I just built a "workhorse" system that I wanted to be as fast as possible for a reasonable outlay. I chose to use two hard disks, one for applications and one for data. For the application drive, I bought a WD Raptor 150 for $299.
I believed what I'd read about RAID 0 not making a big difference in performance, and I didn't like the increased reliability issues involved with striping across two disks. So I went the single drive route. But when I got my rig built, the HD subsystem didn't feel as fast as it should. Definitely not $299 fast. Then I saw a 70 page sticky thread on DFI-Street.com, in the Software forum, dedicated to posting RAID benchmarks. I decided to benchmark my single-drive Raptor.
Let's compare two solutions:
My setup: Raptor 150GB (single drive).
ATTO avg. read-write: 75mb/s
HD Tach random access: 8.1ms
HD Tach CPU utilization: 1%
HD Tach average read: 76.9mb/s
HD Tach burst speed: 137.3mb/s
Poster from post #1020 on DFI-Street RAID benchmark thread:
2xHitachi 80GB SATA II ($50 each at NewEgg, total $100)
ATTO avg. read-write: 110mb/s
HD Tach random access: 12.7ms
HD Tach CPU utilization: 10%
HD Tach Average read: 97.7mb/s
HD Tach burst speed: 325.6mb/s
He gets between a 27% and 47% performance increase over my Raptor for 1/3 the price (not considering the difference in burst speeds). I could have bought a good video card or put the money towards my dream monitor with that $200! Yes, he has some CPU overhead for the RAID, but I have a dual core and wouldn't notice it. Also I have better random access scores than he does, but the type of performance I want most are application loading times and big file transfers. I would rather have his scores!
But benchmarks are artificial, right? Now look at this page where the user timed various computer operations using a stopwatch, with a single-drive Raptor 74GB installed, and with two Raptor 74GBs installed using RAID 0. Assuming he's not lying and this page is not the work of a mentally-ill person, it's obvious that the RAID 0 solution is much faster in the real world. Link: Who Says 2 Raptors Aren't Better Than One
Some additional details:
- I'm out of my RMA period, otherwise I'd spend the $50 in restocking fee and shipping to return the Raptor.
- I can get an exchange drive from WD. They say my drive should have 84mb/s sustained read, and I'm getting 77. Maybe I'll get a better drive if I exchange it?
- I didn't install the Nvidia IDE SW driver, so I don't have any SATA controls in my Device Manager. I haven't been able to determine whether installing these - which have been problematic for some in the past - could help my performance.
I'm posting this for two reasons. (1) To make sure my Raptor's specs are normal, and to see if anyone has any suggestions for making it faster; and (2) to warn people away from the route I took and to encourge them to use a much cheaper, better-performing RAID 0 solution. I wish I'd taken this route.
I'm not posting this to get into a huge flamewar about RAID vs. non-RAID in general. I would ask people to try to address my specific situation if possible. But in case this thread does degenerate before any valuable info is added, I will leave this important message for those who click on this in the future:
DON'T WASTE YOUR MONEY ON A $300 RAPTOR LIKE I DID! GET TWO CHEAP GOOD DISKS AND USE RAID 0!
I believed what I'd read about RAID 0 not making a big difference in performance, and I didn't like the increased reliability issues involved with striping across two disks. So I went the single drive route. But when I got my rig built, the HD subsystem didn't feel as fast as it should. Definitely not $299 fast. Then I saw a 70 page sticky thread on DFI-Street.com, in the Software forum, dedicated to posting RAID benchmarks. I decided to benchmark my single-drive Raptor.
Let's compare two solutions:
My setup: Raptor 150GB (single drive).
ATTO avg. read-write: 75mb/s
HD Tach random access: 8.1ms
HD Tach CPU utilization: 1%
HD Tach average read: 76.9mb/s
HD Tach burst speed: 137.3mb/s
Poster from post #1020 on DFI-Street RAID benchmark thread:
2xHitachi 80GB SATA II ($50 each at NewEgg, total $100)
ATTO avg. read-write: 110mb/s
HD Tach random access: 12.7ms
HD Tach CPU utilization: 10%
HD Tach Average read: 97.7mb/s
HD Tach burst speed: 325.6mb/s
He gets between a 27% and 47% performance increase over my Raptor for 1/3 the price (not considering the difference in burst speeds). I could have bought a good video card or put the money towards my dream monitor with that $200! Yes, he has some CPU overhead for the RAID, but I have a dual core and wouldn't notice it. Also I have better random access scores than he does, but the type of performance I want most are application loading times and big file transfers. I would rather have his scores!
But benchmarks are artificial, right? Now look at this page where the user timed various computer operations using a stopwatch, with a single-drive Raptor 74GB installed, and with two Raptor 74GBs installed using RAID 0. Assuming he's not lying and this page is not the work of a mentally-ill person, it's obvious that the RAID 0 solution is much faster in the real world. Link: Who Says 2 Raptors Aren't Better Than One
Some additional details:
- I'm out of my RMA period, otherwise I'd spend the $50 in restocking fee and shipping to return the Raptor.
- I can get an exchange drive from WD. They say my drive should have 84mb/s sustained read, and I'm getting 77. Maybe I'll get a better drive if I exchange it?
- I didn't install the Nvidia IDE SW driver, so I don't have any SATA controls in my Device Manager. I haven't been able to determine whether installing these - which have been problematic for some in the past - could help my performance.
I'm posting this for two reasons. (1) To make sure my Raptor's specs are normal, and to see if anyone has any suggestions for making it faster; and (2) to warn people away from the route I took and to encourge them to use a much cheaper, better-performing RAID 0 solution. I wish I'd taken this route.
I'm not posting this to get into a huge flamewar about RAID vs. non-RAID in general. I would ask people to try to address my specific situation if possible. But in case this thread does degenerate before any valuable info is added, I will leave this important message for those who click on this in the future:
DON'T WASTE YOUR MONEY ON A $300 RAPTOR LIKE I DID! GET TWO CHEAP GOOD DISKS AND USE RAID 0!