diff options
author | Jerry Zhao <jerryz123@berkeley.edu> | 2023-02-14 15:29:15 -0800 |
---|---|---|
committer | Jerry Zhao <jerryz123@berkeley.edu> | 2023-05-22 14:58:36 -0700 |
commit | 9c8677c39e7959d3acd425f0c8032edda51375f0 (patch) | |
tree | d1c1d78d21311d3f78a8ebfbcedc4c51268c65e4 /benchmarks | |
parent | e7c3232eb35e63b7b6ad1439a93a7e1f47db8d60 (diff) | |
download | riscv-tests-9c8677c39e7959d3acd425f0c8032edda51375f0.zip riscv-tests-9c8677c39e7959d3acd425f0c8032edda51375f0.tar.gz riscv-tests-9c8677c39e7959d3acd425f0c8032edda51375f0.tar.bz2 |
Add vec-sgemm
Diffstat (limited to 'benchmarks')
-rw-r--r-- | benchmarks/Makefile | 1 | ||||
-rw-r--r-- | benchmarks/common/util.h | 16 | ||||
-rw-r--r-- | benchmarks/vec-sgemm/dataset1.h | 179 | ||||
-rwxr-xr-x | benchmarks/vec-sgemm/sgemm_gendata.pl | 205 | ||||
-rw-r--r-- | benchmarks/vec-sgemm/vec-sgemm.S | 223 | ||||
-rw-r--r-- | benchmarks/vec-sgemm/vec-sgemm_main.c | 39 |
6 files changed, 663 insertions, 0 deletions
diff --git a/benchmarks/Makefile b/benchmarks/Makefile index 36f749a..4875111 100644 --- a/benchmarks/Makefile +++ b/benchmarks/Makefile @@ -34,6 +34,7 @@ bmarks = \ pmp \ vec-memcpy \ vec-daxpy \ + vec-sgemm \ #-------------------------------------------------------------------- # Build rules diff --git a/benchmarks/common/util.h b/benchmarks/common/util.h index 081cfd6..6258709 100644 --- a/benchmarks/common/util.h +++ b/benchmarks/common/util.h @@ -41,6 +41,22 @@ static int verifyDouble(int n, const volatile double* test, const double* verify return 0; } +static int verifyFloat(int n, const volatile float* test, const float* verify) +{ + int i; + // Unrolled for faster verification + for (i = 0; i < n/2*2; i+=2) + { + float t0 = test[i], t1 = test[i+1]; + float v0 = verify[i], v1 = verify[i+1]; + int eq1 = t0 == v0, eq2 = t1 == v1; + if (!(eq1 & eq2)) return i+1+eq1; + } + if (n % 2 != 0 && test[n-1] != verify[n-1]) + return n; + return 0; +} + static void __attribute__((noinline)) barrier(int ncores) { static volatile int sense; diff --git a/benchmarks/vec-sgemm/dataset1.h b/benchmarks/vec-sgemm/dataset1.h new file mode 100644 index 0000000..656d2f5 --- /dev/null +++ b/benchmarks/vec-sgemm/dataset1.h @@ -0,0 +1,179 @@ + +#ifndef __DATASET_H +#define __DATASET_H +#define ARRAY_SIZE 1024 + + +#define DIM_SIZE 32 + + +typedef float data_t;static data_t input1_data[ARRAY_SIZE] = +{ + 0, 3, 2, 0, 3, 1, 0, 3, 2, 3, 2, 0, 3, 3, 1, 2, 3, 0, 0, 1, + 1, 1, 2, 3, 1, 2, 3, 1, 1, 3, 2, 2, 0, 1, 3, 2, 2, 2, 0, 0, + 1, 0, 1, 3, 3, 0, 3, 3, 3, 3, 0, 3, 2, 1, 2, 2, 0, 0, 3, 0, + 1, 1, 0, 3, 3, 1, 2, 3, 3, 0, 1, 2, 1, 0, 1, 2, 2, 1, 0, 3, + 1, 0, 2, 2, 1, 1, 1, 1, 1, 1, 2, 0, 3, 1, 1, 2, 2, 3, 3, 1, + 3, 2, 0, 0, 0, 3, 3, 3, 2, 1, 2, 3, 1, 0, 0, 0, 0, 1, 2, 2, + 1, 1, 3, 3, 3, 1, 1, 2, 3, 1, 3, 3, 2, 3, 2, 1, 2, 3, 0, 2, + 2, 1, 1, 0, 0, 0, 0, 0, 1, 3, 3, 1, 1, 1, 2, 2, 3, 2, 1, 1, + 1, 1, 3, 0, 2, 2, 1, 3, 2, 1, 2, 2, 1, 3, 1, 3, 1, 3, 2, 3, + 1, 2, 1, 3, 2, 2, 0, 1, 0, 0, 1, 2, 3, 3, 1, 0, 0, 0, 3, 1, + 2, 3, 2, 3, 2, 0, 0, 0, 0, 0, 3, 1, 3, 0, 0, 0, 3, 1, 1, 1, + 1, 2, 1, 2, 3, 2, 0, 0, 2, 2, 3, 0, 3, 0, 0, 3, 0, 3, 1, 3, + 3, 1, 1, 1, 2, 2, 1, 3, 0, 3, 3, 1, 0, 0, 3, 2, 1, 3, 3, 3, + 1, 0, 