maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/R600
History
Chandler Carruth 9a0051cd59 [SDAG] Make the DAGCombine worklist not grow endlessly due to duplicate 
insertions.

The old behavior could cause arbitrarily bad memory usage in the DAG
combiner if there was heavy traffic of adding nodes already on the
worklist to it. This commit switches the DAG combine worklist to work
the same way as the instcombine worklist where we null-out removed
entries and only add new entries to the worklist. My measurements of
codegen time shows slight improvement. The memory utilization is
unsurprisingly dominated by other factors (the IR and DAG itself
I suspect).

This change results in subtle, frustrating churn in the particular order
in which DAG combines are applied which causes a number of minor
regressions where we fail to match a pattern previously matched by
accident. AFAICT, all of these should be using AddToWorklist to directly
or should be written in a less brittle way. None of the changes seem
drastically bad, and a few of the changes seem distinctly better.

A major change required to make this work is to significantly harden the
way in which the DAG combiner handle nodes which become dead
(zero-uses). Previously, we relied on the ability to "priority-bump"
them on the combine worklist to achieve recursive deletion of these
nodes and ensure that the frontier of remaining live nodes all were
added to the worklist. Instead, I've introduced a routine to just
implement that precise logic with no indirection. It is a significantly
simpler operation than that of the combiner worklist proper. I suspect
this will also fix some other problems with the combiner.

I think the x86 changes are really minor and uninteresting, but the
avx512 change at least is hiding a "regression" (despite the test case
being just noise, not testing some performance invariant) that might be
looked into. Not sure if any of the others impact specific "important"
code paths, but they didn't look terribly interesting to me, or the
changes were really minor. The consensus in review is to fix any
regressions that show up after the fact here.

Thanks to the other reviewers for checking the output on other
architectures. There is a specific regression on ARM that Tim already
has a fix prepped to commit.

Differential Revision: http://reviews.llvm.org/D4616

llvm-svn: 213727
 		2014-07-23 07:08:53 +00:00
..
32-bit-local-address-space.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
64bit-kernel-args.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
128bit-kernel-args.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
README R600: Reorganize lit tests and document how they should be organized 2013-04-19 02:10:53 +00:00
add.ll R600/SI: Only use SALU instructions for 64-bit add in a block of CF depth 0 2014-05-15 14:41:54 +00:00
add_i64.ll R600: Promote i64 loads to v2i32 2014-07-02 20:53:54 +00:00
address-space.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
and.ll R600/SI: Fix select on i1 2014-07-15 21:44:37 +00:00
anyext.ll R600/SI: Add a pattern for i32 anyext 2014-02-13 23:34:13 +00:00
array-ptr-calc-i32.ll R600/SI: Use scratch memory for large private arrays 2014-07-21 15:45:01 +00:00
array-ptr-calc-i64.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
atomic_cmp_swap_local.ll IR: add "cmpxchg weak" variant to support permitted failure. 2014-06-13 14:24:07 +00:00
atomic_load_add.ll R600/SI: Fix backwards names for local atomic instructions. 2014-06-11 18:08:37 +00:00
atomic_load_sub.ll R600/SI: Fix backwards names for local atomic instructions. 2014-06-11 18:08:37 +00:00
basic-branch.ll R600: Add failing control flow tests. 2014-03-01 21:45:41 +00:00
basic-loop.ll R600: Add failing control flow tests. 2014-03-01 21:45:41 +00:00
