EngineX-Ascend/enginex-ascend-910-llama.cpp
10
0
Fork 0
Code Issues Pull Requests Actions 4 Projects Releases Wiki Activity

musa: add GGML_UNUSED_VARS (#15446)

Browse Source 
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
This commit is contained in:
R0CKSTAR
2025-08-21 11:06:05 +08:00
committed by GitHub
parent 5682a3745f
commit 8ad038c0fd
17 changed files with 113 additions and 148 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
ggml/src/ggml-cuda/fattn-mma-f16.cuh
View File

@@ -767,14 +767,11 @@ static __device__ __forceinline__ void flash_attn_ext_f16_iter(
}
}
#else
GGML_UNUSED(Q_f2); GGML_UNUSED(K_h2); GGML_UNUSED(V_h2);
GGML_UNUSED(mask_h2); GGML_UNUSED(dstk); GGML_UNUSED(dstk_fixup);
GGML_UNUSED(scale); GGML_UNUSED(slope); GGML_UNUSED(logit_softcap);
GGML_UNUSED(ne01); GGML_UNUSED(ne02); GGML_UNUSED(stride_K); GGML_UNUSED(stride_V);
GGML_UNUSED(stride_mask); GGML_UNUSED(tile_K);
GGML_UNUSED(tile_V); GGML_UNUSED(tile_mask); GGML_UNUSED(Q_B);
GGML_UNUSED(VKQ_C); GGML_UNUSED(KQ_max); GGML_UNUSED(KQ_rowsum);
GGML_UNUSED(kb0); GGML_UNUSED(tile_Q);
GGML_UNUSED_VARS(Q_f2, K_h2, V_h2, mask_h2, dstk, dstk_fixup,
scale, slope, logit_softcap, ne01, ne02,
stride_K, stride_V, stride_mask,
tile_Q, tile_K, tile_V, tile_mask,
Q_B, VKQ_C, KQ_max, KQ_rowsum, kb0);
NO_DEVICE_CODE;
#endif // TURING_MMA_AVAILABLE
}
@@ -1236,14 +1233,10 @@ static __device__ __forceinline__ void flash_attn_ext_f16_process_tile(
}
}
#else
GGML_UNUSED(Q_f2); GGML_UNUSED(K_h2); GGML_UNUSED(V_h2);
GGML_UNUSED(mask_h2); GGML_UNUSED(sinks_f);
GGML_UNUSED(dstk); GGML_UNUSED(dstk_fixup);
GGML_UNUSED(scale); GGML_UNUSED(slope); GGML_UNUSED(logit_softcap);
GGML_UNUSED(ne01); GGML_UNUSED(ne02);
GGML_UNUSED(stride_Q1); GGML_UNUSED(stride_Q2);
GGML_UNUSED(stride_K); GGML_UNUSED(stride_V); GGML_UNUSED(stride_mask);
GGML_UNUSED(jt); GGML_UNUSED(kb0_start); GGML_UNUSED(kb0_stop);
GGML_UNUSED_VARS(Q_f2, K_h2, V_h2, mask_h2, sinks_f, dstk, dstk_fixup,
scale, slope, logit_softcap, ne01, ne02,
stride_Q1, stride_Q2, stride_K, stride_V, stride_mask,
jt, kb0_start, kb0_stop);
NO_DEVICE_CODE;
#endif // TURING_MMA_AVAILABLE
}
@@ -1397,17 +1390,15 @@ static __global__ void flash_attn_ext_f16(
(Q_f2, K_h2, V_h2, mask_h2, sinks_f, dstk, dst_meta, scale, slope, logit_softcap,
ne01, ne02, stride_Q1, stride_Q2, stride_K, stride_V, stride_mask, jt, kb0_start_kernel, kb0_stop_kernel);
#else
GGML_UNUSED(Q); GGML_UNUSED(K); GGML_UNUSED(V); GGML_UNUSED(mask);
GGML_UNUSED(sinks); GGML_UNUSED(KV_max); GGML_UNUSED(dst); GGML_UNUSED(dst_meta);
GGML_UNUSED(scale); GGML_UNUSED(max_bias); GGML_UNUSED(m0); GGML_UNUSED(m1);
GGML_UNUSED(n_head_log2); GGML_UNUSED(logit_softcap);
GGML_UNUSED(ne00); GGML_UNUSED(ne01); GGML_UNUSED(ne02); GGML_UNUSED(ne03);
GGML_UNUSED(nb01); GGML_UNUSED(nb02); GGML_UNUSED(nb03);
GGML_UNUSED(ne10); GGML_UNUSED(ne11); GGML_UNUSED(ne12); GGML_UNUSED(ne13);
GGML_UNUSED(nb11); GGML_UNUSED(nb12); GGML_UNUSED(nb13);
GGML_UNUSED(nb21); GGML_UNUSED(nb22); GGML_UNUSED(nb23);
GGML_UNUSED(ne31); GGML_UNUSED(ne32); GGML_UNUSED(ne33);
GGML_UNUSED(nb31); GGML_UNUSED(nb32); GGML_UNUSED(nb33);
GGML_UNUSED_VARS(Q, K, V, mask, sinks, KV_max, dst, dst_meta, scale,
max_bias, m0, m1, n_head_log2, logit_softcap,
ne00, ne01, ne02, ne03,
nb01, nb02, nb03,
ne10, ne11, ne12, ne13,
nb11, nb12, nb13,
nb21, nb22, nb23,
ne31, ne32, ne33,
nb31, nb32, nb33);
NO_DEVICE_CODE;
#endif // defined(FLASH_ATTN_AVAILABLE) && defined(TURING_MMA_AVAILABLE)
}