1, 1, 2, 1, 0, 1, 1, 2, 2, 3, 1, 2, 2, 2, 0, 1, 3, 3, + 3, 2, 2, 1, 0, 1, 2, 0, 1, 1, 1, 1, 2, 3, 2, 2, 3, 3, 0, 0, + 2, 0, 0, 0, 3, 0, 1, 0, 3, 0, 0, 0, 3, 0, 0, 2, 0, 2, 0, 0, + 2, 3, 2, 0, 0, 3, 3, 2, 1, 1, 0, 2, 0, 0, 3, 3, 2, 3, 3, 0, + 1, 0, 2, 2, 0, 3, 3, 1, 1, 0, 2, 3, 2, 1, 1, 0, 1, 2, 1, 2, + 2, 0, 0, 1, 0, 1, 1, 0, 1, 0, 2, 3, 3, 2, 0, 0, 1, 3, 0, 3, + 3, 0, 0, 0, 0, 3, 3, 1, 0, 0, 3, 3, 2, 1, 2, 1, 3, 3, 0, 1, + 3, 0, 2, 3, 1, 3, 3, 3, 3, 3, 0, 1, 1, 3, 0, 2, 2, 3, 1, 2, + 2, 2, 1, 3, 3, 0, 3, 0, 0, 2, 0, 2, 3, 0, 1, 3, 2, 2, 0, 0, + 2, 3, 0, 2, 2, 2, 3, 1, 0, 3, 3, 3, 3, 1, 0, 3, 3, 2, 0, 3, + 2, 0, 3, 0, 2, 0, 0, 2, 2, 1, 0, 2, 3, 1, 1, 1, 1, 2, 3, 3, + 3, 0, 0, 3, 3, 3, 2, 3, 3, 1, 2, 2, 3, 1, 2, 1, 1, 3, 0, 1, + 2, 0, 2, 0, 0, 1, 3, 2, 0, 1, 3, 2, 3, 3, 0, 0, 0, 1, 0, 3, + 3, 2, 2, 2, 1, 1, 2, 2, 1, 3, 2, 0, 1, 3, 2, 0, 2, 1, 3, 0, + 0, 0, 1, 3, 3, 2, 2, 2, 3, 1, 0, 0, 1, 1, 2, 1, 3, 1, 1, 2, + 2, 3, 2, 3, 0, 2, 3, 3, 0, 3, 0, 0, 1, 0, 0, 0, 1, 3, 1, 1, + 2, 3, 2, 1, 1, 2, 2, 2, 3, 0, 1, 1, 2, 1, 2, 0, 2, 3, 1, 3, + 0, 1, 1, 3, 0, 2, 3, 0, 1, 2, 3, 2, 0, 0, 3, 3, 2, 1, 1, 2, + 3, 0, 1, 1, 1, 1, 2, 0, 1, 2, 0, 1, 1, 1, 0, 1, 3, 2, 3, 1, + 0, 2, 1, 2, 1, 3, 3, 1, 0, 2, 2, 3, 1, 3, 1, 3, 0, 1, 0, 3, + 0, 3, 2, 0, 3, 3, 3, 0, 3, 2, 2, 2, 1, 3, 0, 0, 1, 1, 3, 0, + 1, 2, 1, 0, 0, 0, 3, 2, 2, 0, 0, 2, 1, 3, 0, 0, 3, 0, 0, 2, + 1, 1, 2, 2, 1, 3, 2, 2, 1, 1, 2, 1, 3, 2, 1, 1, 3, 0, 1, 3, + 3, 2, 2, 1, 0, 3, 2, 2, 2, 3, 0, 1, 3, 3, 2, 3, 0, 3, 2, 3, + 1, 1, 0, 0, 0, 2, 3, 0, 3, 0, 1, 1, 3, 1, 3, 2, 1, 1, 2, 1, + 3, 2, 0, 2, 1, 0, 2, 3, 2, 3, 2, 1, 2, 3, 0, 0, 1, 1, 0, 0, + 2, 1, 0, 1, 2, 2, 2, 2, 0, 3, 3, 1, 0, 0, 0, 0, 3, 1, 1, 0, + 0, 0, 0, 1, 2, 2, 1, 3, 0, 2, 3, 2, 3, 2, 2, 1, 2, 2, 3, 3, + 1, 3, 0, 2, 2, 3, 3, 1, 2, 2, 2, 3, 1, 1, 1, 0, 0, 0, 3, 0, + 1, 0, 3, 1, 1, 3, 0, 1, 2, 2, 0, 0, 3, 3, 3, 3, 2, 1, 0, 0, + 1, 0, 2, 0, 1, 1, 0, 0, 3, 3, 2, 1, 1, 1, 0, 1, 1, 2, 2, 1, + 1, 2, 0, 3, 1, 3, 1, 0, 3, 0, 3, 1, 1, 1, 0, 2, 0, 3, 1, 0, + 1, 0, 2, 0, 2, 3, 3, 3, 1, 2, 3, 2, 2, 0, 1, 1, 0, 3, 3, 1, + 3, 3, 2, 0, 2, 0, 2, 2, 3, 3, 3, 0, 2, 3, 3, 1, 3, 2, 2, 2, + 0, 2, 3, 0, 2, 0, 3, 2, 2, 1, 1, 0, 2, 2, 2, 0, 2, 2, 0, 1, + 3, 2, 1, 3, 2, 2, 0, 3, 3, 1, 2, 2, 0, 0, 3, 2, 1, 2, 2, 1, + 3, 1, 2, 0, 0, 1, 1, 2, 1, 3, 2, 2, 3, 0, 2, 1, 3, 2, 1, 3, + 2, 3, 3, 1, 2, 1, 2, 2, 0, 0, 0, 3, 0, 2, 3, 1, 0, 0, 2, 3, + 3, 2, 2, 1 +}; + +static data_t input2_data[ARRAY_SIZE] = +{ + 1, 1, 0, 3, 1, 2, 0, 0, 0, 0, 0, 2, 1, 2, 3, 0, 0, 3, 3, 2, + 2, 1, 2, 3, 3, 0, 2, 2, 1, 1, 2, 2, 0, 2, 2, 1, 2, 3, 2, 2, + 3, 3, 2, 2, 1, 1, 1, 1, 2, 1, 2, 2, 3, 3, 3, 0, 0, 3, 2, 3, + 2, 3, 1, 2, 1, 1, 2, 2, 0, 1, 0, 3, 2, 1, 1, 1, 2, 0, 1, 2, + 2, 0, 2, 1, 3, 3, 2, 3, 2, 0, 3, 1, 3, 3, 2, 0, 1, 0, 1, 1, + 2, 2, 1, 1, 2, 2, 1, 2, 3, 3, 1, 3, 2, 2, 2, 3, 3, 1, 0, 2, + 1, 0, 0, 0, 1, 1, 2, 0, 3, 2, 3, 3, 0, 2, 3, 1, 0, 0, 2, 1, + 2, 0, 2, 1, 1, 2, 3, 1, 3, 2, 1, 0, 0, 0, 0, 0, 2, 2, 0, 2, + 1, 2, 0, 3, 2, 2, 0, 0, 3, 2, 1, 1, 3, 0, 2, 0, 0, 1, 0, 2, + 3, 3, 1, 3, 3, 0, 0, 2, 2, 0, 0, 0, 1, 0, 0, 1, 3, 0, 2, 1, + 3, 2, 2, 1, 3, 2, 0, 1, 2, 2, 3, 2, 1, 1, 1, 1, 3, 0, 1, 3, + 2, 2, 3, 1, 1, 2, 0, 2, 1, 1, 2, 3, 1, 0, 1, 0, 1, 1, 0, 0, + 2, 0, 3, 0, 3, 0, 3, 2, 2, 3, 3, 2, 1, 0, 2, 2, 