rayb@altos86.Altos.COM ( Ray Barbieri) (05/09/91)
We're running the 12/01/84 "C/1" version of the dhrystone benchmark and I'm investigating both Green Hills and Metware to see who is fastest. Is "C/1" equivlent to what everyone calls dhrystone 2.1? I know you purists out there snub your nose at this benchmark and I'd agree in principal. However, our company has decided that we need to show how much "better" than everone else we are by publishing benchmark results. I know that dhrystones on a standard X86 AT is not very interesting in that the system architecture is pretty old and the processors are just faster versions of another old architecture (no fancy parallel processing or the like). However, there is a tremendous market out there of people who's basic understanding of a computer is epitomized in the IBM PC/AT. When they want to upgrade, what do they use to tell them who's machine is "better" than who's? It's pretty impractical to ask them to run their company's accounting package on each of 100 or more vendors to see who's is the best performer. Today, the majority of these people use the dhrystone benchmark program along with others to determine what the new machine is capable. I agree that this is not optimal, but remember that the PC world is defined by standards. Until there is a standard, no one will invest (or believe) in it. Benchmarks also fall prey to this. It's a different sort of challenge to find a benchmark to satisfy the PC user. Unfortunately, once a standard is adopted, it hangs around forever it seems because standards tend to carry a tremedous amount of inertia. Ray Barbieri rayb@altos.com
martin@cbmvax.commodore.com (Martin Hunt) (05/11/91)
In article <305@altos86.Altos.COM> rayb@altos86.Altos.COM ( Ray Barbieri) writes: >We're running the 12/01/84 "C/1" version of the dhrystone >benchmark and I'm investigating both Green Hills and Metware to see who >is fastest. Is "C/1" equivlent to what everyone calls dhrystone 2.1? > No. C/1 indicates you have version 1 of the C language dhrystone. Version C 2.1 is dated May 1988. You should really grab a copy of this. >I know you purists out there snub your nose at this benchmark and I'd agree >in principal. However, our company has decided that we need to show >how much "better" than everone else we are by publishing benchmark >results. I know that dhrystones on a standard X86 AT is not very >interesting in that the system architecture is pretty old and the processors >are just faster versions of another old architecture (no fancy parallel >processing or the like). However, there is a tremendous market out >there of people who's basic understanding of a computer is >epitomized in the IBM PC/AT. When they want to upgrade, what do they >use to tell them who's machine is "better" than who's? It's pretty >impractical to ask them to run their company's accounting package on >each of 100 or more vendors to see who's is the best performer. > >Today, the majority of these people use the dhrystone benchmark program >along with others to determine what the new machine is capable. I >agree that this is not optimal, but remember that the PC world is >defined by standards. Until there is a standard, no one will invest >(or believe) in it. Benchmarks also fall prey to this. > >It's a different sort of challenge to find a benchmark to satisfy the >PC user. Unfortunately, once a standard is adopted, it hangs around >forever it seems because standards tend to carry a tremedous amount of >inertia. > >Ray Barbieri >rayb@altos.com Dhrystone is useful, as long as you know what it is really measuring. It is not an overall indication is system worth, as most PC users assume. At best, it measures how good your C compiler is at optimization, what speed your CPU is running at and the speed of your cache. It indicates nothing about what happens when your program does not fit in cache, how fast your computer can update the screen, read and write to disk, perform various floating point operations, or do integer math. Because of this, it is unsuitable as a measure for system performance, which is unfortunately what most uninformed users use it for. Martin Hunt Commodore-Amiga martin@cbmvax.commodore.com "Windows 3.0 is hot because it's really fun. It has brought some excitement back into the PC industry" - Microsoft I wonder who took the excitement out in the first place?
davidsen@yeti.crd.GE.COM (william E Davidsen) (05/14/91)
In article <305@altos86.Altos.COM>, rayb@altos86.Altos.COM ( Ray Barbieri) writes: |> I know you purists out there snub your nose at this benchmark and I'd agree |> in principal. However, our company has decided that we need to show |> how much "better" than everone else we are by publishing benchmark |> results. |> Today, the majority of these people use the dhrystone benchmark program |> along with others to determine what the new machine is capable. I |> agree that this is not optimal, but remember that the PC world is |> defined by standards. Until there is a standard, no one will invest |> (or believe) in it. Benchmarks also fall prey to this. And everyone has pretty much standardised on C2.1 because 1.x gave unrealistic results. Since it's free I would upgrade, since you will not be comparing yourselves to the same benchmark most reputable vendors are using. -- Bill Davidsen (davidsen@crdos1.crd.ge.com, uunet!crdgw1!crdos1!davidsen) GE Corp R&D Center, Schenectady NY Moderator of comp.binaries.ibm.pc and 386users mailing list "This is your PC. This is your PC on OS/2. Any questions?"