red930 Posted October 17, 2004 Posted October 17, 2004 When I started planning my current system I did a lot of research. One of the things that surprised me the most was the lack of “realworld” performance increases derived from RAID 0 for gaming. Please see this investigation... http://www.anandtech.com/storage/showdoc.aspx?i=2101 My computers are used for a variety of functions. About 20% of the time is spent web browsing for news and information(like dfi-street). Another 20% is used for content creation like ripping CDs or encoding captured audio and video. Gaming consumes about 50% of my computer time and it looks like the final 10% us just idle time. Based upon this usage I could not justify a RAID 0 array. Doubling my cost and failure rate was not worth the small increase in performance. It’s a single drive for daily use and another for my online backup. I was wondering how other people in the forum use their computers. Quote Share this post Link to post Share on other sites More sharing options...
Angry_Games Posted October 17, 2004 Posted October 17, 2004 about 20% gaming 50% doing 'work' (fixing pics, doing forums, writing emails, checking tech sites, making crappy web pages etc) 20% fooling with audio/video editing 10% hodgepodge RAID-0 doesn't help gaming other than loading levels/textures/caching from the hdd faster RAID-0 is tremendous when doing real-time editing of audio/video etc i have a server with 240GB of backup space...plus i can always hook up an 80GB drive to mirror what my RAID-0 stripes if i want to (i just backup daily to my server, and use it to pull all audio, video, cd images etc from) Quote Share this post Link to post Share on other sites More sharing options...
red930 Posted October 17, 2004 Posted October 17, 2004 Angry, you just made the statement that RAID 0 loads games faster. But at what cost? Here are the results of my real world testing. All tests were performed on a clean install of Windows XP Pro Service Pack 2 with no Hot Fixes, imaged from another drive. Far Cry was installed with the 1.1 patch. I perform five test then throw out the highest and lowest and average the remaining three results. I should have made that clear from the start. The computer was rebooted after every test and allowed to idle for five minutes before starting the test. Lanparty Ultra B specs in my sig. Silicon Image Sil 3114r SATARaid Driver 1.0.0.1 9/22/2003 Silicon Image Sil 3114 STATLink Driver 1.2.0.5 9/3/2003 NVIDIA nForce2 ATA (v2.6) Driver 5.10.2600.446 6/3/2004 Maxtor DiamondMax Plus 9 160 PIDE $91.00 Far Cry load time = 42.9 seconds Maxtor DiamondMax Plus 9 160 SATA $97.00 Far Cry load time = 42.1 seconds Western Digital Raptor 74GB 10,000RPM SATA $180.00 Far Cry load time = 39.1 seconds Two Western Digital Raptors 74GB 10,000RPM SATA 148GB RAID-0 $360.00 Far Cry load time = 40.1 seconds The single Raptor was faster than the RAID 0 array. I’m looking forward to running the test when my new Mobo and CPU arrive(see sig). Am I making a mistake in my testing methods? Quote Share this post Link to post Share on other sites More sharing options...
Angry_Games Posted October 17, 2004 Posted October 17, 2004 use Atto to test hdd speeds/transfer rates as well as HD Tach Quote Share this post Link to post Share on other sites More sharing options...
red930 Posted October 17, 2004 Posted October 17, 2004 I have ATTO ver 2.02. Is that the version you use? If you know of any others I would appreciate it. If you look at my post you will see that I quoted HD Tach but I used several programs like Sandra, ATTO and Performance Test v5.0. All of the results were within 0.5% within the applications. I chose one to represent my results instead of trying to quote all of them. Additionally, I try to use real applications for performance test. Like the game loading I presented in another thread. The synthetic benchmarks are just that, synthetic. I learned long ago not to rely solely on them when tweaking. Quote Share this post Link to post Share on other sites More sharing options...
Poisonsnak Posted October 17, 2004 Posted October 17, 2004 Hehe another RAID thread. I would like to point out the advantage of having all your data on one 'disk' with RAID 0. I wanted 400GB of space and that meant I was buying 2 200GB drives, I used RAID 0 because what I really wanted was 1 400GB drive. Back when I thought partitioning was fun (this lasted like a week) I was always annoyed moving data from one partition to the other. I dunno about that synthetic benchmark part, I mean they are misused but not in that way. They are meant for tweaking, for comparing your own results to your own results after tweak, not comparing to other people's computers. If HD tach says your HD is faster then it is, it just may not have as big an effect as you say. How many times did you run the test with each drive? I would be interested to see the margin of error for each test. Like HD Tach for example is almost useless for calculating CPU usage I get anywhere from 5 - 20% on consecutive runs. Quote Share this post Link to post Share on other sites More sharing options...
red930 Posted October 17, 2004 Posted October 17, 2004 Synthetic versus real world performance is exactly why I perform test with applications. The benchmark programs are great for an apple to apple comparison when you’re tweaking for performance or trying to get to the root of a problem. In this thread I used Far Cry for testing. As I stated in both post, I provided only a representative sample of the testing I actually performed. These variations in performance are accurate as far as my testing goes. BTW I perform five test then throw out the highest and lowest and average the remaining three results. I should have made that clear from the start. Quote Share this post Link to post Share on other sites More sharing options...
Poisonsnak Posted October 17, 2004 Posted October 17, 2004 Hmm while I would like to compare myself sata raid 0 vs. not raided there's no way I am erasing these drives hehe. That's interesting (though disappointing hehe) that your Far Cry tests are accurate. (still trying to poke holes!!) Which driver did you use for the raid 0, 1.0.0.7? Are you using SW-IDE? Also did your synthetic benchmarks, if any, agree with your Far Cry tests? edit: what block size did you use for raid? This is actually very important: http://www.hardocp.com/article.html?art=NTU4LDQ= The 128k RAID 0 is slower read, faster write than the standalone. Quote Share this post Link to post Share on other sites More sharing options...
red930 Posted October 17, 2004 Posted October 17, 2004 I think I found a problem with the RAID driver I'm using. Silicon Image Sil 3114r SATARaid Driver 1.0.0.1 9/22/2003 I downloaded the latest from the DFI site but the system says it can't find a better driver when I do the update. More testing is needed. I may just have to start the RAID 0 testing again if I can get the same driver as you. For these test I used strip sizes from 32K to 128K and found 64K was the best for reads and writes. 64K performance is what I reported. Keep poking dude. There’s nothing worse than thinking you’re right when you don’t have all the angles covered. If I missed something I want to know about it. BTW I’ll edit that post and add the RAID driver that I missed. Thanks Quote Share this post Link to post Share on other sites More sharing options...
Angry_Games Posted October 17, 2004 Posted October 17, 2004 thats unfortunately half your problem...the Si3114 was never a strong controller...in fact the older Si3112 outperforms it, as does the VIA VT8237, and the Intel ICH5/6. Quote Share this post Link to post Share on other sites More sharing options...
red930 Posted October 17, 2004 Posted October 17, 2004 How does the nVidia RAID perform on the LP UT nF3 250Gb? Quote Share this post Link to post Share on other sites More sharing options...
red930 Posted October 19, 2004 Posted October 19, 2004 Now that my new mobo is up and running I can answer the question. A whole lot faster. My early SATA benchmarks are 20% faster with the same drive on the UT nF3 250Gb than they were with the LP Ultra B. PIDE performance has stayed exactly the same. After I get everything setup and tweaked I'll start benchmarking again. Well, if I can get my hands on the two WD Raptors again. Thanks everyone. Quote Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.