Fastcode CompareMem B&V

Hi All,

I'm busy writing my entry for this challenge, and will post it later today.
However, since I've started I've noticed that there is a lot of room for
improvement in the benchmarks:

1) Arrays to be compared are not initialized, so the number of compare
iterations needed to hit the first mismatch is down to luck. This badly
affects the scoring depending in which order the entries are benchmarked.
2) The largest compare is up to 1000 bytes. There should be a benchmark with
a larger range.
3) P1 and P2 are always aligned exactly the same, so if you align reads from
P1 then they will automatically also be aligned for P2.
4) The number of bytes to compare (the third parameter) is monotonically
increased in each benchmark run, thus putting no pressure on the branch
predictor in places where the CompareMem routine needs to branch depending
on the number of bytes to compare.
5) The data to compare is exactly the same for all compares in a given
benchmark (of which there are only 2), all that differs is the compare
length. There should be greater variety in the input data.
6) There should be benchmarks with purposely constructed long sequences of
data that are the same with the exception of a single mismatch somewhere.

Validation seems to be in a much much better state than the benchmarks.

I think these are points we should look at in the coming year. For now I am
happy to take the points ;-).

Regards,
Pierre

--
Fastcode Project: http://www.fastcodeproject.org/


.

Relevant Pages

  • Re: java performance on machines
    ... Primarily the goal is to compare cpu/memory power on different ... The idea is to write a series of tests in java, both integer and float ... If synthetic benchmarks are ok then you could look at the SPEC benchmarks ...
    (comp.lang.java.help)
  • Calculating quarterly investment returns---XIRR or another function??
    ... I am trying to compare the rates of return on an investment account to other ... The calculated XIRR for the account for one year is 20.11%. ... For comparison, here are returns for the Vanguard Wellington mutual fund, ... fact provide quarterly results that can be accurately compared to benchmarks ...
    (microsoft.public.excel.worksheet.functions)
  • Re: SPB Benchmarks
    ... I didn't compare just final overall numbers. ... > The h19xx series scores unusually high on graphics benchmarks. ... > no one has been able to explain the reasons why. ...
    (microsoft.public.pocketpc)
  • benchmarking parallel chips
    ... look forward to some very high-quality benchmarks for the SEAforth chip ... Some of the problems I have been dealing with are how to compare ... How does one compare task switch time on a system with one proceesor ... with fewer tasks per processor and multiple tasks running at one time? ...
    (comp.lang.forth)