[ 
https://issues.apache.org/jira/browse/IGNITE-29099?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18123511#comment-18123511
 ] 

Nikita Amelchev commented on IGNITE-29099:
------------------------------------------

h3. Benchmark results

{{JmhCacheFullSyncPutBenchmark}}: 3 server nodes in one JVM, FULL_SYNC, 2 
backups. Allocations per operation ({{gc.alloc.rate.norm}}), mean of 8 JVM 
launches.

||Machine||Operation||master, B/op||patch, B/op||Change||
|Laptop, 10 cores|ATOMIC put|6 888|6 593|−4.3%|
|Laptop, 10 cores|TRANSACTIONAL put|28 576|28 001|−2.0%|
|Server, 16 vCPU|ATOMIC put|6 818|6 599|−3.2%|
|Server, 16 vCPU|TRANSACTIONAL put|29 188|28 646|−1.9%|

> Avoid boxing of cache ID and message direct type on the cache operation hot 
> path
> --------------------------------------------------------------------------------
>
>                 Key: IGNITE-29099
>                 URL: https://issues.apache.org/jira/browse/IGNITE-29099
>             Project: Ignite
>          Issue Type: Improvement
>            Reporter: Nikita Amelchev
>            Assignee: Nikita Amelchev
>            Priority: Minor
>         Attachments: JmhCacheFullSyncPutBenchmark.java, 
> JmhCacheTxFullSyncBenchmark.java
>
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> Every cache update boxes primitive map keys on its hot path:
>  * {{GridCacheSharedContext#cacheContext(int)}} and 
> {{#cacheObjectContext(int)}} look up a {{{}ConcurrentHashMap<Integer, 
> GridCacheContext>{}}}. A cache ID is a hash of the cache name, so it misses 
> the {{Integer}} cache and a new object is allocated on each call. The methods 
> are called several times per operation on every node involved.
>  * {{TcpCommunicationMetricsListener#updateMessageTypeMap}} checks a 
> {{Map<Short, String>}} for every sent and received message.
> These allocations are pure garbage produced on NIO and striped pool threads 
> of all nodes. Both lookups should use primitive keys.
> {\{JmhCacheFullSyncPutBenchmark}} (3 server nodes, FULL_SYNC, 2 backups), 
> \{{gc.alloc.rate.norm}}: allocations per operation drop by 3-4% for an ATOMIC 
> put and by 2% for a transactional put. Throughput is unchanged.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to