<html><head><meta http-equiv="Content-Type" content="text/html charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;">Hi Nadav,<div><br></div><div>Here is a patch that adds the different broadcast instructions to the <span style="font-family: Menlo; font-size: 11px;">ReplaceableInstrsAVX2 </span>table.</div><div>That way the ExeDepsFix pass can take better decisions when AVX2 broadcasts are across domain (int <-> float).</div><div><br></div><div>In particular, prior to this patch we were generating:</div><div><div><span class="Apple-tab-span" style="white-space:pre"> </span>vpbroadcastd<span class="Apple-tab-span" style="white-space:pre"> </span>LCPI1_0(%rip), %ymm2</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>vpand<span class="Apple-tab-span" style="white-space:pre"> </span>%ymm2, %ymm0, %ymm0</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>vmaxps<span class="Apple-tab-span" style="white-space:pre"> </span>%ymm1, %ymm0, %ymm0 ## <- domain change penalty</div></div><div><br></div><div>Now, we generate the following nice sequence where everything is in the float domain:</div><div><div><span class="Apple-tab-span" style="white-space:pre"> </span>vbroadcastss<span class="Apple-tab-span" style="white-space:pre"> </span>LCPI1_0(%rip), %ymm2</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>vandps<span class="Apple-tab-span" style="white-space:pre"> </span>%ymm2, %ymm0, %ymm0</div><div><span class="Apple-tab-span" style="white-space:pre"> </span>vmaxps<span class="Apple-tab-span" style="white-space:pre"> </span>%ymm1, %ymm0, %ymm0</div></div><div><br></div><div>This patch gives a few speed-ups across the llvm test-suite + spec (-O3 -march=core-avx2).</div><div>I have reported here only the tests that show a difference in the disassembly and that run for more than 1s to avoid noise.</div><div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">Benchmark_ID <span class="Apple-tab-span" style="white-space:pre"> </span>Reference<span class="Apple-tab-span" style="white-space:pre"> </span>Test <span class="Apple-tab-span" style="white-space:pre"> </span>Speedup <span class="Apple-tab-span" style="white-space:pre"> </span>Percent</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">-------------------------------------------------------------------------------</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">ASCI_Purple/SMG2000/smg2000 <span class="Apple-tab-span" style="white-space:pre"> </span> 1.5325<span class="Apple-tab-span" style="white-space:pre"> </span> 1.5386<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">CINT2000/181.mcf/181.mcf <span class="Apple-tab-span" style="white-space:pre"> </span> 3.7871<span class="Apple-tab-span" style="white-space:pre"> </span> 3.7725<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">CINT2006/401.bzip2/401.bzip2 <span class="Apple-tab-span" style="white-space:pre"> </span> 1.5876<span class="Apple-tab-span" style="white-space:pre"> </span> 1.5811<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">CINT2006/456.hmmer/456.hmmer <span class="Apple-tab-span" style="white-space:pre"> </span> 1.9414<span class="Apple-tab-span" style="white-space:pre"> </span> 1.941<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">Misc/salsa20 <span class="Apple-tab-span" style="white-space:pre"> </span> 4.7333<span class="Apple-tab-span" style="white-space:pre"> </span> 4.7319<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">PAQ8p/paq8p <span class="Apple-tab-span" style="white-space:pre"> </span> 28.187<span class="Apple-tab-span" style="white-space:pre"> </span> 28.2113<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">Polybench/linear-algebra/kernels/sy<span class="Apple-tab-span" style="white-space:pre"> </span> 11.5787<span class="Apple-tab-span" style="white-space:pre"> </span> 11.5834<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">Polybench/linear-algebra/kernels/sy<span class="Apple-tab-span" style="white-space:pre"> </span> 2.9281<span class="Apple-tab-span" style="white-space:pre"> </span> 2.9271<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/ControlFlow-dbl/ControlFlow-db<span class="Apple-tab-span" style="white-space:pre"> </span> 2.589<span class="Apple-tab-span" style="white-space:pre"> </span> 2.5916<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/ControlFlow-flt/ControlFlow-fl<span class="Apple-tab-span" style="white-space:pre"> </span> 2.2556<span class="Apple-tab-span" style="white-space:pre"> </span> 2.1948<span class="Apple-tab-span" style="white-space:pre"> </span> 1.03<span class="Apple-tab-span" style="white-space:pre"> </span> +3%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/ControlLoops-flt/ControlLoops-<span class="Apple-tab-span" style="white-space:pre"> </span> 1.6671<span class="Apple-tab-span" style="white-space:pre"> </span> 1.6693<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/Equivalencing-dbl/Equivalencin<span class="Apple-tab-span" style="white-space:pre"> </span> 1.5151<span class="Apple-tab-span" style="white-space:pre"> </span> 1.4524<span class="Apple-tab-span" style="white-space:pre"> </span> 1.04<span class="Apple-tab-span" style="white-space:pre"> </span> +4%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/Expansion-dbl/Expansion-dbl <span class="Apple-tab-span" style="white-space:pre"> </span> 2.483<span class="Apple-tab-span" style="white-space:pre"> </span> 2.4818<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/IndirectAddressing-flt/Indirec<span class="Apple-tab-span" style="white-space:pre"> </span> 2.2959<span class="Apple-tab-span" style="white-space:pre"> </span> 2.2353<span class="Apple-tab-span" style="white-space:pre"> </span> 1.03<span