From owner-freebsd-bugs Tue May 2 10:47:11 1995 Return-Path: bugs-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id KAA11002 for bugs-outgoing; Tue, 2 May 1995 10:47:11 -0700 Received: from gndrsh.aac.dev.com (gndrsh.aac.dev.com [198.145.92.241]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id KAA10969 for ; Tue, 2 May 1995 10:46:38 -0700 Received: (from rgrimes@localhost) by gndrsh.aac.dev.com (8.6.11/8.6.9) id KAA01564; Tue, 2 May 1995 10:40:51 -0700 From: "Rodney W. Grimes" Message-Id: <199505021740.KAA01564@gndrsh.aac.dev.com> Subject: Re: 53-MHz Pentium To: wollman@halloran-eldar.lcs.mit.edu (Garrett Wollman) Date: Tue, 2 May 1995 10:40:50 -0700 (PDT) Cc: hsu@clinet.fi, freebsd-bugs@freefall.cdrom.com, bde@FreeBSD.org In-Reply-To: <9505021724.AA00501@halloran-eldar.lcs.mit.edu> from "Garrett Wollman" at May 2, 95 01:24:02 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1823 Sender: bugs-owner@FreeBSD.org Precedence: bulk > > < said: > > > Just wondering what is a 53-MHz Pentium 510? The motherboard is made > > by MSI, CPU was swapped a week ago to bugfree version (but it didn't > > affect this problem). > > The speed of a Pentium chip is calculated by examining the cycle > counter register, DELAY(1000000), and then subtracting the old cycle > counter reading from the new and rounding to the nearest million. So, > your reading indicates that either your CPU clock is running slower > than its nominal rate (I assume it's supposed to be 60-MHz), or the > DELAY function is running too fast. > > My 60-MHz machine (Intel Premiere OEM system) correctly reports the > speed and processor type: > > cpu0 Busy Pentium 510\60 Paul Richards and I have both done a some testing here with some code Bruce Evans sent out. There is infact a problem in DELAY, and I have seen several systems when repeatedly rebooted report values from 83 to 90 Mhz (this is a 90Mhz machine). Paul was able to duplicate it on his system. I need to go spend some more time looking at the code, but it has been low priority for me as it only happens about 1 in 5 boots. We do have a serious problem for DELAY(N) N<20 on these fast machines, Paul measure this using Bruce program and found it to be almost an order of magnitude in error :-(. It was also found that the DELAY skew value that I put into the floppy code to fix a floppy tape problem is almost exactly what Paul measured to be the error in DELAY(N) for N > 50. Bruce, do you have access to a P54C-90 system? Or should I go and try and fix this problem? -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Custom computers for FreeBSD