AVS Forum banner

Status
Not open for further replies.
1 - 9 of 9 Posts

·
Banned
Joined
·
1,007 Posts
Discussion Starter #1
Was checking out various reviews on the Barracuda IV, just to make sure my purchase was a wise one. Whew, thats good to see, virtually silent at idle, reliable, and fast. Great.


The Barracuda IV has come to light recently as an ideal purchase for HTPC uses. If you are thinking of using these drives in a RAID0 configuration, however, think again. I was on a hardware forum at tomshardware.com and spotted an email from Seagate stating that the drives are simply too fast for RAID 0 use. This will make them no faster in RAID 0 then using them individually. The fact that the Barracuda drives have to wait for the controllers, makes the whole process go out of sync and throws out the speed advantages of RAID 0.


So for me, the 80GB Barracuda IV will be quiet, fast and reliable. A perfect purchase. But I am not going to put it in a RAID 0 configuration, and according to everyone else, including Seagate, netiher should you.


here's a thread at storagereview.com regarding the issue.
 

·
Registered
Joined
·
386 Posts
I do not understand the need for ide raid for a HTPC. Have you ever thought of going SCSI? 10,000 RPM Ultra 160 is fairly cheap and the benefit of access time is more apparent than throughput. Just a thought.
 

·
Registered
Joined
·
316 Posts
orbillius,

I also agree. I don't think RAID is necessary for HTPC. I've been using RAID since 1994 and although the speed is nice, it is not necessary. Many digital developers use firewire connections to record and playback their video and that is slower than standard ATA/100 (400Mbps = 50MB/s vs. 100MB/s), much less Ultra 160 SCSI. Besides, running RAID-0 will make a mess of your system when one of the drives goes dead (happens fairly often during RAID use).
 

·
Registered
Joined
·
386 Posts
Yea, I had a 60BG raid 0 partition go out on me about a year and a half ago. Thats why I moved to several 10,000 rpm scsi drives.
 

·
Banned
Joined
·
1,007 Posts
Discussion Starter #5
perhaps my post is not very clear. I do not use RAID 0. However many people take the recommendations on this board and others into consideration when making a purchase. If they WERE going to use them in a RAID 0 setup they would be REALLY disappointed.


I believe there are a decent number of members on this forum who use RAID 0.


I agree that Raid 0 is not important to a HTPC setup. Its speed is attractive for all users. But the reliability leaves something to be desired. If there was a keyboard that would help me do things quicker, but occassionally would require reinstalling windows, then I woudn't use it either.
 

·
Registered
Joined
·
3,679 Posts
My friend the video editor has 4 80 GB ATA Barracuda IV's in a RAID 0 using an on-board Promise Raid controller. (MSI with dual PIII 866's oc'd to ???)


Anyway, his performance in real life as well as bench marks is fantastic. He get's almost 35,000 in WinBench Hiighend C++ benchmark (my fav)! The worlds fastest single drive the Cheetah X15-LP gets less than 30,000.


The IV's do RAID0 quite nicely :)


Ken
 

·
Banned
Joined
·
1,007 Posts
Discussion Starter #7
quote from the link on my first post.
Quote:
The Barracuda ATA IV has the fastest internal transfer rate of any of its competitors. This is why it is the highest performance drive in its class in certain applications when used by itself. Evidently, when the drive is used in some RAID 0 environments, it can supply data to the interface faster than the host system can request it. Under some circumstances, such

as reading sequential data, this can cause the drive to incur a latency. This means when the host request comes too late -- after the data's initial immediate availability -- the drive must wait for the disc to rotate up to one revolution for the requested data to be available again. Under these circumstances, the drive appears to be slow in performance when actually it is too fast. This is not a new phenomenon. Because Seagate is the leader in new technologies and products, and the increased performance

they bring, we sometimes have to wait for the rest of the industry to catch up. When Seagate introduced the Cheetah X15, over one million drives ago, there was a similar issue with a few SCSI RAID controllers. Like the SCSI RAID environment, we anticipate optimized controllers will become available. In the meanwhile, it appears some 2-drive ATA RAID 0s can't keep up with the Barracuda ATA IV. There is no alternate slower firmware

available to accommodate the problem.



Best Regards,


Ron Stacy

Seagate Product Support
 

·
Registered
Joined
·
2,912 Posts
I think the idea of these drives being 'too fast' for RAID is highly unlikely. The only way I could see this being true is in the case of one of the firmware-only RAID cards used on a system with a slow CPU. Since the firmware-only RAID adapters use the host-CPU instead of having a dedicated processor, you wouldn't want to use one in a Celeron 366. But these days 1.2Ghz and up is not at all uncommon, and in such a system the firmware-only RAID controllers can actually outperform true hardware RAID adapters for RAID-0 because the host CPU is so damn fast. I think the explanation offered by Seagate is more CYA/finger-pointing than anything else. After all, high-end SCSI drives are used in RAID configurations all the time.


While RAID-0 may be overkill some some HTPC's, it does have its advantages for those of us who are recording HDTV. In addition to performance, it also offers the advantage of treating multiple drives as a single large drive. My accessDTV wants to put all recordings in the same folder, so having a 45GB C: drive and a 45GB D: drive wouldn't do me much good.


And I hardly think that 10K-rpm SCSI drives are a good alternative for HTPC. Cost aside (and those drives are not cheap compared to IDE), 10K SCSI drives are too noisy and too hot to be worth the trouble for most HTPC's.
 

·
Registered
Joined
·
182 Posts
Quote:
Evidently, when the drive is used in some RAID 0 environments, it can supply data to the interface faster than the host system can request it. Under some circumstances...
Sounds like a lot of ifs to me. This might be true if you're using slower controllers like Ultra 33s or your processor is slow or busy. Seagate states an internal transfer rate of 555 Mbits/sec with a 24 to 41 Mbytes/sec average external sustained rate. It has a Ultra ATA/100 interface that leads to it's max burst rate of 100 Mbytes/sec. But it would only be able to burst whats in it's 2MB cache at that rate. 2MB at 100/sec doesn't take very long. As you can see this drive can not flood the Ultra 100 specification. And theoretically two drives on the same channel couldn't flood the ultra 100 spec. This is probably the first drive that could flood a ultra 33 channel. If you had a well designed ultra 100 controller I can't see how the drives would have to wait for the controller. Next to floppy drives and cd/dvd drives, hard disks are the slowest device in your PC. Why do you think it still take several minutes to boot a PC. Hard drives haven't even come close to keeping up with the speed increases in processors and memory. Heck, these GHz processors don't do you much good for general purpose stuff when hard disks are still as slow as they are. My general purpose PC that I'm typing this reply on is still a 266 PII with Win 2K, and I think it does just fine.


Rob
 
1 - 9 of 9 Posts
Status
Not open for further replies.
Top