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

Add SSE4 target optimized for computation with 8-bit datatypes.

Browse Source 
This change adds a new 'sse4-8' target, where programCount is 16 and
the mask element size is 8-bits.  (i.e. the most appropriate sizing of
the mask for SIMD computation with 8-bit datatypes.)
This commit is contained in:
Matt Pharr
2013-07-23 17:30:32 -07:00
parent 15a3ef370a
commit 53414f12e6
7 changed files with 578 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
ispc.cpp
View File

@@ -310,6 +310,14 @@ Target::Target(const char *arch, const char *cpu, const char *isa, bool pic) :
this->m_maskingIsFree = false;
this->m_maskBitCount = 32;
}
else if (!strcasecmp(isa, "sse4-8")) {
this->m_isa = Target::SSE4;
this->m_nativeVectorWidth = 16;
this->m_vectorWidth = 16;
this->m_attributes = "+sse,+sse2,+sse3,+sse41,-sse42,-sse4a,+ssse3,-popcnt,+cmov";
this->m_maskingIsFree = false;
this->m_maskBitCount = 8;
}
else if (!strcasecmp(isa, "generic-4")) {
this->m_isa = Target::GENERIC;
this->m_nativeVectorWidth = 4;