mirror of
https://github.com/zenlm/engine.git
synced 2026-10-30 15:00:05 +00:00
Eric BuehlerandGitHub
863f33de71
* Add chunking to naive attn * Add it to cublaslt as well * Even better! * Clippy * Remove prompt chunksize apis * Prompt chunksize logging * Format * Determine based on mask type * Max seq len in device map * Max seq len in device map * Format
42 lines
1.5 KiB
Plaintext
42 lines
1.5 KiB
Plaintext
#include <cassert>
|
|
#include <cstdint>
|
|
#include <cuda.h>
|
|
|
|
extern "C" void launch_fp8_to_f32_kernel(const uint8_t *d_input,
|
|
float *d_output, size_t num_elements,
|
|
cudaStream_t stream) {
|
|
assert(false);
|
|
}
|
|
|
|
extern "C" void launch_fp8_to_f16_kernel(const uint8_t *d_input,
|
|
uint16_t *d_output,
|
|
size_t num_elements,
|
|
cudaStream_t stream) {
|
|
assert(false);
|
|
}
|
|
|
|
extern "C" void launch_fp8_to_bf16_kernel(const uint8_t *d_input,
|
|
uint16_t *d_output,
|
|
size_t num_elements,
|
|
cudaStream_t stream) {
|
|
assert(false);
|
|
}
|
|
|
|
extern "C" void launch_f32_to_fp8_kernel(const float *d_input,
|
|
uint8_t *d_output, size_t num_elements,
|
|
cudaStream_t stream) {
|
|
assert(false);
|
|
}
|
|
|
|
extern "C" void launch_f16_to_fp8_kernel(const uint16_t *d_input,
|
|
uint8_t *d_output, size_t num_elements,
|
|
cudaStream_t stream) {
|
|
assert(false);
|
|
}
|
|
|
|
extern "C" void launch_bf16_to_fp8_kernel(const uint16_t *d_input,
|
|
uint8_t *d_output,
|
|
size_t num_elements,
|
|
cudaStream_t stream) {
|
|
assert(false);
|
|
} |