aaron/ispc
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
2,650 Commits 6 Branches 36 Tags
89f6fdd29cd01116c3fa6e03064d43feabc2cf77
Commit Graph

339 Commits

Author SHA1 Message Date
evghenii
ddecdeb834 move remaining int64 from knc.h some of fails to pass tests, grep for evghenii::fails to find out which functions fail and on what tests 2013-09-20 14:55:15 +03:00
evghenii
5cabf0bef0 adding int64 support form knc.h, phase 1. bugs: __lshr & __ashr fail idiv.ispc test, __equal_i64 & __equal_i64_and_mask fails reduce_equal_8.ispc test 2013-09-20 14:13:40 +03:00
evghenii
0ed89e93fa added fails info 2013-09-19 16:34:06 +03:00
egaburov
d68dbbc7bc Merge remote-tracking branch 'upstream/master' into knc-modes 2013-09-19 15:08:17 +02:00
evghenii
0c274212c2 performance tuning for knc-i1x8.h. this gives goed enough performance for double only. float performance is terrible 2013-09-19 16:07:22 +03:00
evghenii
dbef4fd7d7 fixed notation 2013-09-19 14:52:22 +03:00
evghenii
6a21218c13 fix warrning and add KNC 1 2013-09-19 13:45:31 +03:00
Ilia Filippov
00cd90c6b0 test system 2013-09-19 12:26:57 +04:00
evghenii
3cf63362a4 small tuning 2013-09-18 20:03:08 +03:00
evghenii
e4b1f58595 performance fix.. still some issues left with equal_i1 for __vec8_i1 2013-09-18 19:14:41 +03:00
evghenii
4b1a0b4bc4 added fails 2013-09-18 18:41:22 +03:00
evghenii
922edb1128 completed knc-i1x16.h and added knc-i1x8.h with knc-i1x8unsafe_fast.h that doesnt pass several tests.. 2013-09-18 18:14:07 +03:00
Dmitry Babokin
191d9dede5 Merge pull request #585 from tkoziara/master 
Sort description.
 2013-09-16 10:08:49 -07:00
Tomasz Koziara
6e0b9ddc74 Sort description. 2013-09-16 18:02:07 +01:00
Dmitry Babokin
b258027061 Merge pull request #582 from tkoziara/master 
Uniform memory allocation in sort example is fixed.
 2013-09-16 03:29:43 -07:00
Tomasz Koziara
97068765e8 Copyright reversed. 2013-09-14 18:09:04 +01:00
Tomasz Koziara
ed825b3773 Uniform memory allocation fixed. 2013-09-13 13:14:31 +01:00
Ilia Filippov
f620cdbaa1 Changes in perf.py functionality, unification of examples, correction build warnings 2013-08-26 14:04:59 +04:00
Matt Pharr
2b2905b567 Fix (preexisting) bugs in generic-32/64.h with type of "__any", etc. 
This should be a bool, not a one-wide vector of bools.  The equivalent
fix was previously made in generic-16.h, but not made here.  (Note that
many tests are still failing with these targets, but at least they
compile properly now.)
 2013-08-20 09:05:50 -07:00
Matt Pharr
e7f067d70c Fix handling of __clock() builtin for "generic" targets. 2013-08-20 09:04:52 -07:00
Matt Pharr
7ab4c5391c Fix build with LLVM 3.2 and generic-4 / examples/sse4.h target. 2013-08-09 19:56:43 -07:00
Matt Pharr
cd9afe946c Merge branch 'master' into arm 
Conflicts:
	Makefile
	builtins.cpp
	ispc.cpp
	ispc.h
	ispc.vcxproj
	opt.cpp
 2013-08-06 17:39:21 -07:00
Dmitry Babokin
43423c276f Merge pull request #560 from ifilippov/perf 
Supporting perf.py on Mac OS
 2013-08-01 13:20:01 -07:00
Ilia Filippov
3c06924a02 Supporting perf.py on Mac OS 2013-08-01 12:47:37 +04:00
Dmitry Babokin
220f0b0b40 Renaming mandelbrot_tasks files to be different from mandelbrot 2013-07-30 19:53:12 -07:00
Dmitry Babokin
fa93cb7d0b InterlockedAdd -> InterlockedExchangeAdd for better portability (InterlockedAdd is not always supported) 2013-07-29 22:46:36 -07:00
Matt Pharr
b6df447b55 Add reduce_add() for int8 and int16 types. 
This maps to specialized instructions (e.g. PSADBW) when available.
 2013-07-25 09:46:01 -07:00
Matt Pharr
d7b0c5794e Add support for ARM NEON targets. 
Initial support for ARM NEON on Cortex-A9 and A15 CPUs.  All but ~10 tests
pass, and all examples compile and run correctly.  Most of the examples
show a ~2x speedup on a single A15 core versus scalar code.

