JIT: enable devirtualization/inlining of other array interface methods #109209

AndyAyersMS · 2024-10-25T00:28:21Z

The JIT recently enabled devirtualization of GetEnumerator, but other methods were inhibited from devirtualization because the runtime was returning an instantiating stub instead of the actual method. This blocked inlining and the JIT currently will not GDV unless it can also inline.

So for instance ICollection<T>.Count would not devirtualize.

We think we know enough to pass the right inst parameter (the exact method desc) so enable this for the array case, at least for normal jitting.

For NAOT array devirtualization happens via normal paths thanks to Array<T> so should already fpr these cases. For R2R we don't do array interface devirt (yet).

There was an existing field on CORINFO_DEVIRTUALIZATION_INFO to record the need for an inst parameter, but it was unused and so I renamed it and use it for this case.

Contributes to #108913.

The JIT recently enabled devirtualization of `GetEnumerator`, but other methods were inhibited from devirtualization because the runtime was returning an instantiating stub instead of the actual method. This blocked inlining and the JIT currently will not GDV unless it can also inline. So for instance `ICollection<T>.Count` would not devirtualize. We think we know enough to pass the right inst parameter (the exact method desc) so enable this for the array case, at least for normal jitting. For NAOT array devirtualization happens via normal paths thanks to `Array<T>` so should already fpr these cases. For R2R we don't do array interface devirt (yet). There was an existing field on `CORINFO_DEVIRTUALIZATION_INFO` to record the need for an inst parameter, but it was unused and so I renamed it and use it for this case. Contributes to dotnet#108913.

AndyAyersMS · 2024-10-25T00:31:20Z

FYI @EgorBo

@jkotas is this hopelessly naive? It seems to work but I suspect most of the SZArrayHelper methods ignore the inst param.

SPMI will not see diffs from this; it needs to see new behavior from the VM side.

EgorBo · 2024-10-25T00:47:15Z

@MihuBot

AndyAyersMS · 2024-10-25T01:25:01Z

@EgorBot

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Running;
using System.Runtime.CompilerServices;

BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args);

public class Bench
{
    static byte[] Data = new byte[512];
    [Benchmark]
    public int Test() => TestInner(Data);
    [MethodImpl(MethodImplOptions.NoInlining)]
    int TestInner(ICollection<byte> c) => c.Count;
}

AndyAyersMS · 2024-10-25T16:12:22Z

EgorBot results:

BenchmarkDotNet v0.14.0, Ubuntu 24.04 LTS (Noble Numbat)
AMD EPYC 9R14, 1 CPU, 8 logical and 8 physical cores
Job-RSCWGO : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
Job-MXOKVG : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

Method	Toolchain	Mean	Error	Ratio
Test	Main	0.5454 ns	0.0002 ns	1.00
Test	PR	0.5456 ns	0.0001 ns	1.00

When I run locally though:

BenchmarkDotNet v0.14.0, Windows 11 (10.0.26100.2033)
Intel Core i7-8700 CPU 3.20GHz (Coffee Lake), 1 CPU, 12 logical and 6 physical cores
.NET SDK 9.0.100-rc.2.24474.11
[Host] : .NET 9.0.0 (9.0.24.47305), X64 RyuJIT AVX2
Job-BZVIXB : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX2
Job-TXYRTJ : .NET 9.0.0 (9.0.24.47305), X64 RyuJIT AVX2

Runtime=.NET 9.0

Method	Job	Toolchain	Mean	Error	StdDev	Ratio	RatioSD
Test	Job-BZVIXB	CoreRun	0.5370 ns	0.0405 ns	0.0666 ns	0.21	0.03
Test	Job-TXYRTJ	net9.0	2.5593 ns	0.0303 ns	0.0269 ns	1.00	0.01

Maybe my baselines are out of date...

EgorBo · 2024-10-25T16:28:03Z

@EgorBot -intel -arm64 -profiler --envvars DOTNET_JitDisasm:TestInner

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Running;
using System.Runtime.CompilerServices;

BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args);

public class Bench
{
    static byte[] Data = new byte[512];

    [Benchmark]
    public int Test() => TestInner(Data);

    [MethodImpl(MethodImplOptions.NoInlining)]
    int TestInner(ICollection<byte> c) => c.Count;
}

jkotas · 2024-10-25T16:48:51Z

It seems to work but I suspect most of the SZArrayHelper methods ignore the inst param.

