[snaggs]

Im getting a new PowerMac, and need a 2nd hard disk on which to put my Aperture library. I was wondering if I should;

1) Buy a Raptor 74gb and move my system onto here (how do I do this btw?)'
2) Get another 250gig hard-drive and use RAID 1 mirroring for faster reads (do you get faster reads with software raid? I know you did with hardware raid on PC's)

Its just that the drive has been around for a while, so I was wondering if it was still substantionly faster than say the 16meg cache WD 250gig drive.


Daniel.

[mduell]

I know a few people who bought Raptors, and they say they are a bit faster but not worth the speed-size-price tradeoff.

RAID1 doesn't improve read speeds (unless you're using one of those newfangled RAID10-on-two-disks "hacks", but I don't think OSX supports that) and software RAID is going to hurt performance instead of helping.

16MB cache matters for at best the first 107 milliseconds; after that you're stuck reading from the disk.

I'd buy a reputable brand (Hitachi, Seagate, WD, Maxtor) 250 or 300GB drive (they're about the same in price/size these days) and make regular backups instead of using RAID.

[svtcontour]

I personally feel its definitly worth it. I dont use a Raptor, I use 15K SCSI drives in my main box but non the less, I've also experienced 10K SCSI and the raptor is about the same speed as current 10K SCSI which is fast! I say grab the raptor, make that your system drive and applications drive. Put all your data and work in progress on the 7200RPM drive.

[new newton]

SATA and SCSI are very different beasts. A 10K SCSI setup is going to feel quite a bit faster than anything on a flavor of IDE.

[snaggs]

Im leaning towards the Raptor, since it will only be a apps drive.. and 70gig should be enough for that.. all pictures will go on the second drive.

When I get my new machine, how would I move the system etc to the new drive? Just bootup with the CD that comes with the machine? Will I loose all the iLife, iWork software if I do that?

Daniel.

PS. How does Firewire 800 compare to SATA?

[mduell]

Originally Posted by new newton

SATA and SCSI are very different beasts. A 10K SCSI setup is going to feel quite a bit faster than anything on a flavor of IDE.

I don't think the PowerMacs even support NCQ for SATA.

Originally Posted by snaggs

Im leaning towards the Raptor, since it will only be a apps drive.. and 70gig should be enough for that.. all pictures will go on the second drive.

When I get my new machine, how would I move the system etc to the new drive? Just bootup with the CD that comes with the machine? Will I loose all the iLife, iWork software if I do that?

How does Firewire 800 compare to SATA?

You can clone your drive using something like CarbonCopyCloner (or dd if you're unix-inclined).

FW800 is slower than SATA. The theoretical bandwidth is there, but you can't get the same bandwidth or latency that SATA provides. Barefeats compared the two here (look at the 1x1 configurations, since you're not using RAID).

[snaggs]

On another forum I found these links, which I thought you might find interesting.. seems the Raptor is only marginally faster than the current generation of 7200's;



Check here for the rest..

Daniel.

[svtcontour]

Originally Posted by snaggs

On another forum I found these links, which I thought you might find interesting.. seems the Raptor is only marginally faster than the current generation of 7200's;



Check here for the rest..

Daniel.

I dont think judging the performance of the drive on the basis of a reboot really makes sense. Its kind of judging the performance of a car which will be racing on a race course based on its 0-60 time. The best way is to check forums for people who have upgraded and get their input, especially whenit comes to a system that is used heavily in a multitasked environment. The 10K drive will pull ahead strongly there.

[snaggs]

On another forum I found these links, which I thought you might find interesting.. seems the Raptor is only marginally faster than the current generation of 7200's;



Check here for the rest..

Daniel.

[svtcontour]

LOL Ok I know I only posted that once. What the hell is going on

[Kristoff]

It's the stoopid macnn refactor *BARF*
Wilco Tango Foxtrot???


Anyway. I have Raptors in one of my machines and the seat-of-the-pants meter indicates that they contribute to an overall increase in snappiness.

