jaapv@accucx.cc.ruu.nl (Jaap Verhage) (12/12/90)
I've got a strange little problem - to me, that is. A while ago, I ftp'ed some of the RAM refresh rate-twiddling programs from Simtel. The idea is that by refreshing your RAM chips less often than the standard rate for your PC, you cut down on CPU overhead, thus increasing the speed at which the thing operates - seemingly, at least. Don't expect anything gorgeous - if speed increases by anything like 5 % or a bit more, that's it. Still, every little bit helps, and I like fooling around, so I tried. The Landmark test for CPU speed showed my work-based Olivetti M28, a 8-MHz AT, to start running at 8.5 MHz; my home-based Laser AT went from 11.6 to 12.5 MHz. Wow! You can imagine I noticed this *instantly* in running applications :-). But. The Laser's 5.25" HD diskettedrive came up with horrendously long seek times on disks! The Olivetti had no problems at all. Seek times on the Laser's diskettes grew as I cranked up the speed of the CPU by lowering the RAM refresh rates. Now what *is* this? May be stuff for a nice little discussion. If you don't think so, but do have suggestions, email me and I'll summarize. Thanks for your time and trouble. -- Regards, Jaap. Jaap Verhage, Academic Computer Centre, State University at Utrecht, Holland. jaapv@cc.ruu.nl +<-*|*->+ I claim *every*thing and speak for myself
jaapv@accucx.cc.ruu.nl (Jaap Verhage) (01/03/91)
A while ago, in article <995@accucx.cc.ruu.nl> I wrote: >I've got a strange little problem - to me, that is. A while ago, I >ftp'ed some of the RAM refresh rate-twiddling programs from Simtel. >The idea is that by refreshing your RAM chips less often than the >standard rate for your PC, you cut down on CPU overhead, thus >increasing the speed at which the thing operates - seemingly, at >least. Don't expect anything gorgeous - if speed increases by >anything like 5 % or a bit more, that's it. Still, every little bit >helps, and I like fooling around, so I tried. The Landmark test for >CPU speed showed my work-based Olivetti M28, a 8-MHz AT, to start >running at 8.5 MHz; my home-based Laser AT went from 11.6 to 12.5 >MHz. Wow! You can imagine I noticed this *instantly* in running >applications :-). But. The Laser's 5.25" HD diskettedrive came up >with horrendously long seek times on disks! The Olivetti had no >problems at all. Seek times on the Laser's diskettes grew as I >cranked up the speed of the CPU by lowering the RAM refresh rates. >Now what *is* this? >May be stuff for a nice little discussion. If you don't think so, >but do have suggestions, email me and I'll summarize. Thanks for >your time and trouble. Reactions to this were as follows: ***** I have experienced the same magic. Would be nice to know why this happens. All thougths are strictly my own. __________________________________________________________ Mats Lordin Infologics AB E-mail: mats@infolog.se P.O. Box 91 Phone int +46-8-92 20 00 S-191 22 Sollentuna Phone fax +46-8-96 08 46 __________________________________________________________ ***** Will you settle for wild speculation? WHile the hard drive is usually accessed via program control, the floppy runs with DMA. If your DMA chip is having a hard time keeping up, this could be the problem. If you have a NEAT chipset, or other configurable system, try adding a wait state to the DMA, or slowing the bus down, as from 10 to 8 MHz. That will slow you back down, but might be an interesting test for information. I would consider running one way for floppy use and one way for normal operation, if (a) you can really tell the diference, and (b) it's easy to switch. --- bill davidsen - davidsen@sixhub.uucp (uunet!crdgw1!sixhub!davidsen) sysop *IX BBS and Public Access UNIX moderator of comp.binaries.ibm.pc and 80386 mailing list "Stupidity, like virtue, is its own reward" -me ***** Yup, I noticed the exact same thing. I slowed the refresh rate so that the floppy still performed properly, but not past. Apparently the BIOS is doing something that is time critical, and is based on the refresh rate. You don't happen to have an AMI bios, do you? (it's what I have) ***** Well ... (me again): slowing the bus down didn't help; I do have an AMI BIOS in the Laser. Putting back the standard refresh rate in the place where it belongs, cures the problem and also cures my wonderful speed increase. Doesn't look like much can be done ... I guess I'll only muddle with the refresh rate for nightlong tasks like ray tracing and such. Thanks to all who responded! -- Regards, Jaap. Jaap Verhage, Academic Computer Centre, State University at Utrecht, Holland. jaapv@cc.ruu.nl +<-*|*->+ I claim *every*thing and speak for myself
rbono@necis.UUCP ( NM1D) (01/05/91)
In article <1041@accucx.cc.ruu.nl>, jaapv@accucx.cc.ruu.nl (Jaap Verhage) writes: > A while ago, in article <995@accucx.cc.ruu.nl> I wrote: > >I've got a strange little problem - to me, that is. A while ago, I > >ftp'ed some of the RAM refresh rate-twiddling programs from Simtel. > >The idea is that by refreshing your RAM chips less often than the > >standard rate for your PC, you cut down on CPU overhead, thus > >increasing the speed at which the thing operates - seemingly, at > >least. Don't expect anything gorgeous - if speed increases by > >anything like 5 % or a bit more, that's it. Still, every little bit > >helps, and I like fooling around, so I tried. The Landmark test for > >CPU speed showed my work-based Olivetti M28, a 8-MHz AT, to start > >running at 8.5 MHz; my home-based Laser AT went from 11.6 to 12.5 > >MHz. Wow! You can imagine I noticed this *instantly* in running > >applications :-). But. The Laser's 5.25" HD diskettedrive came up > >with horrendously long seek times on disks! The Olivetti had no > >problems at all. Seek times on the Laser's diskettes grew as I > >cranked up the speed of the CPU by lowering the RAM refresh rates. > >Now what *is* this? > >May be stuff for a nice little discussion. If you don't think so, > >but do have suggestions, email me and I'll summarize. Thanks for > >your time and trouble. > > Reactions to this were as follows: > Sorry that I missed the original post.... No one seemed to have the correct answer, so I thought that I would give you my few cents worth. From the answers, many of you with various BIOSs have seen this (Award, etc.) First, I will state that it is my belief that messing around with system timing is a dangerous thing to do at best. Ram refresh timing is one of those areas where you could be asking for reliablility problems, not to mention other side effects (as many of you have noticed). Yes, I am aware that many of you have done this, and have not seen (or noticed) any problems as of yet (this is not to say that you will have problems). Suffice to say that if you want a reliable system, don't mess with your refresh timing, if speed is more important to you than reliability, then go ahead and have fun! Now, with that 'dislaimer' here is your answer: This discussion is not relative to any particular type or brand of BIOS, but to the functions any BIOS must perform. There are many times that the BIOS needs to generate accurate timing delays, especially when communicating with periphials (like floppy and hard disks). Of course, this is what the BIOS is designed to do! Timing delays must be generated independantly of CPU speed, as many of todays systems have CPU's that can be switched between multiple speeds (6/8/16/33 MHz, etc). To do this, these BIOSs make use of known hardware that has a particular rate. In general the BIOS can't make use of the hardware timer chip, because that is used by many applications. One thing that is *always* running is the RAM refresh (if it is not running, then the system would not function, DRAM requires refreshing to hold it's data). The refresh rate is known by the BIOS and there is a way for the BIOS to detect when a refresh has occured (I/O port 61h, bit 4). Therefore the BIOS could generate accurate delays by monitoring the refresh rate. By changing the refresh rate you have changed the BIOSs timebase! This can lead to all kinds of problems (could even cause disks to be 'trashed' as improper delays are generated when trying to write to disks). By the symtoms many of you have discussed, my opinion is that many of you have BIOSs that make use of the refresh timing to generate delays. Of course, your BIOS may not use the refresh timing for delays, and then you will probably not see this 'problem'. Keep in mind that other software may also try to use the refresh rate for non CPU dependant timing delays... changing the refresh rate may affect these programs also. My advice? Don't mess with the refresh rate if you wan't a reliable system! Disclaimer: These are my own opinions -- /**************************************************************************\ * Rich Bono (NM1D) Note NEW email address: rbono@necis.ma.nec.com * * (508) 635-6300 NEC Technologies Inc. NM1D@WB1DSW * \**************************************************************************/