Most of them ignore inst param, except GetEnumerator. We take advantage of it to reduce number of methods that need to be instantiated:

runtime/src/coreclr/vm/array.cpp

Lines 984 to 990 in 9228ccf

    
           // OPTIMIZATION: For any method other than GetEnumerator(), we can safely substitute 
        
           // "Object" for reference-type theT's. This causes fewer methods to be instantiated. 
        
           if (startingMethod[inheritanceDepth] != METHOD__SZARRAYHELPER__GETENUMERATOR && 
        
               !theT.IsValueType()) 
        
           { 
        
               theT = TypeHandle(g_pObjectClass); 
        
           }

I think this approach may work for non-shared code, but it looks too simple to handle shared generic code properly.

EgorBo · 2024-10-25T16:59:51Z

When I run locally though:

@AndyAyersMS hm.. weird, it feels like it's already expanded in main according to bot...

… method desc

AndyAyersMS · 2024-10-26T15:36:22Z

@EgorBot -intel -arm64 -profiler --envvars DOTNET_JitDisasm:TestInner

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Running;
using System.Runtime.CompilerServices;

BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args);

public class Bench
{
    static string[] Data = new string[512];

    [Benchmark]
    public int Test() => TestInner(Data);

    [MethodImpl(MethodImplOptions.NoInlining)]
    int TestInner(ICollection<string> c) => c.Count;
}

AndyAyersMS · 2024-10-26T15:39:13Z

When I run locally though:

@AndyAyersMS hm.. weird, it feels like it's already expanded in main according to bot...

The issue was with array devirt for ref-type (or shared) array elements... updated the benchmark to use string[], and hackily fixed things to pass the right method desc as dictionary source. If this works out, I'll need to update the jit interface to do it properly.

AndyAyersMS · 2024-10-26T17:54:48Z

still need to update JIT to expect a method handle instead of class handle in a few places.

AndyAyersMS · 2024-10-27T05:23:53Z

@EgorBot -intel -arm64 -profiler --envvars DOTNET_JitDisasm:TestInner

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Running;
using System.Runtime.CompilerServices;

BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args);

public class Bench
{
    static string[] Data = new string[512];

    [Benchmark]
    public int Test() => TestInner(Data);

    [MethodImpl(MethodImplOptions.NoInlining)]
    int TestInner(ICollection<string> c) => c.Count;
}

AndyAyersMS · 2024-10-27T05:24:09Z

@MihuBot

jkotas · 2024-10-27T15:50:13Z

src/coreclr/vm/jitinterface.cpp

+        THROWS;
+        GC_TRIGGERS;


Suggested change

THROWS;

GC_TRIGGERS;

NOTHROW;

GC_NOTRIGGER;

LEAF JIT/EE interface methods should be nothrow/notrigger.

AndyAyersMS · 2024-10-27T18:21:57Z

@EgorBot -intel -arm64 -profiler --envvars DOTNET_JitDisasm:Foreach Count

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Running;
using System.Runtime.CompilerServices;

BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args);

public class Bench
{
    string[] s_ro_str_array = new string[512];

    [Benchmark]
    public int Foreach()
    {
        IEnumerable<string> e = s_ro_str_array;
        int sum = 0;
        foreach (string s in e) sum += s == null ? 0 : s.Length;
        return sum;
    }

    [Benchmark]
    public int Count() => CountInner(s_ro_str_array);

    [MethodImpl(MethodImplOptions.NoInlining)]
    int CountInner(ICollection<string> c) => c.Count;
}

The Foreach test should now devirtualize and stack allocate the enumerator (but not yet promote, needs #109237).

AndyAyersMS · 2024-10-27T18:49:37Z

@EgorBot -intel -arm64

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Running;
using System.Runtime.CompilerServices;

BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args);

public class Bench
{
    string[] s_ro_str_array = new string[512];

    [Benchmark]
    public int Foreach()
    {
        IEnumerable<string> e = s_ro_str_array;
        int sum = 0;
        foreach (string s in e) sum += s == null ? 0 : s.Length;
        return sum;
    }

    [Benchmark]
    public int Count() => CountInner(s_ro_str_array);

    [MethodImpl(MethodImplOptions.NoInlining)]
    int CountInner(ICollection<string> c) => c.Count;
}

EgorBo · 2024-10-27T18:55:02Z

@AndyAyersMS the bot didn't report anything because of --envvars DOTNET_JitDisasm:Foreach Count (namely, because of whitespace)