[CanadaRAM]

The real point of a Raptor is to use it as a scratch disk for the OS and applications, and to move the data files at least to a second spindle. If you can move the applications to a third drive on a fast buss, that's even better.

The advantages of NCQ and SCSI are negligable for single user desktop computer usage -- these technologies (along with RAID) show their benefit on servers under multi-user loads. See www.storagereview.com for more discussion on this.

[badtz]

Originally Posted by CanadaRAM

The real point of a Raptor is to use it as a scratch disk for the OS and applications, and to move the data files at least to a second spindle. If you can move the applications to a third drive on a fast buss, that's even better.

if you're limited to a Raptor and a 500gb HD, would it be better for the Raptor to be the main HD and the 500gb to be storage/scratch?

If you have all of the libraries of various applications (FCP, Logic, GarageBand, etc) on the 500gb; will it hurt performance by a lot to have your audio/video written to that drive also?

[OreoCookie]

Originally Posted by new newton

SATA and SCSI are very different beasts. A 10K SCSI setup is going to feel quite a bit faster than anything on a flavor of IDE.

This is not true in case of the Raptor. The Raptor is specced like a SCSI drive (5-year warranty, etc.).

[new newton]

And length of warranty is a factor when judging the speed of a storage system? Uh huh...

Read a little about the underlying technologies, rather than superficially looking at specifications that aren't relevant.

[action]

hi gang,

i have dual rev a 2ghz powermac g5 with 4.5 gigs of ram, ati x800 video card.

i have both drive bays installed with a 74 gb raptor, soft raided 0 for a total of 138.5 gigs.
this is my boot drive and i have installed aperture and it's library on it.

i also i have two internal 250 gb maxtor and wd drives with 16mb cache using the sonnet 4+4 card with g5 bracket.

accessing the the aperture library from the raptor drives was noticabley faster on the raptors then the wd or maxtor drives. it was most apparent when i was stacking. i still get a beach ball with the raptors but it's only momentary. with the other two drives which were not raided btw, you have to wait about 5 times longer.

it could be other factors such as the drives being on the card instead of the main bus etc...

booting and shutting down is superfast on raptor raid. i also saw the barefeats benches and i
don't think they represent my real life performance.

just my observations.

chung lee

[Mac_Developer]

Originally Posted by snaggs

2) Get another 250gig hard-drive and use RAID 1 mirroring for faster reads (do you get faster reads with software raid? I know you did with hardware raid on PC's)

One of the benefits of SoftRaid over Apple's RAID is stripped reads from a mirrored drive configuration.

That said, my quad G5 is supposed to arrive tomorrow with one 500gig drive and another on my shelf waiting to be dropped in. I'm still debating whether to configure a RAID 0 or have two independant disks with the system and apps on one and data on the other. Then there is the case for partitioning for running different operating system versions for testing, etc.

[OreoCookie]

Originally Posted by new newton

And length of warranty is a factor when judging the speed of a storage system? Uh huh...

Read a little about the underlying technologies, rather than superficially looking at specifications that aren't relevant.

Excuse me?
I do own a server with SCSI drives, and I am working with Sun hardware at university.

The reason why I brought up the warranty was to illustrate that it is specced as a workstation or server drive and not a desktop drive. Why should a 10k drive with a SCSI connector be any faster? The 10k drives that are out now use full-size platters (unlike earlier models with smaller platters to further reduce the seek time), only the 15k drives have smaller platters (2 inch).

I know the underlying technologies and there are quite a few things that SATA has, but SCSI does not. The total transfer rate is shared with SCSI whereas each SATA drive has a dedicated point-to-point connection. Nominally SCSI is faster, U320 SCSI has a bandwidth of 320 MB/s. Put in two drives and you essentially have the same bandwidth as SATA1 (150 MB/s). Thing is, you usually connect more than two drives to a SCSI adapter

Rather than making brush accusations, you should stick to factual arguments. Other people might know more than you give them credit for

[new newton]

Originally Posted by OreoCookie

I do own a server with SCSI drives, and I am working with Sun hardware at university.

