projects / oota-llvm.git / history
? search:
summary | shortlog | log | commit | commitdiff | tree
first ⋅ prev ⋅ next
Add target hook for whether it is profitable to reduce load widths
[oota-llvm.git] / lib / Target / R600 / AMDGPUISelLowering.cpp
2014-12-12 Matt ArsenaultAdd target hook for whether it is profitable to reduce... blob | commitdiff | raw
2014-12-07 Marek OlsakR600/SI: Update instruction conversions for VI blob | commitdiff | raw | diff to current
2014-11-26 Matt ArsenaultR600/SI: Use ZeroOrNegativeOneBooleanContent blob | commitdiff | raw | diff to current
2014-11-23 Matt ArsenaultR600: Fix assert on copy of an i1 on pre-SI blob | commitdiff | raw | diff to current
2014-11-15 Matt ArsenaultR600: Permute operands when selecting legacy min/max blob | commitdiff | raw | diff to current
2014-11-15 Tom StellardR600: Fix 64-bit integer division blob | commitdiff | raw | diff to current
2014-11-15 Tom StellardR600: Factor i64 UDIVREM lowering into its own fuction blob | commitdiff | raw | diff to current
2014-11-14 Matt ArsenaultR600/SI: Combine min3/max3 instructions blob | commitdiff | raw | diff to current
2014-11-14 Matt ArsenaultR600/SI: Match integer min / max instructions blob | commitdiff | raw | diff to current
2014-11-13 Matt ArsenaultR600/SI: Fix fmin_legacy / fmax_legacy matching for SI blob | commitdiff | raw | diff to current
2014-11-13 Aditya NandakumarWe can get the TLOF from the TargetMachine - so constru... blob | commitdiff | raw | diff to current
2014-11-13 Matt ArsenaultR600: Error on initializer for LDS. blob | commitdiff | raw | diff to current
2014-11-13 Aditya NandakumarThis patch changes the ownership of TLOF from TargetLow... blob | commitdiff | raw | diff to current
2014-10-21 Matt ArsenaultAdd minnum / maxnum codegen blob | commitdiff | raw | diff to current
2014-10-21 Matt ArsenaultR600/SI: Add missing parameter to div_fmas intrinsic blob | commitdiff | raw | diff to current
2014-10-16 Matt ArsenaultR600: Fix nonsensical implementation of computeKnownBit... blob | commitdiff | raw | diff to current
2014-10-16 Matt ArsenaultR600: Remove dead function blob | commitdiff | raw | diff to current
2014-10-15 Matt ArsenaultR600: Remove unnecessary part of computeKnownBitsForTar... blob | commitdiff | raw | diff to current
2014-10-15 Matt ArsenaultMove variable down to use blob | commitdiff | raw | diff to current
2014-10-15 Matt ArsenaultR600: Fix miscompiles when BFE has multiple uses blob | commitdiff | raw | diff to current
2014-10-15 Matt ArsenaultR600: Use existing variable blob | commitdiff | raw | diff to current
2014-10-15 Matt ArsenaultR600: Remove outdated comment blob | commitdiff | raw | diff to current
2014-10-03 Matt ArsenaultR600/SI: Custom lower f64 -> i64 conversions blob | commitdiff | raw | diff to current
2014-10-03 Matt ArsenaultR600: Custom lower [s|u]int_to_fp for i64 -> f64 blob | commitdiff | raw | diff to current
2014-10-03 Matt ArsenaultR600/SI: Fix ftrunc f64 conformance failures. blob | commitdiff | raw | diff to current
2014-09-26 Matt ArsenaultR600/SI: Add a note about the order of the operands... blob | commitdiff | raw | diff to current
2014-09-22 Tom StellardR600: Don't set BypassSlowDiv for 64-bit division blob | commitdiff | raw | diff to current
2014-09-22 Tom StellardR600/SI: Use ISD::MUL instead of ISD::UMULO when loweri... blob | commitdiff | raw | diff to current
2014-09-19 Matt ArsenaultR600: Better fix for bug 20982 blob | commitdiff | raw | diff to current
2014-09-18 Matt ArsenaultR600: Bug 20982 - Avoid undefined left shift of negativ... blob | commitdiff | raw | diff to current
2014-09-10 Matt ArsenaultR600: Custom lower frem blob | commitdiff | raw | diff to current
