Date: Tue, 13 Aug 1996 08:21:51 +0300 (EET DST) From: Mr Operating System <freebsd@shadows.aeon.net> To: andreas@klemm.gtn.com (Andreas Klemm) Cc: freebsd-current@freebsd.org Subject: Re: compile time Message-ID: <199608130521.IAA04794@shadows.aeon.net> In-Reply-To: <Pine.BSF.3.95.960812214954.267J-100000@klemm.gtn.com> from Andreas Klemm at "Aug 12, 96 09:52:25 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
> > i thought i'd share, since people have posted here weird times... > > machine: p133 with async 256Kb cache (later today 512Kb pipeline), oh yeah, i didnt list my ram, i think it's 2 60ns samsung 8M and 2 60ns ibm 4M 72pin SIMMs, i'm not sure about the speed of 4M it's been a while... > > Sun Aug 11 01:55:01 EET DST 1996 make world starts > > Sun Aug 11 06:06:27 EET DST 1996 make world ended > sync burst cache numbers would be interesting in comparison to async yes, i installed yesterday 512Kb 7ns pipeline burst cache and compiled it again... and i didnt expect this much chance... this is _exactly_ same set of source, i wanted to make a test that is "relevant"... sorry, didnt want to put 256Kb cache... =) time: Mon Aug 12 23:34:31 EET DST 1996 make world starts Tue Aug 13 02:45:34 EET DST 1996 make world ends meaning 4:11 changed to 3:11!!!!! i got _one_ hour off with just the pipeline burst cache!!! and i dont consider my board being "fast"... (can anyone verify this with a similar test?) oh, and on both compilations, machine ran X, using about 10 megs of ram for it, it was online (i try _not_ to be off line for more than it takes to reboot). and i run also nntpd (cnews) which has one nonbin usenet group, which gets read by random people, about one nntpd running all the time, sometimes several... machine did not forward packets. > cache ... Otherwise it seems, that my very loud 7200 U/min Quantum > Grand Prix SCSI disk && AHA 2940 && sync burst cache men seems to > speed up things in comparison to your numbers ... Although we are > not exactly comparable, since you installed profiled libs, do you ?! i did 'make world' as it comes from the standard Makefile, only option i added was -DNOGAMES and the said comment out. soon i attempt the same compilation with gigabyte's dual pentium board, which has onboard aha7880, i try to attempt it with again same set of src from eide, narrow and wide scsi2... board is most similar with it's other parts than the one rated faster with p200 than ppro150 by some pcmag few weeks ago, dont remember which one... and i can use atleast 133 and 166 cpu (one only since there's no dual support on freebsd), i will also use another board with only 256Kb pipeline. tho those will happen from another harddrive, i made a copy, cant get those boards to my home... (boards at work, yes, i work for pc hardware seller as a system admin) i also try to make compiling with standard, edo and ecc ram, we just hardly sell anything using edo/ecc... (for the dual cpu devteamers, i get one of those to myself in few months) i'm sorry flooding the list with this, but atleast _i_ am interested in how much exactly one can speed up the machine with swapping hardware. > andreas@klemm.gtn.com /\/\___ Wiechers & Partner Datentechnik GmbH mickey
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199608130521.IAA04794>