Arm64/SVE: Implemented `ConvertToint64` and `ConvertToUInt64` #104069

ebepho · 2024-06-27T05:26:25Z

Contributes to #99957

ConvertToInt64 and ConvertToUInt64 stress test output:

Passed test: _Sve_r::JIT.HardwareIntrinsics.Arm._Sve.Program.Sve_ConvertToInt64_long_float() : 13
Passed test: _Sve_r::JIT.HardwareIntrinsics.Arm._Sve.Program.Sve_ConvertToInt64_long_double() : 13
Passed test: _Sve_r::JIT.HardwareIntrinsics.Arm._Sve.Program.Sve_ConvertToUInt64_ulong_float() : 13
Passed test: _Sve_r::JIT.HardwareIntrinsics.Arm._Sve.Program.Sve_ConvertToUInt64_ulong_double() : 13
===================Running jitstress===================
------------------- {'JitMinOpts': '1'} -------------------
------------------- {'JitStress': '1'} -------------------
------------------- {'JitStress': '2'} -------------------
------------------- {'JitStress': '1', 'TieredCompilation': '1'} -------------------
------------------- {'JitStress': '2', 'TieredCompilation': '1'} -------------------
------------------- {'TailcallStress': '1'} -------------------
------------------- {'ReadyToRun': '0'} -------------------
===================Running jitstressregs===================
------------------- {'JitStressRegs': '1'} -------------------
------------------- {'JitStressRegs': '2'} -------------------
------------------- {'JitStressRegs': '3'} -------------------
------------------- {'JitStressRegs': '4'} -------------------
------------------- {'JitStressRegs': '8'} -------------------
------------------- {'JitStressRegs': '0x10'} -------------------
------------------- {'JitStressRegs': '0x80'} -------------------
------------------- {'JitStressRegs': '0x1000'} -------------------
------------------- {'JitStressRegs': '0x2000'} -------------------
===================Running jitstress2-jitstressregs===================
------------------- {'JitStress': '2', 'JitStressRegs': '1'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '2'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '3'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '4'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '8'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '0x10'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '0x80'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '0x1000'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '0x2000'} -------------------

Passed test: _Sve_ro::JIT.HardwareIntrinsics.Arm._Sve.Program.Sve_ConvertToInt64_long_float() : 13
Passed test: _Sve_ro::JIT.HardwareIntrinsics.Arm._Sve.Program.Sve_ConvertToInt64_long_double() : 13
Passed test: _Sve_ro::JIT.HardwareIntrinsics.Arm._Sve.Program.Sve_ConvertToUInt64_ulong_float() : 13
Passed test: _Sve_ro::JIT.HardwareIntrinsics.Arm._Sve.Program.Sve_ConvertToUInt64_ulong_double() : 13
===================Running jitstress===================
------------------- {'JitMinOpts': '1'} -------------------
------------------- {'JitStress': '1'} -------------------
------------------- {'JitStress': '2'} -------------------
------------------- {'JitStress': '1', 'TieredCompilation': '1'} -------------------
------------------- {'JitStress': '2', 'TieredCompilation': '1'} -------------------
------------------- {'TailcallStress': '1'} -------------------
------------------- {'ReadyToRun': '0'} -------------------
===================Running jitstressregs===================
------------------- {'JitStressRegs': '1'} -------------------
------------------- {'JitStressRegs': '2'} -------------------
------------------- {'JitStressRegs': '3'} -------------------
------------------- {'JitStressRegs': '4'} -------------------
------------------- {'JitStressRegs': '8'} -------------------
------------------- {'JitStressRegs': '0x10'} -------------------
------------------- {'JitStressRegs': '0x80'} -------------------
------------------- {'JitStressRegs': '0x1000'} -------------------
------------------- {'JitStressRegs': '0x2000'} -------------------
===================Running jitstress2-jitstressregs===================
------------------- {'JitStress': '2', 'JitStressRegs': '1'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '2'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '3'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '4'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '8'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '0x10'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '0x80'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '0x1000'} -------------------
------------------- {'JitStress': '2', 'JitStressRegs': '0x2000'} -------------------

cc @dotnet/arm64-contrib

updating branch due to assert issue.

kunalspathak · 2024-06-27T17:13:13Z

src/coreclr/jit/hwintrinsic.cpp

@@ -1731,6 +1731,8 @@ GenTree* Compiler::impHWIntrinsic(NamedIntrinsic        intrinsic,
                {
                    case NI_Sve_ConvertToInt32:
                    case NI_Sve_ConvertToUInt32:
+                    case NI_Sve_ConvertToInt64:


I expected to have entries for ConvertToSingle and ConvertToDouble here as well. Did you confirm this works for scenario where maskSize != operSize in lowering?

kunalspathak · 2024-06-27T17:13:22Z

src/coreclr/jit/lowerarmarch.cpp

-                                   (embOp->GetHWIntrinsicId() == NI_Sve_ConvertToUInt32));
+                            assert((embOp->GetHWIntrinsicId() == NI_Sve_ConvertToInt32)  ||
+                                   (embOp->GetHWIntrinsicId() == NI_Sve_ConvertToUInt32) ||
+                                   (embOp->GetHWIntrinsicId() == NI_Sve_ConvertToInt64)  ||


likewise here.

kunalspathak · 2024-06-27T17:16:25Z

src/coreclr/jit/hwintrinsiccodegenarm64.cpp

-                            opt = intrinEmbMask.baseType == TYP_DOUBLE ? INS_OPTS_D_TO_S : INS_OPTS_SCALABLE_S;
+                            embOpt = intrinEmbMask.baseType == TYP_DOUBLE ? INS_OPTS_D_TO_S : INS_OPTS_SCALABLE_S;
+                            break;
+                        }


This also needs handling of ConvertToSingle

embOpt = intrinEmbMask.baseType == TYP_DOUBLE ? INS_OPTS_D_TO_S : INS_OPTS_SCALABLE_S;

and ConvertToDouble:

embOpt = intrinEmbMask.baseType == TYP_FLOAT ? INS_OPTS_S_TO_D : INS_OPTS_SCALABLE_D;

which means you can combine ConvertToSingle with Convert*32 and ConvertToDouble with Convert*64.

kunalspathak

Added some comments around test. code changes looks good.

kunalspathak · 2024-06-28T01:01:46Z

src/tests/Common/GenerateHWIntrinsicTests/GenerateHWIntrinsicTests_Arm.cs

@@ -78,9 +78,9 @@
                }
            }";

-const string SimpleVecOpTest_ValidationLogicForCndSelForNarrowing = @"for (var i = 0; i < Op1ElementCount; i++)
+const string SimpleVecOpTest_ValidationLogicForCndSelNarrowing = @"for (var i = 0; i < Op1ElementCount; i++)


why this change?

