home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.minix
- Path: sparky!uunet!mcsun!sunic!ericom!ericom!etlmike
- From: etlmike@jerry.ericsson.se (Mike Wilcox)
- Subject: Re: (none given)
- In-Reply-To: hp@vmars.tuwien.ac.at's message of Fri, 11 Sep 1992 09:34:44 GMT
- Message-ID: <ETLMIKE.92Sep11121917@jerry.ericsson.se>
- Sender: news@ericsson.se
- Nntp-Posting-Host: jerry.ericsson.se
- Organization: Ericsson Ltd., Cellular Systems & Special Networks.
- References: <1992Sep9.120116.13849@udel.edu> <ETLMIKE.92Sep10173232@jerry.ericsson.se>
- <1992Sep11.093444.223@email.tuwien.ac.at>
- Date: Fri, 11 Sep 1992 11:19:17 GMT
- Lines: 49
-
-
- In article <1992Sep11.093444.223@email.tuwien.ac.at> hp@vmars.tuwien.ac.at (Peter Holzer) writes:
-
- etlmike@jerry.ericsson.se (Mike Wilcox) writes:
-
- >Unfortunately, I have gone to the effort of changing the Minix
- >Centre's version of Minix-386 precisely to allow GCC to compile the
- >kernel.
- [...]
- >Now, granted I have a lot of other changes in my version of the
- >kernel, but I can say that after using GCC to compile this,
- >my system runs a LOT slower than when BCC compiled it in 32-bit
- >mode.
-
- Can you still compile your system with bcc (or c386)? If so, you could
- easily test if the lack of speed is due to the compiler or your
- changes.
-
- I haven't recently tried - But, I know the changes are slowing the
- system down, AND that GCC slows it down some more.
-
- Also, gcc has gazillions of options to control optimization. Some of
- them (e.g. -fomit_frame_pointer) make a huge difference in speed. The
- code produced by gcc without optimization is quite slow, usually.
-
- Hmmm - it is very slow when no options are used. I have many
- screenful's of benchmark speeds with the kernel compiled with
- different compilers, and different options (especially to GCC).
- Also with different speeds for different versions of GCC!
-
- I think best compilation is with -O -fomit-frame-pointer
- -fstrength-reduce -fcombine-regs.
-
- I seem to remember that using -finline-functions actually made things
- worse. If anyone can explain THAT one to me, I'd be grateful!
-
-
- Peter
- --
- | _ | Peter J. Holzer | Think of it |
- | |_|_) | Technical University Vienna | as evolution |
- | | | | Dept. for Real-Time Systems | in action! |
- | __/ | hp@vmars.tuwien.ac.at | Tony Rand |
-
- Mike
- --
- Mike Wilcox Guildford, Surrey, England.
- CME 20 Design Dept. Tel: +44-483-303666
- Ericsson Cellular Systems. Email: etlmike@deep-thought.ericsson.se
-