summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Simplify all() and any() implementationsErik Schnetter2013-07-056-20/+20
|
* Simplify testing all and anyErik Schnetter2013-07-051-4/+4
|
* Improve commentErik Schnetter2013-07-041-1/+1
|
* Remove superfluous inline declarationsErik Schnetter2013-07-042-2/+2
|
* Correct condition for when to test isfiniteErik Schnetter2013-07-041-1/+1
|
* Simplify prod and sumErik Schnetter2013-07-043-12/+12
|
* Correct pod and sum for VSXErik Schnetter2013-07-041-2/+2
|
* Optimize prod and sumErik Schnetter2013-07-044-8/+31
|
* Remove unused definitionsErik Schnetter2013-07-041-8/+0
|
* Test maxval/minvalErik Schnetter2013-07-041-0/+16
|
* Implement maxval and minval reduction operationsErik Schnetter2013-07-0417-5/+204
|
* Move #defines for OpenCL into the OpenCL source filesErik Schnetter2013-07-041-1/+21
| | | | We do not want to introduce a header file for OpenCL.
* Remove unused codeErik Schnetter2013-07-041-7/+0
|
* Avoid warnings when building pocl library without double precision supportErik Schnetter2013-07-042-2/+12
|
* Correct error: use rint() instead of round()Erik Schnetter2013-07-031-1/+1
|
* Clarify that we won't change rounding modes during optimizationErik Schnetter2013-07-031-1/+1
|
* Add commented-out code that would be a performance improvement for round()Erik Schnetter2013-07-031-1/+10
|
* Do not use type punning for converting between vectors and elements -- gcc ↵Erik Schnetter2013-07-0311-66/+220
| | | | 4.8 doesn't like this any more
* Provide generic functions to extract elements from vectorsErik Schnetter2013-07-031-0/+23
|
* Specify maximum allowed error for many math functionsErik Schnetter2013-07-031-26/+26
|
* Allow specifying the maximum allowed error (in terms of ulp)Erik Schnetter2013-07-031-4/+11
|
* Correct typo in commentErik Schnetter2013-07-031-2/+2
|
* Take maximum vector element into account when testing vector reductionsErik Schnetter2013-07-031-1/+5
|
* Correct frexp and ilogb testsErik Schnetter2013-07-031-2/+3
|
* Remove outdated commented-out codeErik Schnetter2013-07-031-1/+0
|
* More ilogb correctionsErik Schnetter2013-07-033-11/+27
|
* Suggest implementing minval and maxval reductionsErik Schnetter2013-07-031-0/+1
|
* Correct operator!= for AVX instructionsErik Schnetter2013-07-032-2/+2
|
* Change implementation of fmod and remainder to be standard-conformingErik Schnetter2013-07-031-8/+6
|
* Modernize commented-out code for expErik Schnetter2013-07-031-1/+1
|
* Correct frexp/ilogb testsErik Schnetter2013-07-031-6/+5
|
* Rename cpp include guardErik Schnetter2013-07-031-3/+3
|
* Correct some frexp/ilogb implementationsErik Schnetter2013-07-033-11/+43
|
* Denormalized numbers are not yer properly supportedErik Schnetter2013-07-031-1/+1
|
* Turn cpp warnings into TODO itemsErik Schnetter2013-07-031-3/+3
|
* Whitespace change in asm intrinsicsErik Schnetter2013-07-0215-25/+25
|
* Choose FP conformance settings automatically depending on fast-mathErik Schnetter2013-07-021-5/+22
|
* Remove unused automatic int->bool conversionErik Schnetter2013-07-022-4/+0
|
* Improve performance of ilogb and frexpErik Schnetter2013-07-022-6/+9
|
* Correct isnan on x86; improve isnan on other architecturesErik Schnetter2013-07-027-9/+63
|
* Remove wrong commentErik Schnetter2013-07-021-1/+1
|
* Avoid compiler warningErik Schnetter2013-07-021-1/+1
|
* Improve performanc of exp10Erik Schnetter2013-07-021-1/+1
|
* Rewrite rcp (reciprocal) functionErik Schnetter2013-06-281-1/+45
|
* Implement frexpErik Schnetter2013-06-2819-1/+99
|
* Update coefficient listErik Schnetter2013-06-211-0/+895
|
* Revert debugging changes for coefficient creationErik Schnetter2013-06-211-6/+4
|
* Optimize exp and logErik Schnetter2013-06-212-0/+44
|
* Update generated coefficientsErik Schnetter2013-06-201-1282/+20
|
* Allow more flexibility with Padé approximations when determining coefficientsErik Schnetter2013-06-201-26/+42
|
OpenPOWER on IntegriCloud