I reverted this change -- shouldn't have been in this PR

kunalspathak · 2024-06-28T01:02:29Z

src/tests/Common/GenerateHWIntrinsicTests/GenerateHWIntrinsicTests_Arm.cs

-    ("SveSimpleVecOpDifferentRetTypeTest.template",        new Dictionary<string, string> { ["TestName"] = "Sve_ConvertToInt32_int_double",                                                                         ["Isa"] = "Sve",           ["LoadIsa"] = "Sve",     ["Method"] = "ConvertToInt32",                                                       ["RetVectorType"] = "Vector",    ["RetBaseType"] = "Int32",   ["Op1VectorType"] = "Vector",    ["Op1BaseType"] = "Double",                                                                                                                           ["LargestVectorSize"] = "64", ["NextValueOp1"] = "TestLibrary.Generator.GetDouble()",                                                                                                                                                      ["ValidateIterResult"] = "Helpers.ConvertDoubleToInt32(firstOp[i]) != result[i]",  ["GetIterResult"] = "Helpers.ConvertDoubleToInt32(leftOp[i])"}),
-    ("SveSimpleVecOpDifferentRetTypeTest.template",        new Dictionary<string, string> { ["TestName"] = "Sve_ConvertToUInt32_uint_float",                                                                        ["Isa"] = "Sve",           ["LoadIsa"] = "Sve",     ["Method"] = "ConvertToUInt32",                                                      ["RetVectorType"] = "Vector",    ["RetBaseType"] = "UInt32",  ["Op1VectorType"] = "Vector",    ["Op1BaseType"] = "Single",                                                                                                                           ["LargestVectorSize"] = "64", ["NextValueOp1"] = "TestLibrary.Generator.GetSingle()",                                                                                                                                                      ["ValidateIterResult"] = "Helpers.ConvertToUInt32(firstOp[i]) != result[i]",       ["GetIterResult"] = "Helpers.ConvertToUInt32(leftOp[i])"}),
-    ("SveSimpleVecOpDifferentRetTypeTest.template",        new Dictionary<string, string> { ["TestName"] = "Sve_ConvertToUInt32_uint_double",                                                                       ["Isa"] = "Sve",           ["LoadIsa"] = "Sve",     ["Method"] = "ConvertToUInt32",                                                      ["RetVectorType"] = "Vector",    ["RetBaseType"] = "UInt32",  ["Op1VectorType"] = "Vector",    ["Op1BaseType"] = "Double",                                                                                                                           ["LargestVectorSize"] = "64", ["NextValueOp1"] = "TestLibrary.Generator.GetDouble()",                                                                                                                                                      ["ValidateIterResult"] = "Helpers.ConvertDoubleToUInt32(firstOp[i]) != result[i]", ["GetIterResult"] = "Helpers.ConvertDoubleToUInt32(leftOp[i])"}),
+


can you delete the 2 empty lines?

kunalspathak · 2024-06-28T01:03:00Z

src/tests/Common/GenerateHWIntrinsicTests/GenerateHWIntrinsicTests_Arm.cs

@@ -177,7 +177,7 @@
    ("_BinaryOpTestTemplate.template",                   "SecureHashBinOpTest.template",                new Dictionary<string, string> { ["TemplateName"] = "SecureHash", ["TemplateValidationLogic"] = SecureHashOpTest_ValidationLogic }),
    ("_TernaryOpTestTemplate.template",                  "SecureHashTernOpTest.template",               new Dictionary<string, string> { ["TemplateName"] = "SecureHash", ["TemplateValidationLogic"] = SecureHashOpTest_ValidationLogic }),
    ("_SveUnaryOpTestTemplate.template",                 "SveSimpleVecOpTest.template",                 new Dictionary<string, string> { ["TemplateName"] = "Simple",     ["TemplateValidationLogic"] = SimpleVecOpTest_ValidationLogic,             ["TemplateValidationLogicForCndSel"] = SimpleVecOpTest_ValidationLogicForCndSel }),
-    ("_SveUnaryOpDifferentRetTypeTestTemplate.template", "SveSimpleVecOpDifferentRetTypeTest.template", new Dictionary<string, string> { ["TemplateName"] = "Simple",     ["TemplateValidationLogic"] = SimpleVecOpTest_ValidationLogicForNarrowing, ["TemplateValidationLogicForCndSel"] = SimpleVecOpTest_ValidationLogicForCndSelForNarrowing }),
+    ("_SveUnaryOpDifferentRetTypeTestTemplate.template", "SveSimpleVecOpDifferentRetTypeTest.template", new Dictionary<string, string> { ["TemplateName"] = "Simple",     ["TemplateValidationLogic"] = SimpleVecOpTest_ValidationLogicForNarrowing, ["TemplateValidationLogicForCndSel"] = SimpleVecOpTest_ValidationLogicForCndSelNarrowing }),


likewise here.

