OSchip
/
llvm-project
mirror of https://github.com/THU-DSP-LAB/llvm-project.git
13
0
Fork 6
Code Issues Packages Projects Releases Wiki Activity
llvm-project/llvm/test/Analysis/CostModel/X86
History
Craig Topper 8dfb9627b7 [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 
This moves v32i16/v64i8 to a model consistent with how we
treat integer types with avx1.

This does change the ABI for types vXi16/vXi8 vectors larger than
512 bits to pass in multiple zmms instead of multiple ymms. We'd
already hacked some code to make v64i8/v32i16 pass in zmm.

Cost model is still a bit of a mess. In some place I tried to
match existing behavior. But really we need to account for
splitting and concating costs. Cost model for shuffles is
especially pessimistic.

Differential Revision: https://reviews.llvm.org/D76212
 		2020-04-15 12:17:18 -07:00
..
aggregates.ll [CostModel] Model all `extractvalue`s as free. 2019-08-29 11:50:30 +00:00
alternate-shuffle-cost.ll [CostModel][X86] Try to check against common prefixes before using target-specific cpu checks 2020-02-24 11:59:07 +00:00
arith-fix.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
arith-fma.ll
…
arith-fp.ll [CostModel][X86] Improve extract/insert element costs (PR43605) 2020-02-27 15:54:13 +00:00
arith-overflow.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
arith-ssat.ll [CostModel][X86] Try to check against common prefixes before using target-specific cpu checks 2020-02-24 11:59:07 +00:00
arith-usat.ll [CostModel][X86] Try to check against common prefixes before using target-specific cpu checks 2020-02-24 11:59:07 +00:00
arith.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
bitreverse.ll
…
bswap.ll
…
cast.ll [X86] Fix the cost model for v16i16->v16i32 zero_extend/sign_extend with AVX2 2020-01-29 15:52:10 -08:00
costmodel.ll
…
ctlz.ll [CostModel][X86] Add CTLZ scalar costs 2019-10-14 16:30:17 +00:00
ctpop.ll [CostModel][X86] Add CTPOP scalar costs (PR43656) 2019-10-14 14:07:43 +00:00
cttz.ll [CostModel][X86] Improve ISD::CTTZ costs accounting for BSF/TZCNT implementations 2020-03-13 16:51:13 +00:00
div.ll [CostModel][X86] Try to check against common prefixes before using target-specific cpu checks 2020-02-24 11:59:07 +00:00
extend.ll [CostModel][X86] Improve extract/insert element costs (PR43605) 2020-02-27 15:54:13 +00:00
fcmp.ll [CostModel][X86] Try to check against common prefixes before using target-specific cpu checks 2020-02-24 11:59:07 +00:00
fmaxnum.ll [X86][SSE] Add more accurate costs for fmaxnum/fminnum codegen 2020-03-10 11:59:40 +00:00
fminnum.ll [X86][SSE] Add more accurate costs for fmaxnum/fminnum codegen 2020-03-10 11:59:40 +00:00
fptosi.ll [CostModel][X86] Improve InsertElement costs for sub-128bit vectors 2020-04-10 14:55:46 +01:00
fptoui.ll [CostModel][X86] Improve InsertElement costs for sub-128bit vectors 2020-04-10 14:55:46 +01:00
fround.ll [CostModel][X86] Improve extract/insert element costs (PR43605) 2020-02-27 15:54:13 +00:00
fshl.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
fshr.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
gep.ll
…
i32.ll
…
icmp.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
insert-extract-at-zero.ll
…
interleave-load-i32.ll
…
interleave-store-i32.ll
…
interleaved-load-float.ll
…
interleaved-load-i8.ll
…
interleaved-load-store-double.ll
…
interleaved-load-store-i64.ll
…
interleaved-store-i8.ll
…
intrinsic-cost.ll [CostModel][X86] Improve extract/insert element costs (PR43605) 2020-02-27 15:54:13 +00:00
lit.local.cfg
…
load_store.ll [CostModel][X86] Regenerate load_store.ll costs tests 2020-04-15 11:54:39 +01:00
loop_v2.ll
…
masked-intrinsic-cost.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
min-legal-vector-width.ll [X86] Add a more accurate truncate cost for v8i64->v8i8 2020-04-13 21:09:41 -07:00
reduce-add.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduce-and.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduce-fmax.ll [X86] Improve min/max reduction costs. 2020-04-09 17:28:50 -07:00
reduce-fmin.ll [X86] Improve min/max reduction costs. 2020-04-09 17:28:50 -07:00
reduce-mul.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduce-or.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduce-smax.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduce-smin.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduce-umax.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduce-umin.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduce-xor.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
reduction.ll [X86] More accurately model the cost of horizontal reductions. 2020-03-22 14:20:15 -07:00
rem.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
scalarize.ll
…
shuffle-broadcast.ll [CostModel][X86] Add shuffle costs for some common sub-128bit vectors 2020-04-09 19:57:06 +01:00
shuffle-extract_subvector.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
shuffle-insert_subvector.ll [CostModel][X86] Add shuffle costs for some common sub-128bit vectors 2020-04-09 19:57:06 +01:00
shuffle-reverse.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
shuffle-single-src.ll [CostModel][X86] Add shuffle costs for some common sub-128bit vectors 2020-04-09 19:57:06 +01:00
shuffle-transpose.ll [CostModel][X86] Add shuffle costs for some common sub-128bit vectors 2020-04-09 19:57:06 +01:00
shuffle-two-src.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
sitofp.ll [CostModel][X86] Improve extract/insert element costs (PR43605) 2020-02-27 15:54:13 +00:00
size-cost.ll [x86] add size cost tests for casts and binops; NFC 2020-04-06 12:38:15 -04:00
slm-arith-costs.ll Recommit r367901 "[X86] Enable -x86-experimental-vector-widening-legalization by default." 2019-08-07 16:24:26 +00:00
sse-itoi.ll
…
strided-load-i8.ll
…
strided-load-i16.ll
…
strided-load-i32.ll
…
strided-load-i64.ll
…
testshiftashr.ll Recommit r367901 "[X86] Enable -x86-experimental-vector-widening-legalization by default." 2019-08-07 16:24:26 +00:00
testshiftlshr.ll Recommit r367901 "[X86] Enable -x86-experimental-vector-widening-legalization by default." 2019-08-07 16:24:26 +00:00
testshiftshl.ll Recommit r367901 "[X86] Enable -x86-experimental-vector-widening-legalization by default." 2019-08-07 16:24:26 +00:00
tiny.ll
…
trunc.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
uitofp.ll [CostModel][X86] Try to check against common prefixes before using target-specific cpu checks 2020-02-24 11:59:07 +00:00
uniformshift.ll
…
vdiv-cost.ll
…
vector-extract.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
vector-insert.ll [X86] Make v32i16/v64i8 legal types without avx512bw. Use custom splitting instead. 2020-04-15 12:17:18 -07:00
vector_gep.ll
…
vectorized-loop.ll
…
vselect-cost.ll [CostModel][X86] Remove comments that begin with a filecheck prefix. 2020-04-13 18:39:24 +01:00
vshift-ashr-cost.ll [CostModel][X86] Improve extract/insert element costs (PR43605) 2020-02-27 15:54:13 +00:00
vshift-lshr-cost.ll [CostModel][X86] Improve extract/insert element costs (PR43605) 2020-02-27 15:54:13 +00:00
vshift-shl-cost.ll [CostModel][X86] Improve extract/insert element costs (PR43605) 2020-02-27 15:54:13 +00:00