magid@sandstorm.Berkeley.EDU (Paul Magid) (08/08/90)
I bought a 386sx thinking it was a 16mhz machine running at 16mhz. I think the dealer made a mistake and put in the wrong mother board b/c it runs at 20mhz. The cpu is a sx that is rated at 16mhz; however, the chips and technologies chip set is rated at 20mhz. The computer does run at 20mhz, and has ami bios. Now for my question: By how much percent will my 386 chips life be reduced? 10%, 20%? How long is the average life expectancy of a 386sx chip in years? Should I be worried, and should I take the computer back and demand my money back? Paul
mlbarrow@athena.mit.edu (Michael L Barrow) (08/09/90)
In article <1990Aug7.230306.16069@agate.berkeley.edu> magid@sandstorm.Berkeley.EDU (Paul Magid) writes: > I bought a 386sx thinking it was a 16mhz machine running at 16mhz. I > think the dealer made a mistake and put in the wrong mother board b/c it runs at > 20mhz. The cpu is a sx that is rated at 16mhz; however, the chips and > technologies chip set is rated at 20mhz. The computer does run at 20mhz, and > has ami bios. Now for my question: By how much percent will my 386 chips > life be reduced? 10%, 20%? How long is the average life expectancy of a 386sx > chip in years? Should I be worried, and should I take the computer back and > demand my money back? If the CPU is rated for 16MHz, then do _not_ let the machine run at 20MHz! Take it back & yell at them for doing something that silly. You will wear down your CPU if you run it faster than its rated. ----- Michael L Barrow mlbarrow@athena.mit.edu MIT Information Services Consultant MIT Project Athena Volunteer User Consultant Member, Student Information Processing Board (SIPB)
robertb@cs.washington.edu (Robert Bedichek) (08/09/90)
In article <1990Aug9.144532.677@athena.mit.edu> mlbarrow@athena.mit.edu (Michael L Barrow) writes: >In article <1990Aug7.230306.16069@agate.berkeley.edu> >magid@sandstorm.Berkeley.EDU (Paul Magid) writes: > >If the CPU is rated for 16MHz, then do _not_ let the machine run at 20MHz! >Take it back & yell at them for doing something that silly. You will wear >down your CPU if you run it faster than its rated. > Can you cite any reference to support this claim? If not, can you give any explanation for why a CPU chip, or any other chip, should "wear out" if run at a higher-than-rated clock? My understanding of chip speeds, from listening to silicon designers and managers at Intel talk (I used to work there), is that there is no physical difference between chips that are rated at different speeds other than what is stamped on their case. The faster parts have just passed the chip tests at a higher speed. I could be incorrect, of course, let me know. Rob Bedichek
stretch@EMUNIX.EMICH.EDU (Brian Stretch) (08/10/90)
I have a 386SX with a Deico motherboard running at 16/20Mhz. The CPU chip
is rated for 16Mhz, but I have never had any problems with 20Mhz operation,
so I usually leave it on "Turbo".
We used to run Tandy Color Computers with 1 or 1.5Mhz Motorola 6809's at
1.5 or 2Mhz.. no problems. (Some of the 1Mhz machines couldn't handle it,
but they were _really_ old..) As a percentage, that's a much higher speed
jump than 16-20Mhz. 'Course, Motorola, IMHO, makes much better stuff than I
Intel..
---Brian Stretch (stretch@emunix.emich.edu)minar@reed.bitnet (Nelson Minar,L08,x640,7776519) (08/10/90)
In article <12743@june.cs.washington.edu> robertb@june.cs.washington.edu (Robert Bedichek) writes: >In article <1990Aug9.144532.677@athena.mit.edu> mlbarrow@athena.mit.edu (Michael L Barrow) writes: >>In article <1990Aug7.230306.16069@agate.berkeley.edu> >>magid@sandstorm.Berkeley.EDU (Paul Magid) writes: >> >>If the CPU is rated for 16MHz, then do _not_ let the machine run at 20MHz! >>Take it back & yell at them for doing something that silly. You will wear >>down your CPU if you run it faster than its rated. > >Can you cite any reference to support this claim? If not, can you give >any explanation for why a CPU chip, or any other chip, should "wear >out" if run at a higher-than-rated clock? Well, you see there are these little gerbils in the chip. They run on this little wheel called the Arithmetic Logic Unit, and they do all the addition. Chips that run faster have happier gerbils. But, if you try to push the gerbils too hard, they get tired. No, you probably cannot hurt a chip by running it faster than it is rated for. What you will do is experience all sorts of nasty failures. Back in the days of the first IBM AT, IBM released 2 versions - a 6MHz version and an 8MHz version. No difference, other than a new crystal. So, smart people bought the 6MHz AT, popped in a new crystal, and poof! A cheaper 8MHz AT. Some people even put in adjustable crystals, and could run their machines up at 9 or 10 MHz. This led to the fabled speed trap in the BIOS, but thats a nasty story. Disclaimer: I know nothing. Do not make hardware modifications to your computer on the basis of this article.
