Browse Source

optimizations for trmm

tags/v0.2.11^2
wernsaar 11 years ago
parent
commit
1b10ff129a
2 changed files with 30 additions and 10 deletions
  1. +12
    -4
      driver/level3/trmm_L.c
  2. +18
    -6
      driver/level3/trmm_R.c

+ 12
- 4
driver/level3/trmm_L.c View File

@@ -135,7 +135,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = js; jjs < js + min_j; jjs += min_jj){
min_jj = min_j + js - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

START_RPCC();

@@ -199,7 +201,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = js; jjs < js + min_j; jjs += min_jj){
min_jj = min_j + js - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

START_RPCC();

@@ -288,7 +292,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = js; jjs < js + min_j; jjs += min_jj){
min_jj = min_j + js - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

START_RPCC();

@@ -352,7 +358,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = js; jjs < js + min_j; jjs += min_jj){
min_jj = min_j + js - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

START_RPCC();



+ 18
- 6
driver/level3/trmm_R.c View File

@@ -119,7 +119,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = 0; jjs < ls - js; jjs += min_jj){
min_jj = ls - js - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

#ifndef TRANSA
GEMM_ONCOPY(min_l, min_jj, a + (ls + (js + jjs) * lda) * COMPSIZE, lda, sb + min_l * jjs * COMPSIZE);
@@ -137,7 +139,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = 0; jjs < min_l; jjs += min_jj){
min_jj = min_l - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

#ifndef TRANSA
TRMM_OLNCOPY(min_l, min_jj, a, lda, ls, ls + jjs, sb + min_l * (ls - js + jjs) * COMPSIZE);
@@ -188,7 +192,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = js; jjs < js + min_j; jjs += min_jj){
min_jj = min_j + js - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

#ifndef TRANSA
GEMM_ONCOPY(min_l, min_jj, a + (ls + jjs * lda) * COMPSIZE, lda, sb + min_l * (jjs - js) * COMPSIZE);
@@ -239,7 +245,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = 0; jjs < min_l; jjs += min_jj){
min_jj = min_l - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

#ifndef TRANSA
TRMM_OUNCOPY(min_l, min_jj, a, lda, ls, ls + jjs, sb + min_l * jjs * COMPSIZE);
@@ -258,7 +266,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = 0; jjs < js - ls - min_l; jjs += min_jj){
min_jj = js - ls - min_l - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

#ifndef TRANSA
GEMM_ONCOPY(min_l, min_jj, a + (ls + (ls + min_l + jjs) * lda) * COMPSIZE, lda,
@@ -313,7 +323,9 @@ int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n, FLOAT *sa, FLO

for(jjs = js; jjs < js + min_j; jjs += min_jj){
min_jj = min_j + js - jjs;
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
if (min_jj > GEMM_UNROLL_N*3) min_jj = GEMM_UNROLL_N*3;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;

#ifndef TRANSA
GEMM_ONCOPY(min_l, min_jj, a + (ls + (jjs - min_j) * lda) * COMPSIZE, lda, sb + min_l * (jjs - js) * COMPSIZE);


Loading…
Cancel
Save