Implement Ziccid (I$ coherence) #2117
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
aswaterman
force-pushed
the
ziccid
branch
3 times, most recently
from
c4b14ab to
d96724b
Compare
The intention is to simplify future handling of I$ coherence data structures.
aswaterman
force-pushed
the
ziccid
branch
2 times, most recently
from
eafb7f9 to
1054bf9
Compare
The algorithm is as follows: - An instruction-side Bloom filter knows which PPNs might be in the I$ - A store-side Bloom filter knows which PPNs might get store TLB hits - When refilling the store TLB, search the I-Bloom filter; on a hit, remove the ITLB entry, rebuild the I-Bloom filter, and flush the I$ - When refilling the ITLB, search the store-Bloom filter; on a hit, remove the store TLB entry and rebuild the store-Bloom filter The effect is that any word that can be stored-to cannot be in any I$. The old scheme (periodically flush I$) was hacky and didn't correctly respect the in-order fetch rule (i.e. fetches are in-order wrt. each other, and so they see ordered stores in order).
aswaterman
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The algorithm is as follows:
The effect is that any word that can be stored-to cannot be in any I$.
The old scheme (periodically flush I$) was hacky and didn't correctly respect the in-order fetch rule (i.e. fetches are in-order wrt. each other, and so they see ordered stores in order).
cc @jerryz123 - this is just a proof of concept for the moment. It is possible to be smarter with the data structures, e.g. build two snoop filters (one global set that remembers all PPNs that might be in any I$, and another global set that remembers all PPNs that might be in any store TLB). If you have any cleverer ideas, LMK.