I guess it has to be wrapped with quotes, but not sure it will work either

AndyAyersMS · 2024-10-28T00:13:21Z

BenchmarkDotNet v0.14.0, Ubuntu 24.04 LTS (Noble Numbat)
Intel Xeon Platinum 8488C, 1 CPU, 8 logical and 4 physical cores
.NET SDK 9.0.100-rc.2.24474.11
  [Host]     : .NET 9.0.0 (9.0.24.47305), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-OEFINY : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-VDHZTA : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

Method	Toolchain	Mean	Error	Ratio
Foreach	/core_root_base/corerun	397.2479 ns	2.3665 ns	1.00
Foreach	/core_root_diff/corerun	442.7244 ns	4.5621 ns	1.11

Count	/core_root_base/corerun	3.8438 ns	0.0332 ns	1.00
Count	/core_root_diff/corerun	0.5762 ns	0.0139 ns	0.15

Locally I have intel as faster (at least on windows):

BenchmarkDotNet v0.14.0, Windows 11 (10.0.26100.2033)
Intel Core i7-8700 CPU 3.20GHz (Coffee Lake), 1 CPU, 12 logical and 6 physical cores
.NET SDK 9.0.100-rc.2.24474.11
  [Host]     : .NET 9.0.0 (9.0.24.47305), X64 RyuJIT AVX2
  Job-GVMQPB : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX2
  Job-DZYGUG : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX2
  Job-LLHZLG : .NET 9.0.0 (9.0.24.47305), X64 RyuJIT AVX2

Runtime=.NET 9.0

Method	Job	Toolchain	Mean	Error	StdDev	Ratio	Gen0	Allocated	Alloc Ratio
foreach_static_readonly_ref_array_via_interface	Job-GVMQPB	base	738.4 ns	4.63 ns	4.34 ns	0.77	0.0048	32 B	1.00
foreach_static_readonly_ref_array_via_interface	Job-DZYGUG	diff	622.7 ns	3.43 ns	3.04 ns	0.65	-	-	0.00
foreach_static_readonly_ref_array_via_interface	Job-LLHZLG	net9.0	959.1 ns	11.22 ns	10.50 ns	1.00	0.0038	32 B	1.00

Loop code is similar too:

;;; base inner loop (heap allocated enumerator)

G_M47389_IG03:  ;; offset=0x0026
       mov      edx, dword ptr [rcx+0x08]
						;; size=3 bbWeight=0.91 PerfScore 1.82
G_M47389_IG04:  ;; offset=0x0029
       add      ebx, edx
						;; size=2 bbWeight=516.86 PerfScore 129.22
G_M47389_IG05:  ;; offset=0x002B
       mov      ecx, dword ptr [rax+0x08]
       inc      ecx
       mov      edx, ecx
       mov      esi, dword ptr [rax+0x0C]
       cmp      edx, esi
       jae      SHORT G_M47389_IG08
						;; size=14 bbWeight=517.86 PerfScore 2977.70
G_M47389_IG06:  ;; offset=0x0039
       mov      dword ptr [rax+0x08], edx
       cmp      ecx, esi
       jae      SHORT G_M47389_IG10
       mov      rdx, gword ptr [rax+0x10]
       cmp      ecx, dword ptr [rdx+0x08]
       jae      SHORT G_M47389_IG11
       mov      ecx, ecx
       mov      rcx, gword ptr [rdx+8*rcx+0x10]
       test     rcx, rcx
       jne      SHORT G_M47389_IG03

;;; diff inner loop (stack allocated enumerator / not promoted)

G_M47389_IG03:  ;; offset=0x0053
       mov      r8d, dword ptr [rcx+0x08]
						;; size=4 bbWeight=0.76 PerfScore 1.52
G_M47389_IG04:  ;; offset=0x0057
       add      eax, r8d
						;; size=3 bbWeight=511.71 PerfScore 127.93
G_M47389_IG05:  ;; offset=0x005A
       mov      ecx, dword ptr [rdx+0x08]
       inc      ecx
       mov      ebx, dword ptr [rdx+0x0C]
       cmp      ecx, ebx
       jae      SHORT G_M47389_IG08
						;; size=12 bbWeight=512.71 PerfScore 2819.93
G_M47389_IG06:  ;; offset=0x0066
       mov      dword ptr [rdx+0x08], ecx
       mov      ecx, dword ptr [rdx+0x08]
       cmp      ecx, dword ptr [rdx+0x0C]
       jae      SHORT G_M47389_IG10
       mov      r8, gword ptr [rdx+0x10]
       cmp      ecx, dword ptr [r8+0x08]
       jae      SHORT G_M47389_IG11
       mov      rcx, gword ptr [r8+8*rcx+0x10]
       test     rcx, rcx
       jne      SHORT G_M47389_IG03

AndyAyersMS · 2024-10-28T00:16:58Z

Now that #109237 is merged, the perf of Foreach with this PR should be significantly better.

