Automatic estimation of overhead in memory/compute architecture#2760
Merged
Automatic estimation of overhead in memory/compute architecture#2760
Conversation
…ce estimation that qubits are partitioned into memory and compute qubits. Based on the capacity of the compute qubits, resource estimation will track the number of READ and WRITE operations into the memory qubits based on eviction strategies for the compute qubits (LRU and LFU).
msoeken
requested review from
DmitryVasilevsky,
billti,
ivanbasov and
swernli
as code owners
|
Change in memory usage detected by benchmark. Memory Report for 64d99c5
|
ivanbasov
reviewed
Oct 29, 2025
ivanbasov
approved these changes
Oct 29, 2025
swernli
reviewed
Oct 29, 2025
|
Change in memory usage detected by benchmark. Memory Report for 24ffa8e
|
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
billti
pushed a commit
that referenced
this pull request
Nov 14, 2025
This adds a RE intrinsic called `EnableMemoryComputeArchitecture` that emulates a compute/memory architecture with _compute qubits_ that can be operated on with logical operations and _memory qubits_ that can be turned into compute qubits in order to compute on them. The intrinsic takes as parameter an upper bound `k` on the number of compute qubits. The first `k` qubits are assumed to be compute qubits from the start. However, when an operation is applied to a qubit that is not a compute qubit and there is no more space for compute qubits, one needs to evict a compute qubit and move it into memory. Similarly, if the qubit to be acted on is in memory, it has to be moved into the compute qubit space. The logical resource counter counts the number of times a qubit is read from memory and written to memory. Further, one can specify an eviction strategy (either `LeastRecentlyUsed` that evicts a compute qubit that has been used least recently or `LeastFrequentlyUsed` that evicts a compute qubit that has been used least frequently).
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
This adds a RE intrinsic called
EnableMemoryComputeArchitecturethat emulates a compute/memory architecture with compute qubits that can be operated on with logical operations and memory qubits that can be turned into compute qubits in order to compute on them. The intrinsic takes as parameter an upper boundkon the number of compute qubits. The firstkqubits are assumed to be compute qubits from the start. However, when an operation is applied to a qubit that is not a compute qubit and there is no more space for compute qubits, one needs to evict a compute qubit and move it into memory. Similarly, if the qubit to be acted on is in memory, it has to be moved into the compute qubit space. The logical resource counter counts the number of times a qubit is read from memory and written to memory. Further, one can specify an eviction strategy (eitherLeastRecentlyUsedthat evicts a compute qubit that has been used least recently orLeastFrequentlyUsedthat evicts a compute qubit that has been used least frequently).