Support at Alden Hosting wrote:
The ataraid-list members might be interested in some basic performance testing I did between the Promise's FastTrak driver and the atarad/pdcraid drivers that are suppose to replace the FastTrak.o driver.
Hardware: Intel SCB2 motherboard Promise 20267 onboard chip Dual PIII 1.13 Gig Hz CPUs 1 Gig RAM Dual Seagate 80 Gig drives (RAID 1 Mirror)
A disk Stress/Performance Test was executed with two versions of the Linux kernel (2.4.18-3smp and 2.4.20-0smp) and the Promise FastTrak vs. ataraid/pdcraid drivers. The test script created three 2 gigabyte disk files in parallel.
=================== TEST script =============================== date echo this will create about 2gb file in ~myhome named largefile1 `dd if=/dev/zero of=largefile1 bs=16384 count=131072 &` echo this will create about 2gb file in ~myhome named largefile2 `dd if=/dev/zero of=largefile2 bs=16384 count=131072 &` echo this will create about 2gb file in ~myhome named largefile3 `dd if=/dev/zero of=largefile3 bs=16384 count=131072 ` date
============= RESULTS ==================================== Linux 2.4.18-3smp (Red Hat 7.3) FastTrack.o (v1.02.0.22) 5-6 minutes (Promise's Pre-Compiled module)
Linux 2.4.20-0smp (Custom Kernel) FastTrack.o (v1.02.0.21) 5-6 minutes (Promise's Partial Open Source Code. Compiled with custom kernel)
ataraid/pdcraid 30 minutes (Test stopped half way done. Only created three 1 gig files)
The Promise FastTrak driver was at least 10 times faster then the GPL ataraid/pdcraid driver. The above testes were execute three times each. Each test result was the same.
...... Note ..... I could not get the Promise's Partial Open Source Code V 1.0.2.0.25 to boot. It compiled OK, but during boot up it always dumped the Stack/Registers in their bInitRaid routine.
The Promise's Partial Open Source Code V 1.0.2.0.21 compiled and booted with no problems.
It would be interesting to see the difference between the driver and Promise's driver.
-- There is no such thing as obsolete hardware. Merely hardware that other people don't want. (The Second Rule of Hardware Acquisition) Sam Flory <sflory@xxxxxxxxxxxx>