kunalspathak · 2024-06-28T15:07:36Z

src/tests/Common/GenerateHWIntrinsicTests/GenerateHWIntrinsicTests_Arm.cs

@@ -80,7 +80,7 @@

 const string SimpleVecOpTest_ValidationLogicForCndSelForNarrowing = @"for (var i = 0; i < Op1ElementCount; i++)
            {
-                {Op1BaseType} iterResult = (mask[i] != 0) ? {GetIterResult} : falseVal[i];
+                {RetBaseType} iterResult = (mask[i] != 0) ? {GetIterResult} : falseVal[i];


Is this needed too?

Suggested change

{RetBaseType} iterResult = (mask[i] != 0) ? {GetIterResult} : falseVal[i];

{Op1BaseType} iterResult = (mask[i] != 0) ? {GetIterResult} : falseVal[i];

I'm going to test this version one more time. Not sure why these changes even appeared, but I think they were supposed to be part of the ConvertToInt32 commit. These changes may be necessary for the tests to run properly.

kunalspathak

LGTM

kunalspathak · 2024-06-28T22:09:21Z

/azp

azure-pipelines · 2024-06-28T22:09:27Z

Supported commands help: Get descriptions, examples and documentation about supported commands Example: help "command_name" list: List all pipelines for this repository using a comment. Example: "list" run: Run all pipelines or specific pipelines for this repository using a comment. Use this command by itself to trigger all related pipelines, or specify specific pipelines to run. Example: "run" or "run pipeline_name, pipeline_name, pipeline_name" where: Report back the Azure DevOps orgs that are related to this repository and org Example: "where" See additional documentation.

ebepho and others added 30 commits June 5, 2024 14:30

Added ConverToInt32 and ConvertToUInt32 for float inputs.

c738b77

Added flags to handle only low predicate registers.

35d39d9

Fixed merge conflicts.

7a781e1

Fix whitespace

1378d60

Remove special codegen flag

10c7a15

Added new test template for operations with different return types.

8004868

Merge branch 'main' into ConvertToInt32

af7ccd4

updating branch due to assert issue.

Add new test template.

8cb76da

Added api for ConvertToInt32 and ConvertToUInt 32 for double.

abe25fc

fix merge conflicts.

0f51f38

Merge branch 'dotnet:main' into main

7fabb91

Completed SVE Apis for ConvertToInt64 and ConvertToUInt64.

11affed

Merge branch 'main' of github.com:ebepho/runtime

d5374ca

Merge branch 'main' into converttoint64

711b28a

ConvertToSingle for int and uint.

fe32a2f

ConvertToSingle for long and ulong.

478b969

Merge branch 'main' of github.com:ebepho/runtime

4aa224d

Merge branch 'main' into ConvertToSingleDouble

cc63edf

Started ConvertToDouble.

ff54068

Merge branch 'main' of github.com:ebepho/runtime

56601b4

Merge branch 'main' into converttoint64

422068b

Changed Validation Template Test name.

5b4c4f3

Merge branch 'main' of github.com:ebepho/runtime

04071a3

Merge branch 'main' of github.com:ebepho/runtime

ffcd267

Merge branch 'main' into converttoint64

d4b8dc3

Merge branch 'main' into converttosingledouble

5ac4a05

Merge branch 'converttosingledouble' into ConvertTo_Int64_Single_Double

f055d0c

Merge branch 'main' of github.com:ebepho/runtime

33626b3

Merge branch 'main' into ConvertTo_Int64_Single_Double

0327fa6

Merge branch 'main' of github.com:ebepho/runtime

da441d1

ebepho added the arm-sve Work related to arm64 SVE/SVE2 support label Jun 27, 2024

kunalspathak mentioned this pull request Jun 27, 2024

Arm64: Implement SVE APIs #99957

Closed

kunalspathak reviewed Jun 27, 2024

View reviewed changes

ebepho added 4 commits June 27, 2024 10:37

Added cases for ConvertToSingle and ConvertToDouble.

a0c7333

double or long to 32 bit value.

