
|
#include "generic_unary_head.comp"
|
|
|
|
#extension GL_EXT_control_flow_attributes : require
|
|
|
|
const uint num_threads = 128;
|
|
|
|
layout(local_size_x = num_threads, local_size_y = 1, local_size_z = 1) in;
|
|
|
|
void main() {
|
|
uint idx = get_idx();
|
|
|
|
// num_threads * num_iter must equal 512, to match the wg_denoms and get_idx calculation
|
|
const uint num_iter = 4;
|
|
|
|
// fast path for when all four iterations are in-bounds
|
|
if (idx + (num_iter-1)*num_threads < p.ne) {
|
|
[[unroll]] for (uint i = 0; i < num_iter; ++i) {
|
|
|
|
#if defined(DATA_D_BF16)
|
|
float f = float(data_a[get_aoffset() + idx]);
|
|
data_d[get_doffset() + idx] = D_TYPE(fp32_to_bf16(f));
|
|
#elif !defined(OPTIMIZATION_ERROR_WORKAROUND)
|
|
data_d[get_doffset() + idx] = D_TYPE(data_a[get_aoffset() + idx]);
|
|
#else
|
|
data_d[get_doffset() + idx] = data_a[get_aoffset() + idx];
|
|
#endif
|
|
idx += num_threads;
|
|
}
|
|
} else {
|
|
[[unroll]] for (uint i = 0; i < num_iter; ++i) {
|
|
if (idx >= p.ne) {
|
|
continue;
|
|
}
|
|
|
|
#if defined(DATA_D_BF16)
|
|
float f = float(data_a[get_aoffset() + idx]);
|
|
data_d[get_doffset() + idx] = D_TYPE(fp32_to_bf16(f));
|
|
#elif !defined(OPTIMIZATION_ERROR_WORKAROUND)
|
|
data_d[get_doffset() + idx] = D_TYPE(data_a[get_aoffset() + idx]);
|
|
#else
|
|
data_d[get_doffset() + idx] = data_a[get_aoffset() + idx];
|
|
#endif
|
|
idx += num_threads;
|
|
}
|
|
}
|
|
}
|