On Fri, 2007-07-06 at 19:54 +0530, Rohan Kulkarni wrote: > SNIP!! > > Hello, > My SATA drive provides me a copy speed of only 15MB/s while > one of my friends using > opensuse 10.2 with an Intel P4 and a Seagate SATA 80GB HDD > gets almost double the > copy speed.I use Fedora 7.Why is there such a large speed > difference? > Thank you > > > Copy speed is affected by many things in addition to processor speed, memory, and drive physics. The head seek time, rotational speed, burst write speed, interface to the system, and internal buss speed to name just a few of the physics of the drives, all impact the drive performance and are not related to the number of "bits per second". File size vs buffer size on the drive, whether the copy was on a single drive or across drives, whether the drives were on a single buss or different busses, and other esoteric details will affect the speed. The format will also affect the speed. The raid form, whether it is stripped, or blocked, and so forth. Basically to learn what affects the speed of your particular system, you need to do lots of leg work on discovering the physical connections, the drive characteristics and so forth. Is this badly impacting your daily use of the system? If it is important to you to get the highest speed, you will want to look at possibly spending large on hardware. That is the basis of high speed, and results from the price of the hardware, and precision of the electronics to controll the servo systems for the drive speed and head movement. The other bits, which bus interfaces are used, and how they are accessed are functions of the Motherboard and/or subsystems such as add on boards either 16,32, or 64 bits, and what the interface paths to these boards are like. Again more money probably will give you a better designed MB with attending gains in speed. Processor speed is secondary to this business. Most processors today are plenty fast enough to overrun the drives, so look to other areas of the hardware. Especially on copy operations. Some systems can implement a dma-dma copy and the processor doesn't even get involved once the paths are set up, other than yeilding memory cycles for DMA (which are often not used externally anyway.) In short, none of us can tell you why there are speed differences without a lot of data about both systems. Some folks here with raid experience may be able to suggest places to look for some "quick checks", but speed is never easy, whether it is boats, cars, motorcycles or computers. Regards, Les H