1, 1, 0, 3, + 3, 2, 2, 0, 0, 3, 0, 1, 0, 0, 1, 2, 0, 1, 3, 0, 1, 2, 2, 0, + 0, 3, 0, 3, 0, 1, 1, 2, 0, 0, 0, 3, 0, 0, 2, 1, 1, 1, 0, 2, + 1, 3, 1, 2, 0, 3, 0, 3, 1, 3, 0, 0, 2, 2, 2, 2, 3, 3, 2, 1, + 2, 2, 1, 1, 2, 2, 2, 2, 0, 3, 0, 0, 2, 0, 1, 2, 0, 3, 2, 3, + 2, 0, 2, 1, 2, 1, 0, 2, 1, 1, 3, 2, 2, 3, 1, 0, 3, 3, 1, 0, + 3, 2, 2, 0, 0, 3, 0, 0, 2, 0, 3, 2, 3, 1, 1, 0, 0, 2, 3, 0, + 0, 1, 1, 1, 2, 1, 3, 2, 1, 3, 0, 1, 3, 3, 1, 1, 1, 1, 1, 1, + 0, 0, 2, 3, 2, 2, 2, 3, 2, 3, 1, 2, 3, 2, 2, 2, 0, 1, 3, 0, + 1, 1, 0, 1, 0, 1, 1, 3, 3, 1, 2, 2, 3, 2, 0, 2, 2, 0, 1, 3, + 0, 1, 3, 2, 1, 3, 3, 2, 0, 1, 3, 2, 0, 2, 1, 1, 0, 3, 0, 1, + 1, 1, 1, 1, 3, 0, 0, 1, 0, 2, 3, 1, 3, 0, 2, 1, 3, 0, 3, 0, + 3, 2, 2, 0, 0, 2, 1, 3, 3, 2, 3, 2, 2, 1, 2, 2, 3, 0, 3, 2, + 2, 0, 3, 2, 3, 2, 0, 0, 1, 2, 0, 0, 2, 0, 0, 3, 3, 2, 0, 0, + 3, 3, 0, 2, 3, 1, 0, 1, 0, 2, 1, 0, 2, 1, 0, 1, 0, 3, 0, 2, + 2, 3, 0, 0, 2, 1, 0, 1, 0, 0, 0, 2, 2, 3, 2, 0, 3, 3, 2, 1, + 0, 0, 3, 1, 2, 3, 3, 1, 0, 3, 1, 1, 0, 3, 3, 3, 2, 2, 2, 0, + 1, 2, 0, 3, 0, 1, 0, 1, 1, 0, 1, 2, 0, 3, 2, 0, 1, 2, 2, 0, + 2, 0, 0, 1, 0, 3, 0, 3, 2, 1, 1, 1, 1, 3, 2, 1, 1, 1, 1, 0, + 2, 1, 1, 3, 2, 0, 2, 1, 1, 0, 2, 2, 1, 3, 0, 2, 1, 0, 1, 2, + 0, 1, 3, 2, 3, 2, 1, 0, 2, 0, 2, 2, 3, 1, 1, 3, 2, 3, 2, 2, + 0, 2, 0, 0, 0, 3, 2, 0, 2, 2, 3, 3, 3, 2, 1, 2, 0, 0, 3, 0, + 2, 0, 3, 2, 2, 3, 0, 3, 2, 1, 2, 2, 1, 2, 0, 0, 3, 1, 2, 0, + 2, 3, 2, 2, 1, 1, 1, 3, 3, 3, 3, 3, 1, 3, 0, 1, 3, 2, 2, 1, + 0, 1, 1, 2, 1, 2, 3, 1, 2, 2, 1, 2, 1, 1, 0, 3, 3, 1, 1, 3, + 2, 0, 0, 1, 2, 0, 1, 3, 1, 0, 0, 2, 2, 3, 3, 0, 2, 3, 2, 1, + 1, 3, 0, 2, 2, 3, 3, 1, 2, 3, 3, 3, 1, 3, 0, 3, 1, 1, 2, 2, + 2, 1, 0, 3, 2, 3, 0, 2, 3, 2, 3, 1, 2, 3, 3, 1, 2, 1, 0, 0, + 0, 3, 3, 3, 3, 0, 3, 3, 3, 3, 2, 1, 0, 3, 0, 3, 2, 3, 1, 0, + 0, 1, 3, 1, 0, 2, 2, 3, 1, 0, 2, 1, 1, 3, 1, 1, 3, 1, 2, 1, + 0, 0, 3, 2, 1, 1, 1, 1, 3, 2, 1, 3, 3, 1, 0, 3, 1, 1, 2, 0, + 0, 0, 2, 3, 3, 2, 2, 3, 0, 2, 3, 1, 3, 3, 0, 2, 1, 2, 2, 2, + 1, 0, 1, 3, 2, 3, 1, 1, 2, 1, 1, 0, 0, 2, 3, 2, 1, 0, 3, 1, + 3, 0, 1, 1, 2, 2, 1, 3, 3, 1, 1, 0, 0, 3, 3, 0, 0, 0, 0, 0, + 3, 1, 3, 0, 0, 0, 3, 3, 2, 1, 3, 0, 1, 3, 1, 1, 1, 0, 1, 0, + 1, 2, 2, 2, 3, 3, 0, 2, 3, 2, 1, 3, 3, 1, 1, 3, 0, 3, 3, 2, + 1, 1, 2, 0, 3, 0, 1, 2, 1, 1, 0, 0, 1, 2, 2, 0, 3, 1, 1, 1, + 3, 3, 3, 1, 0, 3, 3, 2, 2, 2, 1, 2, 0, 1, 1, 3, 0, 3, 1, 0, + 2, 2, 0, 1, 2, 3, 2, 1, 2, 0, 3, 2, 1, 3, 0, 1, 2, 0, 3, 0, + 1, 1, 2, 1 +}; + +static data_t verify_data[ARRAY_SIZE] = +{ + 72, 75, 88, 101, 80, 88, 73, 75, 80, 81, 58, 75, 86, 65, 60, 80, 84, 83, 87, 83, + 108, 93, 85, 76, 72, 98, 79, 86, 80, 96, 91, 85, 72, 64, 70, 83, 68, 92, 51, 54, + 85, 85, 60, 58, 90, 64, 55, 69, 72, 48, 94, 77, 91, 83, 70, 69, 67, 77, 59, 50, + 67, 74, 77, 67, 67, 62, 72, 71, 68, 79, 54, 61, 67, 61, 55, 62, 78, 60, 53, 64, + 67, 69, 99, 68, 88, 60, 66, 63, 70, 62, 65, 50, 53, 66, 70, 72, 75, 78, 85, 95, + 71, 89, 70, 68, 86, 88, 58, 77, 84, 70, 65, 68, 73, 75, 91, 96, 105, 92, 76, 68, + 86, 69, 80, 59, 73, 83, 88, 75, 64, 63, 71, 99, 77, 77, 69, 55, 80, 73, 54, 73, + 87, 78, 60, 69, 65, 78, 86, 89, 95, 92, 63, 69, 89, 61, 80, 65, 70, 77, 89, 77, + 79, 79, 73, 92, 64, 81, 60, 78, 81, 80, 61, 63, 89, 65, 56, 83, 77, 65, 102, 70, + 98, 86, 96, 68, 72, 89, 73, 