Congratulations. Every long-time Mac user has owned SCSI hardware. Working with Sun hardware adds absolutely nothing to your credibility. Why would you think it does?!

The reason why I brought up the warranty was to illustrate that it is specced as a workstation or server drive and not a desktop drive.

Which, of course, has zilch to do with performance. You're wrapped up in marketing, but light on application.

Why should a 10k drive with a SCSI connector be any faster?

Because of the nature of SCSI as compared to any sort of IDE implementation. SCSI uses considerably fewer system resources. This is why I said it's a bad idea to compare the two based on spindle speed, but you didn't grasp that.

Rather than making brush accusations, you should stick to factual arguments. Other people might know more than you give them credit for

The word is brash. I have stuck to factual arguments--I just haven't lowered myself to believing that spindle speed and warranty length are the differentiating factors between different storage systems. I'm also quite certain that there are a great many people that know quite a bit--even if not all of them have Sun hardware at their disposal.

[OreoCookie]

Originally Posted by new newton

Congratulations. Every long-time Mac user has owned SCSI hardware. Working with Sun hardware adds absolutely nothing to your credibility. Why would you think it does?!

First of all, my first Mac was a PowerBook G3, so this and all of my other Macs (iBooks and PowerBooks) had no SCSI drives whatsoever. (Although my veryfirst computer ever had SCSI-1 (Amiga 500) ) No, it's a FreeBSD server (IBM Netfinity hardware) and not a Mac.

Originally Posted by new newton

Which, of course, has zilch to do with performance. You're wrapped up in marketing, but light on application.

Nope, not really. It just says the manufacturer is ready to give a longer warranty which is appreciated when you get even old drives replaced. This is a criterion when you buy computers that are supposed to last longer. We even had 2.1 GB SCSI drives (in some older Dec AlphaStations) replaced free of charge).

Originally Posted by new newton

Because of the nature of SCSI as compared to any sort of IDE implementation. SCSI uses considerably fewer system resources. This is why I said it's a bad idea to compare the two based on spindle speed, but you didn't grasp that.

You're living in ancient history. Ever since IDE learned DMA (direct memory access), this is a non-issue, especially with today's cpu power. The fact that you still refer to it as IDE shows what you are thinking of here

Serial ATA does not slow down any drive you can buy nowadays. And no, spindle speed is not a measure for performance, but it's part of the performance picture (the other ingredient is data density per area). Serial ATA can do NCQ (which does not make a difference for most desktop applications anyway), so feature-wise the interfaces are on-par. I don't feel like lecturing you on this, though. You can read it up yourself here.

Originally Posted by new newton

The word is brash. I have stuck to factual arguments--I just haven't lowered myself to believing that spindle speed and warranty length are the differentiating factors between different storage systems. I'm also quite certain that there are a great many people that know quite a bit--even if not all of them have Sun hardware at their disposal.

Excuse me? I've never claimed that. The Raptor is aimed at servers and workstations. Take a look at the price and the features (warranty being one of them, note the `etc.'). This has nothing to do with brain wash. It's just a factual argument.

[new newton]

Perhaps English isn't your first language, but what I said is that all long-time Mac users have owned SCSI devices. You don't fit that category, apparently. Why do you keep making a big deal out things that really don't lend you any sort of credibility?

I said any IDE implementation, and I'm correct in that. You may not agree, but you're free to be wrong. Nothing can prevent that, apparently. While it's great that SATA has NCQ instructions, that's completely irrelevant when talking about a stock G5--the G5 can't make use of NCQ.

SATA and SCSI aren't in the same class. You can yap about warranties until the cows come home, but it won't change that. Please, feel free to continue your exploration of things that don't make you credible (is it a long list?), irrelevant capabilities, and comparing apples to oranges. I'm sure it'll all be very factual. Useless, but factual.

[snaggs]

Originally Posted by Mac_Developer

One of the benefits of SoftRaid over Apple's RAID is stripped reads from a mirrored drive configuration.

