|
|
|
@@ -338,7 +338,7 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, |
|
|
|
|
|
|
|
for(jjs = js; jjs < js + min_j; jjs += min_jj){ |
|
|
|
min_jj = min_j + js - jjs; |
|
|
|
if (min_jj > GEMM3M_UNROLL_N) min_jj = GEMM3M_UNROLL_N; |
|
|
|
if (min_jj > GEMM3M_UNROLL_N*3) min_jj = GEMM3M_UNROLL_N*3; |
|
|
|
|
|
|
|
START_RPCC(); |
|
|
|
|
|
|
|
@@ -398,7 +398,7 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, |
|
|
|
|
|
|
|
for(jjs = js; jjs < js + min_j; jjs += min_jj){ |
|
|
|
min_jj = min_j + js - jjs; |
|
|
|
if (min_jj > GEMM3M_UNROLL_N) min_jj = GEMM3M_UNROLL_N; |
|
|
|
if (min_jj > GEMM3M_UNROLL_N*3) min_jj = GEMM3M_UNROLL_N*3; |
|
|
|
|
|
|
|
START_RPCC(); |
|
|
|
|
|
|
|
@@ -463,7 +463,7 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, |
|
|
|
|
|
|
|
for(jjs = js; jjs < js + min_j; jjs += min_jj){ |
|
|
|
min_jj = min_j + js - jjs; |
|
|
|
if (min_jj > GEMM3M_UNROLL_N) min_jj = GEMM3M_UNROLL_N; |
|
|
|
if (min_jj > GEMM3M_UNROLL_N*3) min_jj = GEMM3M_UNROLL_N*3; |
|
|
|
|
|
|
|
START_RPCC(); |
|
|
|
|
|
|
|
|