37e1da1

Removed ConvertToDouble and ConvertToSingle.

2c13be7

Removed more of ConvertToSingle and ConvertToDouble.

2f3c901

ebepho changed the title ~~Arm64/SVE: Implemented ConvertToint64, ConvertToUInt64, ConvertToDouble and ConvertToSingle~~ Arm64/SVE: Implemented ConvertToint64 and ConvertToUInt64 Jun 28, 2024

ebepho marked this pull request as ready for review June 28, 2024 00:54

kunalspathak requested changes Jun 28, 2024

View reviewed changes

ebepho added 2 commits June 27, 2024 18:51

all tests pass.

990c75b

addressed comments.

eaf3905

kunalspathak reviewed Jun 28, 2024

View reviewed changes

jit format:

81f11cb

kunalspathak approved these changes Jun 28, 2024

View reviewed changes

Remove trailing space

c2d5d15

This was referenced Jun 28, 2024

Crash in Microsoft.Extensions.Logging.Generators.Roslyn4.0.Tests.WorkItemExecution #90019

Open

tracing/eventpipe/providervalidation/providervalidation/providervalidation.cmd test failure #101759

Closed

kunalspathak closed this Jun 28, 2024

kunalspathak reopened this Jun 28, 2024

kunalspathak closed this Jun 28, 2024

kunalspathak reopened this Jun 28, 2024

This was referenced Jun 29, 2024

[browser][MT] System.Net.Http.Functional.Tests.WorkItemExecution timeout #102385

Closed

[browser][MT] test failure: Tests timed out. Killing driver service #103524

Open

kunalspathak merged commit 6f1d8c5 into dotnet:main Jun 29, 2024
159 of 167 checks passed

github-actions bot locked and limited conversation to collaborators Jul 29, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Arm64/SVE: Implemented `ConvertToint64` and `ConvertToUInt64` #104069

Arm64/SVE: Implemented `ConvertToint64` and `ConvertToUInt64` #104069

ebepho commented Jun 27, 2024 •

edited

Loading

kunalspathak Jun 27, 2024

kunalspathak Jun 27, 2024

kunalspathak Jun 27, 2024

kunalspathak left a comment

kunalspathak Jun 28, 2024

ebepho Jun 28, 2024

kunalspathak Jun 28, 2024

kunalspathak Jun 28, 2024

kunalspathak Jun 28, 2024

kunalspathak Jun 28, 2024

kunalspathak Jun 28, 2024

ebepho Jun 28, 2024 •

edited

Loading

kunalspathak left a comment

kunalspathak commented Jun 28, 2024

azure-pipelines bot commented Jun 28, 2024

	{RetBaseType} iterResult = (mask[i] != 0) ? {GetIterResult} : falseVal[i];
	{Op1BaseType} iterResult = (mask[i] != 0) ? {GetIterResult} : falseVal[i];

Arm64/SVE: Implemented ConvertToint64 and ConvertToUInt64 #104069

Arm64/SVE: Implemented ConvertToint64 and ConvertToUInt64 #104069

Conversation

ebepho commented Jun 27, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kunalspathak left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ebepho Jun 28, 2024 • edited Loading

Choose a reason for hiding this comment

kunalspathak left a comment

Choose a reason for hiding this comment

kunalspathak commented Jun 28, 2024

azure-pipelines bot commented Jun 28, 2024

Arm64/SVE: Implemented `ConvertToint64` and `ConvertToUInt64` #104069

Arm64/SVE: Implemented `ConvertToint64` and `ConvertToUInt64` #104069

ebepho commented Jun 27, 2024 •

edited

Loading

ebepho Jun 28, 2024 •

edited

Loading