With my next computer, I want to focus more on harddrive performance (something I completely ignored on this machine). So, I'm in my planning stage and trying to decide what to do. Right now, my two options are this:
Use my two current 20gb 5400rpm 2mb cache drives in a raid 0 array using the onboard raid that will come with whatever motherboard I choose (I'm not 100% sure, but probably <A HREF="http://www.gameve.com/store/gameve_viewitem.asp?idprodu..." target="_new">this</A> one). I will put my OS, and maybe swapfile on it. They will be on the same channel. While, at the same time, using a "Western Digital Caviar Special Edition 120.0GB Internal Hard Drive" SE (duh) (BB has a great deal on them right now) as my game and file storage. Or. . .
Use a "Seagate ST19171WC 9.1GB SCSI Hard Drive" 7200rpm (for OS), with the other 120gb drive from above still as my game/storage drive.
So, which of the two would get better loading performance? Also, what kind of converter would I need to use the SCSI (they sell (80-PIN SCA to 68-PIN) adapters, and (80-PIN SCA to 50-PIN)).
No it doesn't, each cable can only access one drive at a time, the thing would have to stop writing to one drive to write to the other. Of course drive cache helps.
<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>
so your saying that if i take two drives, on the same channel and stripe them i will see no performance gain? yea right. that would mean all these servers out there with lets say 4 scsi drives on a single scsi channel are getting no performance gain by striping those drives? please tell me i am misunderstandng what your getting at.
He was talking about IDE drives. IDE decives communicate differently to SCSI devices.
In a Master/Slave setup the controller on the Master drive tells the Slave when it can go ahead and transfer data, and that is only when the Master itself is idle. I guess in theory the data could be transferred at 133MB/s from the MOBO to Master where it is lodged in cache and written out at 30MB/s and whislst that is happening data could be transferred to the Slave. But I have no idea if that actually occurs in practice and would depend how the RAID controller handled the data.
The SCSI bus however, in addition to caching, supports multitasking, command queing and reordering which means multiple commands can be sent to different disks in the array at any one time before the previous task is completed and the disks can even sort those commands into the fastest access order before transferring the data back.
irregardless of what kind of system it is, you mean to say, if i take 4 drives and put them on my ide raid i will see only half the performance? i dont think so. i have done it and benched it. 4 drives gave me 4x the write performance over one drive.
hang on, let me clear my throat *cough*bvllsh!t*cough* *cough*selfrighteous*cough* excuse me, it must be some one in the forums lately. your going to sit there and say blank faced that there is no difference between 2 hard drives on separate channels, and 4 hard drives on 2 separate channels striped?
36gb raptor "second fastest after the RAID 0 array" xbitlabs
In the WD Raptor review, xbitlabs points out how much faster a RAID 0 array is compared to other drives including SCSI. I got 2 36gb raptors in raid0 and man 2 in Raid vs 2 in non-raid is quite a bit quicker.
I got 2 x Raptors in Raid 0, on my MSI K8T Neo2 mobo, I get better performance than a 15000rpm 8MB SCSI device (via Sandra 2003). Not too shabby - Drives really perform well. Installing software is bliss I don't even think it's worth putting me swap and temp files onto a second 7200RPM ide drive.
Well, I actually wrote 'not much of an improvement'. And I meant to say that 2 drives on 2 IDE channels perform about twice of 1 drive (on 1 channel of course), whereas 2 drives on 1 IDE channel perform much less better than 1 drive. How much... depends, as usual, on circumstances, like cache size, access method, etc. This is in contrast with SCSI where 4 drives are about 4x faster than a single drive, regardless the channels these drives are connected to (as long as the channel bandwidth isn't saturated).
LOL. I always get a kick out of this. About the only thing this drive is good for is a large outdated paperweight. Just because a drive is "SCSI" does not make it fast. The drive you are refering to is several generations old and is no match for any of the current IDE drives. Needless to say to buy anything special to use <font color=red> this</font color=red> would be completely pointless.
For those of those not familar with this drive its peak transfer rate is around 12,000 KB/s.
It's not what they tell you, its what they don't tell you!<P ID="edit"><FONT SIZE=-1><EM>Edited by ncogneto on 10/20/03 08:04 PM.</EM></FONT></P>
I got 2 36gb raptors in raid0 and man 2 in Raid vs 2 in non-raid is quite a bit quicker
In benchmarks.....yes. In real life.....not for normal desktop useage. Raid 0 is incredilbly overrated and just simply the latest trend. Head over to storagereview ( the one and only storage review site)and have a gander about what they say about raid 0 and its limited benefits.
It's not what they tell you, its what they don't tell you!