@EgorBot -mono --filter System.Memory.ReadOnlyMemory.ToArray

@matouskozak PTAL, it fixes the issue according to the benchmark in#106764 (comment)

/azp run runtime-extra-platforms

We do unroll copies, inmini_emit_memcpy, but we might not be hitting that path for whatever reason, or we might emit some redundant instructions. I think this fix is ok, but it is a sign that our copy function is not optimal.

mini_emit_memcpy

We do unroll copies, inmini_emit_memcpy, but we might not be hitting that path for whatever reason, or we might emit some redundant instructions. I think this fix is ok, but it is a sign that our copy function is not optimal.

Looking at the codepaths for[Write/Read]Unaligned we are going thrumini_emit_memory_[store/load] that doesn't pass thrumini_emit_memcpy but goes straight toEMIT_NEW_[STORE/LOAD]_MEMBASE_TYPE. So we are not even trying to take advantage of themini_emit_memcpy copy unroll.

That should still emit aOP_STOREV_MEMBASE that would later be decomposed to some optimal copy sequence inmono_decompose_vtype_opts.

We might want to backport this fix to .NET 9 due to the severity of the perf regressionsdotnet/perf-autofiling-issues#33182@jkurdek@vitek-karas.

That should still emit aOP_STOREV_MEMBASE that would later be decomposed to some optimal copy sequence inmono_decompose_vtype_opts.

Thank you, you're right. However, we are taking this path

/backport to release/9.0

Started backporting to release/9.0:https://github.com/dotnet/runtime/actions/runs/10505141188

Looking at the codepaths for[Write/Read]Unaligned we are going thrumini_emit_memory_[store/load] that doesn't pass thrumini_emit_memcpy but goes straight toEMIT_NEW_[STORE/LOAD]_MEMBASE_TYPE. So we are not even trying to take advantage of themini_emit_memcpy copy unroll.

Is there an issue tracking this missed optimization?

Looking at the codepaths for[Write/Read]Unaligned we are going thrumini_emit_memory_[store/load] that doesn't pass thrumini_emit_memcpy but goes straight toEMIT_NEW_[STORE/LOAD]_MEMBASE_TYPE. So we are not even trying to take advantage of themini_emit_memcpy copy unroll.

Is there an issue tracking this missed optimization?

I'm not aware of it so I created it#106822.