jones@acsu.buffalo.edu (terry a jones) (08/10/90)
>>If the CPU is rated for 16MHz, then do _not_ let the machine run at 20MHz! >>Take it back & yell at them for doing something that silly. You will wear >>down your CPU if you run it faster than its rated. > >Can you cite any reference to support this claim? If not, can you give >any explanation for why a CPU chip, or any other chip, should "wear >out" if run at a higher-than-rated clock? > >My understanding of chip speeds, from listening to silicon designers >and managers at Intel talk (I used to work there), is that there is no >physical difference between chips that are rated at different speeds >other than what is stamped on their case. The faster parts have just >passed the chip tests at a higher speed. > >I could be incorrect, of course, let me know. > > Rob Bedichek Rob is correct.. Manufacturers will sort CPU chips by speed as they pass testing at that speed. I have had good luck running a number of CPUs at elevated clock rates. There is no justification to the theory that the CPU chip will "wear out" running at an elevated clock rate. I would be concerned that increased temperature would cause the CPU to be more prone to intermittent failure, since it is running beyond guaranteed speeds. Personally, I'd get the box warm, set some diagnostics running, and let it fly for a few days. If there was no flakiness, keep it, being happy with the deal I got...otherwise send it back an complain. Terry -- Terry Jones {rutgers,uunet}!acsu.buffalo.edu!jones SUNY at Buffalo ECE Dept. or: rutgers!ub!jones You are in a maze of twisty little compiler features, all different.
sonny@charybdis.harris-atd.com (Bob Davis) (08/10/90)
In article <1990Aug9.144532.677@athena.mit.edu> mlbarrow@athena.mit.edu (Michael L Barrow) writes: [Deletions] > >If the CPU is rated for 16MHz, then do _not_ let the machine run at 20MHz! >Take it back & yell at them for doing something that silly. You will wear >down your CPU if you run it faster than its rated. > Hmmmm. I bet if you change the CPU oil and filter frequently though, you can really minimize the wear. Most folks just NEVER change it. ______________________________________________________________________________ Bob Davis \\ INTERNET : sonny@trantor.harris-atd.com | _ _ | Harris Corporation, ESS \\ UUCP : ...!uunet!x102a!trantor!sonny |_| |_| | | Advanced Technology Dept.\\ AETHER : K4VNO |==============|_/\/\/\|_| PO Box 37, MS 3A/1912 \\ VOICE : (407) 727-5886 | I SPEAK ONLY | |_| |_| | Melbourne, FL 32902 \\ FAX : (407) 729-2537 | FOR MYSELF. |_________|
strike@clmqt.marquette.Mi.US (Tim Bowser) (08/11/90)
jones@acsu.buffalo.edu (terry a jones) writes: >Rob is correct.. Manufacturers will sort CPU chips by speed as they pass >testing at that speed. I have had good luck running a number of CPUs at >elevated clock rates. There is no justification to the theory that the >CPU chip will "wear out" running at an elevated clock rate. I would be >concerned that increased temperature would cause the CPU to be more prone >to intermittent failure, since it is running beyond guaranteed speeds. >Personally, I'd get the box warm, set some diagnostics running, and let it >fly for a few days. If there was no flakiness, keep it, being happy with >the deal I got...otherwise send it back an complain. This is *precisely* what I have here, in a Pioneer 386sx 20mHz motherboard. The CPU is stamped 16mHz, but the manufacturer of the board stressed the chip at 20 with no failure. The distributor of the motherboard claims to do the same thing, and from six months of continuous operations in the Xenix environment, I have zero complaints. Because it is running at above the stamped spec, I take pains to make sure the system receives as much cooling air as it can get. Heck, this is good advice for *any* system, no matter what the internals are doing. Keep the machine. -- Tim Bowser ("Strikemaster") | Standard | mailrus!sharkey!clmqt!strike Enterprise Information System | Disclaimer | strike@clmqt.marquette.Mi.US Marquette, Mi. USA | Here | Voice:(906)-346-6735 => UNIX: The Adventure Begins... To vi, or not to vi, that is the question. <=