AndyAyersMS · 2024-10-28T00:17:33Z

@EgorBot -intel -arm64

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Running;
using System.Runtime.CompilerServices;

BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args);

public class Bench
{
    string[] s_ro_str_array = new string[512];

    [Benchmark]
    public int Foreach()
    {
        IEnumerable<string> e = s_ro_str_array;
        int sum = 0;
        foreach (string s in e) sum += s == null ? 0 : s.Length;
        return sum;
    }

    [Benchmark]
    public int Count() => CountInner(s_ro_str_array);

    [MethodImpl(MethodImplOptions.NoInlining)]
    int CountInner(ICollection<string> c) => c.Count;
}

AndyAyersMS · 2024-10-28T01:07:29Z

BenchmarkDotNet v0.14.0, Ubuntu 24.04 LTS (Noble Numbat)
Unknown processor
.NET SDK 9.0.100-rc.2.24474.11
  [Host]     : .NET 9.0.0 (9.0.24.47305), Arm64 RyuJIT AdvSIMD
  Job-UXBTJA : .NET 10.0.0 (42.42.42.42424), Arm64 RyuJIT AdvSIMD
  Job-VJTWZI : .NET 10.0.0 (42.42.42.42424), Arm64 RyuJIT AdvSIMD

Method	Toolchain	Mean	Error	Ratio
Foreach	/core_root_base/corerun	1,297.675 ns	0.0915 ns	1.00
Foreach	/core_root_diff/corerun	282.593 ns	0.8442 ns	0.22

Count	/core_root_base/corerun	1.111 ns	0.0003 ns	1.00
Count	/core_root_diff/corerun	1.071 ns	0.0003 ns	0.96

BenchmarkDotNet v0.14.0, Ubuntu 24.04 LTS (Noble Numbat)
Intel Xeon Platinum 8488C, 1 CPU, 8 logical and 4 physical cores
.NET SDK 9.0.100-rc.2.24474.11
  [Host]     : .NET 9.0.0 (9.0.24.47305), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-GNMMMF : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-ZDWMNG : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

Method	Toolchain	Mean	Error	Ratio
Foreach	/core_root_base/corerun	382.5067 ns	1.2694 ns	1.00
Foreach	/core_root_diff/corerun	298.1094 ns	1.5452 ns	0.78

Count	/core_root_base/corerun	3.7048 ns	0.0246 ns	1.00
Count	/core_root_diff/corerun	0.5671 ns	0.0118 ns	0.15

AndyAyersMS · 2024-10-28T14:33:29Z

NAOT failure looks like #104340.

Unhandled exception. System.Threading.SynchronizationLockException: Object synchronization method was called from an unsynchronized block of code.
   at System.Threading.Monitor.Exit(Object) + 0x170
   at System.Runtime.CompilerServices.ConditionalWeakTable`2.Container.Finalize() + 0xdc
   at System.Runtime.__Finalizer.DrainQueue() + 0x4c
   at System.Runtime.__Finalizer.ProcessFinalizers() + 0x8c
./RunTests.sh: line 176:    23 Aborted

AndyAyersMS · 2024-10-28T14:41:07Z

Updated diff inner loop for Foreach above. Enumerator is gone, now we just need to make this loop shape fit in with our normal loop opts.

       jmp      SHORT G_M47389_IG05
       align    [12 bytes for IG03]
						;; size=34 bbWeight=1.84 PerfScore 8.74
G_M47389_IG03:  ;; offset=0x0030
       mov      r10d, dword ptr [r8+0x08]
						;; size=4 bbWeight=4.58 PerfScore 9.16
G_M47389_IG04:  ;; offset=0x0034
       add      eax, r10d
						;; size=3 bbWeight=944.53 PerfScore 236.13
G_M47389_IG05:  ;; offset=0x0037
       inc      edx
       cmp      edx, 512
       jae      SHORT G_M47389_IG08
						;; size=10 bbWeight=946.37 PerfScore 1419.55
G_M47389_IG06:  ;; offset=0x0041
       mov      edx, edx
       mov      r8, gword ptr [rcx+8*rdx+0x10]
       test     r8, r8
       jne      SHORT G_M47389_IG03

AndyAyersMS · 2024-10-28T14:46:36Z

@EgorBo @jkotas PTAL
cc @dotnet/jit-contrib

The JIT now "inlines" the instantiating stub and the instantiated method.

