HD: SATAII 7200 vs. SATA 10,000?

[SunziBingfa]

I'm currently working on building a computer gaming rig, and doing it right this time with extensive research and planning.

One question I can't find an answer to concerns hard drives:

SATAII is faster than SATA in terms of data transfer. A 10k rpm is faster than a 7200 rpm in terms of quickly accessing data, but it also seems like this would have an impact on data read/write times.

I'm thinking higher cache size is superior and that NCQ is ideal.

I'm confused when it comes to making comparisons between HDs when looking SATA/SATAII and RPMs.

How can I decide if a SATAII 7200 is better or worse than a SATA 10000 in terms of system performance (read/write/load times)??

I have checked into some reviews and looked at a few HD related forums without coming across answers to what strikes me as a crucial question.

I also take it that the SCSI interface drives will lag behind SATA in performance. Please let me know if any of my thinking here is in error.

 

[Jiggz]

You're basically in the "ballpark" in your understanding of SATA. But there is a catch wherein actual performance cannot accurately be based on RPM, Interface (SATA/SATA II) or cache. Although SATA has twice the available bandwidth of SATA I, presently this is insignificant because exisiting hdd cannot even saturate ATA-133 worse yet SATA I which has 150mb/s! Although for the future, an SATA II will be nice.

As for the RPM, this has a direct impact on the data transfer and access rate of the hdd. The higher RPM the faster is the data transfer rate and also the access when compared to a hdd of the same density of different RPM. For example, a 7200 RPM, 200GB HDD will be faster then a 5400 RPM 200GB hdd both in transfer rate and access. But a 7200 RPM, 80GB HDD may not necessarily be faster than a 5400 RPM 250GB HDD because of the difference in platter density. Although the 7200RPM hdd may have a faster access time. As for the cache, the higher the better although it's only a slight improvement. Meaning comparing 8MB cache with 16MB, obviously the higher cache will perform better.

Comparing SCSI with SATA is a little harder because there are so many versions of SCSI. But the best SCSI interface will outperform the best SATA, at present. But it cost a lot more.

Presently in the SATA world, the best hdd in a single configuration is WD's Raptor 150.

 

[fire400]

if you get an NCQ drive from Seagate, you're pretty much gravy with the speedier bunch of enthusiasts who enjoy cheap priced hard drives.

 

[bob4432]

ncq is really not going to matter much for your setup.

jiggz is pretty much right on.

don't let the interface fool you as even a single, current gen 15k u320 scsi hdd will not go over ~90MB/s str. current gen 7.2krpm pata/sata will be ~60MB/s str

 

[Bobthelost]

NCQ is bad, i have no idea what fire400 is on about as every bench i've seen that NCQ hurts performance, also since the seagates aren't the fastest line of 7200rpm drives (both the hitachi deskstar and the WD caviar ranges are faster) i think he's talking crap here.

The interface is rather irrelevant, if you want some good reviews of drives then check out this place: http://www.storagereview.com/comparison.html

Jiggz has it right.

 

[SunziBingfa]

Thanks for your responses!

I think I have a better grasp of the distinction now. Besides Bob's link to Storage Review, this link:

http://www23.tomshardware.com/storage.h...delx=33&model1=117&model2=124&chart=32

also should be very helpful in determining which drive to select.

 

[Bobthelost]

I don't trust toms all that far, their reviews vary from fantastic to utter bilge. Not sure which is applicable here, but all of thier benches are artifical or low level ones, and if you compare them to the ones at storage review you can see that it doesn't always work out that the best speced ones are the best performing.

 

[GOREGRINDER]

just get a raptor and be happy:laugh:

 

[t3h l337 n3wb]

Originally posted by: GOREGRINDER
just get a raptor and be happy:laugh:

Raptors are a huge waste of money. Newer SATA drives with 16MB of cache and denser platters are almost as fast as a 74GB Raptor. The 150GB costs about $2 per GB, while you can find other large drives for about $0.35 per GB. Do you want to pay nearly 6 times the amount per GB just for a little boost in load times?

 

[markkleb]

I have used raptor drives and to tell you the truth the 16 mb buffer 7200 rpm drives seem faster.