cgordon@vpnet.chi.il.us (Gordon Hlavenka) (08/13/90)
>Well, you see there are these little gerbils in the chip. They run on this >little wheel called the Arithmetic Logic Unit, and they do all the addition. This is strange. I had gerbils in my ALU but they didn't do addition, they only multiplied. CAVEAT: keep the gerbils well fed, otherwise they will eat your random number seed, and all your BASICA RND() functions will underflow. ---------------------------------------------------------- Gordon S. Hlavenka cgordon@vpnet.chi.il.us Disclaimer: He's lying
rommel@lan.informatik.tu-muenchen.dbp.de (Kai-Uwe Rommel) (08/16/90)
This seems to be a popular practice of clone manufacturers. I have a 24MHz '386DX board with a CPU stamped as 20MHz. It works well, no problems at all. And the machine runs often 12 or 16 hours a day. I know the dealer personally and he does not have any failures with the boards of this type (he sold a lot of them). What method use the people at Intel to determine, if a particular chip can be rated 16, 20, 25 or 33MHz? Kai Uwe Rommel -- /* Kai Uwe Rommel * Munich * rommel@lan.informatik.tu-muenchen.dbp.de */
c37189h@saha.hut.fi (Harri "Haba" Suomalainen) (08/16/90)
>The CPU is stamped 16mHz, but the manufacturer of the board stressed the >chip at 20 with no failure. The distributor of the motherboard claims to ..... >Because it is running at above the stamped spec, I take pains to make sure >the system receives as much cooling air as it can get. Heck, this is good >advice for *any* system, no matter what the internals are doing. I tried a 80387 with my 386. I was happy to see my 16MHz 387 worked fine with my 25MHz machine. All it does is it gets a bit warm. I was prepared to cool it with a piece of metal put on it, but it wasn't needed. I put it under a serious testing for some time and it was ok. I also saved some $$$ with a 'slow' version :-) -hs -- Harri Suomalainen c37189h@saha.hut.fi haba@otax.tky.hut.fi
prk@planet.bt.co.uk (Peter Knight) (08/17/90)
magid@sandstorm.Berkeley.EDU (Paul Magid) writes: > I bought a 386sx thinking it was a 16mhz machine running at 16mhz. I >think the dealer made a mistake and put in the wrong mother board b/c it runs at >20mhz. The cpu is a sx that is rated at 16mhz; however, the chips and >technologies chip set is rated at 20mhz. The computer does run at 20mhz, and >has ami bios. Now for my question: By how much percent will my 386 chips >life be reduced? 10%, 20%? How long is the average life expectancy of a 386sx >chip in years? Should I be worried, and should I take the computer back and >demand my money back? >Paul Dear All, As this is my first outting on the net, I am a bit terrified of doing something wrong, but here goes. There is no difference between 16 and 20MHz chips, except that the faster chips get through the testers faster. However, the tests that are used compare the chip to its speed rating on the data sheet, which includes all kinds of interesting stuff about the pulse widths and how they relate to clock speed. If the board manufacturer does his own tests of incoming parts against not the data sheet but against his board specs, then there *should* be no problem. It shouldn't be done for where the cost of failure is more than an irate user and a crashed hard disk, of course. As for shortening the life, it could do. These chips run hot and the faster they are clocked, the hotter they get. The extra heat *could* shorten the life of the chip, but on the other hand PC's tend to fail because of wires breaking, solder joints giving up, boards disintegrating, etc. Peter Knight BT Research The views in this article are the author's own and are not necessarily endorsed by British Telecom. No liability accepted.