73, 70, 89, 84, 76, 48, 61, 63, 70, 70, 79, 50, 53, + 64, 63, 43, 51, 59, 62, 43, 63, 55, 77, 79, 74, 75, 74, 64, 44, 65, 69, 72, 66, + 54, 71, 74, 72, 69, 76, 68, 89, 94, 75, 65, 53, 85, 79, 65, 74, 82, 73, 58, 70, + 84, 77, 99, 72, 92, 84, 78, 62, 59, 83, 71, 74, 63, 85, 80, 78, 71, 72, 79, 83, + 73, 82, 60, 85, 76, 82, 60, 70, 82, 68, 54, 85, 84, 70, 86, 74, 100, 88, 98, 68, + 67, 87, 69, 73, 68, 88, 76, 71, 47, 43, 47, 80, 54, 65, 40, 37, 59, 53, 33, 48, + 62, 40, 36, 55, 36, 62, 53, 57, 70, 69, 45, 43, 53, 61, 42, 57, 56, 63, 51, 47, + 59, 75, 64, 89, 83, 75, 59, 75, 91, 92, 58, 64, 83, 74, 58, 60, 76, 66, 97, 69, + 90, 95, 92, 64, 78, 75, 77, 73, 65, 78, 82, 75, 47, 54, 59, 71, 59, 56, 53, 42, + 60, 55, 40, 51, 60, 46, 36, 59, 46, 57, 67, 43, 51, 53, 53, 38, 54, 56, 55, 48, + 41, 46, 63, 63, 80, 77, 89, 102, 89, 98, 74, 86, 98, 93, 63, 76, 98, 77, 48, 101, + 86, 88, 100, 82, 102, 90, 95, 75, 86, 103, 83, 98, 80, 104, 98, 86, 71, 74, 80, 90, + 86, 87, 73, 70, 81, 83, 55, 66, 90, 66, 58, 84, 77, 84, 93, 72, 99, 75, 85, 65, + 70, 89, 71, 82, 64, 79, 82, 80, 67, 73, 86, 101, 78, 97, 66, 64, 84, 80, 55, 64, + 79, 73, 51, 79, 89, 68, 94, 77, 109, 102, 82, 61, 66, 93, 88, 70, 82, 82, 85, 69, + 69, 72, 66, 97, 85, 90, 70, 59, 76, 89, 53, 56, 90, 79, 71, 64, 70, 67, 100, 92, + 106, 89, 83, 78, 73, 80, 70, 72, 65, 70, 92, 88, 57, 76, 55, 85, 66, 80, 61, 63, + 63, 78, 54, 58, 71, 73, 54, 63, 63, 62, 89, 76, 86, 81, 83, 54, 70, 81, 78, 64, + 56, 72, 74, 81, 75, 63, 68, 89, 65, 77, 58, 68, 75, 83, 52, 62, 82, 63, 55, 75, + 51, 70, 95, 66, 83, 77, 86, 61, 64, 77, 48, 70, 66, 82, 72, 75, 79, 71, 72, 89, + 78, 78, 66, 59, 91, 80, 55, 64, 79, 68, 54, 71, 67, 75, 87, 84, 100, 101, 76, 58, + 74, 82, 61, 74, 75, 97, 85, 79, 61, 55, 69, 68, 72, 65, 52, 64, 80, 73, 48, 54, + 71, 66, 42, 61, 66, 63, 92, 64, 85, 77, 73, 54, 74, 73, 76, 66, 62, 79, 85, 70, + 71, 84, 87, 81, 88, 86, 77, 77, 93, 88, 78, 71, 101, 89, 58, 84, 95, 81, 89, 97, + 104, 79, 83, 76, 90, 81, 91, 74, 70, 76, 91, 80, 51, 48, 56, 69, 47, 63, 54, 42, + 63, 63, 42, 52, 66, 56, 39, 59, 61, 52, 59, 63, 62, 68, 57, 35, 67, 58, 56, 52, + 61, 63, 60, 47, 85, 75, 89, 106, 88, 95, 74, 82, 107, 107, 64, 78, 98, 90, 62, 91, + 79, 87, 111, 84, 104, 106, 96, 68, 94, 99, 81, 89, 79, 105, 95, 86, 65, 63, 77, 89, + 66, 88, 56, 73, 82, 92, 41, 62, 85, 66, 50, 81, 57, 71, 77, 78, 86, 89, 77, 53, + 67, 78, 61, 63, 72, 82, 69, 66, 59, 46, 55, 70, 56, 64, 45, 50, 65, 64, 42, 56, + 78, 49, 51, 52, 38, 56, 72, 55, 73, 72, 61, 50, 63, 60, 47, 57, 55, 73, 53, 68, + 85, 88, 91, 96, 82, 89, 73, 76, 87, 86, 67, 69, 96, 84, 57, 89, 87, 89, 99, 88, + 104, 90, 85, 75, 88, 92, 85, 75, 74, 87, 103, 94, 55, 48, 56, 65, 72, 50, 45, 51, + 63, 62, 47, 57, 79, 53, 36, 63, 54, 68, 71, 59, 63, 61, 63, 41, 50, 73, 57, 59, + 56, 76, 73, 65, 61, 64, 61, 79, 53, 73, 57, 44, 61, 59, 59, 56, 81, 59, 49, 62, + 65, 55, 69, 72, 79, 70, 58, 57, 68, 61, 62, 50, 57, 60, 66, 66, 63, 77, 81, 89, + 85, 81, 76, 73, 78, 95, 59, 70, 81, 77, 46, 79, 78, 79, 83, 81, 84, 82, 85, 48, + 74, 85, 85, 74, 74, 80, 80, 74, 60, 76, 80, 97, 88, 93, 66, 66, 73, 84, 56, 70, + 90, 63, 58, 78, 73, 93, 90, 78, 94, 88, 82, 67, 85, 70, 81, 86, 74, 82, 88, 82, + 68, 73, 75, 91, 78, 97, 71, 66, 74, 85, 50, 59, 86, 77, 70, 74, 75, 74, 99, 82, + 99, 91, 86, 65, 80, 77, 72, 69, 60, 78, 90, 87, 79, 69, 74, 98, 70, 86, 81, 67, + 69, 78, 48, 65, 88, 70, 70, 70, 69, 72, 96, 90, 99, 82, 81, 76, 98, 73, 74, 71, + 69, 73, 94, 89 +}; + + +#endif //__DATASET_H
