raptors are not SATA 3gb/s, how does this effect performance?

[InCogneato]

Hey guys, I'm planning to get two 74GB Raptors for RAID0 but I notice these are not SATA 3gb/s. Would they be faster in RAID0 if they were 3gb/s or would it not boost their performance? Any info on this subject would be greatly appreciated.

I'm a little confused about this whole thing. I will be using this array simply for my main Apps and OS n stuff, and I will have some other drives for data.

THANKS

[tuskenraider]

It doesn't and is the reason they aren't. SATAII is a theoretical spec that can't be reached by any hardrive's mechanical components by a longshot, so therefore has no performance gain to be had with it in reality over SATAI.

[GLSauron]

SATA is a point to point interface. Until a single drive can hit more than 150MB burst, SATA1/SATA2 is completely irrelevant for transfer speeds. Also, RAID doesn't factor in at all either.

[smoothmove]

today, Bestbuy has the 74 gig Raptors for $120 Retail


Just FYI

[InCogneato]

Thanks guys, so if i understand correctly, there is absolutely no way a consumer hard drive passes the SATA1 maximum? What determines how much the HDD can burst, RPM and Cache? I thought the whole SATA2 speeds were specifically for RAID because I was aware that the HDDs can't reach that level on their own, but I was thinking with RAID0 they might, since they share information.

How about that 15,000 RPM Fujitsu @ Newegg? Would it even come close to passing SATA1? Also, why is everyone always so bent on getting SATA 2? Just because its the latest drive not because it actually matters? Thanks!

[InCogneato]

today, Bestbuy has the 74 gig Raptors for $120 Retail


Just FYIThanks, yes that is why the idea came up in my head, haha.

[mdameron]

Also, why is everyone always so bent on getting SATA 2? Just because its the latest drive not because it actually matters? Thanks!

Well it's a newer tech and is backwards compatible... for the same price. Also, there are more SATA 2 drives available on newegg the last I checked.

There aren't any real world differences to be had between SATA 1 and 2... and hell, PATA for that matter. But with PATA you deal with shitty cables, and have to mess with master and slave BS.

I was using 2 WD 80gb SATA 2 drives in RAID 0 on a SATA 1 board. They were operating in SATA 1, of course. Switched platforms to SATA 2 with the same RAID 0 and stripe size, and scores in HD Tach went up around 40%. This is a benchmark though... notice no difference in real world.

[InCogneato]

Well it's a newer tech and is backwards compatible... for the same price. Also, there are more SATA 2 drives available on newegg the last I checked.

There aren't any real world differences to be had between SATA 1 and 2... and hell, PATA for that matter. But with PATA you deal with shitty cables, and have to mess with master and slave BS.

I was using 2 WD 80gb SATA 2 drives in RAID 0 on a SATA 1 board. They were operating in SATA 1, of course. Switched platforms to SATA 2 with the same RAID 0 and stripe size, and scores in HD Tach went up around 40%. This is a benchmark though... notice no difference in real world.
Interesting! Thank you.

[tuskenraider]

Switched platforms to SATA 2 with the same RAID 0 and stripe size, and scores in HD Tach went up around 40%. While I would expect performance differences on different mobo's, you're saying 40% in average read, access times and burst speed? I'd be interested in seeing the numbers.

[mdameron]

While I would expect performance differences on different mobo's, you're saying 40% in average read, access times and burst speed? I'd be interested in seeing the numbers.

In HDTach, I saw the ~40% increase in those little bars at the bottom left of the screen, doing the "long" benchmark.


Shortly after setting up my first RAID (and SATA for that matter), I ditched my 478 intel rig for the AMD rig in my sig. That's when I moved from SATA I to SATA II on the motherboard. I had SATA II HD's on both rigs.

I'll see if i can find pics from threads I started asking if it was running correctly. They were imageshack pics, so they may be deleted from their server by now. Do not remember the numbers exactly. I think I was in the 50's with the single PATA drive, 90's with the 2 SATA II drives in RAID 0 on the SATA 1 board, and then in the 130's on the current setup.

Remember, I saw that increase after changing mobos (thus SATA/RAID controllers), and CPU's (P4 3.0 w/ HT to 3000+).

I'll start digging. There is no real world increase though that I can tell. I've only ever had a single PATA drive and then RAID 0.

[mdameron]

Current setup. SATA II drives on SATA II board. 64k stripe size. Folding at 100%, IE with like 20 tabs open.

http://img292.imageshack.us/img292/721/hdtachji2.th.png (http://img292.imageshack.us/my.php?image=hdtachji2.png)

I FOUND IT!!!! Turns out it's not same stripe size.

Ok, same SATA II drives, 16k stripe size. SATA I board. Abit AI7.

http://img232.imageshack.us/img232/8045/hdtachvw8.png

Wish they were the same stripe size though.