bfe_uint.ll R600: Disable the BFE pattern 2014-01-23 18:49:33 +00:00
bfi_int.ll R600/SI: Add instruction shrinking pass 2014-07-21 16:55:33 +00:00
big_alu.ll R600/SI: Add intrinsics for various math instructions. 2014-06-19 01:19:19 +00:00
bitcast.ll R600/SI: Fix bitcast between v2i32 and f64 2014-06-11 19:31:13 +00:00
bswap.ll R600: Fix selection failure for vector bswap 2014-06-09 16:20:25 +00:00
build_vector.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
call.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
call_fs.ll R600: Add a test for r183108 2013-06-04 15:03:35 +00:00
cayman-loop-bug.ll R600: Workaround for cayman loop bug 2013-12-02 17:29:37 +00:00
cf-stack-bug.ll R600: Recommit 199842: Add work-around for the CF stack entry HW bug 2014-01-23 16:18:02 +00:00
cf_end.ll R600: Stop emitting the instruction type byte before each instruction 2013-05-06 17:50:44 +00:00
codegen-prepare-addrmode-sext.ll [CodeGenPrepare] Fix the check of the legality of an instruction. 2014-02-22 01:06:41 +00:00
combine_vloads.ll Add target hook to prevent folding some bitcasted loads. 2013-11-15 04:42:23 +00:00
complex-folding.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
concat_vectors.ll R600: Add missing test for concat_vectors 2014-07-20 07:13:17 +00:00
copy-illegal-type.ll R600: Add dag combine for copy of an illegal type. 2014-07-15 02:06:31 +00:00
ctlz_zero_undef.ll R600: Implement zero undef variants of ctlz/cttz 2014-07-15 15:51:09 +00:00
ctpop.ll R600/SI: Add instruction shrinking pass 2014-07-21 16:55:33 +00:00
ctpop64.ll R600/SI: Add patterns for ctpop inside a branch 2014-06-20 17:06:11 +00:00
cttz_zero_undef.ll R600: Implement zero undef variants of ctlz/cttz 2014-07-15 15:51:09 +00:00
cvt_f32_ubyte.ll R600/SI: Clean up some of the unused REGISTER_{LOAD,STORE} code 2014-07-21 15:45:06 +00:00
dagcombiner-bug-illegal-vec4-int-to-fp.ll R600: Prettier asmPrint of Alu 2013-05-02 21:52:30 +00:00
default-fp-mode.ll R600: Add denormal handling subtarget features. 2014-07-14 23:40:49 +00:00
disconnected-predset-break-bug.ll R600: use native for alu 2013-04-30 00:14:38 +00:00
dot4-folding.ll R600: Const/Neg/Abs can be folded to dot4 2013-06-04 23:17:15 +00:00
elf.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
elf.r600.ll R600: Refactor stack size calculation 2014-01-22 21:55:43 +00:00
extload.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
extract_vector_elt_i16.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
fabs.ll R600/SI: Fold fabs/fneg into src input modifier 2014-05-10 19:18:39 +00:00
fadd.ll R600/SI: Expand all v8[if]32 operations 2014-02-13 23:34:15 +00:00
fadd64.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
fceil.ll R600: Expand vector fceil 2014-06-18 17:57:29 +00:00
fceil64.ll R600: Expand vector fceil 2014-06-18 17:57:29 +00:00
fcmp-cnd.ll R600: Non vector only instruction can be scheduled on trans unit 2013-09-04 19:53:46 +00:00
fcmp-cnde-int-args.ll R600: Improve custom lowering of select_cc 2013-03-08 15:37:09 +00:00
fcmp.ll R600: Non vector only instruction can be scheduled on trans unit 2013-09-04 19:53:46 +00:00
fcmp64.ll R600/SI: Add instruction shrinking pass 2014-07-21 16:55:33 +00:00
fconst64.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
fcopysign.f32.ll R600: Handle fcopysign 2014-06-10 19:00:20 +00:00
fcopysign.f64.ll R600: Handle fcopysign 2014-06-10 19:00:20 +00:00
fdiv.ll R600/SI: Implement less wrong f32 fdiv 2014-07-15 20:18:31 +00:00
fdiv64.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
fetch-limits.r600.ll R600: Fix the fetch limits for R600 generation GPUs 2013-06-07 20:28:55 +00:00
fetch-limits.r700+.ll R600: Fix the fetch limits for R600 generation GPUs 2013-06-07 20:28:55 +00:00