That said, my quad G5 is supposed to arrive tomorrow with one 500gig drive and another on my shelf waiting to be dropped in. I'm still debating whether to configure a RAID 0 or have two independant disks with the system and apps on one and data on the other. Then there is the case for partitioning for running different operating system versions for testing, etc.

Aha, now that sounds like what I want.. have you seen any benchmarks on this? Are the write speeds alot slower?

Daniel.

[snaggs]

Just done more reading, it seems that Apples Software RAID does 99% of what SoftRaid does, and I don't have to worry about making special bootup disks.

From what I understand, you can make raid sets out of partitions.. so what I'm thinking is the following setup;

2 x 250 gig drives configured as

1x125gig RAID1 for Apps and system files
1x250gig RAID0 Volume for Aperture master file (photo's) & scratch

and then use the external hard-drives I have as vaults, backing up masterfile every hour. The only question is if the RAID 1 striped reads of SoftRAID are worth $139, which is really very dear for a GUI they plop on top of Apples raid functionality.

Daniel.

[mduell]

Originally Posted by OreoCookie

The reason why I brought up the warranty was to illustrate that it is specced as a workstation or server drive and not a desktop drive.

Might want to tell Seagate they should be slapping workstation labels on all their bargain desktop drives.

[snaggs]

One last question, do the PowerMacs come with the SATA cable inside to connect a second drive? I bought a WD250gig today, and the PowerMac arrives on Monday.

Daniel.

[new newton]

Yes. You'll also find four black mounting screws just to the left of the drive bay. Use them.

[Cadaver]

Originally Posted by snaggs

One last question, do the PowerMacs come with the SATA cable inside to connect a second drive? I bought a WD250gig today, and the PowerMac arrives on Monday.

Daniel.

Yes.

[CatOne]

Originally Posted by new newton

...

SATA and SCSI aren't in the same class. You can yap about warranties until the cows come home, but it won't change that. Please, feel free to continue your exploration of things that don't make you credible (is it a long list?), irrelevant capabilities, and comparing apples to oranges. I'm sure it'll all be very factual. Useless, but factual.

Fine. SCSI drives are better for server use than are SATA drives. How does that make them ANY faster for workstation use? NCQ is pretty much worthless for a "desktop" workload -- it's useful when you have tons of requests coming in from all over the machine simultaneously, because you can sweep the heads a lot less by batching up reads or writes to areas that are nearer each other, reducing thrash.

So seek times of SCSI drives are a bit better than IDE or SATA drives. But sustained throughput is no better -- areal density of IDE/SATA drives is so much higher, that they can push 65 to 70 MB/sec on a 7200 RPM drive. No 10K SCSI drive can do that -- only some of the 15K drives can. And the original comparison was on 10K SCSI to 10K SATA... if you don't have side-by-side benchmarks your "I know SCSI, and you're a fool" high horse is fairly artificial, no?

[CIA]

Just a note here, I have a quad, and am running 2 raptors intenally striped into a software RAID 0 using OSX Disk utility, (for the record, I am booting off this), it is crazy snappy. I am not the best to comment on it tho, as I never even booted off the drive the machine came with, I took it out before I powered it up for the first time and installed 10.4 on the newly created RAID. Coming from a Dual 800 Quicksilver the speed is great.
I got the Raptors for $120 each from Newegg during the day after thanksgiving sale. ($160 with mail-in rebates knocking the price down to $120)
Yes, pricey, but I had a badass new machine coming, I wanted a badass RAID to boot it from!

And Since it's a point, I have been using Macs since before they had SCSI (first one was a 512ke), and through all the SCSI years. My DP800 was the first mac I owned that didn't have it.
512Ke
Powerbook170
Mac LC
Mac LC III
Quadra 6400/180
Powermac G3/266
PM Quicksilver DP800
Powermac Quad
Toss about 4 other random machines in there if you count the ones my parents owned over the years.

I like SCSI, the new flavors are awesome and still fast. But price wins out for me, I'll take a fast striped IDE or SATA RAID (WITH BACKUP, without one you are just looking for disaster) over SCSI anyday now.

