sleepy/llama.cpp
1
0
Fork 0
Code Issues 6 Pull Requests 1 Actions 188 Packages Projects Releases Wiki Activity

hexagon: optimization for HMX mat_mul (#21554)

Browse Source 
* hexagon: add async HMX worker

Introduce hmx-worker (dedicated thread for HMX compute) to overlap HMX
matmul with HVX dequant/DMA stages in the pipeline path, replacing the
previous synchronous HMX calls that blocked the main thread.

* hexagon: cost-based VTCM chunk search for out-stationary matmul

* hexagon: fix futex race in hmx_worker_drain
Store the boolean to local variable avoid atomic load twice

* hex-mm: hmx optimize scatter/transpose and use HMX intrinsics

* hex-vmem: drop vmem limit a touch under 3GB on v73

* hexagon: add fwd declaration of htp_context

* hex-hmx: replace hmx-worker with hmx-queue that mimics dma-queue interface

Simplifies the overall implemantion, reduces thread wakeup roundtrips.

* hex-mm: add debug log to hmx work func called from hmx-queue

* Update hmx-queue.h

Co-authored-by: Max Krasnyansky <max.krasnyansky@gmail.com>

---------

Co-authored-by: Kim-Chyan Gan <kgan@qti.qualcomm.com>
Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com>
Co-authored-by: Max Krasnyansky <max.krasnyansky@gmail.com>
This commit is contained in:
Yiwei Shao
2026-04-14 14:09:03 -07:00
committed by GitHub
parent fae3a28070
commit 5d14e5d19b
10 changed files with 589 additions and 197 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-hexagon/htp/hex-utils.h
+13 -2
View File
@@ -31,6 +31,14 @@ static inline uint64_t hex_get_pktcnt() {
return pktcnt;
}
static inline uint32_t hex_ceil_pow2(uint32_t x) {
if (x <= 1) { return 1; }
int p = 2;
x--;
while (x >>= 1) { p <<= 1; }
return p;
}
static inline size_t hmx_ceil_div(size_t num, size_t den) {
return (num + den - 1) / den;
}
@@ -73,8 +81,7 @@ static inline void hex_l2fetch(const void * p, uint32_t width, uint32_t stride,
#define HEX_L2_LINE_SIZE 64
#define HEX_L2_FLUSH_SIZE (128 * 1024)
static inline void hex_l2flush(void * addr, size_t size)
{
static inline void hex_l2flush(void * addr, size_t size) {
if (size > HEX_L2_FLUSH_SIZE) {
qurt_mem_cache_clean((qurt_addr_t) 0, 0, QURT_MEM_CACHE_FLUSH_INVALIDATE_ALL, QURT_MEM_DCACHE);
} else {
@@ -89,4 +96,8 @@ static inline void hex_l2flush(void * addr, size_t size)
}
}
static inline void hex_pause() {
asm volatile(" pause(#255)\n");
}
#endif /* HEX_UTILS_H */