Repository navigation
PGO: redundant Tier0-instr promotions #84517
Description
static void Main()
{
for (int i = 0; i < 100; i++)
{
Test();
var _ = Test2;
Thread.Sleep(16);
}
}
[MethodImpl(MethodImplOptions.NoInlining)]
static int Test1()
{
int sum = 0;
for (int i = 0; i < 10; i++)
sum += i;
return sum;
}
static int Test2 { get; set; }Actual result:
147: JIT compiled Prog:Test1():int [Instrumented Tier0, IL size=21, code size=109, hash=0x792ae54b]
311: JIT compiled Prog:Test1():int [Instrumented Tier0, IL size=21, code size=109, hash=0x792ae54b]
313: JIT compiled Prog:Test1():int [Tier1 with Dynamic PGO, IL size=21, code size=14, hash=0x792ae54b]
152: JIT compiled Prog:get_Test2():int [Tier0, IL size=6, code size=12, hash=0xf8310770]
316: JIT compiled Prog:get_Test2():int [Instrumented Tier0, IL size=6, code size=12, hash=0xf8310770]
319: JIT compiled Prog:get_Test2():int [Tier1, IL size=6, code size=7, hash=0xf8310770]
Since Test1 has a loop we always instrument it in Tier0, but then it's promoted to Instrumented Tier0 again - it can be omitted.
Also, all Tier0 methods should warn VM in advance that they don't need instrumentation at all (e.g. simple auto properties like get_Test2).
cc @AndyAyersMS
Activity
AndyAyersMS commentedon Apr 12, 2023
Even more interesting when OSR gets involved:
EgorBo commentedon Apr 12, 2023
Heh indeed funny one but presumably the same issue, I'll fix it this week
AndyAyersMS commentedon Apr 20, 2023
We can also see this when we have a nontrivial intrinsic R2R method; we rejit at tier1-instr but it doesn't actually instrument; then we rejit at tier1 and don't have any dynamic PGO data, so we are jitting twice and probably producing the same code each time.
EgorBo commentedon Aug 5, 2023
Moving to 9.0 as the VM change was not trivial. We think that this doesn't really affect perf or startup but still nice to fix
EgorBo commentedon Jul 18, 2024
Moving to post-P7 as it's not a correctness problem