Configuration and Benchmarks

This content was originally featured on and has been converted to PC Perspective’s website. Some color changes and flaws may appear.

The Configuration:

  • Tyan K7 Thunder (revision “A”, BIOS 2.06) ENPC
  • 2-1.2GHz AMD AthlonMP Processors Monarch Computer
  • 2-ElanVital FSCUG9C-6FC coolers ElanVital
  • 256MB Samsung PC2100 ECC Memory Monarch Computer
  • AMI Elite 1600 RAID Controller (A159 firmware, driver 5.21) AMI
  • 4-18GB Seagate X15 II Hard Drives (ST318452LW) Hyper Microsystems
  • Windows2000 (service pack 2)
  • NTFS partition
  • Striped (RAID 0 performance)
  • Write-Through Cache Enabled
  • 64KB Block Size
Benchmarks Used:
  • Intel’s Iometer (1998.10.08) download
  • Atto’s AttoMark 1.63 download
  • WinBench99 v1.2
    • Business Disk WinMark 99
    • High-End Disk WinMark 99
    • download
  • CliBench MK III SMP (0.7.10, 20.05.2000)


The configuration was comprehensive however uncomplicated. In fact, the Elite’s BIOS provides the user many performance options in which to configure the array. Briefly, one needs to format the hard drives, and then choose the hard drives in which to build the array. Once the array is chosen, you are provided a myriad of performance options. Once the performance options are chosen, you initialize the hard drives.

Write-Back Cache: Many modern disk controllers have several MB of cache on board. Onboard cache gives the controller greater freedom in scheduling reads and writes to disks attached to the controller. Usually, the controller lets a System Administrator determine how the cache is used. In write-back mode, the controller reports a write operation as complete as soon as the data is in the cache. This sequence improves write performance at the expense of reliability. Power failures or system crashes can result in lost data in the cache, possibly corrupting the file system.

Write-Through Cache: The opposite of write-back. When running in a write-through mode, the controller will not report a write as complete until it is written to the disk drives. This sequence reduces read/write performance by forcing the controller to suspend an operation while it satisfies the write request.

Comparing the two policies, in general terms write-back provides better performance, but at the slight risk of memory integrity. Write-back caching saves the system from performing many unnecessary write cycles to the system RAM, which can lead to noticeably faster execution. However, when write-back caching is used, writes to cached memory locations are only placed in cache, and the RAM itself isn’t actually updated until the cache line is booted out to make room for another address to use it.

The only setback that I came across was setting the controller to 32KB stripe size and enabling Write-Back Cache. Traditionally, by enabling these two features, would provide the best performance. After running a couple of cycles of Iometer and AttoMark, I immediately knew there was a problem. The maximum sequential reads and writes were at 45MB/sec. After working with the white-coats at AMI, I was notified that the Elite 1600 was optimized for Write-Through Cache and 64KB stripe size. I then reformatted the hard drives, built the array and initialized the hard drives. The results were enormous. My write speed where upward to 155MB/sec and my read speeds where upward to 175MB/sec. These are the highest results I have ever recorded. So high in fact they unseated the reining champion, Atto’s UL3D.

Power Console Plus:

Power Console Plus is software the provides on the fly RAID migration, creating almost limitless adaptability and expansion of any logical drive while the system remains operational. Power Console Plus is an object-oriented GUI utility that configures and monitors RAID systems locally or over a network with several servers. Power Console Plus can be executed from any workstation. Power Console Plus allows you to control and monitor the status of hard disk drives, tape drives, and CD-ROM drives. Power Console Plus runs under Microsoft Windows2000. Use Power Console Plus to: choose a configuration method, designate drives as hot spares, create physical arrays, define and initialize logical drives, and rebuild failed disk drives. While this utility allows the user to change cache settings, it does not allow the addition or manipulation of RAID arrays without erasing the existing data.

Benchmark Outcome:

With the exception of proprietary benchmarks, Intel’s Iometer, ZD’s WinBench99 v1.2 and Atto’s ATTOMark v2.02 have predominately been the foundation for benchmarking storage performance. While utilities such as HD Tach 2.61, Performance Test 3.4, and SiSoft Sandra’s Disk Test do provide multipurpose utilities to measure disk performance, they do not adequately represent multiple SCSI devices or RAID performance. Often, the results are distorted and misrepresent the actual performance.

« PreviousNext »