2014-08-29 Matt ArsenaultR600/SI: Use mad for fsub + fmul blob | commitdiff | raw | diff to current
2014-08-21 Sanjay Patelname change: isPow2DivCheap -> isPow2SDivCheap blob | commitdiff | raw | diff to current
2014-08-15 Matt ArsenaultR600/SI: Use source modifiers for f64 fneg blob | commitdiff | raw | diff to current
2014-08-15 Matt ArsenaultR600/SI: Use source modifier for f64 fabs blob | commitdiff | raw | diff to current
2014-08-15 Matt ArsenaultR600/SI: Add intrinsic for ldexp blob | commitdiff | raw | diff to current
2014-08-12 Jan VeselyR600: Use optimized 24bit path in udivrem blob | commitdiff | raw | diff to current
2014-08-12 Jan VeselyR600: Remove unused code. blob | commitdiff | raw | diff to current
2014-08-12 Jan VeselyR600: Use i24 optimized path for SREM blob | commitdiff | raw | diff to current
2014-08-09 Matt ArsenaultR600: Disable FP exceptions. blob | commitdiff | raw | diff to current
2014-08-05 Tom StellardR600/SI: Avoid generating REGISTER_LOAD instructions. blob | commitdiff | raw | diff to current
2014-08-04 Eric ChristopherRemove the TargetMachine forwards for TargetSubtargetIn... blob | commitdiff | raw | diff to current
2014-08-04 Matt ArsenaultUse the known address space constant rather than checki... blob | commitdiff | raw | diff to current
2014-08-01 Tom StellardRevert "R600: Move code for generating REGISTER_LOAD... blob | commitdiff | raw | diff to current
2014-08-01 Tom StellardR600: Move code for generating REGISTER_LOAD into R600I... blob | commitdiff | raw | diff to current
2014-07-31 Louis GerbargMake sure no loads resulting from load->switch DAGCombi... blob | commitdiff | raw | diff to current
2014-07-25 Chandler Carruth[SDAG] Enable the new assert for out-of-range result... blob | commitdiff | raw | diff to current
2014-07-24 Matt ArsenaultR600: Add new functions for splitting vector loads... blob | commitdiff | raw | diff to current
2014-07-24 Matt ArsenaultR600: Fix LowerSDIV24 blob | commitdiff | raw | diff to current
2014-07-21 Tom StellardR600/SI: Store constant initializer data in constant... blob | commitdiff | raw | diff to current
2014-07-18 Tim NorthoverR600: support fpext/fptrunc operations to and from... blob | commitdiff | raw | diff to current
2014-07-18 Tim NorthoverR600: support f16 -> f64 conversion intrinsic. blob | commitdiff | raw | diff to current
2014-07-15 Jan VeselyR600: Implement zero undef variants of ctlz/cttz blob | commitdiff | raw | diff to current
2014-07-15 Matt ArsenaultR600: Add dag combine for copy of an illegal type. blob | commitdiff | raw | diff to current
2014-07-10 Jan VeselyR600: Implement float to long/ulong blob | commitdiff | raw | diff to current
2014-07-07 Matt ArsenaultR600: Fix mishandling of load / store chains. blob | commitdiff | raw | diff to current
2014-07-02 Tom StellardR600: Add a comment that llvm.AMDGPU.trunc is a legacy... blob | commitdiff | raw | diff to current
2014-07-02 Tom StellardR600: Promote i64 loads to v2i32 blob | commitdiff | raw | diff to current
2014-07-02 Matt ArsenaultR600: Fix crashes when an illegal type load or store... blob | commitdiff | raw | diff to current
2014-06-30 Matt ArsenaultR600: Move mul combine to separate function blob | commitdiff | raw | diff to current
2014-06-27 Matt ArsenaultR600: Move load/store ReplaceNodeResults to common... blob | commitdiff | raw | diff to current
2014-06-26 Aaron BallmanSilencing a warning about isZExtFree hiding an inherite... blob | commitdiff | raw | diff to current
2014-06-26 Matt ArsenaultR600: Fix vector FMA blob | commitdiff | raw | diff to current
2014-06-24 Tom StellardR600: Promote i64 stores to v2i32 blob | commitdiff | raw | diff to current
2014-06-24 Matt ArsenaultR600: Fix inconsistency in rsq instructions. blob | commitdiff | raw | diff to current