ffloor.ll R600: Implement f64 ftrunc, ffloor and fceil. 2014-06-18 17:05:30 +00:00
floor.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
fma.ll R600: Fix vector FMA 2014-06-26 01:28:05 +00:00
fmad.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
fmax.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
fmin.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
fmul.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
fmul64.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
fmuladd.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
fnearbyint.ll R600: Handle fnearbyint 2014-06-18 22:03:45 +00:00
fneg-fabs.ll R600/SI: Fix fneg for 0.0 2014-02-04 07:12:38 +00:00
fneg.ll R600/SI: Fold fabs/fneg into src input modifier 2014-05-10 19:18:39 +00:00
fp16_to_fp.ll R600: support f16 -> f64 conversion intrinsic. 2014-07-18 08:43:24 +00:00
fp32_to_fp16.ll R600: rename misleading fp16 test. 2014-07-18 08:43:30 +00:00
fp64_to_sint.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
fp_to_sint.ll R600: Implement float to long/ulong 2014-07-10 22:40:21 +00:00
fp_to_uint.f64.ll R600/SI: Match fp_to_uint / uint_to_fp for f64 2014-05-22 03:20:30 +00:00
fp_to_uint.ll R600: Implement float to long/ulong 2014-07-10 22:40:21 +00:00
fpext.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
fptrunc.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
fsqrt.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
fsub.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
fsub64.ll R600/SI: Prettier operand printing for 64-bit ops. 2014-06-18 17:13:51 +00:00
ftrunc.ll R600/SI: Add a pattern for f32 ftrunc 2014-06-20 17:06:09 +00:00
gep-address-space.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
gv-const-addrspace-fail.ll R600: Partially fix constant initializers for structs and vectors. 2014-05-21 22:42:42 +00:00
gv-const-addrspace.ll R600/SI: Use scratch memory for large private arrays 2014-07-21 15:45:01 +00:00
half.ll R600: support fpext/fptrunc operations to and from f16. 2014-07-18 13:01:37 +00:00
i8-to-double-to-float.ll R600: Use KCache for kernel arguments 2013-07-23 01:48:18 +00:00
icmp-select-sete-reverse-args.ll R600: Non vector only instruction can be scheduled on trans unit 2013-09-04 19:53:46 +00:00
icmp64.ll R600/SI: Add i64 cmp tests 2013-12-10 21:11:55 +00:00
imm.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
indirect-addressing-si.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
indirect-private-64.ll R600/SI: Clean up some of the unused REGISTER_{LOAD,STORE} code 2014-07-21 15:45:06 +00:00
infinite-loop-evergreen.ll R600: Add failing control flow tests. 2014-03-01 21:45:41 +00:00
infinite-loop.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
input-mods.ll R600: Expand vector fexp2 2014-06-20 17:06:05 +00:00
insert_vector_elt.ll R600/SI: Fix verifier error with pseudo store instructions. 2014-05-01 16:37:52 +00:00
insert_vector_elt_f64.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
jump-address.ll Add a RequireStructuredCFG Field to TargetMachine. 2013-12-07 01:49:19 +00:00
kcache-fold.ll R600: Fix handling of NAN in comparison instructions 2013-09-28 02:50:50 +00:00
kernel-args.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
large-alloca.ll R600: Use LDS and vectors for private memory 2014-06-17 16:53:14 +00:00
large-constant-initializer.ll R600/SI: Store constant initializer data in constant memory 2014-07-21 14:01:14 +00:00
lds-oqap-crash.ll R600: LDS instructions shouldn't implicitly define OQAP 2014-03-13 17:13:04 +00:00
lds-output-queue.ll Allow aliases to be unnamed_addr. 2014-06-06 01:20:28 +00:00
lds-size.ll Allow aliases to be unnamed_addr. 2014-06-06 01:20:28 +00:00
legalizedag-bug-expand-setcc.ll LegalizeDAG: Respect the result of TLI.getBooleanContents() when expanding SETCC 2013-03-08 15:37:02 +00:00
