summaryrefslogtreecommitdiffstats
path: root/src/gallium/auxiliary/gallivm
Commit message (Collapse)AuthorAgeFilesLines
* gallivm: rename a var: s/val/array_size/Brian Paul2010-05-201-3/+3
|
* gallivm: fix spelling in commentBrian Paul2010-05-201-1/+1
|
* gallivm: handle pointer types in lp_dump_llvmtype()Brian Paul2010-05-201-0/+5
|
* gallivm: Tweak ret_mask handling.José Fonseca2010-05-171-24/+15
|
* gallivm: implement function calls by inliningZack Rusin2010-05-171-11/+134
| | | | with this approach we inline the entire function body in the caller
* gallivm: Centralize some of the LLVM debugging options.José Fonseca2010-05-152-0/+31
|
* gallivm: Temporarily remove function call supportJosé Fonseca2010-05-151-438/+53
| | | | | Commits moved to the gallivm-call feature branch for further experimentation and stabilization.
* gallivm: use our util_snprintfZack Rusin2010-05-151-1/+2
|
* gallivm: implement function callsZack Rusin2010-05-151-53/+437
|
* gallivm: added some vector length sanity checksBrian Paul2010-05-141-0/+2
|
* gallivm: added lp_sizeof_llvm_type()Brian Paul2010-05-142-0/+41
|
* gallivm: comments, const qualifiersBrian Paul2010-05-141-4/+11
|
* tgsi: clean up in emit_fetch()Brian Paul2010-05-141-43/+41
|
* gallivm: silence uninitialized var warningBrian Paul2010-05-131-1/+1
|
* gallivm: silence casting warningBrian Paul2010-05-131-1/+17
|
* gallivm: silence uninitialized var warningBrian Paul2010-05-131-1/+1
|
* llvmpipe: silence uninitialized var warningsBrian Paul2010-05-131-1/+2
|
* gallivm: added lp_dump_llvmtype() debug helperBrian Paul2010-05-132-0/+40
|
* llvmpipe: additional commentsBrian Paul2010-05-131-3/+5
|
* gallivm: rename a varBrian Paul2010-05-131-3/+4
|
* Conditionally revert "gallivm: Use a more compact approach for ↵José Fonseca2010-05-131-0/+10
| | | | | | lp_build_broadcast_scalar()." This reverts commit a09e46c72461183c879d8472b44fe740ecc79b9f.
* gallivm: Ensure PIPE_OS_xxx are defined.José Fonseca2010-05-121-0/+1
|
* gallivm: Use a more compact approach for lp_build_broadcast_scalar().José Fonseca2010-05-121-8/+10
| | | | | | | It produces exactly the same machine code, but it cuts 5% of the number of instructions generated for a typical shader. Also, preserve the scalar when length is 1.
* gallivm: additional commentsBrian Paul2010-05-122-6/+39
|
* gallivm: s/bool/booleanBrian Paul2010-05-121-3/+3
|
* gallivm: rename vars, update comments, etcBrian Paul2010-05-122-21/+23
|
* gallivm: comment out LLVMUnionTypeKind caseBrian Paul2010-05-121-0/+2
|
* gallivm: rename texel result paramBrian Paul2010-05-122-17/+22
| | | | Be clear that this parameter returns four texel channel results.
* gallivm: add a few commentsBrian Paul2010-05-121-3/+3
|
* gallivm/llvmpipe: add const qualifiersBrian Paul2010-05-123-11/+11
|
* gallivm: added lp_typekind_name() util functionBrian Paul2010-05-122-0/+49
|
* gallivm/llvmpipe: move an old comment to a better locationBrian Paul2010-05-101-0/+11
|
* gallivm: Silent warning.José Fonseca2010-05-101-1/+1
|
* gallivm: Add missing lvalue.José Fonseca2010-05-101-1/+1
|
* gallivm: Temporarily disable custom LLVMDumpValue on MSVC.José Fonseca2010-05-101-2/+3
| | | | | | | Seeing very weird crashes during std::cout initialization. The fault probably lies in the way I build LLVM on MSVC, but disable for now to allow more time to investigate.
* gallivm: cosf/sinf are macros on MSVC.José Fonseca2010-05-101-2/+12
| | | | So taking the function address does not work.
* gallivm: Add an alternative to LLVMDumpValue that works with Windows GUI apps.José Fonseca2010-05-083-1/+103
|
* gallivm: Fix mipfiltering with negative lod bias.José Fonseca2010-05-081-2/+4
| | | | | | | In particular, don't use the clamped lod to compute level + 1, or lod in [-1, 0] range will actually interpolate with level 1. This makes Mipfilter DCT pass 100%.
* gallivm: Actually do floor/ceil/trunc for scalars.José Fonseca2010-05-083-166/+34
| | | | | | Also start axing the code duplication for scalar case. The olution is to treat the scalar case specially in a few innermost functions, and leave outer functions untouched.
* gallivm: Import the code to compute the minimax polynomials.José Fonseca2010-05-081-0/+85
| | | | | | It's quite a pain to remember the details after a while, and it is quite likely we'll want to use this again, either for different polynomial orders or different functions, so commit it here.
* gallivm: Use a minimax polynomial for exp2 in range [0,1] instead [-0.5,5].José Fonseca2010-05-081-14/+41
| | | | | | | | | | The advantage of range[-0.5, 0.5] is that it doesn't require floor (for which intrinsics are only available in SSE4.1). But the EXP opcode pretty much forces us to use floor, and there is a good floor approximation around truncation available anyway. This fixes EXP failures in VShader DCT.
* gallivm: Don't hardcode number of args twice.José Fonseca2010-05-081-1/+1
|
* gallivm: The the JIT engine to use our sinf()/cosf() on Windows.José Fonseca2010-05-081-18/+79
| | | | | A quick hack to get the right results, as there are many DCT tests which use these opcodes to generate data to test other opcodes.
* gallivm: Fix BREAK/CONT translation.José Fonseca2010-05-081-44/+63
| | | | | | | | | | The cont_mask must be restored and exec mask recomputed in order to decide whether to repeat the loop or not. Unlike the continue mask, the break_mask must be preserved across loop iterations. Fixes several VShader DCT cases, and no regressions with glean.
* gallivm: Fix segfaul when inserting allocas in an empty function.José Fonseca2010-05-081-3/+10
|
* gallivm: Centralize SoA swizzling into a single place.José Fonseca2010-05-085-70/+128
|
* gallivm: Support predicates.José Fonseca2010-05-082-8/+101
|
* gallivm: Proper implementation of TXL opcode.José Fonseca2010-05-044-68/+96
|
* gallivm: Fix several glitches introduced in the prev commit.José Fonseca2010-05-043-6/+7
|
* gallivm: Implement TXD.José Fonseca2010-05-044-37/+58
|