\ No newline at end of file diff --git a/benchmarks/vec-sgemm/sgemm_gendata.pl b/benchmarks/vec-sgemm/sgemm_gendata.pl new file mode 100755 index 0000000..7d7abad --- /dev/null +++ b/benchmarks/vec-sgemm/sgemm_gendata.pl @@ -0,0 +1,205 @@ +#!/usr/bin/perl -w +#========================================================================== +# matmul_gendata.pl +# +# Author : Christopher Batten (cbatten@mit.edu) +# Date : April 29, 2005 +# +(our $usageMsg = <<'ENDMSG') =~ s/^\#//gm; +# +# Simple script which creates an input data set and the reference data +# for the matmul benchmark. +# +ENDMSG + +use strict "vars"; +use warnings; +no warnings("once"); +use Getopt::Long; + +#-------------------------------------------------------------------------- +# Command line processing +#-------------------------------------------------------------------------- + +our %opts; + +sub usage() +{ + + print "\n"; + print " Usage: matmul_gendata.pl [options] \n"; + print "\n"; + print " Options:\n"; + print " --help print this message\n"; + print " --size size of input data [1000]\n"; + print " --seed random seed [1]\n"; + print "$usageMsg"; + + exit(); +} + +sub processCommandLine() +{ + + $opts{"help"} = 0; + $opts{"size"} = 1000; + $opts{"seed"} = 1; + Getopt::Long::GetOptions( \%opts, 'help|?', 'size:i', 'seed:i' ) or usage(); + $opts{"help"} and usage(); + +} + +#-------------------------------------------------------------------------- +# Helper Functions +#-------------------------------------------------------------------------- + +sub printArray +{ + my $arrayName = $_[0]; + my $arrayRef = $_[1]; + + my $numCols = 20; + my $arrayLen = scalar(@{$arrayRef}); + + print "static data_t ".$arrayName."[ARRAY_SIZE] = \n"; + print "{\n"; + + if ( $arrayLen <= $numCols ) { + print " "; + for ( my $i = 0; $i < $arrayLen; $i++ ) { + print sprintf("%3d",$arrayRef->[$i]); + if ( $i != $arrayLen-1 ) { + print ", "; + } + } + print "\n"; + } + + else { + my $numRows = int($arrayLen/$numCols); + for ( my $j = 0; $j < $numRows; $j++ ) { + print " "; + for ( my $i = 0; $i < $numCols; $i++ ) { + my $index = $j*$numCols + $i; + print sprintf("%3d",$arrayRef->[$index]); + if ( $index != $arrayLen-1 ) { + print ", "; + } + } + print "\n"; + } + + if ( $arrayLen > ($numRows*$numCols) ) { + print " "; + for ( my $i = 0; $i < ($arrayLen-($numRows*$numCols)); $i++ ) { + my $index = $numCols*$numRows + $i; + print sprintf("%3d",$arrayRef->[$index]); + if ( $index != $arrayLen-1 ) { + print ", "; + } + } + print "\n"; + } + + } + + print "};\n\n"; +} + + + +#-------------------------------------------------------------------------- +# Matmul +#-------------------------------------------------------------------------- + +# http://answers.oreilly.com/topic/418-how-to-multiply-matrices-in-perl/ + +sub mmult { + my ($m1,$m2) = @_; + my ($m1rows,$m1cols) = matdim($m1); + my ($m2rows,$m2cols) = matdim($m2); + + my $result = [ ]; + my ($i, $j, $k); + + for $i (range($m1rows)) { + for $j (range($m2cols)) { + for $k (range($m1cols)) { + $result->[$i][$j] += $m1->[$i][$k] * $m2->[$k][$j]; + } + } + } + return $result; +} + +sub range { 0 .. ($_[0] - 1) } + + +sub veclen { + my $ary_ref = $_[0]; + my $type = ref $ary_ref; + if ($type ne "ARRAY") { die "$type is bad array ref for $ary_ref" } + return scalar(@$ary_ref); +} + +sub matdim { + my $matrix = $_[0]; + my $rows = veclen($matrix); + my $cols = veclen($matrix->[0]); + return ($rows, $cols); +} + +#-------------------------------------------------------------------------- +# Main +#-------------------------------------------------------------------------- + +sub main() +{ + + processCommandLine(); + srand($opts{"seed"}); + + # create random input arrays + my $mat_values1; + my $mat_values2; + for ( my $i = 0; $i < $opts{"size"}; $i++ ) { + for ( my $j = 0; $j < $opts{"size"}; $j++ ) { + $mat_values1->[$i][$j] = int(rand(4)); + $mat_values2->[$i][$j] = int(rand(4)); + } + } + + # perform matmul + my $mat_results = mmult( $mat_values1, $mat_values2 ); + + # translate 2d arrays to 1d-somethings (I don't know how to code in perl - Chris) + my @values1; + my @values2; + my @results; + for ( my $i = 0; $i < $opts{"size"}; $i++ ) { + for ( my $j = 0; $j < $opts{"size"}; $j++ ) { + my $value1 = $mat_values1->[$i][$j]; + my $value2 = $mat_values2->[$i][$j]; + my $result = $mat_results->[$i][$j]; + push( @values1, $value1 ); + push( @values2, $value2 ); + push( @results, $result ); + } + } + + print "\n#ifndef __DATASET_H"; + print "\n#define __DATASET_H"; + print "\n\#define ARRAY_SIZE ".($opts{"size"}*$opts{"size"})." \n\n"; + print "\n\#define DIM_SIZE ".$opts{"size"}." \n\n"; + print "\ntypedef float data_t;"; + + printArray( "input1_data", \@values1 ); + printArray( "input2_data", \@values2 ); + printArray( "verify_data", \@results); + + print "\n#endif //__DATASET_H"; + +} + +main(); + diff --git a/benchmarks/vec-sgemm/vec-sgemm.S b/benchmarks/vec-sgemm/vec-sgemm.S new file mode 100644 index 0000000..290f83c --- /dev/null +++ b/benchmarks/vec-sgemm/vec-sgemm.S @@ -0,0 +1,223 @@ + .text + .balign 4 + .global vec_sgemm_nn +# RV64IDV system +# +# void +# vec_sgemm_nn(size_t n, +# size_t m, +# size_t k, +# const float*a, // m * k matrix +# size_t lda, +# const float*b, // k * n matrix +# size_t ldb, +# float*c, // m * n matrix +# size_t ldc) +# +# c += a*b (alpha=1, no transpose on input matrices) +# matrices stored in C row-major order + +#define n a0 +#define m a1 +#define k a2 +#define ap a3 +#define astride a4 +#define bp a5 +#define bstride a6 +#define cp a7 +#define cstride t0 +#define kt t1 +#define nt t2 +#define bnp t3 +#define cnp t4 +#define akp t5 +#define bkp s0 +#define nvl s1 +#define ccp s2 +#define amp s3 + +# Use args as additional temporaries +#define ft12 fa0 +#define ft13 fa1 +#define ft14 fa2 +#define ft15 fa3 + +#define FRAMESIZE 32 + +# This version holds a 16*VLMAX block of C matrix in vector registers +# in inner loop, but otherwise does not cache or TLB tiling. + +vec_sgemm_nn: + ld cstride, 0(sp) # Get arg from stack frame + addi sp, sp, -FRAMESIZE + sd s0, 0(sp) + sd s1, 8(sp) + sd s2, 