[svtcontour]

One thing I will say about small fast drives. I think its definitly worth having, especially anything that can offload the CPU. I'm not sure exactly what is giving me the most benefit but my system is a monster and I can do tons of disk intensive tasks and the system still feels snappy enough even though my hard disk activity light is on solid and I can hear the drive going nuts.

Granted I'm running 15k U320 drives (on an U160 controller) in my dual xeon box but I think a 10K raptor would give at least 50% the boost I felt going from IDE to my setup and that would be astronomical. Again this is only an issue if you've got multiple disk intensive things on the go.....

[mduell]

Originally Posted by CatOne

So seek times of SCSI drives are a bit better than IDE or SATA drives. But sustained throughput is no better -- areal density of IDE/SATA drives is so much higher, that they can push 65 to 70 MB/sec on a 7200 RPM drive. No 10K SCSI drive can do that -- only some of the 15K drives can.

Excuse me, you're spewing.
10KRPM SCSI drives from Seagate, Hitachi, Fujitsu, and Maxtor are all faster than the fastest 10KRPM SATA drive, much less any 7.2KRPM SATA drive (source), and they have similar areal densities.

[OreoCookie]

Originally Posted by mduell

Excuse me, you're spewing.
10KRPM SCSI drives from Seagate, Hitachi, Fujitsu, and Maxtor are all faster than the fastest 10KRPM SATA drive, much less any 7.2KRPM SATA drive (source), and they have similar areal densities.

Have you taken a look at your link?
If we pretend that a single figure can somehow measure the performance of a drive, the page you have linked to says basically the opposite: only one 10k drive is faster While I would not want to measure these drives with desktop performance figures (here it was some sort of office benchmark), your link should at least be consistent with what you're trying to argue.

Alternatively, there is a head-to-head benchmark between the latest-gen Raptor and a 10k SCSI drive here (also a storagereview article).

While WD has delivered a solution that can match a SCSI-based solution's speed and scalability, one must also keep in mind the key factors of infrastructure and reliability. As with TCQ itself, SATA's support hardware such as backplanes, all-in-one solutions, and the like remain in their infancy when contrasted to the maturity and longevity of SCSI hardware. Also keep in mind that while Western Digital claims an enterprise-class 1.2 million hour MTTF spec and backs the Raptors with a 5-year warranty, the line is still new and remains relatively unproven compared to established solutions such as Seagate's Cheetah series. Finally, remember that the prices listed above represent the cost of the storage subsystem alone- factoring in the total cost of server hardware when motherboards, CPU, and RAM are considered can dilute the difference significantly.

[new newton]

It's pretty amazing how easily the point can be lost. Someone stated that the two are equivalent, and those of us who know what we're talking about stated that they're very different storage systems. How tough was that?

[OreoCookie]

Originally Posted by new newton

It's pretty amazing how easily the point can be lost. Someone stated that the two are equivalent, and those of us who know what we're talking about stated that they're very different storage systems. How tough was that?

I have no idea what you want to say here. And besides if you want to say it's me who claimed something, don't say `someone'. I also wonder what makes you an authority on this topic

The Raptor is aimed at low to mid-range servers and workstations and storagereview as well as other reliable sources say that they are equivalent for that.

I think it's rather you who's buying into the marketing of the other harddrive manufacturers about SCSI and all (Ever wonder why no other harddrive manufacturer produces 10k SATA drives? Hint: which harddrive manufacturers do produce SCSI drives?)

Also, I'm sure you are aware that the next-gen SCSI (SAS) is based on SATA. Yes, that's right, if you have a SAS adapter, you can use SATA drives (not the other way around, though). I think this pretty much deals with your claim that SATA is technologically inferior to SCSI

[new newton]

You have no idea what I'm saying? What I'm saying--and what I've been saying--is very simple. SCSI and SATA are not equivalents, and should not be viewed as such. Why on Earth is that so difficult for some of you?

[mduell]

Originally Posted by OreoCookie

Have you taken a look at your link?
If we pretend that a single figure can somehow measure the performance of a drive, the page you have linked to says basically the opposite: only one 10k drive is faster While I would not want to measure these drives with desktop performance figures (here it was some sort of office benchmark), your link should at least be consistent with what you're trying to argue.