lit.local.cfg Reduce verbiage of lit.local.cfg files 2014-06-09 22:42:55 +00:00
literals.ll R600: Move code handling literal folding into R600ISelLowering. 2013-09-12 23:44:53 +00:00
llvm.AMDGPU.abs.ll R600: Add a few tests I forgot to add. 2014-06-19 04:24:43 +00:00
llvm.AMDGPU.barrier.global.ll R600: Add missing tests for some intrinsics 2014-07-12 00:36:19 +00:00
llvm.AMDGPU.barrier.local.ll R600: Add missing tests for some intrinsics 2014-07-12 00:36:19 +00:00
llvm.AMDGPU.bfe.i32.ll R600: Try to convert BFE back to standard bit ops when possible. 2014-05-22 18:09:12 +00:00
llvm.AMDGPU.bfe.u32.ll R600: Try to convert BFE back to standard bit ops when possible. 2014-05-22 18:09:12 +00:00
llvm.AMDGPU.bfi.ll R600: Add BFE, BFI, and BFM intrinsics to help with writing tests. 2014-03-31 18:21:18 +00:00
llvm.AMDGPU.bfm.ll R600: Add BFE, BFI, and BFM intrinsics to help with writing tests. 2014-03-31 18:21:18 +00:00
llvm.AMDGPU.brev.ll R600/SI: Add intrinsics for brev instructions 2014-06-18 17:13:57 +00:00
llvm.AMDGPU.clamp.ll R600: Add a few tests I forgot to add. 2014-06-19 04:24:43 +00:00
llvm.AMDGPU.cube.ll R600: Use new getNamedOperandIdx function generated by TableGen 2013-06-25 21:22:18 +00:00
llvm.AMDGPU.cvt_f32_ubyte.ll R600/SI: Use v_cvt_f32_ubyte* instructions 2014-06-11 17:50:44 +00:00
llvm.AMDGPU.div_fixup.ll R600/SI: Add intrinsics for various math instructions. 2014-06-19 01:19:19 +00:00
llvm.AMDGPU.div_fmas.ll R600/SI: Add intrinsics for various math instructions. 2014-06-19 01:19:19 +00:00
llvm.AMDGPU.div_scale.ll R600/SI: Fix div_scale intrinsic. 2014-06-23 18:28:28 +00:00
llvm.AMDGPU.fract.ll R600: Add a few tests I forgot to add. 2014-06-19 04:24:43 +00:00
llvm.AMDGPU.imad24.ll R600: Expand mad24 for GPUs without it 2014-05-22 18:00:20 +00:00
llvm.AMDGPU.imax.ll R600/SI: Move instruction patterns to scalar versions. 2014-03-21 18:01:18 +00:00
llvm.AMDGPU.imin.ll R600/SI: Move instruction patterns to scalar versions. 2014-03-21 18:01:18 +00:00
llvm.AMDGPU.imul24.ll R600: Expand mul24 for GPUs without it 2014-05-22 18:00:24 +00:00
llvm.AMDGPU.kill.ll R600: Add missing tests for some intrinsics 2014-07-12 00:36:19 +00:00
llvm.AMDGPU.legacy.rsq.ll R600: Fix inconsistency in rsq instructions. 2014-06-24 22:13:39 +00:00
llvm.AMDGPU.mul.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
llvm.AMDGPU.rcp.f64.ll R600/SI: Allow using f32 rcp / rsq when denormals not handled. 2014-07-15 23:50:10 +00:00
llvm.AMDGPU.rcp.ll R600/SI: Allow using f32 rcp / rsq when denormals not handled. 2014-07-15 23:50:10 +00:00
llvm.AMDGPU.rsq.clamped.f64.ll R600: Fix inconsistency in rsq instructions. 2014-06-24 22:13:39 +00:00
llvm.AMDGPU.rsq.clamped.ll R600: Fix inconsistency in rsq instructions. 2014-06-24 22:13:39 +00:00
llvm.AMDGPU.rsq.ll R600: Fix inconsistency in rsq instructions. 2014-06-24 22:13:39 +00:00
llvm.AMDGPU.tex.ll R600: Swizzle texture/export instructions 2013-06-04 15:04:53 +00:00
llvm.AMDGPU.trig_preop.ll R600/SI: Add intrinsics for various math instructions. 2014-06-19 01:19:19 +00:00
llvm.AMDGPU.trunc.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
llvm.AMDGPU.umad24.ll R600: Expand mad24 for GPUs without it 2014-05-22 18:00:20 +00:00
llvm.AMDGPU.umax.ll R600: Compute masked bits for min and max 2014-03-31 19:35:33 +00:00
llvm.AMDGPU.umin.ll R600: Compute masked bits for min and max 2014-03-31 19:35:33 +00:00
llvm.AMDGPU.umul24.ll R600: Expand mul24 for GPUs without it 2014-05-22 18:00:24 +00:00
llvm.SI.fs.interp.constant.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
llvm.SI.gather4.ll R600/SI: add gather4 and getlod intrinsics (v3) 2014-06-18 22:00:29 +00:00
llvm.SI.getlod.ll R600/SI: add gather4 and getlod intrinsics (v3) 2014-06-18 22:00:29 +00:00