16(sp) + + # Check for zero size matrices + beqz n, exit + beqz m, exit + beqz k, exit + + # Convert elements strides to byte strides. + slli astride, astride, 2 + slli bstride, bstride, 2 + slli cstride, cstride, 2 + + slti t6, m, 16 + bnez t6, end_rows + +c_row_loop: # Loop across rows of C blocks + + mv nt, n # Initialize n counter for next row of C blocks + + mv bnp, bp # Initialize B n-loop pointer to start + mv cnp, cp # Initialize C n-loop pointer + +c_col_loop: # Loop across one row of C blocks + vsetvli nvl, nt, e32, ta, ma # 32-bit vectors, LMUL=1 + + mv akp, ap # reset pointer into A to beginning + mv bkp, bnp # step to next column in B matrix + + # Initalize current C submatrix block from memory. + vle32.v v0, (cnp); add ccp, cnp, cstride; + vle32.v v1, (ccp); add ccp, ccp, cstride; + vle32.v v2, (ccp); add ccp, ccp, cstride; + vle32.v v3, (ccp); add ccp, ccp, cstride; + vle32.v v4, (ccp); add ccp, ccp, cstride; + vle32.v v5, (ccp); add ccp, ccp, cstride; + vle32.v v6, (ccp); add ccp, ccp, cstride; + vle32.v v7, (ccp); add ccp, ccp, cstride; + vle32.v v8, (ccp); add ccp, ccp, cstride; + vle32.v v9, (ccp); add ccp, ccp, cstride; + vle32.v v10, (ccp); add ccp, ccp, cstride; + vle32.v v11, (ccp); add ccp, ccp, cstride; + vle32.v v12, (ccp); add ccp, ccp, cstride; + vle32.v v13, (ccp); add ccp, ccp, cstride; + vle32.v v14, (ccp); add ccp, ccp, cstride; + vle32.v v15, (ccp) + + + mv kt, k # Initialize inner loop counter + + # Inner loop scheduled assuming 4-clock occupancy of vfmacc instruction and single-issue pipeline + # Software pipeline loads + flw ft0, (akp); add amp, akp, astride; + flw ft1, (amp); add amp, amp, astride; + flw ft2, (amp); add amp, amp, astride; + flw ft3, (amp); add amp, amp, astride; + # Get vector from B matrix + vle32.v v16, (bkp) + + # Loop on inner dimension for current C block + k_loop: + vfmacc.vf v0, ft0, v16 + add bkp, bkp, bstride + flw ft4, (amp) + add amp, amp, astride + vfmacc.vf v1, ft1, v16 + addi kt, kt, -1 # Decrement k counter + flw ft5, (amp) + add amp, amp, astride + vfmacc.vf v2, ft2, v16 + flw ft6, (amp) + add amp, amp, astride + flw ft7, (amp) + vfmacc.vf v3, ft3, v16 + add amp, amp, astride + flw ft8, (amp) + add amp, amp, astride + vfmacc.vf v4, ft4, v16 + flw ft9, (amp) + add amp, amp, astride + vfmacc.vf v5, ft5, v16 + flw ft10, (amp) + add amp, amp, astride + vfmacc.vf v6, ft6, v16 + flw ft11, (amp) + add amp, amp, astride + vfmacc.vf v7, ft7, v16 + flw ft12, (amp) + add amp, amp, astride + vfmacc.vf v8, ft8, v16 + flw ft13, (amp) + add amp, amp, astride + vfmacc.vf v9, ft9, v16 + flw ft14, (amp) + add amp, amp, astride + vfmacc.vf v10, ft10, v16 + flw ft15, (amp) + add amp, amp, astride + addi akp, akp, 4 # Move to next column of a + vfmacc.vf v11, ft11, v16 + beqz kt, 1f # Don't load past end of matrix + flw ft0, (akp) + add amp, akp, astride +1: vfmacc.vf v12, ft12, v16 + beqz kt, 1f + flw ft1, (amp) + add amp, amp, astride +1: vfmacc.vf v13, ft13, v16 + beqz kt, 1f + flw ft2, (amp) + add amp, amp, astride +1: vfmacc.vf v14, ft14, v16 + beqz kt, 1f # Exit out of loop + flw ft3, (amp) + add amp, amp, astride + vfmacc.vf v15, ft15, v16 + vle32.v v16, (bkp) # Get next vector from B matrix, overlap