Have you taken a look at the post that I was replying to?

Originally Posted by CatOne

So seek times of SCSI drives are a bit better than IDE or SATA drives. But sustained throughput is no better -- areal density of IDE/SATA drives is so much higher, that they can push 65 to 70 MB/sec on a 7200 RPM drive. No 10K SCSI drive can do that -- only some of the 15K drives can.

I've bolded the performance metric he wanted to compare.

Now from StorageReview's transfer rate test:
Maxtor Atlas 10K V (300 GB Ultra320 SCSI) - 89.5 |
Fujitsu MAT (300 GB Ultra320 SCSI) - 88.5 |
Hitachi Ultrastar 10K300 (300 GB Ultra320 SCSI) - 86.3 |
Seagate Cheetah 10K.7 Server Mode (300 GB Ultra320 SCSI) - 76.5 |
Seagate Cheetah 10K.7 Desktop Mode (300 GB Ultra320 SCSI) - 76.5 |
Western Digital Raptor WD740GD-00FLA1 with TCQ (74 GB SATA) - 71.8 |
Western Digital Raptor WD740GD-00FLA1 no TCQ (74 GB SATA) - 71.8 |
Seagate Barracuda 7200.8 no NCQ (400 GB SATA) - 69.8 |
Western Digital Caviar WD3200JD (320 GB SATA) - 66.5 |
Maxtor MaXLine III with NCQ (300 GB SATA) - 65.7 |

As I said, 10KRPM SCSI drives from Seagate, Hitachi, Fujitsu, and Maxtor are all faster than the fastest 10KRPM SATA drive, much less any 7.2KRPM SATA drive.

[mduell]

Originally Posted by OreoCookie

Have you taken a look at your link?
If we pretend that a single figure can somehow measure the performance of a drive, the page you have linked to says basically the opposite: only one 10k drive is faster While I would not want to measure these drives with desktop performance figures (here it was some sort of office benchmark), your link should at least be consistent with what you're trying to argue.

Have you taken a look at the post that I was replying to?

Originally Posted by CatOne

So seek times of SCSI drives are a bit better than IDE or SATA drives. But sustained throughput is no better -- areal density of IDE/SATA drives is so much higher, that they can push 65 to 70 MB/sec on a 7200 RPM drive. No 10K SCSI drive can do that -- only some of the 15K drives can.

I've bolded the performance metric he wanted to compare.

Now from StorageReview's transfer rate test:
Maxtor Atlas 10K V (300 GB Ultra320 SCSI) - 89.5 |
Fujitsu MAT (300 GB Ultra320 SCSI) - 88.5 |
Hitachi Ultrastar 10K300 (300 GB Ultra320 SCSI) - 86.3 |
Seagate Cheetah 10K.7 Server Mode (300 GB Ultra320 SCSI) - 76.5 |
Seagate Cheetah 10K.7 Desktop Mode (300 GB Ultra320 SCSI) - 76.5 |
Western Digital Raptor WD740GD-00FLA1 with TCQ (74 GB SATA) - 71.8 |
Western Digital Raptor WD740GD-00FLA1 no TCQ (74 GB SATA) - 71.8 |
Seagate Barracuda 7200.8 no NCQ (400 GB SATA) - 69.8 |
Western Digital Caviar WD3200JD (320 GB SATA) - 66.5 |
Maxtor MaXLine III with NCQ (300 GB SATA) - 65.7 |

As I said, 10KRPM SCSI drives from Seagate, Hitachi, Fujitsu, and Maxtor are all faster than the fastest 10KRPM SATA drive, much less any 7.2KRPM SATA drive.

[CatOne]

Originally Posted by mduell

Excuse me, you're spewing.
10KRPM SCSI drives from Seagate, Hitachi, Fujitsu, and Maxtor are all faster than the fastest 10KRPM SATA drive, much less any 7.2KRPM SATA drive (source), and they have similar areal densities.

