char-fork/llama.cpp

git clone https://git.t4t.associates/char-fork/llama.cpp

Max Krasnyanskyhexagon: MUL_MAT and MUL_MAT_ID rework : 32x32 tiled weight repack, kernel-params, cached graphs (#24954)8be759e6f

char/aliceai-t5-moe
brancheschar/aliceai-t5-moeall refs →
675 B23 linesraw
1// FastRPC IDL interface for GGML HTP
2
3#ifndef HTP_IDL
4#define HTP_IDL
5
6#include "AEEStdDef.idl"
7#include "remote.idl"
8
9struct htp_iface_pmu_conf {
10    uint32 events[8];
11};
12
13interface htp_iface : remote_handle64 {
14    AEEResult start(in uint32 sess_id, in uint64 dsp_queue_id, in uint32 n_hvx, in uint32 n_hmx, in uint64 max_vmem);
15    AEEResult stop();
16    AEEResult mmap(in uint32 fd, in uint32 size);
17    AEEResult munmap(in uint32 fd);
18    AEEResult profiler(in uint32 mode, in htp_iface_pmu_conf pmu);
19    AEEResult etm(in uint32 enable);
20    AEEResult hwinfo(rout uint32 n_threads, rout uint32 n_hvx, rout uint32 n_hmx, rout uint64 vtcm_size);
21};
22
23#endif /* HTP_IDL */