From owner-freebsd-hardware Mon Jun 9 19:03:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA00111 for hardware-outgoing; Mon, 9 Jun 1997 19:03:33 -0700 (PDT) Received: from www2.shoppersnet.com (shoppersnet.com [204.156.152.112]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA29999 for ; Mon, 9 Jun 1997 19:03:21 -0700 (PDT) Received: (from hlew@localhost) by www2.shoppersnet.com (8.8.5/8.8.5) id TAA20786; Mon, 9 Jun 1997 19:12:07 -0700 (PDT) Date: Mon, 9 Jun 1997 19:12:07 -0700 (PDT) From: Howard Lew To: Craig Johnston cc: hardware@freebsd.org Subject: Re: poor memory bandwidth on ABIT IT5H rev 1.5 In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-hardware@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Fri, 6 Jun 1997, Craig Johnston wrote: > > Interesting -- > > 2.2.2 on the ABIT with an AMD K5-PR166 gives a result of > 60 megs/sec maximum via the naive memory bandwidth benchmark: > dd if=/dev/zero of=/dev/null bs=1m count=100 > > while 2.2.2 on an Asus mobo (the HX one, t2p4? or something?) > gives 86 meg/sec by the same benchmark. > > The ABIT board has EDO dram @60ns, the Asus FPM @ 60ns. > > Both kernels have 'flags 0x7' for npx0 -- as their slower (than intel) > FPUs don't speed up memory moves. > > I'd be extremely interested in seeing your results for 2.2.2 with > the K5 and the above flags on the ABIT and other motherboards. > > Anyone else have reason to believe the ABITs memory bandwidth > sucks? > I thought the TX was better than the VX, so I compared several motherboards with dd if=/dev/zero of=/dev/null bs=1m count=1000 CPU is AMD K5-PR133 FIC PT2006 (Intel VX) with 256K PB Cache & FreeBSD-2.1.7.1R Got 65MB/s FreeTech F63T (Intel VX) with 256K PB Cache & FreeBSD-2.1.7.1R Got 62MB/s FreeTech F79 (Intel TX) with 512K PB Cache & FreeBSD-3.0SNAP-6/6/97 Got 43MB/s Is there something different about 3.0SNAP6/6/97? Is there any option parameter for the kernel config for the K5-PR133? Unfortunately, 2 parameters changed, so I can't tell if it is the MB or the OS. I decided to go with 3.0snap because of this missing TX PCI & IDE drivers. All three tests were done with 32MB EDO memory (set to best memory settings).