shonle@csu-cs.UUCP (Greg Shonle) (11/01/84)
[]
I was just reading in InfoWorld that it is possible to speed up
the PC/AT. According to the latest InfoWorld, somebody pulled out
the 12 mhz chip, and put an 18 mhz crystal in (the crystal is socketed,
not soldered). It worked, except for the 80287, a diagnostics program,
and some clock-sensitive software, such as communication programs.
It looks like IBM can speed the AT up by 50% whenever they want to.
-------------------------------------------
Greg Shonle
...!{boulder,cires,denelcor,hao,hplabs,mit-vax,unmvax}!csu-cs!shonlegreenber@acf4.UUCP (11/11/84)
<> Interestingly enough, the only socketed "chip" on the moma-board is the clock chip. IBM states (from a usually reliable source) : "We are waiting for the higher speed FP chip to be fully debugged!" Ross M. Greenberg @ NYU ----> allegra!cmcl2!acf4!greenber <----
kfl@hoxna.UUCP (Kenton Lee) (11/12/84)
xxx When changing crystals, I would be especially careful of how the increased CPU speed affects RAM access and disk access, especially floppy disks.
toby@apple.UUCP (Toby Farrand) (11/13/84)
Sticking a faster clock in one machine is no proof that *all* machines will Distribution: net Organization: Apple Education Research Group, Cupertino CA Lines: 15 work at the faster clock. Will the machine still work at higher than normal temperature? Does the machine still pass RFI standards at the higher speed. Does the machine consume more power now? There was some mention of some diagnostics failures, which ones failed? How easy is it going to be to speed up the 80287? Who gave info world the idea to do this? Who gave them the machine? Who loves to spread rumors about future product directions that may or may not be true? I'll believe a cheap 80286 system can run at 18MHz when I see it. toby (The opinions expressed here are mine and do not reflect those of my employer)
topher@cyb-eng.UUCP (Topher Eliot) (11/15/84)
> I was just reading in InfoWorld that it is possible to speed up > the PC/AT. According to the latest InfoWorld, somebody pulled out > the 12 mhz chip, and put an 18 mhz crystal in (the crystal is socketed, > not soldered). It worked, except for the 80287, a diagnostics program, > and some clock-sensitive software, such as communication programs. > It looks like IBM can speed the AT up by 50% whenever they want to. This may be true. However, before anyone out there either tries to soup up their own AT, or reaches the conclusion that IBM is playing marketing/demand games, consider whether or not it will run at 18MHz when: the ambient temperature reaches 95F, or 40F, or the line voltage falls 10%, or you get a processor chip that is somehow marginal, or you get a memory chip that is just a tad on the slow side, or you plug in one more board, loading the bus drivers that much more, blah, blah, blah. Anyone who has every gone through the rigors of preparing a product for mass production will recognize that broadcasting the conclusion that 'it can be speeded up 50% any time you want' on the basis of one casual experiment is a bit premature.
jec@iuvax.UUCP (11/28/84)
[] I seriously doubt you are really running at 18Mhz. I suspect that the clock rate is divided at some point although I don't have the schematics to back this claim. If anyone knows for sure you might let us know. -- James Conley Indiana University 68K Education Board Project ..iuvax!jec
dan@rna.UUCP (Dan Ts'o) (12/02/84)
I believe that the PC/AT's clock is divided by three. Thus a 12 Mhz clock is really clocking the CPU and FPU at 4Mhz. 18Mhz -> 6Mhz which is better than the PC's clock (4.7Mhz ?). But the 80287 at 4Mhz is worst than the PC's 8087 at 4.7Mhz - essentially same FPU at a slower clock.
b-davis@utah-cs.UUCP (Brad Davis) (12/03/84)
The standard AT uses a 12 mhz clock that is divided by 2 (6 mhz) for the 80286 and 3 (4 mhz) for the 80287. If an 18 mhz crystal were used the clock speeds would be 9 mhz and 6 mhz. 80287's can't run at 6 mhz yet (as far as I know).
shonle@csu-cs.UUCP (Greg Shonle) (02/21/85)
[fnord!]
Just to set the record straight, the 80286 ALWAYS divides the external
clock speed by 2. Ie, the current 12mhz clock gives a 6mhz effective
speed. 18mhz->9mhz effective. The 80287 can either use the external
clock line directly, or it can divide by 3. The AT divides by 3. This
is controlled by one of the pins. Therefore, the AT's 287 runs at 4mhz.
Speaking of speeding things up, does anyone know where one can
buy 8-10mhz 286's and 6mhz 287's in small quantities?
Greg Shonle
...!{boulder,cires,denelcor,hao,hplabs,mit-vax,unmvax}!csu-cs!shonle