summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* Update build instructionsErik Schnetter2013-11-251-1/+2
|
* Change order of functions (asin before acos)Erik Schnetter2013-09-101-6/+6
|
* Acknowledge SLEEF in asin and friendsErik Schnetter2013-09-101-0/+6
|
* Use SLEEF's algorithms for asin, acos, atan, atan2Erik Schnetter2013-09-101-105/+158
|
* Use SLEEF's algorithms for sin, cos, tanErik Schnetter2013-09-101-63/+198
|
* Add copy of SLEEF library for referenceErik Schnetter2013-09-1029-0/+24342
|
* Use mad() instead of fma()Erik Schnetter2013-09-103-81/+81
|
* Introduce mad(), as low-accuracy version of fma()Erik Schnetter2013-09-1019-7/+109
|
* Use __builtin_clzs if availableErik Schnetter2013-09-091-2/+4
|
* Provide vector types implemented via (scalar) compiler builtinsErik Schnetter2013-09-094-282/+866
|
* Prefer "long" over "long long" if both types are the sameErik Schnetter2013-09-091-0/+8
|
* SSE/AVX: Use nicer type namesErik Schnetter2013-09-026-626/+674
|
* Use nicer type names in vec_pseudoErik Schnetter2013-09-021-129/+129
|
* Use nicer type names in vec_testErik Schnetter2013-09-021-150/+164
|
* Add build instruction for Pandora (Power7 at LSU)Erik Schnetter2013-09-021-1/+4
|
* interp.cc: Use best double precision vector typeErik Schnetter2013-09-021-1/+1
|
* rsqrt: Do not require strictly positive argumentsErik Schnetter2013-09-021-1/+1
|
* SSE: Check whether abs/max/min are available as intrinsics before using themErik Schnetter2013-09-022-4/+39
|
* Use strictly positive arguments or testing rsqrtErik Schnetter2013-09-021-1/+1
|
* Improve white space in test.cc random()Erik Schnetter2013-09-021-3/+2
|
* Support Altivec with ClangErik Schnetter2013-09-031-3/+34
|
* Altivec, VSX: Make build with IBM compilerErik Schnetter2013-09-022-121/+90
|
* Remove invalid commentErik Schnetter2013-09-021-4/+0
|
* Protect __builtin_clz and __builtin_popcount with #ifdefsErik Schnetter2013-09-021-0/+8
|
* Allow (again) larger errors for rounding functionsErik Schnetter2013-09-021-25/+26
|
* Provide clean set of libm elemental functions in namespace vml_stdErik Schnetter2013-09-0215-243/+308
| | | | Different C++ standard versions and different compilers provide different sets of elemental functions in std. Thus, we create a clean set in a new namespace vml_std.
* Correct auto-detecting VSXErik Schnetter2013-09-021-1/+1
|
* Implement new integer functions for NEONErik Schnetter2013-09-022-13/+65
|
* Correct new integer functions for Altivec/VSXErik Schnetter2013-09-022-13/+69
|
* SSE: Rename function argumentErik Schnetter2013-09-022-6/+7
|
* QPX: Support additional integerErik Schnetter2013-09-021-8/+61
|
* Correct SSE frexp()Erik Schnetter2013-09-021-4/+5
|
* Reduce accuracy requirement for fmaErik Schnetter2013-09-021-1/+1
|
* Add build instructions for Compute (at Perimeter)Erik Schnetter2013-09-011-0/+3
|
* Correct error in debug instantiationsErik Schnetter2013-09-011-1/+1
|
* Implement more integer functions: abs bitifthen clz isignbit max min ↵Erik Schnetter2013-09-0122-110/+896
| | | | | | popcount rotate Rename integer signbit to isignbit to avoid name conflicts.
* NEON: Describe idea for prod() improvementErik Schnetter2013-08-261-0/+1
|
* Describe how to build on Stampede (MIC) and Raspberry Pi (ARM)Erik Schnetter2013-08-261-0/+9
|
* Provide a string describing the configuration optionsErik Schnetter2013-08-262-38/+74
|
* Make frexp take a pointer argument instead of a referenceErik Schnetter2013-08-2120-31/+34
|
* Add float example to build scriptErik Schnetter2013-08-201-0/+1
|
* Use better timers for benchmarking loopsErik Schnetter2013-08-201-4/+9
|
* Benchmark more math functionsErik Schnetter2013-08-201-55/+151
|
* Correct detecting ARM NEON intrinsicsErik Schnetter2013-08-201-2/+2
|
* Add single-precision exampleErik Schnetter2013-08-202-0/+41
|
* Provide trivial fallback "best vector" if no "best vector" is availableErik Schnetter2013-08-201-14/+22
|
* Neon: Implement unaligned load/store operationsErik Schnetter2013-08-202-10/+16
|
* Explicitly align allocated memoryErik Schnetter2013-08-191-1/+6
|
* Add potential optimization for SSE/AVX all/any functionsErik Schnetter2013-08-195-5/+5
|
* SSE: Correct float4 all/anyErik Schnetter2013-08-191-2/+2
|
OpenPOWER on IntegriCloud