loads with jump stalls + j k_loop + +1: vfmacc.vf v15, ft15, v16 + + # Save C matrix block back to memory + vse32.v v0, (cnp); add ccp, cnp, cstride; + vse32.v v1, (ccp); add ccp, ccp, cstride; + vse32.v v2, (ccp); add ccp, ccp, cstride; + vse32.v v3, (ccp); add ccp, ccp, cstride; + vse32.v v4, (ccp); add ccp, ccp, cstride; + vse32.v v5, (ccp); add ccp, ccp, cstride; + vse32.v v6, (ccp); add ccp, ccp, cstride; + vse32.v v7, (ccp); add ccp, ccp, cstride; + vse32.v v8, (ccp); add ccp, ccp, cstride; + vse32.v v9, (ccp); add ccp, ccp, cstride; + vse32.v v10, (ccp); add ccp, ccp, cstride; + vse32.v v11, (ccp); add ccp, ccp, cstride; + vse32.v v12, (ccp); add ccp, ccp, cstride; + vse32.v v13, (ccp); add ccp, ccp, cstride; + vse32.v v14, (ccp); add ccp, ccp, cstride; + vse32.v v15, (ccp) + + # Following tail instructions should be scheduled earlier in free slots during C block save. + # Leaving here for clarity. + + # Bump pointers for loop across blocks in one row + slli t6, nvl, 2 + add cnp, cnp, t6 # Move C block pointer over + add bnp, bnp, t6 # Move B block pointer over + sub nt, nt, nvl # Decrement element count in n dimension + bnez nt, c_col_loop # Any more to do? + + # Move to next set of rows + addi m, m, -16 # Did 16 rows above + slli t6, astride, 4 # Multiply astride by 16 + add ap, ap, t6 # Move A matrix pointer down 16 rows + slli t6, cstride, 4 # Multiply cstride by 16 + add cp, cp, t6 # Move C matrix pointer down 16 rows + + slti t6, m, 16 + beqz t6, c_row_loop + + # Handle end of matrix with fewer than 16 rows. + # Can use smaller versions of above decreasing in powers-of-2 depending on code-size concerns. +end_rows: + # Not done. + +exit: + ld s0, 0(sp) + ld s1, 8(sp) + ld s2, 16(sp) + addi sp, sp, FRAMESIZE + ret diff --git a/benchmarks/vec-sgemm/vec-sgemm_main.c b/benchmarks/vec-sgemm/vec-sgemm_main.c new file mode 100644 index 0000000..b0a3939 --- /dev/null +++ b/benchmarks/vec-sgemm/vec-sgemm_main.c @@ -0,0 +1,39 @@ +// See LICENSE for license details. + +//************************************************************************** +// SGEMM benchmark +//-------------------------------------------------------------------------- +// +// This benchmark tests a vectorized sgemm implementation. + +#include <string.h> +#include "util.h" + +//-------------------------------------------------------------------------- +// Input/Reference Data + +#include "dataset1.h" + +//-------------------------------------------------------------------------- +// Main + +void *vec_sgemm_nn (size_t, size_t, size_t, const float*, size_t, const float*, size_t, float*, size_t); + +int main( int argc, char* argv[] ) +{ + float results_data[ARRAY_SIZE] = {0}; + +#if PREALLOCATE + // If needed we preallocate everything in the caches + vec_sgemm_nn(DIM_SIZE, DIM_SIZE, DIM_SIZE, input1_data, DIM_SIZE, input2_data, DIM_SIZE, results_data, DIM_SIZE); + memset(results_data, 0, sizeof(results_data)); +#endif + + // Do the sgemm + setStats(1); + vec_sgemm_nn(DIM_SIZE, DIM_SIZE, DIM_SIZE, input1_data, DIM_SIZE, input2_data, DIM_SIZE, results_data, DIM_SIZE); + setStats(0); + + // Check the results + return verifyFloat( ARRAY_SIZE, results_data, verify_data ); +} |