Re: FastCode - IntToStr B&V v0.12

Hi Avatar

> Yes, i like this too. And what about a B&V that can also post results
> by itself, either to the newsgroup or to a WebService that registers
> the results automatically for Dennis?

Actually Sander made this for us, but I did not find energy and time to work
more on it.

Remember that we do generally not want benchmark runs from a bunch of PC's.

We always benchmark on the official set of reference PC's here in my office.

The "Blended" target always consist of results from the same machines.

We always want many validations, but do not save results - just react on
failures.

The MM benchmark is something different. Here we want to record both
benchmark and validation results from as many machines as possible.

Regards
Dennis


.

Relevant Pages

  • Re: Fastcode MM B&V 0.51
    ... Dennis wrote: ... > through validations because of 1) ... Fix benchmark batch files to only test Delphi 2005. ... - Replace batch files with a native test coordinator app. ...
    (borland.public.delphi.language.basm)
  • Re: Ten Strike Bowling Game? (repro)
    ... Bowling Machines and although they bare the "Benchmark Games" name on ... eventually blow the main game fuse in the bottom of the cabinet. ... Further more after playing the game repeatedly the manquin figure ...
    (rec.games.pinball)
  • Re: Apple 2 benchmark?
    ... accelerator architectures--for example, caching. ... Then there's the special case of the Byte benchmark: ... Sieve, which was dutifully run on dozens of machines in the 1980s, ... The problem, of course, is broad comparability and verifiability. ...
    (comp.sys.apple2)
  • Re: "Complete exercise" of x86 example?
    ... purpose, then what is that very specific purpose? ... The benchmark is targeted towards vintage computing hobbyists who ... collect, restore, and use 808x- to 80486-based machines. ... I will be coming up with performance metrics for a CPU instruction mix ...
    (comp.lang.asm.x86)
  • Re: Alternative SZ IntToStrB&V v0.04
    ... Dennis, please do not write that I never wrote. ... on calculation for one element deviation. ... > When we use the benchmark, we do one run on each of the target PC's. ... That is reasonable argument against showing maximum spreading value. ...
    (borland.public.delphi.language.basm)