EgorBo · 2024-10-28T15:42:56Z

src/coreclr/jit/gentree.cpp

@@ -14238,6 +14238,52 @@ CORINFO_CLASS_HANDLE Compiler::gtGetHelperArgClassHandle(GenTree* tree)
    return result;
 }

+//------------------------------------------------------------------------
+// gtGetHelperArgMethodHandle: find the compile time method handle from


I presume the impl could be merged with gtGetHelperArgClassHandle to reduce copy-paste, but not sure it's worth it

EgorBo · 2024-10-28T15:44:25Z

src/coreclr/tools/Common/JitInterface/CorInfoImpl.cs

+        {
+            *methodArg = null;
+            *classArg = null;
+            return null;


worth leaving a note why it's no-op for NAOT/R2R?

Yes, good point.

Also this should not be too hard to implement, but I don't know how to do it, and it won't get used. Currently R2R lacks the devirtualization support, and NAOT doesn't handle arrays the same way.

EgorBo

Should we kick off a PGO pipeline?

AndyAyersMS · 2024-10-28T15:55:00Z

/azp run runtime-coreclr libraries-pgo, runtime-coreclr pgo, runtime-coreclr pgostress

azure-pipelines · 2024-10-28T15:55:26Z

Azure Pipelines successfully started running 3 pipeline(s).

dotnet-issue-labeler bot added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Oct 25, 2024

dotnet-policy-service bot assigned AndyAyersMS Oct 25, 2024

MihuBot mentioned this pull request Oct 25, 2024

[JitDiff X64] [AndyAyersMS] JIT: enable devirtualization/inlining of other array interface methods MihuBot/runtime-utils#755

Open

EgorBot mentioned this pull request Oct 25, 2024

EgorBot for AndyAyersMS in #109209 EgorBot/runtime-utils#134

Open

This was referenced Oct 25, 2024

LibraryImportGenerator.Unit.Tests crashing on linux-x64 mono interpreter #100800

Open

System.Security.Cryptography.Csp.Tests failing on macOS x64 Debug #104172

Open

EgorBot mentioned this pull request Oct 25, 2024

EgorBot for EgorBo in #109209 EgorBot/runtime-utils#135

Open

AndyAyersMS mentioned this pull request Oct 25, 2024

JIT: De-abstraction in .NET 10 #108913

Open

need to pass the wrapper method desc as the inst arg, not the wrapped…

c58bb4a

… method desc

EgorBot mentioned this pull request Oct 26, 2024

EgorBot for AndyAyersMS in #109209 EgorBot/runtime-utils#137

Open

extend jit interface

6afc434

EgorBot mentioned this pull request Oct 27, 2024

EgorBot for AndyAyersMS in #109209 EgorBot/runtime-utils#138

Open

MihuBot mentioned this pull request Oct 27, 2024

[JitDiff X64] [AndyAyersMS] JIT: enable devirtualization/inlining of other array interface methods MihuBot/runtime-utils#763

Open

unblock inlining for instantiating stubs + gdvs

aecf6a0

jkotas reviewed Oct 27, 2024

View reviewed changes

AndyAyersMS added 2 commits October 27, 2024 10:48

fix embedding of inst param

5d21416

fix contracts

baec53f

This comment was marked as outdated.

