Information on RAID 0 and Serial ATA

By Merle Nicholson, Tampa PC Users Group
merle@merlenicholson.com

For the May meeting, Kevan Sheridan gave an excellent and fascinating talk from Intel material on Pentium 4 Hyper-Threading. Several points of discussion were raised, and during that I realized that I needed to do some research.

One interesting slide was given showing an optimal P-4 system using two Serial ATA RAID 0 drives that suggested that if you had, say, two 80G drives, you would have a total of 160G. And that is the case. I looked it up at an excellent site at Advanced Computer & Network Corporation http://www.acnc.com/04_01_00.html.

RAID 0 uses a minimum of two drives and is used to maximize throughput:

Characteristics/Advantages
  • RAID 0 implements a striped disk array, the data is broken down into blocks, and each block is written to a separate disk drive.
  • I/O performance is greatly improved by spreading the I/O load across many channels and drives.
  • Best performance is achieved when data is striped across multiple controllers with only one drive per controller.

Disadvantages
  • Not a "True" RAID because it is NOT fault-tolerant
  • The failure of just one drive will result in all data in an array being lost
  • Should never be used in mission critical environments

This is interesting to me, since my own personal computer has one SATA drive and one unused channel, with RAID 0 built onto the motherboard.

Serial ATA is a standard newer than the usual IDE (ATA and Ultra ATA) drive standard. SATA has become very common, with RAID SATA support built onto the motherboard at a very small cost premium.

Over the past couple of years we’ve seen the “burst rate” rating for hard drives go from 33MB/sec to 66, 100 and finally 133. I tried to find some direct speed comparisons – Parallel ATA to Serial ATA, but gave up on it due to conflicting terminology. The Seagate website claimed only a 1 to 5% improvement in throughput, but also cited the smaller cable used improving air flow (see picture), lower production costs due to the easier-handling connectors, and lower power consumption.

Processors

Kevan mentioned a big discrepancy in megahertz ratings of processors and noted that you just can’t trust a megahertz number any more, and that’s absolutely correct. An Intel P-4 3000 runs at 3000MHz, and an AMD XP 3000+ runs at 2167MHz. Looking at the specifications, the most glaring difference is that the AMD processes nine instructions per cycle to the Intel’s four.

Also there’s a difference in the number of integer registers and floating point registers. AMD does offer benchmark comparisons for both AMD and Intel.

Documentation is offered on the site that explains the benchmarks. One interesting thing is that the numbers suggest that for some benchmarks, the P-4 may have better performance with Hyper-Threading turned off. u

Comparison of serial and parallel cables.  The parallel cable is the larger one