Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!news.unimelb.EDU.AU!munnari.OZ.AU!spool.mu.edu!howland.reston.ans.net!newsfeed.internetmci.com!gatech!news.emf.net!overload.lbl.gov!eubie.lbl.gov!jin From: jin@eubie.lbl.gov (Jin Guojun[ITG]) Newsgroups: alt.comp.periphs.mainboard.asus,comp.unix.bsd.freebsd.misc,comp.unix.bsd.netbsd.misc Subject: Re: Triton vs. SiS Pentium chipsets Date: 30 Jan 1996 03:43:21 GMT Organization: Lawrence Berkeley Laboratory Lines: 83 Message-ID: <4ek44p$oib@overload.lbl.gov> References: <MICHAELV.96Jan15233000@MindBender.HeadCandy.com> <DLGoEE.1p9@madsoft.lonestar.org> <4duqa2$5v9@fountain.mindlink.net> NNTP-Posting-Host: eubie.lbl.gov Xref: euryale.cc.adfa.oz.au alt.comp.periphs.mainboard.asus:9336 comp.unix.bsd.freebsd.misc:13003 comp.unix.bsd.netbsd.misc:2106 In article <4duqa2$5v9@fountain.mindlink.net>, Steve Tang <steve_tang@mindlink.bc.ca> wrote: >In article <DLGoEE.1p9@madsoft.lonestar.org>, mike@madsoft.lonestar.org (Mike) says: >: >:Michael L. VanLoon (michaelv@MindBender.HeadCandy.com) wrote: >:>What exactly are the differences between the Triton and SiS >:>Pentium-class chipsets? Does the SiS have any disadvantages versus >:>the Triton? Does the SiS demand special hardware (cache, RAM, etc.) >:>that the Triton doesn't? Which performs better? >: >:Get the January 8th (or the following week) copy of Infoworld and >:find the comparison of five or six 133MHz Pentium systems. All but >:one had about the same performance. The fast one were Tritons and the >:slow one was based on SIS. The chipset is probably not the sole >:cause of lower performance, but it is critical to good performance. > >Unless the five or six P133 systems are all by the same company, this >is still not a fair comparison. If same type of CPU is used, the comparison will show the difference. >What I want to see are benchmarks from two P133 systems with exactly >the same configuration with the only difference being one using the >P55TP4XE and the other using the P55SP4. > >That would be a fair comparison. > >Steve Tang If you want to see what is the different, below is some sent to me in acouple of weeks ago: --------------------------- included message ------------------------- Back on the second of January I wrote: I have an ASUSTek PCI/I-P54SP4 rev 1.5 motherboard with a 90MHz Pentium .... we got another 32MB of RAM... Now the machine runs like dog bait. I had several interesting suggestions from people at BSDI and the bsdi-users mailing list about only caching the first 32MB of memory and other possible problems. I was starting to believe the system was only caching the lower 32MB of main memory. Then I received the following from Jin Guojun at LBL: The problem is not cache amount. It is in SiS PCI controller design. If you installed 4 - 1MB memory SIMM, that is, total 4 MB memory installed, you will get same result. PCI/I-P54SP4 makes all BSD based O.S. having same problem. See www-itg.lbl.gov/ISS/performance.ps (p6-7) for more details. This message (and the reference to problems with any four-SIMM memory configuration) swayed us in the direction of getting a motherboard with a non-SiS chipset (cheaper than a pair of 32MB SIMMS). We purchased an ASUSTek P/I-P55TP4XEG/90 motherboard, moved the CPU and plunked in the 64MB. Everything's working fine now (except for this disk problem that has cropped up - I tell ya, if it's not one thing, it's another...) Thanks to Jin Guojun for the solution. Moral of the story: stay away from the SiS chipset, at least for large mmory configurations running BSDish operating systems. One other thing worth noting. While ASUS apparently makes good products and has a pretty good Web site (www.asus.com.tw), they were absolutely no help in this situation. When I spoke with a tech support person on the phone he was no help and didn't seem particularly interested in finding a solution or even assigning a problem report number and taking a description of the problem. It was something like, "Gee, we've never heard of any problems like that with our products. Sorry. The system should cache 64MB of memory okay. Bye." There wasn't even any attempt to suggest other possible sources for the problem. Furthermore, mail sent to either of their tech support email addresses (one to the Taiwan address and two to the US address listed on their Web site - not including this message) never received replies - not even an autoresponder like those used by companies like BSDi and Netscape. signature is omitted --------------------------------------------------- -- /-------------- Jin Guojun ------------ v ---- Internet: g_jin@lbl.gov ----\ | Imaging & Distributed Computing | Usenet: ucbvax!g_jin@lbl.gov | | Lawrence Berkeley Laboratory | Bitnet: -- | | 50B-2239, Berkeley, CA 94720 - jin%george.lbl.gov@Csa3.LBL.Gov |