Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!purdue!ames!attctc!chasm From: chasm@attctc.Dallas.TX.US (Charles Marslett) Newsgroups: comp.sys.ibm.pc Subject: Re: V20 vs. 8088: Is there an Summary: New coprocessors are from IIT, not Witek Message-ID: <8961@attctc.Dallas.TX.US> Date: 12 Aug 89 05:40:20 GMT References: <89080907301922@masnet.uucp> Organization: The Unix(R) Connection, Dallas, Texas Lines: 36 In article <89080907301922@masnet.uucp>, william.richardson@canremote.uucp (WILLIAM RICHARDSON) writes: > . . . Has anyone noticed the new 80287/80387 from > Witec? They claim 200% increase with the same clock rate on the 80287. I don't know about such from Weitek, but a startup company (the founders are from Weitek, if I remember my rumors correctly) called IIT has started makeing 287 and 387 clones. I got one! And it is fast -- one funny, though, is that ... > Fun note: most ATs' run their Co-Processer at 2/3 the system speed! > That can be fixed though, I'm running mine & 16Mhz on a 12Mhz AT ! As you mentioned, a signal can be used to divide the incoming clock (usually twice the CPU clock or equal to the CPU clock) by 3 before using it in the 287 -- the IIT chip divides the clock by 2 rather than by 3, so you get an additional speedup if the box runs with the divide-by-3 enabled. I wrote a benchmark suite (that includes one demo written by IIT programmers and it indicates you can get anywhere from a few percent improvement to about 4x depending on the type of code (if you load the FPU and bang away -- like my fractal demo -- it runs away with you, but if you just do the usual plodding C code -- load, add, store sequences -- then you'll get something more like a 10 or 20% improvement over a similar speed 287). Most programs are somewhere in between. > --- > ~ EZ-Reader 1.20 ~ -ex > > * QNet 1.03a3: CanConfMail: . HQBBS / ~ Toronto ~ Canada ~ 416y566y1765 =========================================================================== Charles Marslett STB Systems, Inc. <== Apply all standard disclaimers Wordmark Systems <== No disclaimers required -- that's just me chasm@attctc.dallas.tx.us