On 08/25/09 05:05, Vladimir Kotal wrote: > ... > The results of the AES microbenchmark runs (5 repetitions of each > test) on T5220 machine (64 HW threads, each 1167MHz) are rather > interesting:
The single thread performance number is the important one because we are worried about the latency here. The AES 64 test uses AES software provider because it falls under the 512-byte threshold. So, we don't see any regression as expected. I think the slight improvement could be from other factors like the cache. The AES 512 test uses n2cp. The -0.26% regression for the single thread case is fine since the number is not significant. > However, there are 2 interesting anomalies: > - the AES encrypting 64 bytes of data with CKM_AES_CBC case actually > reports small performance improvement in average (mainly because of > improvement with 16 threads) > - I don't know how to interpret this > - the AES encrypting 512 bytes of data with CKM_AES_CBC case has an > interesting quirk with 16 threads > - possibly related to the 8 MAU processing units in UltraSPARC-T2 This could be because of the cache or the variation in thread scheduling. In any case, this fix does not add any locking. So, it is OK to let this pass. -Krishna