Current open issues/TODOs
- Code quality looks decent, but hasn't been carefully examined.  Known
  issues/opportunities for improvement include:
  - fp32 vector divide is done as a series of scalar divides rather than
    a vector divide (which I believe exists, but I may be mistaken.)
    This is particularly harmful to examples/rt, which only runs ~1.5x
    faster with ispc, likely due to long chains of scalar divides.
  - The compiler isn't generating a vmin.f32 for e.g. the final scalar
    min in reduce_min(); instead it's generating a compare and then a
    select instruction (and similarly elsewhere).
  - There are some additional FIXMEs in builtins/target-neon.ll that
    include both a few pieces of missing functionality (e.g. rounding
    doubles) as well as places that deserve attention for possible
    code quality improvements.

- Currently only the "cortex-a9" and "cortex-15" CPU targets are
  supported; LLVM supports many other ARM CPUs and ispc should provide
  access to all of the ones that have NEON support (and aren't too
  obscure.)

- ~5 of the reduce-* tests hit an assertion inside LLVM (unfortunately
   only when the compiler runs on an ARM host, though).

- The Windows build hasn't been tested (though I've tried to update
  ispc.vcxproj appropriately).  It may just work, but will more likely
  have various small issues.)

- Anything related to 64-bit ARM has seen no attention.
 2013-07-19 23:07:24 -07:00
Matt Pharr
b007bba59f Replace inline assembly in task system with equivalent gcc intrinsics. 
gcc/icc build only: the Windows build still uses the Win32 calls for
these.
 2013-07-19 23:07:24 -07:00
Ilia Filippov
fd7f87b55e Supporting perf.py on Windows and some small corrections in it 2013-07-02 19:23:18 +04:00
Dmitry Babokin
8be4128c5a Merge pull request #534 from ifilippov/perf 
add script for measuring performance
 2013-07-01 05:09:03 -07:00
Ilia Filippov
806e37338c add script for measuring performance 2013-07-01 13:30:49 +04:00
Dmitry Babokin
ec1095624a Merge pull request #527 from tkoziara/master 
examples/sort added
 2013-06-25 10:11:39 -07:00
Tomasz Koziara
a23d69ebe8 Copyright changed to simplify legal matters. 2013-06-25 17:28:27 +01:00
Tomasz Koziara
86ee8db778 Parallel prefix sum added + minor amendements. 2013-06-25 12:45:51 +01:00
Ilia Filippov
9fb981e9a0 correction of --instrument option support 2013-06-25 12:33:23 +04:00
Tomasz Koziara
f2452f040d First commit of the radix sort example. 2013-06-24 18:37:44 +01:00
james.brodman
6211966c55 Change mask to use __mmask16 instead of a struct. 2013-05-30 16:04:44 -04:00
james.brodman
7b2eaf63af knc.h cleanup 2013-05-10 13:36:18 -04:00
Dmitry Babokin
1069a3c77e Removing some sources of warnings sse4.h and trailing spaces 2013-04-25 03:40:32 +04:00
james.brodman
52dcbf087a Implemented 3 more intrinsics on double precision vectors 2013-03-28 11:55:53 -04:00
james.brodman
ef1af547e2 Change sse4.h to enable inlining. 2013-03-13 10:55:53 -04:00
Jean-Luc Duprat
24087ff3cc Expose none() in the ISPC standard library. 
On KNC: all(), any() and none() do not generate a redundant movmsk instruction.
 2012-11-27 13:38:28 -08:00
Jean-Luc Duprat
2129b1e27d knc.h: Fixed __rsqrt_varying_float() to use _mm512_invsqrt_ps() instead of _mm512_invsqrt_pd() 
This was a typo.
 2012-11-21 15:40:35 -08:00
Jean-Luc Duprat
d3b86dcc90 KNC: fix implementation of __all() to use KNCni mask test instructions... 2012-11-14 09:24:01 -08:00
Jean-Luc Duprat
b601331362 Approximation for inverse sqrt and reciprocal provided in fast math mode. 
RCP was actually slow in fast math mode
   Inverse sqrt did not expose fast approximation
 2012-11-13 14:01:35 -08:00
james.brodman
97ddc1ed10 Fixed =/== error in __all() 2012-11-08 16:30:12 -05:00
jbrodman
e323b1d0ad Fixed compile error: == instead of = 2012-10-26 16:55:28 -04:00
Matt Pharr
406fbab40e Fix bugs in declarations of __any, __all, and __none in examples/intrinsics. 
They return bool, not vector of bool.
 2012-10-17 10:55:50 -07:00
Matt Pharr
9002837750 Remove incorrect assert in tasksys.cpp 2012-10-15 10:43:46 -07:00