home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.coherent
- Path: sparky!uunet!spool.mu.edu!wupost!cs.utexas.edu!torn!maccs!mcshub!csx.cciw.ca!hcp
- From: hcp@csx.cciw.ca (H.C. Pulley)
- Subject: Re: Poor performance.
- Organization: Canada Centre for Inland Waters
- Date: Fri, 4 Sep 1992 14:37:43 GMT
- Message-ID: <1992Sep4.143743.4041@csx.cciw.ca>
- References: <01GOD0KWMK4Y90N81N@delphi.com>
- Lines: 55
-
- In article <01GOD0KWMK4Y90N81N@delphi.com> "Coherent operating system" <COHERENT@indycms.bitnet> writes:
- >Pred Bundalo writes:
- >> COHERENT 4.0.1 fine with both internal and external cache's enabled as well
- >> as ROM video shadowing.
- >> When I run a dhrystone program on the DOS side, I get 25000 dhrystones/s
- >> on average. When I run the same program on the COHERENT side, I get a
- >> horrible 2200 dhrystones/s--consistantly.
- >> I'm perplexed. Why would the performance be so poor on the COHERENT side?
- >> Also, it appears the problem is specific to my system/motherboard since I'm
- >> somewhat sure I had at least 2000 dhrystones/sec when I used my old 386-20
- >> motherboard.
- >> Does COHERENT do some disabling I'm not aware of?
- >
- >You have to be very careful with benchmarks. One thing I've noticed over
- >the years is that dhrystone is wonderful for benchmarking compilers, but
- >it doesn't tell you squat about the speed of the computer. For example,
- >in one case it appeared that a 10MHz 80286-based system ran circles
- >around a 20MHz 68020-based system. The differences were very deceiving
- >due to the compilers used. I had used one of Borland's compilers on
- >the 80286 and the compiler packaged with the 68020 system with no explicit
- >optimizations on either system. All this proved was that Borland creates
- >better compilers than the other vendor. When full optimizations were
- >turned on, there was literally no comparison between the two.
-
- Dhrystone 2.1 running under Coherent 3.2.1a got about 2600 on my 286-12 and
- 10800 on my 386-40c64. I got 3300 and 12500 under DOS (Turbo C++ v1.01)
- respectively. I hope that the 386 improves under Coherent 4.0.1 with the 32
- bit instructions and registers.
-
- Although it is true that Dhrystone tests BOTH the compiler and the CPU, if
- MWC's C compiler only got 2200 Dhrystones on a 486, I would suspect a problem.
-
- My copy of 4.0.1 has not yet arrived (only ordered it 9 business days ago) I
- cannot test the new compiler.
-
- Are you sure that your Dhrystone is working under Coherent? 2200 sounds a lot
- like 22000/10; 22000 would be quite reasonable. Could it be that you have an
- order of magnitude problem? (When I say reasonable here I am still skeptical;
- I would hope that, with 32 bit instructions and registers, the DOS Dhrystone
- should be lower than under Coherent, not higher.)
-
- The make for Dhrystone is quite different under DOS and Coherent. The timing
- is much more precise under Coherent. Make sure that your defines are correct,
- etc.
-
- After all this, it could of course, still be a motherboard problem but I doubt
- it. 2000 for a 386/20 still sounds quite low (note the numbers for my 286-12
- above).
-
- Harry
- --
- hcp@csx.cciw.ca | This message | It takes all kinds,
- hcpiv@grumpy.cis.uoguelph.ca | released to the | and to each his own.
- -----------------------------------| PUBLIC DOMAIN. | This thought in mind,
- Stay away from the DOS side Luke! | | I walk alone.
-