red930 Posted November 13, 2005 Posted November 13, 2005 NO! You need to read the thread to learn how to set the benchmarks so you can compare to the others. is my score any good? i dont wanna read through 45 pages http://i31.photobucket.com/albums/c360/ooeric/hd.jpg Share this post Link to post Share on other sites More sharing options...
red930 Posted November 13, 2005 Posted November 13, 2005 The first post will show you how to configure both programs. is my score any good? i dont wanna read through 45 pages http://i31.photobucket.com/albums/c360/ooeric/hd.jpg Share this post Link to post Share on other sites More sharing options...
ooeric Posted November 14, 2005 Posted November 14, 2005 ok ok is this better now? http://i31.photobucket.com/albums/c360/ooeric/newhdscore.jpg Share this post Link to post Share on other sites More sharing options...
red930 Posted November 14, 2005 Posted November 14, 2005 Yessiree ! Share this post Link to post Share on other sites More sharing options...
ooeric Posted November 14, 2005 Posted November 14, 2005 but i dont know how to read those chart numbers. especially comparing them is that on par with what regular satas do? nothing extradinary right? Share this post Link to post Share on other sites More sharing options...
soundx98 Posted November 14, 2005 Posted November 14, 2005 actually pretty low. Stripe Cluster Ratio of 4:1 is usually best. 64/4 will cost you performance. Share this post Link to post Share on other sites More sharing options...
ooeric Posted November 14, 2005 Posted November 14, 2005 what does that mean? lol im thinking about copying off all my data and formatting the drives. what settings should i use?? and from what the pic looks like 128 has the best score? Share this post Link to post Share on other sites More sharing options...
soundx98 Posted November 14, 2005 Posted November 14, 2005 A lot depends on the what the file sizes are on the drives. Windoz is usually smaller files so the default cluster size of 4 works well. If a 4 to 1 Stripe to cluster size generally provides best performance you would set the Stripe size to 16. 16 stripe and 4 cluster is usually very good on those drives. Now if your drive concern was mostly larger files like DVDs and MP3s a larger stripe of maybe 64 or 32 and a cluster size of 16 or 8 may be optimum for you. Many of the geekier among us 1st set up windows on an IDE drive and then set up RAID 0 and test the different strip/cluster combinations for maximum speed. ie, random access time and average speed in HD Tach. The burst test in HD Tach has problems reading RAID 0 configs so don't make yourself crazy considering it. Always run the tests 3-4 times so that you get an average. Hope that helps. (It was covered 4-5 times in the thread but yes, it is a long one to read. Although it probably would have saved you some time on best setup) Share this post Link to post Share on other sites More sharing options...
Fight Game Posted November 15, 2005 Posted November 15, 2005 Many of the geekier among us 1st set up windows on an IDE drive and then set up RAID 0 and test the different strip/cluster combinations for maximum speed. ie, random access time and average speed in HD Tach. But if you test a drive with nothing on it, won't the results change once there is data there, depending on if the files are small or large? I would assume that if all the files are small (or atleast the files in the area that it's running the test) then 16:4 would be best, and if the test area has very large files, then 64:16 would be best? Correct? I'm kinda confused by this, and have asked before w/o response. If this is correct, then it doesn't make much since to run the benchmark's on a blank partition, unless you sure that all the files going there will be very small. I think it would be helpful in the benchmark program, to specify what area of the disk (folders, etc) to test. I mean, if I run the benchmarks on a blank drive, several different times, to find the best combo of stripe/cluster, then it all changes once I get data there, then that's alot of wasted time. You are supposed to use larger stripe/cluster for larger file sizes, but how do you know what files are being tested in the benchmark? I mostly play games, so it would be nice to optimize disk performance based on those file sizes, but since I have no idea what area of the disk is being tested, who knows... Share this post Link to post Share on other sites More sharing options...
soundx98 Posted November 15, 2005 Posted November 15, 2005 You've raised some good thoughts as always FG. yes of course what is eventually placed on the drives will make a difference. But if your primary concern is a "snappy" OS than 16/4 is usually nice and safe. But if you start rippin DVDs like crazy (like me) you may want to increase it to a 32/8. (And you can have different cluster sizes on the array as long as it's partitioned.) Not many arrays that don't find a 32/8 very good and fast for overall usage. I suggest you run multiple times as there is really no way of telling where on the drive/partiton ATTO or HD Tach will write the test files. And just like a Futuremark test it's never the same twice in a row. Guys often test multiple 8 hour prime95 sessions but don't have time to test their RAID Setup. Got a big OS partiton or a just a little one, is the RAID dedicated to movies and MP3s, just some thinks to think about is all. And if nothing else it will invariably show that 64/4 is seldom the fastest/optimal and give the user an idea on what happens to average speed, CPU usage, and access time after doing a little testing. Not hard and fast rules and just a suggestion as most have a spare IDE drive they can load an OS on real quick like (don't need graphics drivers and all the programs and such), and then do some real testing of the stripes and clusters. Some of the geekconians and nerdites may even find it interesting. Oh, always load the OS on the test drive with the RAID drives not connected. Share this post Link to post Share on other sites More sharing options...
red930 Posted November 15, 2005 Posted November 15, 2005 But if you test a drive with nothing on it, won't the results change once there is data there, Both ATTO and HD Tach are OS independent. The benchmark results are the same whether the drive is populated or not according to their documentation and the empirical results I have gotten. Share this post Link to post Share on other sites More sharing options...
ooeric Posted November 15, 2005 Posted November 15, 2005 i see. so even tho i set it to 64/4 right now i guess its ok since my 400gb partition is holding alot of movies animes etc. would i see that big of a gain if i move off all that data, reformat with 16/4 ? or would it only be some minute thing? Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now