class="Apple-tab-span" style="white-space:pre"> </span> +3%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/InductionVariable-dbl/Inductio<span class="Apple-tab-span" style="white-space:pre"> </span> 2.7742<span class="Apple-tab-span" style="white-space:pre"> </span> 2.7731<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/LinearDependence-dbl/LinearDep<span class="Apple-tab-span" style="white-space:pre"> </span> 2.2146<span class="Apple-tab-span" style="white-space:pre"> </span> 2.2224<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/LinearDependence-flt/LinearDep<span class="Apple-tab-span" style="white-space:pre"> </span> 1.5389<span class="Apple-tab-span" style="white-space:pre"> </span> 1.5225<span class="Apple-tab-span" style="white-space:pre"> </span> 1.01<span class="Apple-tab-span" style="white-space:pre"> </span> +1%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/NodeSplitting-dbl/NodeSplittin<span class="Apple-tab-span" style="white-space:pre"> </span> 2.4763<span class="Apple-tab-span" style="white-space:pre"> </span> 2.4753<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/Packing-dbl/Packing-dbl <span class="Apple-tab-span" style="white-space:pre"> </span> 2.595<span class="Apple-tab-span" style="white-space:pre"> </span> 2.5093<span class="Apple-tab-span" style="white-space:pre"> </span> 1.03<span class="Apple-tab-span" style="white-space:pre"> </span> +3%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/Packing-flt/Packing-flt <span class="Apple-tab-span" style="white-space:pre"> </span> 2.2906<span class="Apple-tab-span" style="white-space:pre"> </span> 2.2796<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/Searching-dbl/Searching-dbl <span class="Apple-tab-span" style="white-space:pre"> </span> 2.9092<span class="Apple-tab-span" style="white-space:pre"> </span> 2.8992<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/Searching-flt/Searching-flt <span class="Apple-tab-span" style="white-space:pre"> </span> 2.893<span class="Apple-tab-span" style="white-space:pre"> </span> 2.8579<span class="Apple-tab-span" style="white-space:pre"> </span> 1.01<span class="Apple-tab-span" style="white-space:pre"> </span> +1%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">TSVC/StatementReordering-dbl/Statem<span class="Apple-tab-span" style="white-space:pre"> </span> 2.5175<span class="Apple-tab-span" style="white-space:pre"> </span> 2.5113<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">nbench/nbench <span class="Apple-tab-span" style="white-space:pre"> </span> 5.2255<span class="Apple-tab-span" style="white-space:pre"> </span> 5.2213<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">sqlite3/sqlite3 <span class="Apple-tab-span" style="white-space:pre"> </span> 2.113<span class="Apple-tab-span" style="white-space:pre"> </span> 2.1165<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">-------------------------------------------------------------------------------</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">Min (25) <span class="Apple-tab-span" style="white-space:pre"> </span> -<span class="Apple-tab-span" style="white-space:pre"> </span> -<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> -</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">-------------------------------------------------------------------------------</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">Max (25) <span class="Apple-tab-span" style="white-space:pre"> </span> -<span class="Apple-tab-span" style="white-space:pre"> </span> -<span class="Apple-tab-span" style="white-space:pre"> </span> 1.04<span class="Apple-tab-span" style="white-space:pre"> </span> -</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">-------------------------------------------------------------------------------</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">Sum (25) <span class="Apple-tab-span" style="white-space:pre"> </span> 99<span class="Apple-tab-span" style="white-space:pre"> </span> 98<span class="Apple-tab-span" style="white-space:pre"> </span> 1<span class="Apple-tab-span" style="white-space:pre"> </span> +0%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">-------------------------------------------------------------------------------</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">A.Mean (25) <span class="Apple-tab-span" style="white-space:pre"> </span> -<span class="Apple-tab-span" style="white-space:pre"> </span> -<span class="Apple-tab-span" style="white-space:pre"> </span> 1.01<span class="Apple-tab-span" style="white-space:pre"> </span> +1%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">-------------------------------------------------------------------------------</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">G.Mean 2 (25) <span class="Apple-tab-span" style="white-space:pre"> </span> -<span class="Apple-tab-span" style="white-space:pre"> </span> -<span class="Apple-tab-span" style="white-space:pre"> </span> 1.01<span class="Apple-tab-span" style="white-space:pre"> </span> +1%</div><div style="margin: 0px; font-size: 11px; font-family: Menlo;">-------------------------------------------------------------------------------</div></div><div><br></div><div>Thanks,</div><div><div apple-content-edited="true">
<div style="color: rgb(0, 0, 0); font-family: Helvetica; font-style: normal; font-variant: normal; font-weight: normal; letter-spacing: normal; line-height: normal; orphans: 2; text-align: -webkit-auto; text-indent: 0px; text-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webkit-text-size-adjust: auto; -webkit-text-stroke-width: 0px; word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; ">-Quentin</div>
</div>
</div></body></html>