summaryrefslogtreecommitdiff
AgeCommit message (Expand)AuthorFilesLines
2016-04-19ir_to_mesa: Emit smarter ir_binop_logic_or for vertex programsptn-narrowingIan Romanick1-15/+11
2016-04-19prog: Delete all remains of OPCODE_SNE, OPCODE_SEQ, OPCODE_SGT, and OPCODE_SLEIan Romanick8-428/+0
2016-04-19ir_to_mesa: Do not emit OPCODE_SEQ or OPCODE_SNEIan Romanick1-9/+64
2016-04-19ir_to_mesa: Do not emit OPCODE_SLE or OPCODE_SGTIan Romanick1-2/+12
2016-04-19radeonsi: enable TGSI support cap for compute shadersBas Nieuwenhuizen4-9/+33
2016-04-19radeonsi: Consider input SGPR count for compute shader SGPR count.Bas Nieuwenhuizen2-6/+13
2016-04-19radeonsi: Add CE synchronization for compute dispatches.Bas Nieuwenhuizen3-2/+8
2016-04-19mesa/st: enable compute shaders if images are also supportedBas Nieuwenhuizen1-3/+4
2016-04-19radeonsi: clean up compute flushBas Nieuwenhuizen2-18/+8
2016-04-19radeonsi: do not do two full flushes on every compute dispatchBas Nieuwenhuizen5-22/+17
2016-04-19radeonsi: split setting graphics and compute descriptorsBas Nieuwenhuizen4-14/+59
2016-04-19radeonsi: split texture decompression for compute shadersBas Nieuwenhuizen4-4/+16
2016-04-19radeonsi: update predicate condition for compute dispatchesBas Nieuwenhuizen2-0/+15
2016-04-19radeonsi: implement TGSI compute dispatchBas Nieuwenhuizen1-27/+77
2016-04-19radeonsi: only emit compute shader state when switching shadersBas Nieuwenhuizen2-59/+88
2016-04-19radeonsi: rework compute scratch bufferBas Nieuwenhuizen3-93/+47
2016-04-19radeonsi: do per cs setup for compute shaders once per csBas Nieuwenhuizen3-32/+48
2016-04-19radeonsi: don't pass scratch buffer to user SGPRsBas Nieuwenhuizen1-8/+0
2016-04-19radeonsi: split input upload off from si_launch_gridBas Nieuwenhuizen1-41/+52
2016-04-19radeonsi: implement TGSI compute shader creationBas Nieuwenhuizen1-18/+58
2016-04-19radeonsi: update shader count for compute shadersBas Nieuwenhuizen1-1/+2
2016-04-19radeonsi: set maximum work group size based on block sizeBas Nieuwenhuizen1-0/+12
2016-04-19radeonsi: implement shared atomicsBas Nieuwenhuizen1-1/+76
2016-04-19radeonsi: implement shared memory load/storeBas Nieuwenhuizen1-2/+82
2016-04-19radeonsi: add shared memoryBas Nieuwenhuizen4-0/+37
2016-04-19radeonsi: lower compute shader argumentsBas Nieuwenhuizen2-0/+50
2016-04-19radeonsi: Use CE for all descriptors.Bas Nieuwenhuizen1-10/+64
2016-04-19gallium/util: Add u_bit_scan_consecutive_range64.Bas Nieuwenhuizen1-0/+14
2016-04-19radeonsi: Replace list_dirty with a mask.Bas Nieuwenhuizen2-17/+29
2016-04-19radeonsi: Add CE uploader.Bas Nieuwenhuizen3-0/+37
2016-04-19radeonsi: Allocate chunks of CE ram.Bas Nieuwenhuizen2-9/+27
2016-04-19radeonsi: Add CE synchronization.Bas Nieuwenhuizen2-0/+27
2016-04-19radeonsi: Add CE packet definitions.Bas Nieuwenhuizen1-0/+6
2016-04-19radeonsi: Create CE IB.Bas Nieuwenhuizen5-1/+54
2016-04-19winsys/amdgpu: Enlarge const IB size.Bas Nieuwenhuizen1-8/+20
2016-04-19winsys/amdgpu: Add support for const IB.Marek Olšák3-5/+124
2016-04-19winsys/amdgpu: split IB data into a new structure in preparation for CEMarek Olšák4-47/+48
2016-04-19gallium/radeon: move ring_type into winsysesMarek Olšák5-10/+11
2016-04-19llvmpipe: Call LLVMShutdown before exiting.Jose Fonseca1-0/+2
2016-04-19llvmpipe: Avoid LLVMGetGlobalContext in tests.Jose Fonseca5-6/+24
2016-04-19llvmpipe: Skip false exp2 failure in lp_test_arit due to buggy MSVCRT.Jose Fonseca1-4/+34
2016-04-19llvmpipe: Test more vector lengths.Jose Fonseca1-13/+30
2016-04-19gallivm: Avoid llvm::sys::getProcessTriple().Jose Fonseca1-3/+3
2016-04-19gallivm: Remove lp_get_module_id.Jose Fonseca4-12/+15
2016-04-19gallivm: Fix MCJIT with LLVM 3.3.Jose Fonseca1-3/+3
2016-04-19gallivm: Make MCJIT a runtime option.Jose Fonseca1-75/+72
2016-04-19scons: Show the unit test full path.Jose Fonseca1-1/+1
2016-04-19gallivm: Use LLVMSetTarget.Jose Fonseca1-3/+9
2016-04-19gallivm: Use LLVMPrintValueToString where available.Jose Fonseca1-35/+10
2016-04-19gallium/tests: Update UTIL_FORMAT_MAX_* defines.Jose Fonseca2-3/+7