Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

hexagon: add neg, exp, sigmoid, softplus ops, cont, repeat ops (#20701)

Browse Source 
Add element-wise unary ops needed by Qwen 3.5's DeltaNet linear
attention layers. These ops follow the existing unary-ops pattern
with VTCM DMA double-buffering.

- neg: negate via scale by -1.0
- exp: uses existing hvx_exp_f32 HVX intrinsics
- sigmoid: uses existing hvx_sigmoid_f32_aa HVX intrinsics
- softplus: log(1 + exp(x)) scalar fallback
- CONT reuses the existing CPY infrastructure since making a tensor
  contiguous is equivalent to a same-type copy.
- REPEAT implements tiled memory copy with multi-threaded execution via
  the worker pool, supporting f32 and f16 types. The kernel parallelizes
  across output rows and uses memcpy for each tile.

Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com>
This commit is contained in:
Krishna Sridhar
2026-03-17 15:34:36 -07:00
committed by GitHub
co-authored by Max Krasnyansky
parent 892e3c333a
commit cf23ee2447
11 changed files with 441 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-hexagon/htp/main.c
+45
View File
@@ -516,6 +516,39 @@ static void proc_cpy_req(struct htp_context * ctx, struct htp_general_req * req,
send_htp_rsp(ctx, req->op, rsp_status, rsp_bufs, 1, &prof);
}
static void proc_repeat_req(struct htp_context * ctx, struct htp_general_req * req, struct dspqueue_buffer * bufs) {
struct dspqueue_buffer rsp_bufs[1];
// We had written to the output buffer, we'd also need to flush it
rsp_bufs[0].fd = bufs[1].fd;
rsp_bufs[0].ptr = bufs[1].ptr;
rsp_bufs[0].offset = bufs[1].offset;
rsp_bufs[0].size = bufs[1].size;
rsp_bufs[0].flags = (DSPQUEUE_BUFFER_FLAG_FLUSH_SENDER | // Flush HTP
DSPQUEUE_BUFFER_FLAG_INVALIDATE_RECIPIENT); // Invalidate CPU
// Setup Op context
struct htp_ops_context octx = { 0 };
octx.ctx = ctx;
octx.src0 = req->src0;
octx.dst = req->dst;
octx.flags = req->flags;
octx.op = req->op;
// Update data pointers
octx.src0.data = (uint32_t) bufs[0].ptr;
octx.dst.data = (uint32_t) bufs[1].ptr;
octx.n_threads = ctx->n_threads;
struct profile_data prof;
profile_start(&prof);
uint32_t rsp_status = op_repeat(&octx);
profile_stop(&prof);
send_htp_rsp(ctx, req->op, rsp_status, rsp_bufs, 1, &prof);
}
static void proc_get_rows_req(struct htp_context * ctx, struct htp_general_req * req, struct dspqueue_buffer * bufs) {
struct dspqueue_buffer rsp_bufs[1];
@@ -1090,6 +1123,10 @@ static void htp_packet_callback(dspqueue_t queue, int error, void * context) {
case HTP_OP_SQR:
case HTP_OP_SQRT:
case HTP_OP_UNARY_NEG:
case HTP_OP_UNARY_EXP:
case HTP_OP_UNARY_SIGMOID:
case HTP_OP_UNARY_SOFTPLUS:
if (n_bufs != 2) {
FARF(ERROR, "Bad unary-req buffer list");
continue;
@@ -1175,6 +1212,14 @@ static void htp_packet_callback(dspqueue_t queue, int error, void * context) {
proc_cpy_req(ctx, &req, bufs);
break;
case HTP_OP_REPEAT:
if (n_bufs != 2) {
FARF(ERROR, "Bad repeat-req buffer list");
continue;
}
proc_repeat_req(ctx, &req, bufs);
break;
case HTP_OP_ARGSORT:
if (n_bufs != 2) {
FARF(ERROR, "Bad argsort-req buffer list");