2014-06-24 Matt ArsenaultR600: Remove DIV_INF blob | commitdiff | raw | diff to current
2014-06-23 Matt ArsenaultR600/SI: Fix div_scale intrinsic. blob | commitdiff | raw | diff to current
2014-06-23 Matt ArsenaultR600: Remove AMDILISelLowering blob | commitdiff | raw | diff to current
2014-06-23 Matt ArsenaultR600: Select is not expensive. blob | commitdiff | raw | diff to current
2014-06-23 Matt ArsenaultR600: Move add/sub with overflow out of AMDILISelLowering blob | commitdiff | raw | diff to current
2014-06-23 Matt ArsenaultR600: Move more out of AMDILISelLowering blob | commitdiff | raw | diff to current
2014-06-23 Matt ArsenaultR600/SI: Handle i64 sub. blob | commitdiff | raw | diff to current
2014-06-23 Matt ArsenaultR600: Rename AMDIL file blob | commitdiff | raw | diff to current
2014-06-22 Jan VeselyR600: Use LowerSDIVREM for i64 node replace blob | commitdiff | raw | diff to current
2014-06-22 Jan VeselyR600: Implement custom SDIVREM. blob | commitdiff | raw | diff to current
2014-06-20 Tom StellardR600/SI: Add a pattern for f32 ftrunc blob | commitdiff | raw | diff to current
2014-06-20 Tom StellardR600: Expand vector flog2 blob | commitdiff | raw | diff to current
2014-06-20 Tom StellardR600: Expand vector fexp2 blob | commitdiff | raw | diff to current
2014-06-19 Matt ArsenaultR600/SI: Add intrinsics for various math instructions. blob | commitdiff | raw | diff to current
2014-06-18 Matt ArsenaultUse stdint macros for specifying size of constants blob | commitdiff | raw | diff to current
2014-06-18 Matt ArsenaultR600: Handle fnearbyint blob | commitdiff | raw | diff to current
2014-06-18 Matt ArsenaultUse LL suffix for literal that should be 64-bits. blob | commitdiff | raw | diff to current
2014-06-18 Jan VeselyR600: Expand vector fceil blob | commitdiff | raw | diff to current
2014-06-18 Matt ArsenaultWork around ridiculous warning. blob | commitdiff | raw | diff to current
2014-06-18 Matt ArsenaultR600/SI: Add intrinsics for brev instructions blob | commitdiff | raw | diff to current
2014-06-18 Matt ArsenaultR600: Implement f64 ftrunc, ffloor and fceil. blob | commitdiff | raw | diff to current
2014-06-18 Matt ArsenaultR600: Custom lower f64 frint for pre-CI blob | commitdiff | raw | diff to current
2014-06-17 Matt ArsenaultR600/SI: Match ctlz_zero_undef blob | commitdiff | raw | diff to current
2014-06-17 Tom StellardR600: Use LDS and vectors for private memory blob | commitdiff | raw | diff to current
2014-06-17 Tom StellardSelectionDAG: Expand i64 = FP_TO_SINT i32 blob | commitdiff | raw | diff to current
2014-06-15 Matt ArsenaultFix copy paste error blob | commitdiff | raw | diff to current
2014-06-15 Matt ArsenaultR600: Remove a few more things from AMDILISelLowering blob | commitdiff | raw | diff to current
2014-06-15 Matt ArsenaultR600: Fix assert on vector sdiv blob | commitdiff | raw | diff to current
2014-06-15 Matt ArsenaultR600: Move / cleanup more leftover AMDIL stuff. blob | commitdiff | raw | diff to current
2014-06-15 Matt ArsenaultR600: Move division custom lowering out of AMDILISelLow... blob | commitdiff | raw | diff to current
2014-06-15 Matt ArsenaultR600: Report that integer division is expensive. blob | commitdiff | raw | diff to current
2014-06-14 Matt ArsenaultR600: Fix asserts related to constant initializers blob | commitdiff | raw | diff to current
2014-06-14 Matt ArsenaultR600: Use address space enum instead of value blob | commitdiff | raw | diff to current
2014-06-13 Matt ArsenaultR600: Cleanup some old AMDIL stuff. blob | commitdiff | raw | diff to current
2014-06-13 Matt ArsenaultR600/SI: Fix selection error on i64 rotl / rotr. blob | commitdiff | raw | diff to current
next
C/C++ LLVM-based compilers that forbids OOTA behaviors