llvm.SI.image.ll R600/SI: Use i32 vectors for resources and samplers 2014-07-11 17:11:52 +00:00
llvm.SI.image.sample.ll R600/SI: Use i32 vectors for resources and samplers 2014-07-11 17:11:52 +00:00
llvm.SI.image.sample.o.ll R600/SI: Use i32 vectors for resources and samplers 2014-07-11 17:11:52 +00:00
llvm.SI.imageload.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
llvm.SI.load.dword.ll R600/SI: Add intrinsic for BUFFER_LOAD_DWORD* instructions 2014-01-27 07:20:51 +00:00
llvm.SI.resinfo.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
llvm.SI.sample-masked.ll R600/SI: Add ShaderType attribute to some tests 2014-02-13 23:34:07 +00:00
llvm.SI.sample.ll R600/SI: Add ShaderType attribute to some tests 2014-02-13 23:34:07 +00:00
llvm.SI.sampled.ll R600/SI: Add ShaderType attribute to some tests 2014-02-13 23:34:07 +00:00
llvm.SI.sendmsg.ll R600/SI: Add intrinsic for S_SENDMSG instruction 2014-01-27 07:20:44 +00:00
llvm.SI.tbuffer.store.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
llvm.SI.tid.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
llvm.amdgpu.dp4.ll R600: Add missing tests for some intrinsics 2014-07-12 00:36:19 +00:00
llvm.amdgpu.kilp.ll R600: Add missing tests for some intrinsics 2014-07-12 00:36:19 +00:00
llvm.amdgpu.lrp.ll R600: Add missing tests for some intrinsics 2014-07-12 00:36:19 +00:00
llvm.cos.ll R600: Expand vector sin and cos. 2014-05-02 15:41:47 +00:00
llvm.exp2.ll R600: Expand vector fexp2 2014-06-20 17:06:05 +00:00
llvm.floor.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
llvm.log2.ll R600: Expand vector flog2 2014-06-20 17:06:07 +00:00
llvm.pow.ll R600: Enable vector fpow. 2014-02-04 17:18:37 +00:00
llvm.rint.f64.ll R600: Custom lower f64 frint for pre-CI 2014-06-18 17:05:26 +00:00
llvm.rint.ll R600: Mostly remove remaining AMDIL intrinsics. 2014-06-12 21:15:44 +00:00
llvm.round.ll R600: Add support for ISD::FROUND 2013-11-27 21:23:20 +00:00
llvm.sin.ll R600/SI: implement range reduction for sin/cos 2014-07-19 18:44:39 +00:00
llvm.sqrt.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
llvm.trunc.ll R600: Allow ftrunc 2013-12-20 05:11:55 +00:00
load-i1.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
load-input-fold.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
load.ll R600: Add dag combine for copy of an illegal type. 2014-07-15 02:06:31 +00:00
load.vec.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
load64.ll R600/SI: Minor improvements to test. 2013-12-14 00:38:04 +00:00
local-64.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
local-atomics.ll R600/SI: Use a register set to -1 for data0 on ds_inc*/ds_dec* 2014-06-12 08:21:54 +00:00
local-atomics64.ll R600/SI: Use a register set to -1 for data0 on ds_inc*/ds_dec* 2014-06-12 08:21:54 +00:00
local-memory-two-objects.ll Allow aliases to be unnamed_addr. 2014-06-06 01:20:28 +00:00
local-memory.ll Allow aliases to be unnamed_addr. 2014-06-06 01:20:28 +00:00
loop-address.ll R600: add a pass that merges clauses. 2013-10-01 19:32:58 +00:00
loop-idiom.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
lshl.ll R600/SI: Prefer SALU instructions for bit shift operations 2013-11-13 23:36:37 +00:00
lshr.ll R600/SI: Prefer SALU instructions for bit shift operations 2013-11-13 23:36:37 +00:00
mad_int24.ll R600: Match 24-bit arithmetic patterns in a Target DAGCombine 2014-04-07 19:45:41 +00:00
mad_uint24.ll R600: Match 24-bit arithmetic patterns in a Target DAGCombine 2014-04-07 19:45:41 +00:00
max-literals.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
mubuf.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
mul.ll R600: Fix test. Using wrong check prefix. 2014-06-05 08:00:36 +00:00
mul_int24.ll R600: Match 24-bit arithmetic patterns in a Target DAGCombine 2014-04-07 19:45:41 +00:00
