Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Split CACHE.SNOOP.REMOTE_REQ_LOCAL_HITM event per cache level #12

Merged
merged 1 commit into from
Oct 25, 2024

Conversation

dr-sc
Copy link
Contributor

@dr-sc dr-sc commented Oct 16, 2024

No description provided.

Signed-off-by: Dmitriy Ryabtsev <d.ryabtsev@yadro.com>
@dr-sc dr-sc mentioned this pull request Oct 16, 2024
@@ -36,8 +36,10 @@
|CACHE.L2.PREF.MERGE |L2 cache hits caused by prefetcher where data is not yet in cache but was already requested by preceding miss.
|CACHE.L2.PREF.UNUSED |Number of cachelines brought into L2 by prefetcher and evicted without being accessed even once.
|CACHE.L2.WB |Writebacks to next level cache or memory.
|CACHE.SNOOP.LOCAL_REQ_REMOTE_HITM.SPEC |Private cache misses where data was found in another core cache in modified state. This event can be used to accout for contested accesses cases where several cores read/write the same cachelines.
|CACHE.SNOOP.REMOTE_REQ_LOCAL_HITM |Snoop requests which found cacheline in the core cache in modified state. This event can be used to accout for contested accesses cases where several cores read/write the same cachelines.
|CACHE.SNOOP.LOCAL_REQ_REMOTE_HITM.SPEC |Private cache misses where data was found in another core cache in modified state. This event can be used to account for contested accesses cases where several cores read/write the same cachelines.
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

One thing to keep in mind with this event is that it may only reflect HITMs in a subset of caches. E.g., if the L3/LLC is external, this may only count HITMs in the L1D and L2. I guess the implication is that LOCAL is only guaranteed to cover caches local to (within) the hart. I think that's okay, the kind of sharing cases we're looking for are likely to be getting HITMs in the L1D's of the conflicting harts.

@bcstrongx bcstrongx merged commit ab34ec3 into riscv:main Oct 25, 2024
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants