Read-ahead optimization

[mjc]

Hi everybody,

Last night looking up some info I came across a tip that suggested turning "read ahead optimization" in Win9x off. It went on to state that most modern hard drives have on board caching capabilities that were more efficient than Windows. So I tried it. I ran HDTach before and after, I saw a modest (1-3%) diffrence in the times but the area that I saw the most improvement was HDTachs "CPU Usage" reading from 2-5% before to zero after. It seems to me that by using thre drives on board capabilities it frees up some cpu cycles.

I was just wondering if anyone had any thoughts about this?

I'm going to run this way for a few days and see if I notice any improvements/degradations and post back with my test results.

------------------
mjc

[This message has been edited by mjc (edited 01-27-2001).]

[bassman]

I find this very interesting. I would like to know how to run this test on my machine. (HDTach) What steps and so on. thanks Frank

------------------
I want to be like Pete

[mjc]

HDTach is a HD benchmarking tool: "HD Tach is a physical performance hard drive test for Windows 95/98 and Windows NT"

I couldn't get the link before so here it is:HDTach (http://www.tcdlabs.com/hdtach.htm)

------------------
mjc

[ixl]

The best place for finding people who really know a lot about hard disk benchmarking is the General forum at http://www.storagereview.com. I haven't seen anyone mention read-ahead caching. However, HDTach isn't considered the most reliable hard disk benchmark.
How many times did you try it with caching both enabled and disabled? You should always average a few trials; differences of only a couple of points are generally just noise.

------------------
Charles M. Kozierok
Webslave, The PC Guide (http://www.PCGuide.com)
Comprehensive PC Reference, Troubleshooting, Optimization and Buyer's Guides...
Note: Please reply to my forum postings here on the forums. Thanks.

[mjc]

I've been running a few days now and haven't seen any real difference except in CPU usage. It's like the drive now doesn't use many clock cycles for caching operations. I used HDtach because it is what I had and ran a couple of runs of five or six run throughs each and averaged the results both before and after.

It is one of those things you always hear about: to get the bes from your harddrive you need rea ahead caching enabled in Windows...... This tip I found just seemed to be the opposite but it kind of makes sense if the drive has the ability to do caching why not use it? Just one of those thimgs I thought I'd try and see if it works or if it is just a fluke. Thought I'd see if anyone had any other thoughts on the matter, too.

------------------
mjc

[mjc]

Well Charles,
Thanks for the link, I couldn't find anything specific but a lot of information on caching in general. It seems that caching operations are very dependant upon drive, CPU configuration and quality of the drive's onboard cache/control circuitry so what works well with one setup may not work well with another. So since I found that I free up a little CPU time without losing any performance I think that I'm going to stay with this configuration. It seems on my system that the drive was doing a good job with read-ahead and Windows was just doing it over, and by doing so using up a little more CPU cycles than necessary.

------------------
mjc

[Randy_tx]

I wonder if I could EVER tax a 900 mhz processor enough to "feel" the difference?......oh well, might as well try it! http://www.PCGuide.com/ubb/biggrin.gif

------------------
Os2 wasn't a bad idea...

[mjc]

Maybe not, but I've got a 500 Celeron and a couple of points does make adifference

------------------
mjc