summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* Begin to add MIC supportErik Schnetter2013-08-193-1/+663
|
* QPX: Undo using const& to pass QPX vectorsErik Schnetter2013-08-181-66/+66
| | | | I hear that the QPX API does allow passing structs containing a single vector in a vector register.
* Replace tabs by spacesErik Schnetter2013-08-173-31/+31
|
* QPX: Correct various errorsErik Schnetter2013-08-171-12/+31
|
* QPX: Correct some type errorsErik Schnetter2013-08-171-3/+6
|
* QPX: Pass vector arguments as const&, not by valueErik Schnetter2013-08-171-83/+95
| | | | This should improve performance, since passing by value involves scalarizing the vector on QPX.
* QPX: Correct type errors in any/allErik Schnetter2013-08-171-6/+6
|
* QPX: Remove debug outputErik Schnetter2013-08-171-2/+0
|
* QPX: Correct copysignErik Schnetter2013-08-171-1/+1
|
* QPX: Optimize all/anyErik Schnetter2013-08-171-3/+11
|
* QPX: Remove debug outputErik Schnetter2013-08-171-2/+0
|
* Remove pocl subdirectory. This is now instead part of pocl.Erik Schnetter2013-08-1510-1089/+0
|
* Update build instructionsErik Schnetter2013-08-152-15/+4
|
* Remove GPL and LGPL licences. This makes Vecmathlib a pure MIT licence project.Erik Schnetter2013-08-152-839/+0
|
* Add MIT licenceErik Schnetter2013-08-151-0/+19
|
* Clean up build mechanismErik Schnetter2013-08-154-60/+34
|
* Simplify example, add commentsErik Schnetter2013-07-282-19/+57
|
* Merge branch 'master' of bitbucket.org:eschnett/vecmathlibErik Schnetter2013-07-281-4/+4
|\
| * Do not build pocl subdirectory by defaultErik Schnetter2013-07-271-4/+4
| |
* | Do not use std_ prefix for fma, fmax, or fminErik Schnetter2013-07-281-7/+0
|/
* Correct and simplify convert_int and convert_float for SSE doublesErik Schnetter2013-07-252-16/+16
|
* Sort captured std:: functions alphabeticallyErik Schnetter2013-07-251-2/+2
|
* Update assert handlingErik Schnetter2013-07-251-1/+2
|
* Output mask value when describing memory access errorsErik Schnetter2013-07-251-0/+1
|
* Correct QPX ifthenErik Schnetter2013-07-141-1/+1
|
* Pre-declare pair_* structs to avoid compiler warningsErik Schnetter2013-07-102-14/+44
|
* Mark helper functions defined in a header file as inlineErik Schnetter2013-07-101-5/+5
|
* Handle both C++11 and C99Erik Schnetter2013-07-091-14/+4
|
* Remove erroneous vecmathlib:: prefix when calling libm functionsErik Schnetter2013-07-091-1/+1
|
* Add space between consecutive closing templace angle bracketsErik Schnetter2013-07-091-1/+1
|
* Add _cl_ prefix for all functionsErik Schnetter2013-07-091-2/+2
|
* Implement ifthen with boolvec argumentsErik Schnetter2013-07-0717-56/+194
|
* Simplify return type specificationsErik Schnetter2013-07-0713-56/+53
|
* Use stdint.h's types instead of guessing types' sizesErik Schnetter2013-07-071-8/+8
|
* Handle clang's way of handling Altivec intrinsicsErik Schnetter2013-07-071-3/+9
|
* Test operator[] for boolvecErik Schnetter2013-07-071-0/+5
|
* Do not use 0b syntax for binary values -- this is not supported by IBM's xlCErik Schnetter2013-07-071-1/+1
|
* Correct QPX iota()Erik Schnetter2013-07-071-1/+1
|
* Always use std:: qualifier to accessl libm math functionsErik Schnetter2013-07-071-4/+4
|
* Ensure that isnan etc. are indeed macros when using C99 math functionsErik Schnetter2013-07-071-0/+15
|
* Use math.h instead of cmath when C99 math function are neededErik Schnetter2013-07-071-1/+5
|
* Use single precision versions for all functions with math.hErik Schnetter2013-07-071-3/+3
|
* Don't #define restrict, expect this to be done via compiler optionsErik Schnetter2013-07-071-2/+0
|
* Do not require C++11 any moreErik Schnetter2013-07-0614-182/+217
| | | | | | Do not use any C++11-specific features, such as the auto keyword. Use C99 math functions when C++11 is not available.
* Ignore "interp" executableErik Schnetter2013-07-061-0/+1
|
* Add example that interpolates from an arrayErik Schnetter2013-07-061-0/+63
|
* Improve SSE/AVX convert_intErik Schnetter2013-07-062-2/+24
|
* Introduce a truly unaligned exampleErik Schnetter2013-07-051-9/+57
| | | | Also, add facility for an explicit delay calculation to de-emphasize cache effects.
* Add comments to loop exampleErik Schnetter2013-07-051-4/+32
|
* Improve all() and any() implementationsErik Schnetter2013-07-052-12/+4
|
OpenPOWER on IntegriCloud