CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

cont : tmp

Browse Source
This commit is contained in:
Georgi Gerganov
2025-07-29 07:42:55 +03:00
parent 477d43988a
commit 381879e0ac
1 changed files with 8 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
ggml/src/ggml-cpu/repack.cpp
View File

@@ -1534,15 +1534,22 @@ template <typename BLOC_TYPE, int64_t INTER_SIZE, int64_t NB_COLS, ggml_type PAR
continue;
}
#if 1
memcpy( htmp + i02*nbht2 + htpe[i02]*nbht1,
(char *) src1->data + i12*nb12 + (i20%ne11)*nb11,
ggml_row_size(GGML_TYPE_F32, ne10));
#else
from_float(
(float *)((char *) src1->data + i12*nb12 + (i20%ne11)*nb11),
(void *) (hsrc1 + htpe[i02]*nbh11 + i02*nbh12), ne10);
#endif
hids[i12*ne20 + i20] = i02*ne12 + htpe[i02];
htpe[i02]++;
}
}
#if 1
// htmp (float32) => hsrc1 (param type)
for (int64_t i02 = 0; i02 < ne02; ++i02) { // n_expert
if (i02 % nth != ith) {
@@ -1563,6 +1570,7 @@ template <typename BLOC_TYPE, int64_t INTER_SIZE, int64_t NB_COLS, ggml_type PAR
(void *) (hsrc1 + i11*nbh11 + i02*nbh12), ne10);
}
}
#endif
ggml_barrier(params->threadpool);