I'm spewing?

Thanks for the link... it shows the WD SATA drive FIRMLY parked amongst the 10K RPM SCSI drives. The ONLY drives that are substantially (more than 10%) faster are the 15K SCSI drives.

When you provide a link, it's best if it works FOR you, rather than against you!

Here's a little cut and paste of the SR Office DriveMark 2002 IOPS, where you see the *only* 10K SCSI drives faster than the Raptor are the Maxtor Atlas 10K and the Fujitsu MAS3735. A number of 15K drives and other 10K SCSI drives are slower:

Use the checkboxes on the left to choose devices for the head-to-head comparison

Maxtor Atlas 15K II (147 GB Ultra320 SCSI) - 725 |
Fujitsu MAU (147 GB Ultra320 SCSI) - 719 |
Seagate Cheetah 15K.4 Desktop Mode (147 GB Ultra320 SCSI) - 633 |
Maxtor Atlas 15k (73 GB Ultra320 SCSI) - 621 |
Maxtor Atlas 10K V (300 GB Ultra320 SCSI) - 613 |
Fujitsu MAS3735 (73 GB Ultra320 SCSI) - 610 |
Western Digital Raptor WD740GD-00FLA1 no TCQ (74 GB SATA) - 585 |
Seagate Cheetah 15K.3 (73 GB Ultra320 SCSI) - 578 |
Fujitsu MAT (300 GB Ultra320 SCSI) - 568 |
Maxtor Atlas 10k IV (147 GB Ultra320 SCSI) - 503 |
Fujitsu MAP3147 (146 GB Ultra320 SCSI) - 490 |
Seagate Cheetah X15-36LP (36.7 GB Ultra160/m SCSI) - 485 |
Maxtor MaXLine III with NCQ (300 GB SATA) - 485 |
Western Digital Raptor WD740GD-00FLA1 with TCQ (74 GB SATA) - 485 |
Western Digital Raptor WD360GD (36 GB SATA) - 483 |
Seagate Savvio 10K.1 Desktop Mode (73 GB Ultra320 SCSI) - 483 |
Maxtor MaXLine III no NCQ (300 GB SATA) - 472 |
Hitachi Deskstar 7K400 no TCQ (400 GB SATA) - 472 |
Western Digital Caviar WD2500KS (250 GB SATA) - 461

[mduell]

Originally Posted by CatOne

I'm spewing?

Thanks for the link... it shows the WD SATA drive FIRMLY parked amongst the 10K RPM SCSI drives. The ONLY drives that are substantially (more than 10%) faster are the 15K SCSI drives.

When you provide a link, it's best if it works FOR you, rather than against you!

<snip irrelevant benchmark>

Yes, you're spewing. Your original post involved a specific claim about a specific performance metric. That claim is not true.
To refresh everyone's memory:

Originally Posted by CatOne

So seek times of SCSI drives are a bit better than IDE or SATA drives. But sustained throughput is no better -- areal density of IDE/SATA drives is so much higher, that they can push 65 to 70 MB/sec on a 7200 RPM drive. No 10K SCSI drive can do that -- only some of the 15K drives can.

My link backed up my assertion that 10K SCSI drives provide higher sustained throughput than any SATA drives do and SCSI drives that can push more than 65-70MBps sustained throughput.

You didn't say anything about 10% faster. You didn't say anything about IOPS.
You asserted that the sustained throughput of 10KRPM SCSI was no better than SATA. You asserted that no 10KRPM SCSI drive can push 65-70MBps. Both of those assertions are false.
Please graciously accept this correction.

[ghporter]

Zealots... what do you do with them?

SCSI and SATA have their own roles, and sometimes they get to swap. That's cool. There's no reason to start asking each other to "step out back and settle it."

The biggest problems I've ever seen with SCSI is finding quality interface cards with drivers that work. This is on Wintel platforms, by the way, as I understand Macs have been SCSI-friendly for ages. No matter, SCSI still demands a premium for "modern capacity" drives, though you can often find 4GB to 15GB SCSI drives at computer shows for cheap. I don't need what SCSI provides enough to want to pay those premium prices.

