aaron/ispc
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge pull request #612 from dbabokin/missing_target

Browse Source 
Adding support for avx1-i64x4 target in test system.
This commit is contained in:
Dmitry Babokin
2013-10-01 07:43:26 -07:00
parent af95188f29 49cefc2e97
commit 959a741c14
4 changed files with 24 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
alloy.py
View File

@@ -200,7 +200,7 @@ def check_targets():
answer = answer + ["sse4-i32x4", "sse4-i32x8", "sse4-i16x8", "sse4-i8x16"]
if AVX == False and "avx" in f_lines[i]:
AVX = True;
answer = answer + ["avx1-i32x8", "avx1-i32x16"]
answer = answer + ["avx1-i32x8", "avx1-i32x16", "avx1-i64x4"]
if AVX11 == False and "rdrand" in f_lines[i]:
AVX11 = True;
answer = answer + ["avx1.1-i32x8", "avx1.1-i32x16"]
@@ -217,7 +217,7 @@ def check_targets():
answer = answer + ["sse4-i32x4", "sse4-i32x8", "sse4-i16x8", "sse4-i8x16"]
if "AVX1.0" in f_lines:
AVX = True;
answer = answer + ["avx1-i32x8", "avx1-i32x16"]
answer = answer + ["avx1-i32x8", "avx1-i32x16", "avx1-i64x4"]
if "RDRAND" in f_lines:
AVX11 = True;
answer = answer + ["avx1.1-i32x8", "avx1.1-i32x16"]
@@ -246,7 +246,7 @@ def check_targets():
if SSE4 == False and "wsm" in f_lines[i]:
answer_sde = answer_sde + [["-wsm", "sse4-i32x4"], ["-wsm", "sse4-i32x8"], ["-wsm", "sse4-i16x8"], ["-wsm", "sse4-i8x16"]]
if AVX == False and "snb" in f_lines[i]:
answer_sde = answer_sde + [["-snb", "avx1-i32x8"], ["-snb", "avx1-i32x16"]]
answer_sde = answer_sde + [["-snb", "avx1-i32x8"], ["-snb", "avx1-i32x16"], ["-snb", "avx1-i64x4"]]
if AVX11 == False and "ivb" in f_lines[i]:
answer_sde = answer_sde + [["-ivb", "avx1.1-i32x8"], ["-ivb", "avx1.1-i32x16"]]
if AVX2 == False and "hsw" in f_lines[i]:

18
fail_db.txt
View File

@@ -931,3 +931,21 @@
.\tests\switch-12.ispc compfail x86 avx2-i32x16 Windows LLVM 3.3 cl -O2 *
.\tests\switch-8.ispc compfail x86 avx2-i32x16 Windows LLVM 3.3 cl -O2 *
.\tests\switch-9.ispc compfail x86 avx2-i32x16 Windows LLVM 3.3 cl -O2 *
.\tests\exclusive-scan-add-10.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\exclusive-scan-add-9.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\max-uint-1.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\max-uint.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\min-uint-2.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\packed-load-1.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\packed-store.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\reduce-add-uint-1.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\reduce-add-uint.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\reduce-add-uint64-1.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\reduce-add-uint64.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\reduce-max-uint.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\reduce-min-uint64.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\uint64-max-1.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\uint64-max.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\uint64-min-1.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\uint64-min.ispc runfail x86 avx1-i64x4 Windows LLVM 3.3 cl -O2 *
.\tests\reduce-min-uint64.ispc runfail x86 avx1-i64x4 Windows LLVM 3.4 cl -O2 *

4
ispc.cpp
View File

@@ -714,11 +714,11 @@ Target::SupportedTargets() {
#endif
"sse2-i32x4, sse2-i32x8, "
"sse4-i32x4, sse4-i32x8, sse4-i16x8, sse4-i8x16, "
"avx1-i32x8, avx1-i32x16, "
"avx1-i32x8, avx1-i32x16, avx1-i64x4, "
"avx1.1-i32x8, avx1.1-i32x16, "
"avx2-i32x8, avx2-i32x16, "
"generic-x1, generic-x4, generic-x8, generic-x16, "
"generic-x32, generic-x64";
"generic-x32, generic-x64";
}

2
run_tests.py
View File

@@ -449,7 +449,7 @@ def verify():
check = [["g++", "clang", "cl"],["-O0", "-O2"],["x86","x86-64"],
["Linux","Windows","Mac"],["LLVM 3.1","LLVM 3.2","LLVM 3.3","LLVM head"],
["sse2-i32x4", "sse2-i32x8", "sse4-i32x4", "sse4-i32x8", "sse4-i16x8",
"sse4-i8x16", "avx1-i32x8", "avx1-i32x16", "avx1.1-i32x8", "avx1.1-i32x16",
"sse4-i8x16", "avx1-i32x8", "avx1-i32x16", "avx1-i64x4", "avx1.1-i32x8", "avx1.1-i32x16",
"avx2-i32x8", "avx2-i32x16", "generic-1", "generic-4", "generic-8",
"generic-16", "generic-32", "generic-64"]]
for i in range (0,len(f_lines)):