diff --git a/ggml-metal.m b/ggml-metal.m index 97751bacf..a6672692a 100644 --- a/ggml-metal.m +++ b/ggml-metal.m @@ -1017,7 +1017,7 @@ void ggml_metal_graph_compute( [encoder setBytes:&ne00 length:sizeof(ne00) atIndex:2]; [encoder setBytes:&ne01 length:sizeof(ne01) atIndex:3]; [encoder setBytes:&ne02 length:sizeof(ne02) atIndex:4]; - [encoder setThreadgroupMemoryLength:nth/32*sizeof(float) atIndex:0]; + [encoder setThreadgroupMemoryLength:MAX(16, nth/32*sizeof(float)) atIndex:0]; [encoder dispatchThreadgroups:MTLSizeMake(ne01*ne02*ne03, 1, 1) threadsPerThreadgroup:MTLSizeMake(nth, 1, 1)]; } break; @@ -1348,7 +1348,7 @@ void ggml_metal_graph_compute( [encoder setBytes:&ne00 length:sizeof( int64_t) atIndex:2]; [encoder setBytes:&nb01 length:sizeof(uint64_t) atIndex:3]; [encoder setBytes:&eps length:sizeof( float) atIndex:4]; - [encoder setThreadgroupMemoryLength:nth*sizeof(float) atIndex:0]; + [encoder setThreadgroupMemoryLength:MAX(16, nth*sizeof(float)) atIndex:0]; const int64_t nrows = ggml_nrows(src0); @@ -1403,7 +1403,8 @@ void ggml_metal_graph_compute( const int n_past = ((int32_t *) dst->op_params)[0]; const int n_dims = ((int32_t *) dst->op_params)[1]; const int mode = ((int32_t *) dst->op_params)[2]; - const int n_orig_ctx = ((int32_t *) dst->op_params)[3]; + // skip 3, n_ctx, used in GLM RoPE, unimplemented in metal + const int n_orig_ctx = ((int32_t *) dst->op_params)[4]; float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow; memcpy(&freq_base, (int32_t *) dst->op_params + 5, sizeof(float));