The FSX stutter benchmark, storage edition

[cmeeks 8]

Not to beat a dead horse or anything, but I did my own hard drive comparison. I performed the test using the FSXMark11 benchmark and did a fresh install prior to each set of test runs. I compared a Mushkin Callisto Deluxe 120GB SSD and a Samsung Spinpoint F3 1TB. Quite honestly, I saw no difference in min, max, or avg FPS nor did I see any reduction in stutters. The plots below (each taken from the second of three test runs on each hard drive) should convey just about everything you need to know. Once again, an SSD makes no difference any way you look at it.

[dazz 69]

Thanks so much for testing it yourself Corey :smile:What methodology did you use to test smoothness please? observation? fraps also?

[cmeeks 8]

Sorry Dario - I edited that post probably 10 times or more over a period of 20 minutes. Did I end up answering your questions? The data was created using FRAPS frame times just like your memory tests last week or the week before. I had been meaning to test this myself, but you beat me to it! So I figured I would just add one more testament to the fact that an SSD does not help.

[dazz 69]

Sorry Dario - I edited that post probably 10 times or more over a period of 20 minutes. Did I end up answering your questions?

Sure you did. haha, I forgot you keep editing posts like there's no tomorrow Thanks againInterestingly enough, you have the same thing going with the same exact repeatable FPS pattern in both tests. I think it has something to do with some async stuff going on there, cause when I used slew mode instead of pause to start the tests, the stutter pattern would have a variable offset depending on the amount of time I took to turn off slew mode and start the test

[Gary A 8]

Not to beat a dead horse or anything, but I did my own hard drive comparison. I performed the test using the FSXMark11 benchmark and did a fresh install prior to each set of test runs. I compared a Mushkin Callisto Deluxe 120GB SSD and a Samsung Spinpoint F3 1TB. Quite honestly, I saw no difference in min, max, or avg FPS nor did I see any reduction in stutters. The plots below (each taken from the second of three test runs on each hard drive) should convey just about everything you need to know. Once again, an SSD makes no difference any way you look at it.

Corey, any chance you could post access time and CPU utilization? I am just curious.

[cmeeks 8]

Gary,I wasn't sure what you meant by CPU utilization, but I went back to my realtemp log and plotted the CPU usage in the plots I posted above. It's only plotted every 30 seconds, so the first and last data points don't necessarily line up exactly with the start and end of the FSXMark11 benchmark. It also doesn't help that realtemp occasionally skips a log interval, so sometimes the data is only every 60 seconds. Here's an HD Tune benchmark for each of the hard drives to answer your other question.

[Gary A 8]

Gary,I wasn't sure what you meant by CPU utilization, but I went back to my realtemp log and plotted the CPU usage in the plots I posted above. It's only plotted every 30 seconds, so the first and last data points don't necessarily line up exactly with the start and end of the FSXMark11 benchmark. It also doesn't help that realtemp occasionally skips a log interval, so sometimes the data is only every 60 seconds. Here's an HD Tune benchmark for each of the hard drives to answer your other question.

Thank you Corey, that is perfect.The HD Tune results are a little starnge on CPU "usage". I would have expected the Samsung to show some positive use. Also the access time on the Samsung seems a bit long. In comparison my Raid set in which I would expect nearly twice the access time is 12ms.

[cmeeks 8]

It appears right in line with this bit-tech review. I don't fully understand RAID, but from a theoretical standpoint I would expect lower access times since you have two read/write heads doing the work of finding the data.

[ryanbatc 12,859]

I never did think an SSD improved performance, but it does decrease loading times correct?

[cmeeks 8]

It helps with loading times and it helps with noise - two perfectly legitimate reasons to get an SSD in my book if you have the money to blow.

[Guest jahman]

You shouldn't be hearing PC noise (video card, fans, HDDs) if your sound volume is set at a realistic level :-)Cheers,- jahman.

[Gary A 8]

It appears right in line with this bit-tech review. I don't fully understand RAID, but from a theoretical standpoint I would expect lower access times since you have two read/write heads doing the work of finding the data.

Thanks again Corey, I am a dough-head...I am so used to my vrap access time of around 7ms that when I saw 12ms on the Raid I just thought, well I didn't think. You are right, seek time in theory is half (influenced by stripe size and cluster) but access time which is the sum of seek + rotational latency + + + cannot be any faster because those delays are inherent of mechanical drives.

[Guest jahman]

With RAID access times remain exactly the same, but you get an increase in bandwidth (transfer rate) equal to the number of drives you have in a RAID 0 configuration (but if only one drive fails, al data is lost).Other RAID configurations like RAID 5 give you error correction as well, so your RAID array won't fail if a single drive fails, but the speed-up in transfer rate will not be equivalent to the number of drives in the RAID array because of the error-correction data that needs to be read and calculated. Also with RAID 5 and similar configurations the calculation of error correction data can be a significan bottleneck (specifically if done on-board on the MB) unless you use a dedicated (and expensive) RAID PCIe card.Cheers,- jahman.