Similar average read and access, huge difference in burst though. Unless my math's off, ~68%

[WhiteZero]

Probably just the difference between the controllers. Since SATA I maxes out at 192MBps (1.5Gbps) and SATA II maxes out at 384MBps (3.0Gbps).
So unless you were topping out at 192 before and breaking that now, it's no improvment from the jump from SATA I to II.

But of course that limitation is only per drive, once the data exits the SATA bus and onto the RAID controller it can go higher.

[mdameron]

Probably just the difference between the controllers. Since SATA I maxes out at 192MBps (1.5Gbps) and SATA II maxes out at 384MBps (3.0Gbps).
So unless you were topping out at 192 before and breaking that now, it's no improvment from the jump from SATA I to II.

But of course that limitation is only per drive, once the data exits the SATA bus and onto the RAID controller it can go higher.

Pretty much my thinking. Wish I could have eliminated some variables in the process. I'd like a real SATA 1 -> 2 comparison.

[WhiteZero]

Pretty much my thinking. Wish I could have eliminated some variables in the process. I'd like a real SATA 1 -> 2 comparison.

Well, when it comes to single drives, I'm pretty sure even a single Raptor 150 series wont break 77MBps and thats pretty far off from SATA I's max.

I really dont think you would have seen more than a meg or 2 difference between the 2 interfaces.
*shrug*

Heres a pretty awesome lab with 1 to 4 Raptor 74's and 150's in raid 0:
http://www.gamepc.com/labs/view_content.asp?id=raptor150raid&page=5

[mdameron]

Well, when it comes to single drives, I'm pretty sure even a single Raptor 150 series wont break 77MBps and thats pretty far off from SATA I's max.

I really dont think you would have seen more than a meg or 2 difference between the 2 interfaces.
*shrug*

Heres a pretty awesome lab with 1 to 4 Raptor 74's and 150's in raid 0:
http://www.gamepc.com/labs/view_content.asp?id=raptor150raid&page=5

Jesus look at those burst rates.

I wonder what real world application depends on burst the most? And then the others? In gaming with loading big maps and such, which "stat" is the best?

[WhiteZero]

Jesus look at those burst rates.

I wonder what real world application depends on burst the most? And then the others? In gaming with loading big maps and such, which "stat" is the best?
Well, if you read the comments, the burst rates only really represent the RAID controllers speed. And their using like a $500 raid controller.
lol

And the best stat for loading large maps, I'd think would be Sustained. I believe "burst" are pretty small file moves.


EDIT:
Also, inspect thoes screenshot a bit closer. Your burst speed may have shot up (due to your RAID controller change out), but your sequential, average read and seek time are all pretty much the same.

[unhappy_mage]

Probably just the difference between the controllers. Since SATA I maxes out at 192MBps (1.5Gbps) and SATA II maxes out at 384MBps (3.0Gbps).
"Sata 1" at 1.5gbps also has 8/10b coding. So it really is 150 MB/s, not 187.5 (where'd 192 come from?!). Similarly, 3.0gbps with 8/10b coding is 300 MB/s. Just thought I'd point that out.

All that's likely to use the burst transfer rate is a poorly designed app. Reads should happen once, not many times on the same data, and writing the data and then reading it without flushing to disk is usually a sign of poor design.

[WhiteZero]

"Sata 1" at 1.5gbps also has 8/10b coding. So it really is 150 MB/s, not 187.5 (where'd 192 come from?!). Similarly, 3.0gbps with 8/10b coding is 300 MB/s. Just thought I'd point that out.

All that's likely to use the burst transfer rate is a poorly designed app. Reads should happen once, not many times on the same data, and writing the data and then reading it without flushing to disk is usually a sign of poor design.

Where did I get 192 from? lol
1.5Gbp x 1024 = 1536Kbps / 8 = 192KBps

I guess I got it from math...

[unhappy_mage]

gbps (gigabits per second) is usually a base-ten thing. Gigabit network, for example, is 10**9 bits/sec, not 2**30. But I agree, if you do the math like that that's what you get.

I wonder if sata 3.0 is really (nominally) 3*10**9 bits/sec or if it's 3*2**30. That'd be 3e9/10/1024/1024 = 286 megabytes per second versus 3*2**30/10/1024/1024 = 307.2 megabytes per second. In the real world, though, there's probably enough variance in clock rate that it's slightly off anyways.

[drizzt81]

gbps (gigabits per second) is usually a base-ten thing. Gigabit network, for example, is 10**9 bits/sec, not 2**30. But I agree, if you do the math like that that's what you get.

I wonder if sata 3.0 is really (nominally) 3*10**9 bits/sec or if it's 3*2**30. That'd be 3e9/10/1024/1024 = 286 megabytes per second versus 3*2**30/10/1024/1024 = 307.2 megabytes per second. In the real world, though, there's probably enough variance in clock rate that it's slightly off anyways.
SATA 3Gbps is signaling speed, they use a 8/10 encoding from what I recall, that is why 3Gbps -> 100 MiB/s