Disk performance hamstrung by gremlins?
RyanMM
Ferndale, MI Icrontian
I'm running into a situation I can't recall ever having seen before, and all my attempts at diagnostics have proven fruitless.
I'm working on an HP laptop with Win7. Dual 500GB hard drives, 6GB RAM, HM55 chipset, 1.6GHz i7.
When this thing arrived, it was dog slow. I ran through the usual culprits - Disk problems (nope, drives test fine, normal benchmark results hooked to testbed, no difference in performance eliminating the second disk from the equation).
No viruses of any kind, boot sector or otherwise. No more detritus in the startup queue than any normal computer; on a normal boot, there's about 27000 handles in 1200 threads; total RAM use of 1.7GB. CPU idles around 4-5% usage.
But if I run HDTune on this thing as is, I get a max speed of 70MB/s, average of 60MB/s, and a burst speed of merely 30MB/s.
The same drive, disconnected and hooked to an older SATA-II testbed system (nForce 6150, if it matters), pulls down 107MB/s, 82MB/s, and 180MB/s respectively.
The graph from the computer as-is actually looks like the hard drive is hitting an upper limit on performance at the 70MB/s mark, because in the last 10% of the benchmark curve, you can see the performance arc kick in and match up with the graph from the testbed system. And there's that mysterious burst speed deficiency.
So it's gotta be bad hardware or something, right? I wish it was that simple. Booting from UBCD4WIN and running the test gives results that match up to the external testbed results, except for the burst performance. That's 80MB/s, less than half the much older testbed's performance. Is there an issue with the performance of the HM55 chipset?
I've been able to get performance back to normal by disabling certain startup items and services, but it's not consistent from one boot to the next. The max and average speeds line up, but there's still a cap of 50MB/s on the boost performance. And it seems like the more I play with it, the more infrequently I can manage to get the better results.
I'm gonna do a backup shortly and restore to factory to see what performance is like out of the gate. I'm genuinely stymied by these drive performance numbers and the lack of consistent reproducibility.
I absolutely hate situations where I can't find a valid solution, and this is shaping up to be one of those cases if I don't have an aha moment.
I'm working on an HP laptop with Win7. Dual 500GB hard drives, 6GB RAM, HM55 chipset, 1.6GHz i7.
When this thing arrived, it was dog slow. I ran through the usual culprits - Disk problems (nope, drives test fine, normal benchmark results hooked to testbed, no difference in performance eliminating the second disk from the equation).
No viruses of any kind, boot sector or otherwise. No more detritus in the startup queue than any normal computer; on a normal boot, there's about 27000 handles in 1200 threads; total RAM use of 1.7GB. CPU idles around 4-5% usage.
But if I run HDTune on this thing as is, I get a max speed of 70MB/s, average of 60MB/s, and a burst speed of merely 30MB/s.
The same drive, disconnected and hooked to an older SATA-II testbed system (nForce 6150, if it matters), pulls down 107MB/s, 82MB/s, and 180MB/s respectively.
The graph from the computer as-is actually looks like the hard drive is hitting an upper limit on performance at the 70MB/s mark, because in the last 10% of the benchmark curve, you can see the performance arc kick in and match up with the graph from the testbed system. And there's that mysterious burst speed deficiency.
So it's gotta be bad hardware or something, right? I wish it was that simple. Booting from UBCD4WIN and running the test gives results that match up to the external testbed results, except for the burst performance. That's 80MB/s, less than half the much older testbed's performance. Is there an issue with the performance of the HM55 chipset?
I've been able to get performance back to normal by disabling certain startup items and services, but it's not consistent from one boot to the next. The max and average speeds line up, but there's still a cap of 50MB/s on the boost performance. And it seems like the more I play with it, the more infrequently I can manage to get the better results.
I'm gonna do a backup shortly and restore to factory to see what performance is like out of the gate. I'm genuinely stymied by these drive performance numbers and the lack of consistent reproducibility.
I absolutely hate situations where I can't find a valid solution, and this is shaping up to be one of those cases if I don't have an aha moment.
0
Comments
From what is happening here, the automatic background defragmenting in Win 7 is not as aggressively done as in Vista, for example. Vista bogged because the defrag was running a lot(among other things). Looks like Microsoft might have erred the other way some (too passive an auto defrag)with Win 7 Pro.
YMMV but it might be worth a try just to see if that helps.
also see if you get the same symptoms at different performance settings.
I also did a factory restore and verified that the factory performance was equally gimped - 107MB/s max, 83MB/s average, and only 53MB/s burst performance, just like the UBCD4WIN results. Which tells me there's something about the way this HM55 chipset is wired that gives it piss-poor SATA thoroughput, OR, there's some problem with the benchmark when run on this particular chipset. I'm gonna check some other benchmarks to see what kind of results those put up, just to kill that particular cat and close this chapter out.
Thanks again everyone!