My personal take on the 74GB Raptor is that I ain't paying nearly $200 for ANY 74GB drive. The thing can be supersonic for all I care, and I'm not going to pay that much. I just ordered a 250GB SATA-150 Maxtor drive for my Windows desktop, and with shipping it's going to cost less than $100. I cannot imagine an application for a horrendously fast drive that would work with hardware I'm likely to have access to, nor a need for such superfast drives for anything I do have access to. In short, you're not going to find me driving to the grocery store in a Lotus, and there's similarly no reason for me to buy such a high-dollar drive.

[svtcontour]

Originally Posted by ghporter

Zealots... what do you do with them?

SCSI and SATA have their own roles, and sometimes they get to swap. That's cool. There's no reason to start asking each other to "step out back and settle it."

The biggest problems I've ever seen with SCSI is finding quality interface cards with drivers that work. This is on Wintel platforms, by the way, as I understand Macs have been SCSI-friendly for ages. No matter, SCSI still demands a premium for "modern capacity" drives, though you can often find 4GB to 15GB SCSI drives at computer shows for cheap. I don't need what SCSI provides enough to want to pay those premium prices.

My personal take on the 74GB Raptor is that I ain't paying nearly $200 for ANY 74GB drive. The thing can be supersonic for all I care, and I'm not going to pay that much. I just ordered a 250GB SATA-150 Maxtor drive for my Windows desktop, and with shipping it's going to cost less than $100. I cannot imagine an application for a horrendously fast drive that would work with hardware I'm likely to have access to, nor a need for such superfast drives for anything I do have access to. In short, you're not going to find me driving to the grocery store in a Lotus, and there's similarly no reason for me to buy such a high-dollar drive.

SCSI works much better driver and software wise in Wintel land than in Macland. I have had SCSI all my computing life and build machines for people so I have experience with SATA including the Raptor. If a user does not multitask heavily then SATA drives are fine. I find that SCSI drives/controller combos - even ones that have lower benchmarks tend to bog down less under heavy loads.

If for some reason, I could not get SCSI anymore, I'd go with the Raptor. Speed is very important to me and I do not want to feel the computer feeling bogged down even under heavy usage. IDE/ 7200RPM SATA does not allow for this. The Raptor is OK but in the end, SCSI is the only one that feels like its not choking.

[mduell]

Originally Posted by ghporter

My personal take on the 74GB Raptor is that I ain't paying nearly $200 for ANY 74GB drive. The thing can be supersonic for all I care, and I'm not going to pay that much. I just ordered a 250GB SATA-150 Maxtor drive for my Windows desktop, and with shipping it's going to cost less than $100.

Here's to that. When you find yourself choosing between a 36GB 10KRPM and 250GB 7.2KRPM or 74GB 10KRPM and 320GB 7.2KRPM drive at the same price, it's time to step back and ask yourself if the double-digit percentage performance increase or triple-digit percentage capacity increase is a better idea.

[CanadaRAM]

Originally Posted by badtz

if you're limited to a Raptor and a 500gb HD, would it be better for the Raptor to be the main HD and the 500gb to be storage/scratch?

If you have all of the libraries of various applications (FCP, Logic, GarageBand, etc) on the 500gb; will it hurt performance by a lot to have your audio/video written to that drive also?

No, the consensus seems to be that the limiting factors for drive speed is
1) having the data and the swap file on the same spindle, which forces the heads to constantly move between data and scratch reading/writing (hence moving the data away from the System drive speeds performance)
2) the speed of the swap file / scratch disk drive, because the OS and Photoshop scratch files are used constantly, much more often than the data files.

So if you have 2 disks, use your fastest one for the System and scratch, and your largest one for data. I haven't had it shown conclusively to me whether putting the applications onto the System/scratch drive, or onto the data drive, is preferable.

If you can separate scratch, system/applications and data onto three spindles, so much the better. Heavy duty Photoshop users sometimes dedicate a Raptor to Photoshop scratch alone.