mul_uint24.ll SelectionDAG: Use helper function to improve legalization of ISD::MUL 2014-04-11 16:12:01 +00:00
mulhu.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
no-initializer-constant-addrspace.ll R600: Fix asserts related to constant initializers 2014-06-14 04:26:05 +00:00
or.ll R600/SI: Fix select on i1 2014-07-15 21:44:37 +00:00
packetizer.ll R600: Fix typo in R600Schedule.td 2013-06-25 02:39:20 +00:00
parallelandifcollapse.ll R600: Use LDS and vectors for private memory 2014-06-17 16:53:14 +00:00
parallelorifcollapse.ll R600: Use LDS and vectors for private memory 2014-06-17 16:53:14 +00:00
predicate-dp4.ll R600: Make dot_4 instructions predicable 2013-11-16 16:24:41 +00:00
predicates.ll R600: Enable the IR structurizer by default 2013-11-18 19:43:44 +00:00
private-memory-atomics.ll R600: Don't crash on unhandled instruction in promote alloca 2014-06-27 16:52:49 +00:00
private-memory-broken.ll Revert "Temporary hack to try cleaning extra .s file from bots." 2014-06-27 23:11:26 +00:00
private-memory.ll R600/SI: Use scratch memory for large private arrays 2014-07-21 15:45:01 +00:00
pv-packing.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
pv.ll R600/SI: Add intrinsics for various math instructions. 2014-06-19 01:19:19 +00:00
r600-encoding.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
r600-export-fix.ll [SDAG] Make the DAGCombine worklist not grow endlessly due to duplicate 2014-07-23 07:08:53 +00:00
r600-infinite-loop-bug-while-reorganizing-vector.ll R600: Fix an infinite loop when trying to reorganize export/tex vector input 2013-12-10 14:43:31 +00:00
r600cfg.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
reciprocal.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
register-count-comments.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
reorder-stores.ll R600: Fix mishandling of load / store chains. 2014-07-07 18:34:45 +00:00
rotl.i64.ll R600/SI: Fix selection error on i64 rotl / rotr. 2014-06-13 04:00:30 +00:00
rotl.ll R600/SI: Fix selection error on i64 rotl / rotr. 2014-06-13 04:00:30 +00:00
rotr.i64.ll R600: Add a rotr testcase I forgot to add 2014-06-15 21:09:00 +00:00
rotr.ll R600/SI: Fix selection error on i64 rotl / rotr. 2014-06-13 04:00:30 +00:00
rsq.ll R600/SI: Allow using f32 rcp / rsq when denormals not handled. 2014-07-15 23:50:10 +00:00
rv7x0_count3.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
saddo.ll R600: Move add/sub with overflow out of AMDILISelLowering 2014-06-23 18:00:49 +00:00
salu-to-valu.ll R600/SI: Teach SIInstrInfo::moveToVALU() how to move S_LOAD_*_IMM instructions 2014-05-09 16:42:22 +00:00
scalar_to_vector.ll R600/SI: Fix selection failure on scalar_to_vector 2014-06-11 17:40:32 +00:00
schedule-fs-loop-nested-if.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
schedule-fs-loop-nested.ll R600: Enable -verify-machineinstrs in some tests. 2013-10-01 19:32:38 +00:00
schedule-fs-loop.ll R600: Enable -verify-machineinstrs in some tests. 2013-10-01 19:32:38 +00:00
schedule-if-2.ll R600: Enable -verify-machineinstrs in some tests. 2013-10-01 19:32:38 +00:00
schedule-if.ll R600: Enable -verify-machineinstrs in some tests. 2013-10-01 19:32:38 +00:00
schedule-vs-if-nested-loop-failure.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
schedule-vs-if-nested-loop.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
sdiv.ll R600: Fix assert on vector sdiv 2014-06-15 21:08:54 +00:00
select-i1.ll R600/SI: Fix select on i1 2014-07-15 21:44:37 +00:00
select-vectors.ll R600/SI: Expand selects on vectors. 2014-03-06 17:34:03 +00:00
select.ll R600/SI: Fix select on i1 2014-07-15 21:44:37 +00:00
select64.ll Revert "Revert r212640, "Add trunc (select c, a, b) -> select c (trunc a), (trunc b) combine."" 2014-07-10 18:21:04 +00:00
