|
发表于 2010-1-16 12:08:25
|
显示全部楼层
Your approach depends on the access time of individual HDD, it may or may not be better.
If you use both HDD to build a RAID0, it may be faster.
Alternatively, there are several ways to gain a little bit higher performance:
1. Recompile the kernel with the better CPU optimization
2. Use higher version of gcc to compile the kernel
3. Always enable DMA and increase the read ahead buffer, e.g.
hdparm -c1 -d1 -u1 -a1024 /dev/xxx
(DMA for SATA/SCSI HDD may not work but the change of read ahead buffer should always work)
4. Make use of tmpfs |
|