Noticed something after trying out the current beta of Primocache, which uses the 3.0.0.1 kernel. Overall the resulting performance index using Anvil's Storage Utilities shows a decrease from results in 2.7.3 just two weeks ago on the same system. Some tests are faster, overall is slower.
First, the Primocache configuration used in both 2.7.3 and 3.0.0 (.1):
And the performance results from testing both versions:
AUG 28 2017:
SEP 12 2017:
If anyone else has performance comparisons to post, it might be helpful in either validating what I'm seeing, or discrediting it as anomalous.
You were testing the drive with Windows on it? That can be problematic given how Windows monitors and updates itself. You'd have to switch back and forth from a drive image of windows for a real apples to apples comparison. Also, how long was the new version of PrimoCache installed, the whole two weeks between testing, or just a few days? It uses intelligent caching so your old version might have had a near perfect hit rate while your new one was still learning. Given the huge amount of memory you're using, that's very nice btw, that could mean PrimoCache is still working in the background. The difference we're seeing in numbers is low enough to think it's in the margin of error for a SSD.
Which reminds me, a SSD does its own housekeeping all the time. That could be a factor in the benchmarking.
After having bluescreen issues with 3.0.0 beta, I repaired the system and reverted to Primocache 2.7.3. After doing so and allowing the system to reboot and the cache to settle, I re-tested with identical parameters.
The benchmark is back up to 80,545.26. That indicates that (at least for this benchmark), 3.0.0 is inferior from a performance standpoint. Still open to any feedback from others with test scores to discuss.
Last edited by Jaga on Fri Sep 15, 2017 7:14 pm, edited 1 time in total.
Babel17 wrote:You were testing the drive with Windows on it? That can be problematic given how Windows monitors and updates itself. You'd have to switch back and forth from a drive image of windows for a real apples to apples comparison. Also, how long was the new version of PrimoCache installed, the whole two weeks between testing, or just a few days? It uses intelligent caching so your old version might have had a near perfect hit rate while your new one was still learning. Given the huge amount of memory you're using, that's very nice btw, that could mean PrimoCache is still working in the background. The difference we're seeing in numbers is low enough to think it's in the margin of error for a SSD.
Which reminds me, a SSD does its own housekeeping all the time. That could be a factor in the benchmarking.
Either way, nice scores!
No, TRIM isn't affecting the scores at all. Length of time Primo is installed doesn't affect benchmarks at all either. Neither would a drive image matter - the benchmark is using it's own synthetic files to test with. My scores are accurate and valid.