selectcc-cnd.ll R600: Non vector only instruction can be scheduled on trans unit 2013-09-04 19:53:46 +00:00
selectcc-cnde-int.ll R600: Non vector only instruction can be scheduled on trans unit 2013-09-04 19:53:46 +00:00
selectcc-icmp-select-float.ll R600: Prettier asmPrint of Alu 2013-05-02 21:52:30 +00:00
selectcc-opt.ll R600/SI: Remove dead code and add missing tests. 2014-07-20 06:11:02 +00:00
selectcc.ll R600: Expand i64 SELECT_CC 2014-05-09 16:42:19 +00:00
set-dx10.ll R600: Fix handling of NAN in comparison instructions 2013-09-28 02:50:50 +00:00
setcc-equivalent.ll Revert "SelectionDAG: Enable (and (setcc x), (setcc y)) -> (setcc (and x, y)) for vectors" 2014-06-12 16:04:47 +00:00
setcc-opt.ll R600/SI: Remove dead code and add missing tests. 2014-07-20 06:11:02 +00:00
setcc.ll R600/SI: Use VALU instructions for i1 ops 2014-05-15 14:41:50 +00:00
setcc64.ll R600/SI: Use VALU instructions for i1 ops 2014-05-15 14:41:50 +00:00
seto.ll R600/SI: Add instruction shrinking pass 2014-07-21 16:55:33 +00:00
setuo.ll R600/SI: Add instruction shrinking pass 2014-07-21 16:55:33 +00:00
sext-in-reg.ll R600: Try to convert BFE back to standard bit ops when possible. 2014-05-22 18:09:12 +00:00
sgpr-control-flow.ll R600/SI: Only select SALU instructions in the entry or exit block 2014-04-29 23:12:48 +00:00
sgpr-copy-duplicate-operand.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
sgpr-copy.ll R600/SI: Add intrinsics for various math instructions. 2014-06-19 01:19:19 +00:00
shared-op-cycle.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
shl.ll R600: Implement 64bit SHL 2014-06-18 12:27:13 +00:00
si-annotate-cf-assertion.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
si-lod-bias.ll R600/SI: Change formatting of printed registers. 2013-11-12 02:35:51 +00:00
si-sgpr-spill.ll R600/SI: Add intrinsics for various math instructions. 2014-06-19 01:19:19 +00:00
si-vector-hang.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
sign_extend.ll R600/SI: Handle sign_extend and zero_extend to i64 with patterns. 2014-06-10 18:54:59 +00:00
simplify-demanded-bits-build-pair.ll R600: Run more tests with promote alloca disabled. 2014-07-13 02:46:17 +00:00
sint_to_fp.ll R600/SI: Fix [s|u]int_to_fp for i1 2014-05-31 06:47:42 +00:00
sint_to_fp64.ll SelectionDAG: Expand SELECT_CC to SELECT + SETCC 2014-06-10 16:01:22 +00:00
smrd.ll R600/SI: Fix SMRD pattern for offsets > 32 bits 2014-05-09 16:42:21 +00:00
sra.ll R600: Implement 64bit SRA 2014-06-18 12:27:17 +00:00
srem.ll R600: Remove a few more things from AMDILISelLowering 2014-06-15 21:08:58 +00:00
srl.ll R600: Implement 64bit SRL 2014-06-18 12:27:15 +00:00
ssubo.ll R600: Move add/sub with overflow out of AMDILISelLowering 2014-06-23 18:00:49 +00:00
store-v3i32.ll R600: Add failing testcase for <3 x i32> stores. 2014-03-25 16:50:55 +00:00
store-v3i64.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
store-vector-ptrs.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
store.ll R600: Promote i64 stores to v2i32 2014-06-24 23:33:04 +00:00
store.r600.ll R600: Change the RAT instruction assembly names so they match the docs 2013-08-16 01:11:46 +00:00
structurize.ll R600: Enable the IR structurizer by default 2013-11-18 19:43:44 +00:00
structurize1.ll R600: Fix a crash in the AMDILCFGStrucurizer 2013-11-18 19:43:38 +00:00
sub.ll R600/SI: Handle i64 sub. 2014-06-23 18:00:38 +00:00
swizzle-export.ll [SDAG] Make the DAGCombine worklist not grow endlessly due to duplicate 2014-07-23 07:08:53 +00:00
tex-clause-antidep.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
texture-input-merge.ll R600: Use function inputs to represent data stored in gpr 2013-11-11 22:10:24 +00:00
trunc-store-i1.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
