|
DISCLAIMER: All the information contained in this page, or any linked from it, is provided as is, having no warranty or support of any kind, and is used entirely at your own risk.
IDE
I did not expect to have any problems with the IDE drive, or even performance optimization, as the drive is standard IDE running off a Intel PIIX4, both of which are supported by just about every Linux kernel since the early days.
Now with this one its a lot less complex that it could be. In my case I did no tests at all with 2.2.14 Kernel so cannot comment on the performance of this one, if someone want to do some tests please let me know and I will include them here.
TIP: You can ignore this page if you are using my optimized Kernel or .config file as you already have the optimum settings.
Kernel Configuration
Under Kernels 2.4.2 and later the following settings are needed to provide the best possible performance.
|
Under ATA/IDE/MFM/RLL support
|
|
ATA/IDE/MFM/RLL support
|
Yes
|
|
Under IDE,ATA and ATAPI Block Devices.
|
|
Enhanced IDE/MFM/RLL disk/cdrom/tape/floppy support.
|
Yes
|
|
Include IDE/ATA-2 DISK support.
|
Yes
|
|
Use multi-mode by default.
|
Yes
|
|
Generic PCI IDE Chipset support.
|
Yes
|
|
Sharing PCI IDE interrupts support.
|
Yes
|
|
Generic PCI bus-master DMA support.
|
Yes
|
|
Use PCI DMA by default when available.
|
Yes
|
|
Intel PIIXn chipsets support.
|
Yes
|
|
PIIXn Tuning support.
|
Yes
|
NOTE: Remember I do not have the Sony PCMCIA CDROM, so the settings are not included here. If someone wants to send me the Kernel settings for that I will add it here
Performance Optimization
I did extensive tests with hdparm to determine the best possible performance out of the Hitachi drive in this machine. The results of this are the above Kernel settings, which give the following results averaged from 10 iterations of (with a reboot in between) hdparm -Tt /dev/hda :-
Timing buffer-cache reads: 128.00 MB/sec
Timing buffered disk reads: 12.38 MB/sec
Now for the question, can things be made quicker, from all the testing I did the following hdparm settings are the optimum for the drive as far as I can tell based on testing and the drive data sheet.
hdparm -d1 -m8 -u1 -c1 -A1 /dev/hda
Now the bad/good news, with the Kernel settings as shown above, this hdparm line is not actually needed, it makes no difference to the performance.
30GB Toshiba drive performance
After doing the upgrade I ran the same hdparm -Tt test as I talked about above on the new drive with the same physical Kernel as I used above and also with the 2.2.14 Kernel. Here are the results.
2.4.17 Kernel (with my optimized Kernel)
Timing buffer-cache reads: 139.13 MB/sec
Timing buffered disk reads: 19.69MB/sec
2.4.2 Kernel
Timing buffer-cache reads: 121.90 MB/sec
Timing buffered disk reads: 17.83MB/sec
2.2.14 Kernel
Timing buffer-cache reads: 121.90 MB/sec
Timing buffered disk reads: 19.88MB/sec
NB. All of these performance tests where the averaged results on 10 iterations of "hdparm -Tt /dev/hda" with a reboot in between each test.
30GB IBM drive
Here are the average performance numbers for the IBM 30GB drive running under the beta 2.5 Kernel, sent in by Richard Ballantyne.
Timing buffer-cache reads: 151.81 MB/sec
Timing buffered disk reads: 19.49MB/sec
From these I let you draw your own conclusions. If you want to comment on the speed difference between the 2.2 and 2.4 Kernels please mail me
Links and Related Pages
|