Seagate Barracuda 7200.8: 400GBs with NCQ
by Purav Sanghani on April 20, 2005 4:30 PM EST- Posted in
- Storage
The RPM Factor
RPM, or revolutions per minute, is the measure of instances that the motor of the hard drive can rotate the platters by a full 360 degrees. Currently, there are various drives from a few different manufacturers that can rotate their platters 10,000 to 15,000 times per minute, or 15,000RPM. The most common drives today are rated at 7200RPM like our 400GB 7200.8 Seagate Barracuda, and there are still many 5400RPM drives around also.So, does the speed of a drive's motor really make a difference in the performance of a drive? If we look at just the speed of the motor, then yes, there is a great performance boost from a 7200RPM drive to a 10,000 or 15,000RPM drive. The faster that the motor can rotate the platters, the quicker that the read and write heads can do their job on the platters. But there are other factors that come into play when measuring the performance of a hard disk drive. SATA based drives have a maximum data transfer rate of 150MBs/sec (megabytes per second) while IDE drives top out at 133MBs/sec. The transfer rate can be enough to even things out under certain circumstances. The same goes for the amount of cache on the drive. A 7200RPM drive with 16MB of cache has been proven to compete with a 10,000RPM drive with 8MB of cache, again, in certain situations.
We have taken two of the latest drives from Maxtor that we could get our hands on to compare the differences in performance between the DiamondMax 16 series 160GB drive with a 5400RPM motor, and the DiamondMax Plus 9 series 160GB unit with a 7200RPM motor. Both drives are of the PATA/133 flavor, have 8MB buffers on board, and have a total of two 80GB platters each. With the physical specifications being identical in every aspect, let's take a look at how the two units compare in performance based on their spindle speed.
5400RPM vs 7200RPM Spindle Speed |
|||
DiamondMax Plus 9 (7200RPM) |
DiamondMax
16 (5400RPM) |
7200RPM Performance Advantage |
|
SYSMark 2004 - Internet Content Creation Performance |
|||
| Overall | 195 |
191 |
2.09% |
| 3D Content Creation | 174 |
172 |
1.16% |
| 2D Content Creation | 251 |
244 |
2.87% |
| Web Publication | 170 |
167 |
1.8% |
SYSMark 2004 - Office Productivity - Communication Performance |
|||
| Overall | 153 |
144 |
6.25% |
| Communication | 144 |
122 |
18.03% |
SYSMark 2004 - Overall System Performance |
|||
| Overall Performance | 173 |
166 |
4.22% |
| Internet Content Creation | 195 |
191 |
2.09% |
| Office Productivity | 153 |
144 |
6.25% |
Winstone 2004 - Overall System Performance |
|||
| Business | 25.5 |
25 |
2% |
| Multimedia Content Creation | 31.5 |
31.5 |
0 |
Pure Hard Disk Performance - IPEAK, Winstone 2004 |
|||
| Business | 442 |
383 |
15.4% |
| Multimedia Content Creation | 267 |
238 |
12.18% |
Real World Performance - File System Tasks (seconds) |
|||
| File Zip (1 300MB File) | 61.331 |
74.224 |
21.02% |
| File Zip (300 1MB Files) | 62.811 |
72.594 |
15.58% |
| File UnZip (1 300MB File) | 14.383 |
15.500 |
7.77% |
| File UnZip (300 1MB Files) | 14.857 |
20.021 |
34.76% |
| Copy Folder (1 300MB File) | 5.765 |
8.216 |
42.52% |
| Copy Folder (300 1MB Files) | 8.078 |
11.443 |
41.66% |
Real World Performance - Application Load Times (seconds) |
|||
| Photoshop CS | 8.263 |
9.269 |
12.17% |
| Office 2003 - Word | 1.984 |
3.355 |
69.1% |
| Office 2003 - Excel | 2.323 |
2.979 |
28.24% |
| Office 2003 - Access | 1.662 |
3.816 |
29.6% |
| Office 2003 - PowerPoint | 2.289 |
3.823 |
67% |
Real World Performance - Game Level Loading Times (seconds) |
|||
| Half-Life 2 (d1_canals_01) | 23.867 |
21.2 |
-12.58% |
| Doom 3 (caverns1) | 45.667 |
47.567 |
4.16% |
| C&C: Generals (GLA C3S1)* | 34.300 |
34.867 |
1.65% |
Service Time |
|||
| IPEAK Average Read Service Time | 13.82 |
23.31 |
8.13% |
WinBench 99 - Transfer Rate Test |
|||
| Beginning | 59400 |
47200 |
25.85% |
| End | 33800 |
26800 |
26.12% |
*C&C:Generals playing as GLA (campaign 3, stage 1)
The greatest performance increases were seen with our Real World File System Tasks as well as Application Load Time tests. The 7200RPM unit picked up data off its platters much more quickly than the 5400RPM drive. There is no question that a drive's spindle speed has a great effect on the overall performance of the drive. There were certain situations where the spindle speed made no difference like the game level load times, for example. The 5400RPM drive loaded Half-Life 2's d1_canals_01 map more than 2 seconds quicker on average than the 7200RPM mode. Still, this is not a large enough margin to conclude that a higher RPM does not have a positive impact on a drives overall performance.
Facebook
Twitter
RSS
44 Comments
View All Comments
StormGod - Wednesday, April 20, 2005 - link
Hey Anandtech, please make sure your pages are 100% Firefox compatible! While were on the subject, you should really strive to make your pages HTML 4.01 compliant or XHTML 1.0 compliant.cosmotic - Wednesday, April 20, 2005 - link
I was going to comment on the headings too...SLIM - Wednesday, April 20, 2005 - link
The shading and color fill behind the headings and drive names is also missing in firefox. You can highlight the column headings to read what they are supposed to say in firefox. Glad I downloaded that ieview extension now.bigboxes - Wednesday, April 20, 2005 - link
Yup. The column headers for these tables do not show up in Firefox.shoRunner - Wednesday, April 20, 2005 - link
the column labels don't show up in firefox.shoRunner - Wednesday, April 20, 2005 - link
PuravSanghani - Wednesday, April 20, 2005 - link
segagenesis: It seems to be an issue with our sound meter or noise reduction process. We will look into it for our next review. Besides the echo, the recordings should be clear enough to differentiate how each drive sounds.Nighteye2: Your requests will be fulfilled soon. :)
Nighteye2 - Wednesday, April 20, 2005 - link
You know, with all this talk about NCQ, 1 question has not yet been answered: how does it work with RAID? Can you use NCQ on a RAID system?Also, I'd like to see these tests run on a RAID system, see the performance advantage it gives. Maybe compare 2 cheap, somewhat slower drives in a RAID array against a single HD that you can get for the same price?
segagenesis - Wednesday, April 20, 2005 - link
The benchmarks are hardly cut and dry yes, but I do enjoy the fact Seagate has a 5 year warranty on drives. This after seeing the industry at one point was putting out 1-year warranty stock on drives and if you paid extra, 3 years.Raptors are the fastest drives ive ever seen but the lack of space keeps them from being all inclusive. I was kind of suprised that the 7200.8 beat out the Raptor as far as game loading went!
Whats with the weird echo-ish sound recordings of the hard drive noise? What on earth did you use to do this?
FreshPrince - Wednesday, April 20, 2005 - link
man, I need to learn how to use this...anyways...I bought 40 of these drives for my company.16 goes into one raid and another 16 goes into another raid. So far so good, I hear no complaints from my tech guys.
Also, I took 2 and used it as a DFS file server, it's handling 75 users no problems. :-)