Sign in to view

EgorBot mentioned this pull request Oct 27, 2024

EgorBot for AndyAyersMS in #109209 EgorBot/runtime-utils#139

Open

EgorBot mentioned this pull request Oct 27, 2024

EgorBot for AndyAyersMS in #109209 EgorBot/runtime-utils#140

Open

EgorBot mentioned this pull request Oct 27, 2024

EgorBot for AndyAyersMS in #109209 EgorBot/runtime-utils#141

Open

Merge branch 'main' into SZArrayOtherMethodsDevirt

ae86aeb

Merge branch 'main' into SZArrayOtherMethodsDevirt

749a26a

EgorBot mentioned this pull request Oct 28, 2024

EgorBot for AndyAyersMS in #109209 EgorBot/runtime-utils#142

Open

AndyAyersMS marked this pull request as ready for review October 28, 2024 14:44

AndyAyersMS requested a review from MichalStrehovsky as a code owner October 28, 2024 14:44

Merge branch 'main' into SZArrayOtherMethodsDevirt

bec5f62

AndyAyersMS requested a review from EgorBo October 28, 2024 14:46

EgorBo reviewed Oct 28, 2024

View reviewed changes

EgorBo approved these changes Oct 28, 2024

View reviewed changes

jkotas approved these changes Oct 28, 2024

View reviewed changes

AndyAyersMS merged commit 87fea60 into dotnet:main Oct 28, 2024
133 checks passed

AndyAyersMS mentioned this pull request Oct 29, 2024

Array interface method devirtualization #62457

Closed

github-actions bot locked and limited conversation to collaborators Nov 28, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

JIT: enable devirtualization/inlining of other array interface methods #109209

JIT: enable devirtualization/inlining of other array interface methods #109209

AndyAyersMS commented Oct 25, 2024

AndyAyersMS commented Oct 25, 2024

EgorBo commented Oct 25, 2024

AndyAyersMS commented Oct 25, 2024

AndyAyersMS commented Oct 25, 2024

EgorBo commented Oct 25, 2024

jkotas commented Oct 25, 2024

EgorBo commented Oct 25, 2024 •

edited

Loading

AndyAyersMS commented Oct 26, 2024

AndyAyersMS commented Oct 26, 2024

AndyAyersMS commented Oct 26, 2024

AndyAyersMS commented Oct 27, 2024

AndyAyersMS commented Oct 27, 2024

jkotas Oct 27, 2024 •

edited

Loading

This comment was marked as outdated.

AndyAyersMS commented Oct 27, 2024 •

edited

Loading

AndyAyersMS commented Oct 27, 2024

EgorBo commented Oct 27, 2024 •

edited

Loading

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024 •

edited

Loading

AndyAyersMS commented Oct 28, 2024

EgorBo Oct 28, 2024

EgorBo Oct 28, 2024

AndyAyersMS Oct 28, 2024

EgorBo left a comment

AndyAyersMS commented Oct 28, 2024

azure-pipelines bot commented Oct 28, 2024

-        THROWS;
-        GC_TRIGGERS;
+        NOTHROW;
+        GC_NOTRIGGER;

JIT: enable devirtualization/inlining of other array interface methods #109209

JIT: enable devirtualization/inlining of other array interface methods #109209

Conversation

AndyAyersMS commented Oct 25, 2024

AndyAyersMS commented Oct 25, 2024

EgorBo commented Oct 25, 2024

AndyAyersMS commented Oct 25, 2024

AndyAyersMS commented Oct 25, 2024

EgorBo commented Oct 25, 2024

jkotas commented Oct 25, 2024

EgorBo commented Oct 25, 2024 • edited Loading

AndyAyersMS commented Oct 26, 2024

AndyAyersMS commented Oct 26, 2024

AndyAyersMS commented Oct 26, 2024

AndyAyersMS commented Oct 27, 2024

AndyAyersMS commented Oct 27, 2024

jkotas Oct 27, 2024 • edited Loading

Choose a reason for hiding this comment

This comment was marked as outdated.

AndyAyersMS commented Oct 27, 2024 • edited Loading

AndyAyersMS commented Oct 27, 2024

EgorBo commented Oct 27, 2024 • edited Loading

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024

AndyAyersMS commented Oct 28, 2024 • edited Loading

AndyAyersMS commented Oct 28, 2024

EgorBo Oct 28, 2024

Choose a reason for hiding this comment

EgorBo Oct 28, 2024

Choose a reason for hiding this comment

AndyAyersMS Oct 28, 2024

Choose a reason for hiding this comment

EgorBo left a comment

Choose a reason for hiding this comment

AndyAyersMS commented Oct 28, 2024

azure-pipelines bot commented Oct 28, 2024

EgorBo commented Oct 25, 2024 •

edited

Loading

jkotas Oct 27, 2024 •

edited

Loading

AndyAyersMS commented Oct 27, 2024 •

edited

Loading

EgorBo commented Oct 27, 2024 •

edited

Loading

AndyAyersMS commented Oct 28, 2024 •

edited

Loading