From owner-freebsd-hackers Mon May 6 14:43:10 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id OAA12092 for hackers-outgoing; Mon, 6 May 1996 14:43:10 -0700 (PDT) Received: from mail13.digital.com (mail13.digital.com [192.208.46.30]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id OAA12087 for ; Mon, 6 May 1996 14:43:06 -0700 (PDT) Received: from muggsy.lkg.dec.com by mail13.digital.com (5.65v3.2/1.0/WV) id AA19821; Mon, 6 May 1996 17:28:10 -0400 Received: from whydos.lkg.dec.com by muggsy.lkg.dec.com (5.65/DEC-Ultrix/4.3) with SMTP id AA20360; Mon, 6 May 1996 17:28:09 -0400 Received: from localhost (localhost [127.0.0.1]) by whydos.lkg.dec.com (8.6.12/8.6.12) with SMTP id RAA04851 for ; Mon, 6 May 1996 17:32:52 GMT Message-Id: <199605061732.RAA04851@whydos.lkg.dec.com> X-Authentication-Warning: whydos.lkg.dec.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.5 12/11/95 To: hackers@freebsd.org Subject: Pentium Pro impressions Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 06 May 1996 17:32:52 +0000 From: Matt Thomas Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk My faithful 486/66 that I use my as workstation is now a Digital Celebris XL 6150 (aka PentiumPro 150). Note that the PPros that Digital sells do not have 4MB/s PCI->memory limitation. I ran 3 dd (one for each of my disks plus one ttcp over FDDI) and got just under a 12MB/s aggregate transfer rate. FreeBSD 2.1.0-RELEASE #0: Fri May 3 15:57:04 1996 thomas@whydos.lkg.dec.com:/usr/src/sys/compile/DECFDDI CPU: 150-MHz unknown (Pentium-class CPU) Origin = "GenuineIntel" Id = 0x611 Stepping=1 Features=0xf9ff Running ram-speed (on various platforms) gives: Celebris XL 6150 128MB interleaved FreeBSD 2.1.0-RELEASE 49005fb0 0.159 uS/op 6.30e+06 op/S 24.047 Mb/S 8938c0df 0.415 uS/op 2.41e+06 op/S 9.194 Mb/S ASUS TP4P55XE P90, 16MB, 512PB cache FreeBSD 2.1.0-RELEASE 49005fb0 0.242 uS/op 4.13e+06 op/S 15.763 Mb/S 8938c0df 0.092 uS/op 1.08e+07 op/S 41.255 Mb/S DECpc XL 466/d2 32MB, 256KB WT cache FreeBSD 2.1.0-RELEASE 49005fb0 0.307 uS/op 3.26e+06 op/S 12.430 Mb/S 8938c0df 0.124 uS/op 8.07e+06 op/S 30.790 Mb/S Intel OEM 486/50, EISA, 16MB, ??? cache FreeBSD 2.2-960501-SNAP 49005fb0 0.442 uS/op 2.26e+06 op/S 8.634 Mb/S 8938c0df 0.521 uS/op 1.92e+06 op/S 7.318 Mb/S Using a benchmark (and one that I prefer since I compile lots of kernels), is to see how long each system takes to compile a certain config file (in the same order as above): 136.4u 25.8s 3:05.32 87.5% 955+1114k 1006+431io 26pf+0w 275.5u 26.5s 5:21.17 94.0% 1039+1204k 1270+459io 119pf+0w 577.7u 54.9s 10:51.16 97.1% 1019+1192k 997+418io 102pf+0w 655.2u 76.5s 12:39.86 96.3% 913+1120k 1217+560io 126pf+0w Another interesting test is to a simply ttcp to localhost and see what the effective transfer rate is. (this tests both CPU and memory bandwidth; same order as above): ttcp-r: 16777216 bytes in 0.61 real seconds = 209.52 Mbit/sec +++ ttcp-r: 16777216 bytes in 1.51 real seconds = 84.53 Mbit/sec +++ ttcp-r: 16777216 bytes in 3.09 real seconds = 41.37 Mbit/sec +++ ttcp-r: 16777216 bytes in 2.76 real seconds = 46.30 Mbit/sec +++ My impression so far is that the P6-150 is much faster than the P90 for compute intensive tasks (like gcc). It blows away the old 486/66dx2. exmh is visibly faster as are most X11 programs. I'm one happy camper... -- Matt Thomas Internet: matt@3am-software.com 3am Software Foundry WWW URL: http://www.3am-software.com/bio/matt.html Westford, MA Disclaimer: I disavow all knowledge of this message