View Full Version : my new raptor
Ok, i just got my raptor in the mail is there any tricks to make thsi thing run real real fast in single drive setup? i mean i know it should blow my seagate 120gb 8mb cache outta the water but will it rnu at blazing fast speeds out side of a raid setup?
Deathknight
10-25-03, 12:42 PM
Yeah a raptor certainly can run fast outside of raid 0 ;) A single raptor is often faster than a 2 drive raid 0 setup of slower drives.
dreammmatt
10-25-03, 05:32 PM
how much faster are (2) Raptors running in RAID-0, compared to a single Raptor?
donny_paycheck
10-25-03, 11:11 PM
Originally posted by dreammmatt
how much faster are (2) Raptors running in RAID-0, compared to a single Raptor?
For throughput, nearly twice as fast. For everyday use the difference is negligible, and may even be slightly inferior in performance to a single drive. RAID 0 increases seek latency, which is a bulk of general purpose drive performance.
deathman20
10-26-03, 02:23 PM
I figured in my system that 1 72gig Raptor (the new one) would be more then plently to beat most system configs to date in a typical machine, and might give some RIAD 0's a kick in the butt.
Sure it would be faster to run in RAID for throughput but the cost of increased latency doesn't cut it for what the drives kick some butt in.
If you want more space get that new Hitachi drive for storage with a Raptor as the main drive. My idea is to get 1 72gig raptor and possibley (depending on price) another 72gig raptor or a Hitachi 80giger.
deathstar13
10-26-03, 05:35 PM
Originally posted by donny_paycheck
For throughput, nearly twice as fast. For everyday use the difference is negligible, and may even be slightly inferior in performance to a single drive. RAID 0 increases seek latency, which is a bulk of general purpose drive performance. ok i know this has been proven many times about what you said.
but i disagree completly when it comes to the speed and usefulness of raid0.
most people who run raid0 dont go back to single drive setups unless saftey is the issue.
id never build a new "main" rig without raid0,and my last one was with raptors as will be this new build.
even in everyday useage the speed is seen and felt.especially while booting.
all the tests and proof all agree with your statements tho.
deathman20
10-26-03, 08:22 PM
I went from a raid setup to single drive setup, then again I'm wierd. All I know with the raid setup really helped because my drives only have 2mb of cache so its limited otherwise.
Raid 0 is great don't get me wrong even with my statements, but I just choose safey over a speed increase.
donny_paycheck
10-26-03, 10:13 PM
Originally posted by deathstar13
ok i know this has been proven many times about what you said.
but i disagree completly when it comes to the speed and usefulness of raid0.
It's funny because I kind of disagree with myself too, heh. I noticed a large speed increase when I went to RAID 0 back on my KG-7 RAID. I swore I'd never go back. Two years later, I am. I'm using SCSI hardware here, so that's actually my reason for using a single drive. When it's a 15k Quantum Atlas, it's already a lot faster than any ATA RAID could be.
It took a lot of reading benchmarks to convince me that RAID 0 didn't really help much.
I'm a IDE RAID0 user myself & have been for the past 2 years. I do have a spare Single HDD with an OS installed on it just for catastrophic failures on my RAID Array.
The thing is whenever I rebuild the array & forget to burn a CD with updated drivers I use the backup drive to get on the net & DL whatever I need. The speed difference is already felt with mundane tasks that I dont see myself going back to a single drive config for everyday use. The speed is just not there, it's just not as snappy.
I don't run benchmarks on that drive so I don't have numbers, just what I feel when using either configuration, but at the end of the day its what you feel thats more important. Well unless you can get Intel I/O Meter to work then maybe:o
Originally posted by donny_paycheck
It's funny because I kind of disagree with myself too, heh. I noticed a large speed increase when I went to RAID 0 back on my KG-7 RAID. I swore I'd never go back. Two years later, I am. I'm using SCSI hardware here, so that's actually my reason for using a single drive. When it's a 15k Quantum Atlas, it's already a lot faster than any ATA RAID could be.
It took a lot of reading benchmarks to convince me that RAID 0 didn't really help much.
Hmm, are you sure about that. From what I saw on the comparison with the raptor, a 73G Atlas was at most 30% faster and thats if you spent over $600 for the drive alone. The smaller Atlas drives were not tested. I would assume they were still faster, but somewhat less. I can't compare the other tests, I don't have them, but in the winbench disk inspection test my raptor raid setup gets about 120mb/sec compared to the 73G Atlas's 74mb/sec at about 1/3 the cost and that doesn't include the cost of the controller. The raid 0 setup that I have with the SD SE get 100mb/sec on the same test. I realize this is only one benchmark and may not be totally valid, but I wouldn't assume that a single scsi drive is faster than an ATA raid will ever be or even now for that matter. Esp with the new raptor about due and reported to be 30% faster in many areas, which is just about the gap between the old raptor and the Atlas.
Edit: I ran winbench business disk winmark 99 and high end winmark 99 on my WD SE raid 0. In Storage reviews test a WD SE drive by itself scores 10.1mb/sec and 35mb/sec respectively. My raid setup scores 20.3mb/sec and 43.1mb/sec. The 73G Atlas scores 14.9mb/sec and 46.5mb/sec. I'll test the raptors in the next couple of days but they will be faster and probably beat the Atlas in both test.
donny_paycheck
10-27-03, 02:42 PM
One thing to keep in mind though is that seek times impact hugely on a drive's overall performance. An ATA RAID 0 will blow most any single drive away in throughput, which is why if that is what you crave, you should definitely use it.
But seek latency constitutes more of a drive's perceptible speed to the average user (games, bootup, etc.). Faster spindle speeds and larger area densities translate into lower seek times. It also increases throughput for each drive, obviously, but that gain isn't as noticeable.
As a final caveat, a RAID 0 slightly increases seek time, because the controller has to wait for each drive to retreive part of the requested data.
My 15k 73 gig Atlas is being RMA'd right now. If it wasn't I'd post ATTO scores between that and my ATA RAID 0 of 80 gig WD SE drives.
Not to give my age away, but I remember when people use to think that seek time was everything and nothing else matters including rpms. I think we have seen that rpms make a big difference. I think using ATTO is fine as long as you compair the size and type of tranfers that are pertanant to desktop performance. Very few of use run servers, which is one reason that storage review has a separate benchmark for desktop/workstation vs server performance. If I had there benchmark suite I'd run that, but winbench 99 was one of the standards, but of course now the applications that it runs have been updated several times. But considering storage review still reports its results for drives it makes it a nice comparison
Did you get the new FDB-equipped, 4.5ms one?
PS: The enemy is weakened :p
Originally posted by Csybe
PS: The enemy is weakened :p
If you're refering to this discussion then not a chance. If not then never mind, lol
donny_paycheck
10-27-03, 05:26 PM
Originally posted by pkrew
If you're refering to this discussion then not a chance. If not then never mind, lol
I think that's a reference to Return to Castle Wolfenstein...although I have no idea why, heh.
well I tested the raptors using winbench 99 and the got 19.4mb/sec in business winmark and 64.7mb/sec in high end winmark. I'm not sure why they scored lower that the WD SE in the business winmark, but they did outscore the SE in the HE test by around 50% and improved by more than 50% over a single raptor. I thought about what you said about throughput and access time and I would expect this benchmark to test both considering it runs several application in a similar way to users. Of course its a little faster at mouse clicks then we are.
Originally posted by donny_paycheck
I think that's a reference to Return to Castle Wolfenstein...although I have no idea why, heh.
yeah...blunt is a big player, but sorry for being offtopic. :rolleyes:
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.