I sold my raptors and replaced them with 2 16 mb buffer drives and am much happier (cooler too)

 

[GOREGRINDER]

Originally posted by: t3h l337 n3wb

Originally posted by: GOREGRINDER
just get a raptor and be happy:laugh:

Raptors are a huge waste of money. Newer SATA drives with 16MB of cache and denser platters are almost as fast as a 74GB Raptor. The 150GB costs about $2 per GB, while you can find other large drives for about $0.35 per GB. Do you want to pay nearly 6 times the amount per GB just for a little boost in load times?

well for one look at my sig and that answers yer question,..lol and second you get more than load time swiftness,.. in game checkpoints merley flicker if at all,boot times are significantly dropped,accessing the page file isnt even acknowledgable to any gamer and matenence such as formats and re-instals are mere minutes,defragging is done in no time at all and any software scans etc etc etc the list goes on,..and in the end you do get what you pay for as long as the class of hdd matches the class of the rest of the system imo

(for the 36.7gb raptors there are in fact 2 dif models, first editions were quite clunky and didnt live up,updated versions are quite different,i believe middle last year they started updating them if im not mistaken) as a matter a fact sandra sees my getup 4mb/sec faster than my own comparison :Q..lol

 

[DBSX]

I agree, I don't feel the Raptor is "all that" any more. many of the newer drives perform on par. The little faster the Raptor may be doesn't justify the cost. Having said that, I have a Raptor (got it from a friend for cheap) and like it, but I don't really notice that much difference in my overall computing using the Raptor as my boot drive.

Now, to get back on topic, I also agree with Jiggz. Though again I am not sure that the Raptor is really worth the money. It also depends on what you want (space vs speed).

\Dan

 

[SunziBingfa]

Your comments and advice were very helpful!

For me gaming performance, as well as general load/read/write/ times are more important than space considerations. While the Caviar 300GB is much cheaper per GB of space, it looks like the 73GB and 150GB Raptors have its number when it comes to speed considerations.

I'm a little wary about spending ~$300 on a hard drive. On the flip side, the 150GB Raptor does perform better than the 73GB version... decisions decisions...

 

[imperfectcircle25]

The Raptor is currently the fastest drive you can get without going to a 15k SCSI drive
.

 

[Velk]

Originally posted by: GOREGRINDER
well for one look at my sig and that answers yer question,..lol and second you get more than load time swiftness,.. in game checkpoints merley flicker if at all

Subjective, but mostly meaningless without reference.

,boot times are significantly dropped,

True.

accessing the page file isnt even acknowledgable to any gamer

If you are saying that the raptor is so fast that page file access isn't noticeable, that's ridiculous. If you are stating that there's no difference in page file access, but gamers don't care about page file access that's mostly true.

and matenence such as formats and re-instals are mere minutes,defragging is done in no time at all and any software scans

True, but misleading. Most of the advantage in formats, scans and defrags is because the drive is a fraction the size of normal mainstream drives, rather than any speed bonus. I.e. all things being equal, a 36GB drive will take less than 1/10th as much time to defrag as a 400GB drive.

 

[GOREGRINDER]

Originally posted by: imperfectcircle25
The Raptor is currently the fastest drive you can get without going to a 15k SCSI drive
.

thats the bottom line,.. jiggz and bob4432 also hit the nail on the head,..hell, i agree with jiggz right down to his sig quote :laugh:

dont listen to some of the joker's around here,.. alot of em like to discourage what they dont have or dont know,.. youll find that out soon enough sunzi

 

[SunziBingfa]

OK, I've clarified my thinking on this. My ambivalence between the 150 and 73GB Raptor was due to my toying with the idea of getting two 73GB models and they setting them in RAID 1 for easy file backup (no more burning CDs for this purpose).

But I'd rather not pass up the performance benefit of the 150GB model. And I can always buy a super cheap and not so terribly fast hard drive as a backup device.

So, the 150GB Raptor it is!! ... unless something even better comes down in price before I finalize everything and make all the purchases...

On to the CPU research!

 

[BDawg]

I was looking at StorageReview's listing. They have NCQ performing slower than non-NCQ. From what I remember reading about NCQ, it takes a dual core / processor to really shine. Is this not true?

 

[jdkick]

Personally, i've never thought the Raptor's were worth the money... especially now. That might be different if squeezing every last drop of performance from my machine kept me awake at night, but it doesn't. I just don't find that the performance difference sufficient enough to warrant the expense (150GB Raptor vs 160GB 3.0Gb/s SATA represents nearly a 400% price jump). I'd rather go with SATA and put the $$$ towards something else... and not necessarily towards the computer.

 

[Bobthelost]

Originally posted by: BDawg
I was looking at StorageReview's listing. They have NCQ performing slower than non-NCQ. From what I remember reading about NCQ, it takes a dual core / processor to really shine. Is this not true?

As far as i know that's not true, NCQ is a system to speed up data access from hard drives, the thing is that it only really helps randomised access paterns, which only occur in servers. (Heavy mulittasking is not similar btw).

The CPU doesn't make any difference at all.

 

[BDawg]

Originally posted by: Bobthelost

The CPU doesn't make any difference at all.

From the following Anandtech article.

