Sign in to follow this  
Brian Tao

Zoned I/O benchmark figures?

Recommended Posts

Are there any review sites that break down their drive I/O benchmarks by zone (i.e., position of the read/write head)? For instance, how does the outer 10% of a drive compare to the inner 10% of the same drive in terms of read/write throughput, access/seek times, concurrent I/O's, etc. STR is usually the only one that is published as a function of head position.

For any given benchmark, would the first, say, 30 GB on a Raptor 150 (20% of the drive) be faster or slower than the first 30 GB on a 750GB 7200.10 (4% of the drive)? Does the areal density advantage of the 7200.10 make up for the Raptors higher spindle speed? I have not been able to find any benchmarks that break down I/O numbers by zone... I'm guessing they all assume the use of one partition that spans the entire drive?

Share this post


Link to post
Share on other sites

SR's real world benchmark tests use the first 40 GB or so of every drive. It also measures STR at the beginning and end of the drive, and links to a graph showing how STR varies throughout the drive's capacity.

But in practice, a similar rule applies to all real world benchmarks.

If you take a drive, set up one partition, install OS, apps, games, benchmark software, and run your tests... all your data will be at the beginning of the drive, it won't have had the chance to become fragmented or spread halfway across the disk. It's got to hear that a site defrags before every benchmark run, but really, the advantages of large capacity that you're interested in are automatically taken into account (with a greater or lesser degree of scientific accuracy).

Share this post


Link to post
Share on other sites

This "short-stroking" effect of capacity does give bigger HDDs a benefit. Also, latest 7200rpm HDDs have higher density platters. But...

Seagate 7200.10 750GB is not a high performance drive. In fact it simply sucks. Here's an objective comparison. Take that Office DriveMark for example (bigger the better):

Raptor 150: 1010

Raptor X: 990

7K500: 877

T7K500: 870

WD5000YS: 800

ES 750GB: 575

I don't even understand how they can rate that low. Maybe it's related to Seagate's fixation to provide better results for server benchmarks... but I don't understand the reason for that. Maybe because majority of reviewers only perform server benchmarks when they review desktop drives. Well, maybe. And it seems to work, because one time after another Seagate drives get recommended because of their high STR, low access time, etc. low level characteristics. While STR limits some big file-copying scenarios and access time (= seek time + latency) + NCQ defines server performance, caching plays an important role in general purpose, office, gaming, etc. machines. And it's not just cache memory but algorithms as well. Stating a minimum requirement for cache is not uncommon when people ask for advice buying a drive. While it is nice to notice they acknowledge the importance of cache, it's also sad to realize they don't see other factors that make up "caching" other than cache size.

So, if you really want a 7200rpm drive that is close to Raptor, I'd choose a Hitachi. Relatively new T7K500 and de g'olde (E)7K500 are drives to use. For a short while (before recent generation of Raptor) 7K500 was better than Raptor, and also beating vast majority of SCSI drive in both 10krpm and 15krpm categories. (Fujitsu SCSI drives are an exception because they have higher desktop performance than the rest of SCSI attached HDDs.) Sure, T7K500 and (E)7K500 can't beat this newest Raptor. 7K1000 might. My guess is, it's going to rank pretty close but may fall short a bit. But who knows, everything is possible. Hitachi has managed to out-perform others with superior firmware and low cache size. For the first time, Hitachi is actually pushing the cache size up (to 32MB) instead of responding to competitors products.

7K1000 is also available as 750GB variant if you don't need all the capacity. 32MB cache on them as well, unlike 7200.10, but the main thing is, you'll probably get a vastly superior firmware to handle that bigger cache. I say "probably" because it has not been properly reviewed yet. By proper I mean SR, iXBT or something similar, not AnandTech, Tom's Hardware and other mainstream reviewers.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this