🚀 go-pugleaf

In article <1990Mar12.040128.3975@cec1.wustl.edu>, sek9424@cec1.wustl.edu (Scott Eric Keller) wrote:
}
}        I am currently trying to determine what real differences there
}may be between 10Mhz and 12.5Mhz '286 machines. I currently have a 
}Compuadd 286-10 using a 10Mhz '286 processor with 1 wait state and 100ns
}RAM. This machine produces a 9.7Mhz clock rate on both the Lanmark CPU test
}and the PC Tech Journal ATPERF performance test. Out of curiosity, I tried
}operating the machine with 0 wait states. After a few memory errors at 
}power on, the machine will operate for several minutes at 12.7Mhz indicated
}by the Lanmark test. The ATPERF test still shows 9.7.(why?) Eventually this 

Because ATPERF computes the actual speed of the clock signal fed to the
processor, while Landmark and all other benchmarks I know of compute the
"effective throughput" which is good for advertising but not much else.
Landmark is telling you that 9.7MHz at 0 waits is equivalent to 12.7MHz at
one wait state.

}3) Is there a reason for the wait state other than keeping the processor
}   happy? Or is it needed to slow down to keep the memory happy? What is
}   it we are all 'waiting' for?

It is needed to slow down memory accesses enough for the RAM to keep up. Note
that 100ns RAMs *should* be fast enough for zero waits at 10MHz (I used to
have a 9.8 MHz/1ws machine that ran just fine with 150ns RAMs--the wait state
increases the allowable access time on the RAMs by 50%)

--
UUCP: {ucbvax,harvard}!cs.cmu.edu!ralf -=- 412-268-3053 (school) -=- FAX: ask
ARPA: ralf@cs.cmu.edu  BIT: ralf%cs.cmu.edu@CMUCCVMA  FIDO: Ralf Brown 1:129/46
"How to Prove It" by Dana Angluin              Disclaimer? I claimed something?
16. proof by cosmology:
    The negation of the proposition is unimaginable or meaningless.  Popular
    for proofs of the existence of God.