The MaXLine III performs just as well as any of the fastest desktop hard drives available today, but when used with an NCQ-enabling controller, the performance potential is improved tremendously. Although we could only show it in one of our three multitasking tests, NCQ can have some pretty serious performance implications for those users who are running a lot of applications simultaneously.

The benefits to drive-based command reordering are easy to see on paper, but the fact that we were able to reproduce those benefits in a real world benchmark speaks volumes for the technology. As usage patterns become increasingly multithreaded/multitasking oriented, the performance impact of NCQ will improve even further.

 

[Bobthelost]

Originally posted by: BDawg

Originally posted by: Bobthelost

The CPU doesn't make any difference at all.

From the following Anandtech article.

The MaXLine III performs just as well as any of the fastest desktop hard drives available today, but when used with an NCQ-enabling controller, the performance potential is improved tremendously. Although we could only show it in one of our three multitasking tests, NCQ can have some pretty serious performance implications for those users who are running a lot of applications simultaneously.

The benefits to drive-based command reordering are easy to see on paper, but the fact that we were able to reproduce those benefits in a real world benchmark speaks volumes for the technology. As usage patterns become increasingly multithreaded/multitasking oriented, the performance impact of NCQ will improve even further.

http://www.storagereview.com/php/benchm...ves=1&devID_0=269&devID_1=265&devCnt=2

Maxtor is a wierd one, the performance is crap/mediocre across the boards, but almost unchanged in single user performance with/without NCQ. Either way it's not dependant on the CPU you use. You can multitask with a single core CPU, as the article you quote says!

Intel Pentium 4 Extreme Edition 3.4GHz

 

[Matthias99]

Originally posted by: BDawg

Originally posted by: Bobthelost

The CPU doesn't make any difference at all.

From the following Anandtech article.

The MaXLine III performs just as well as any of the fastest desktop hard drives available today, but when used with an NCQ-enabling controller, the performance potential is improved tremendously. Although we could only show it in one of our three multitasking tests, NCQ can have some pretty serious performance implications for those users who are running a lot of applications simultaneously.

The benefits to drive-based command reordering are easy to see on paper, but the fact that we were able to reproduce those benefits in a real world benchmark speaks volumes for the technology. As usage patterns become increasingly multithreaded/multitasking oriented, the performance impact of NCQ will improve even further.

Yes and no.

Any kind of command reordering can theoretically help in situations where you have multiple asynchronous I/Os going at once to the drive and the I/Os are not all linear reads/writes in the same area. NCQ is usually somewhat more effective than TCQ or OS-level reordering, since the OS doesn't know exactly where the data is on disk and the drive's onboard logic does. However, like all predictive algorithms, it can sometimes not be effective, or make bad decisions that make things worse.

With multitasking of different disk-heavy programs (or multithreaded tasks that access files all over the disk), or 'server-like' I/O patterns where you are accessing different pieces of data all over the disk, you'll see the most benefit. It's hard to multitask that much on a 'desktop' system without multiple CPUs (or at least a fairly fast single CPU), and 'normal' games and applications don't usually work like this. But if you have a lot of CPU power (especially with multiple CPUs), and you run multiple disk-heavy apps at once, you could see a significant benefit. If you're not doing that, it will either do nothing or slow you down a bit due to the drive trying to reorder commands when it doesn't really help, or accidentally reordering them such that a critical task has to wait longer.

Even within the 'multitasking' tests in that article, one was helped by ~10% (File Copy + IE/Outlook), one was about the same (Winzip + Excel/Word), and one was ~2% slower (NAV + Winzip/Office). Clearly it is not a panacea.

 

[BDawg]

That article did show that at the time, the MaXLine III performed as well as the fastest hard drive. They claimed that performance should increase with increases in multithreaded/multitasked applications. It wouldn't be too much of a stretch to say that in real-world testing that it would improve with multiprocessors also.

 

[Bobthelost]

Originally posted by: BDawg
That article did show that at the time, the MaXLine III performed as well as the fastest hard drive. They claimed that performance should increase with increases in multithreaded/multitasked applications. It wouldn't be too much of a stretch to say that in real-world testing that it would improve with multiprocessors also.

Because the data access paterns would become more random with a faster CPU? They wouldn't, programs or files are generally stored in the same physical place on the HD, so the time spent seeking from location to location is minimised, as such the locations that the HD will be accessing won't vary depending on how fast the CPU processes the data. But on where the data is stored itself. NCQ works by altering which areas are read in which order. There would be some increase i'm sure, but not all that much.

Now if you were talking about a server then you'd be right, it'd be able to handle more requests for information from more users per second. As such the access paterns would become more random, meaning NCQ would be of more help.