trunc-vector-store-assertion-failure.ll SelectionDAG: Make sure stores are always added to the LegalizedNodes list 2013-08-21 22:42:58 +00:00
trunc.ll R600/SI: Print more immediates in hex format 2014-04-15 22:32:49 +00:00
uaddo.ll R600: Move add/sub with overflow out of AMDILISelLowering 2014-06-23 18:00:49 +00:00
udiv.ll R600: Fix LowerUDIVREM 2013-11-06 17:36:04 +00:00
udivrem.ll R600: Add udivrem test 2014-06-22 21:42:58 +00:00
udivrem64.ll R600: Try to convert BFE back to standard bit ops when possible. 2014-05-22 18:09:12 +00:00
uint_to_fp.f64.ll SelectionDAG: Expand SELECT_CC to SELECT + SETCC 2014-06-10 16:01:22 +00:00
uint_to_fp.ll R600/SI: Fix [s|u]int_to_fp for i1 2014-05-31 06:47:42 +00:00
unaligned-load-store.ll R600/SI: Custom lower SI_IF and SI_ELSE to avoid machine verifier errors 2014-04-29 23:12:53 +00:00
unhandled-loop-condition-assertion.ll R600: Add failing control flow tests. 2014-03-01 21:45:41 +00:00
unroll.ll R600: Unconditionally unroll loops that contain GEPs with alloca pointers 2014-01-23 18:49:28 +00:00
unsupported-cc.ll R600: Fix handling of NAN in comparison instructions 2013-09-28 02:50:50 +00:00
urecip.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
urem.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
usubo.ll R600: Move add/sub with overflow out of AMDILISelLowering 2014-06-23 18:00:49 +00:00
v1i64-kernel-arg.ll R600: Match sign_extend_inreg to BFE instructions 2014-03-17 18:58:11 +00:00
v_cndmask.ll R600/SI: Promote f32 SELECT to i32 2014-05-16 20:56:41 +00:00
valu-i1.ll R600/SI: Use VALU instructions for copying i1 values 2014-04-30 15:31:33 +00:00
vector-alloca.ll R600: Run more tests with promote alloca disabled. 2014-07-13 02:46:17 +00:00
vertex-fetch-encoding.ll R600: Add support for v4i32 stores on Cayman 2013-08-16 01:12:00 +00:00
vselect.ll R600/SI: Use -verify-machineinstrs for most tests 2013-10-10 17:11:46 +00:00
vselect64.ll SelectionDAG: Optimize expansion of vec_type = BITCAST scalar_type 2013-11-22 00:41:05 +00:00
vtx-fetch-branch.ll R600: Correctly handle vertex fetch clauses the precede ENDIFs 2014-01-23 18:49:31 +00:00
vtx-schedule.ll Fix broken CHECK lines. 2014-01-11 21:06:00 +00:00
wait.ll R600: improve dump of S_WAITCNT 2013-10-13 17:56:28 +00:00
work-item-intrinsics.ll R600/SI: Use scratch memory for large private arrays 2014-07-21 15:45:01 +00:00
wrong-transalu-pos-fix.ll TBAA: remove !tbaa from testing cases when they are not needed. 2013-09-30 18:17:35 +00:00
xor.ll R600/SI: Use VALU for i1 XOR 2014-07-21 14:01:10 +00:00
zero_extend.ll R600/SI: Use SReg_64 instead of VSrc_64 when selecting BUILD_PAIR 2014-04-18 00:36:21 +00:00

README 

+==============================================================================+
| How to organize the lit tests                                                |
+==============================================================================+

- If you write a test for matching a single DAG opcode or intrinsic, it should
  go in a file called {opcode_name,intrinsic_name}.ll (e.g. fadd.ll)

- If you write a test that matches several DAG opcodes and checks for a single
  ISA instruction, then that test should go in a file called {ISA_name}.ll (e.g.
  bfi_int.ll

- For all other tests, use your best judgement for organizing tests and naming
  the files.

+==============================================================================+
| Naming conventions                                                           |
+==============================================================================+

- Use dash '-' and not underscore '_' to separate words in file names, unless
  the file is named after a DAG opcode or ISA instruction that has an
  underscore '_' in its name.