[comp.sys.ibm.pc] 387-20 Async vs. 387-25 Sync performance?

luong@grace.cs.washington.edu (Luong La) (02/08/90)

Yo FLOP's guru,

I owned a 386-25 computer w/ option that I can either install an async 387-20
or 387-25 sync. math coprocessor.  My question is :) does the 387-25 sync offers
a good bang for my bucks?  I heard that a 387-25 sync can offer about 
0.6 to 1 MFLOPs, how about 387-20 async?  
   
  (oddity: Crosstalk IV 's timing sped up 3 time if I set the motherboard ready
   for a 387-25 -> 25 Mhz sync operations.  Redemity/suggestion/guess??)


Thanks in advance for using the most powerful multi-everything computer
to process this msg.  :)))))))



=======/\ Zzzzz gosh our planet is depleting/dying... at least I recycle junks /\==



===RLab===Epsilon task takes infinite time=== luong@grace.cs.washington.edu

davidsen@sixhub.UUCP (Wm E. Davidsen Jr) (02/12/90)

In article <10643@june.cs.washington.edu> luong@grace.cs.washington.edu.cs.washington.edu (Luong La) writes:
| Yo FLOP's guru,
| 
| I owned a 386-25 computer w/ option that I can either install an async 387-20
| or 387-25 sync. math coprocessor.  My question is :) does the 387-25 sync offers
| a good bang for my bucks?  I heard that a 387-25 sync can offer about 
| 0.6 to 1 MFLOPs, how about 387-20 async?  

  I think you mean that you can set the clock for 20 or 25MHz, since I
never heard of async or sync versions of the 387 (particularly in the
part catalogs). Typically the f.p. is 20-30% of the total CPU, even on
math intensive applications. If you want better performance you need to
go to IIT or Cyrix (sp?) for 387 clones which run faster at the same
clock speed (perform ops in fewer clocks).
-- 
	bill davidsen - sysop *IX BBS and Public Access UNIX
davidsen@sixhub.uucp		...!uunet!crdgw1!sixhub!davidsen

"Getting old is bad, but it beats the hell out of the alternative" -anon

harper@convex.com (David Harper) (02/13/90)

In article <483@sixhub.UUCP> davidsen@sixhub.UUCP (bill davidsen) writes:
>In article <10643@june.cs.washington.edu> luong@grace.cs.washington.edu.cs.washington.edu (Luong La) writes:
>| Yo FLOP's guru,
>| 
>| I owned a 386-25 computer w/ option that I can either install an async 387-20
>| or 387-25 sync. math coprocessor.  My question is :) does the 387-25 sync offers
>| a good bang for my bucks?  I heard that a 387-25 sync can offer about 
>| 0.6 to 1 MFLOPs, how about 387-20 async?  
>
>  I think you mean that you can set the clock for 20 or 25MHz, since I
>never heard of async or sync versions of the 387 

I missed the original posting on this but I would be willing to bet that Luong 
La is refering to a Micronics motherboard.  This is the way that they describe
their 387 setup stuff in the documentation.  Jumpered for 'sync' operation the
387 can be the same speed as the 386 or slightly slower.  I run a 20 MHz 387
in 'sync' mode in the Micronics 25 MHz motherboard and it works fine, although
I have never benchmarked the performance.  Jumpered for 'async' mode you must
supply an external clock oscillator module to run the coprocessor.  This allows
you to put in any speed coprocessor you want.  Thus, 'asynchronous' in this
context means asynchronous to the 386 clock rather than a mode of operation for
the 387.


Dave Harper    -     Convex Computer Corp.            E-mail address:
3000 Waterview Pky.  Richardson, TX 75081             harper@convex.COM
(214) 497-4525 (W)   (214) 727-4206 (H)