Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
Prices: Mainboards ABIT ASUS Chaintech Shuttle Soyo Tyan CPU Intel P4 2.4C-800 P4 2.6C-800 P4 2.8C-800 P4 3.0-800 P4 3.2-800 AMD AthlonXP XP 1700+ XP 2000+ XP 2400+ XP 2500+ XP 2700+ XP 3000+ XP 3200+ Athlon64 Athlon64 3200+ Athlon64 FX-51 Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Memory Corsair Crucial Kingston Mushkin OCZ Search Prices:

Seektime

As we explained in our second part of "As The HDD Spins", "Seek paradigms" can be defined in a number of different ways on the HDD level, based on either precision and the lack thereof or else a "Don't Care" algoritm. On the level of benchmarks, seek execution can either be totally random, which does not make sense in the context of desktop drives or else, using defined parameters such as the track-to-track seek time or the 1/3 stroke which are the most representative of a drive's seek performance - provided that they are really executed.

Seek performance [in milliseconds] for both Seagate Barracuda 7200.7 and Maxtor DiamondMax Plus9 (lower is better). Whether the speed difference between the two drives is real or not is questionable but the results were consistent. Seek times of 33 vs 34 µsec do not sound like anything was really executed, though. For more background information on seeks, check these explanations

Average Random Transfer Rate

Probably by far the most important "synthetic test" for any real world performance is the "Average Random Transfer Rate" of a drive, not to be mixed up with the Average Sequential Transfer. The Average Random Transfer Rate reflects a combination of random accesses and I/O performance which is about the closest approximation of any real world situation I can think of. For the test pattern, the block length was set to 128 blocks (64 KBytes).

      

Average Random Read Transfer Rates for the DiamondMax Plus9 and the Barracuda 7200.7. In the case of the DiamondMax Plus9 periodically, the high transfer rate shot up to roughly 100 MB/sec, which compresses the scale at the bottom of the graph. This happened irrespective of whether MaxBoost was enabled or not. Keep in mind that the "Max" transfer rate is not a sustained value but a transfer rate that was "accidentally hit somewhere along the way" for a single transfer.

The SCSI Toolbox graph of the DiamondMax is a bit hard to read, therefore, we have plotted the the actual random transfer rates for the DiamondMax Plus 9 with and without MaxBoost below.

Random average transfer rates in kBytes/sec: As in some instances before, we see a performance hit when MaxBoost is enabled. The Barracuda 7200.7 takes the lead here.

For those interested more in I/O performance, we calculated the average random I/Os per second from the data above

next page:    => Reliability - Conclusion =>