summaryrefslogtreecommitdiff
path: root/backend/src/backend
Commit message (Expand)AuthorAgeFilesLines
* Started to implement OpenCL barrier() functionBenjamin Segovia2012-11-128-5/+72
* Started the boiler plate for barrier (and fences) instructionsBenjamin Segovia2012-11-091-0/+30
* Removed MAD instruction in the IRBenjamin Segovia2012-11-091-20/+0
* Added support for __local pointer in kernel arguments. This require to add moreBenjamin Segovia2012-11-097-20/+57
* Fixed more warning with LLVM 3.2Benjamin Segovia2012-11-091-1/+2
* Fixed a warning when compiling with LLVM 3.1Benjamin Segovia2012-11-092-5/+11
* Fixed preprocessor defines for LLVM 3.0Benjamin Segovia2012-11-081-0/+10
* Cleaned some old bitrot code in llvm_gen_backendBenjamin Segovia2012-11-081-1/+5
* Added support for llvm 3.1Benjamin Segovia2012-11-085-8/+8
* Made a quick pass to clean up a bit the headers (there is way too many "INLINE"Benjamin Segovia2012-11-075-44/+47
* Started to revamp ir::Instruction class to make it use intrusive_list insteadBenjamin Segovia2012-11-071-1/+1
* Added support for select instructions with embedded comparesBenjamin Segovia2012-11-066-20/+137
* Changed style of the comment for insn schedulingBenjamin Segovia2012-11-061-5/+6
* Made the zero cycle strategy work. This stronly limits the register pressure.Benjamin Segovia2012-11-064-5/+96
* Added a additional rule for MADs. When we try to avoid spilling, we do notBenjamin Segovia2012-11-061-0/+7
* Added more features to the instruction scheduler to support both pre-registerBenjamin Segovia2012-11-067-35/+111
* Fixed a potential bug with cvarBenjamin Segovia2012-11-051-3/+3
* Added julia setBenjamin Segovia2012-11-051-1/+1
* Removed the remaining bits of the Gen specific extensions. Much cleaner codeBenjamin Segovia2012-11-057-421/+2
* Started to remove support for gen extensions. This was basically a quick fireBenjamin Segovia2012-11-051-1/+2
* Removed the remaining bits of the c++ simulator. We only need the support forBenjamin Segovia2012-11-0511-3519/+2
* - Added reference images for all the remaining tests that need themBenjamin Segovia2012-11-021-11/+106
* - Made ribbon test pass (still need to generate the golden image)Benjamin Segovia2012-11-022-1/+15
* Made compiler_clod pass. The image is now properly computed.Benjamin Segovia2012-11-026-9/+23
* Removed use of exceptions since LLVM may not support them (depending on how itBenjamin Segovia2012-10-129-34/+37
* Added support for various math functionsBenjamin Segovia2012-10-124-10/+48
* Pushed back original headers we partially lost during the code base merge. NowBenjamin Segovia2012-10-126-12/+140
* Fixed a bug in the flag register allocatorBenjamin Segovia2012-10-092-9/+14
* Separate instruction data from the rest to prepare intrusive_list use forBenjamin Segovia2012-10-081-1/+1
* Used intrusive_list for SelectionVectorBenjamin Segovia2012-10-083-21/+10
* Used intrusive_list for SelectionBlockBenjamin Segovia2012-10-085-93/+34
* Used intrusive_list for SelectionInstructionBenjamin Segovia2012-10-083-72/+38
* Used intrusive_list for forward list instruction schedulingBenjamin Segovia2012-10-081-122/+69
* Added intrusive list implementation from Maciej Sinilo (MIT licensed)Benjamin Segovia2012-10-081-1/+3
* Removed utest kernels used for the (removed) internal testsBenjamin Segovia2012-10-082-3038/+0
* Cleaned up disassembly code to make it used the new files (and not the mesaBenjamin Segovia2012-10-082-408/+203
* Fixed issue with flag register encoding in instruction schedulingBenjamin Segovia2012-10-081-2/+3
* Activated instruction schedulingBenjamin Segovia2012-10-085-14/+59
* Fixed compilation issue with gcc 4.6Benjamin Segovia2012-10-081-1/+1
* Handled null register when schedulingBenjamin Segovia2012-10-055-16/+29
* Almost finished instruction scheduling. Just need to handle accumulator writeBenjamin Segovia2012-10-041-1/+1
* Fixed bug in growing pool allocatorBenjamin Segovia2012-10-042-3/+3
* Fixed read/write dependencies for the insn schedulerBenjamin Segovia2012-10-043-6/+6
* Started to make instruction scheduling workBenjamin Segovia2012-10-023-29/+57
* First implementation (untested) of instruction scheduling. Now it is prettyBenjamin Segovia2012-09-215-64/+440
* Removed unused filesBenjamin Segovia2012-09-194-23/+23
* Cleaned up instruction selection interface. Push everything back to the cppBenjamin Segovia2012-09-195-507/+592
* Cleaned up register allocator interface to limit dependenciesBenjamin Segovia2012-09-192-67/+106
* Made SelectionInstruction variable sizedBenjamin Segovia2012-09-194-100/+116
* Implemented linear allocator to have both fast instruction allocation andBenjamin Segovia2012-09-195-217/+151
View Lorry Controller Status