Path: news.apana.org.au!goliath.apana.org.au!sysx.apana.org.au!sleeper.apana.org.au!greathan.apana.org.au!news-mail-gateway From: fluido@marktest.pt (Carlo Emilio Prelz) Newsgroups: apana.lists.os.linux.kernel Subject: Benchmarks - revs 1.3.50 & 51 Date: Thu, 28 Dec 1995 08:05:17 +0100 (MET) Organization: Mail-to-News Gateway Lines: 158 Sender: daemon@greathan.apana.org.au Approved: usenet@greathan.apana.org.au Distribution: apana Message-ID: <Pine.LNX.3.91.951228075225.11217A-100000@pimpinel.fluido.org> NNTP-Posting-Host: greathan.apana.org.au Hi. Here follow the benchmark comparisons w.r.t. revisions 50 & 51. I decided to wait before posting rev 50 before because the file read & write results again bounced up with it, and I spent some (very little) time trying to see if I could have done anything to cause that. With no results, as usual. Now I also have results for the "greased weasel", and I decided to send out my numbers for what they're worth. Not much has changed with the latest release... I repeat my query: if anybody has suggestions about these huge jumps in disk performance, please let me know. I am ALWAYS running the benchmark test after a reboot. I login as root, start the process in background and log out, and then generally go to sleep. The machine is NOT online, and is not doing anything important at night. I have not changed anything in the kernel config in the latest 3 revisions. I have a plain old ISA disk controller, and two generally fullish disks, but again, the situation did not change notably between 1.3.49 and 1.3.50. And, I now have to admit that I find it faster, for example, to load a huge C file into emacs, and have it do the syntax coloring. I noticed it this morning. No problems in patching & compiling. --8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<-- From: * Linux pimpinel.fluido.org 1.3.49 #77 Fri Dec 22 21:05:57 MET 1995 i486 * Fri Dec 22 21:41:03 MET 1995 To: * Linux pimpinel.fluido.org 1.3.50 #78 Mon Dec 25 11:18:00 MET 1995 i486 * Mon Dec 25 12:11:11 MET 1995 ******* Results ******* File Write (30 seconds) || 1000.0 -> 9460.0 +846.00% File Write (10 seconds) || 1200.0 -> 8070.0 +572.50% File Copy (30 seconds) || 968.0 -> 2636.0 +172.31% File Copy (10 seconds) || 1024.0 -> 2645.0 +158.30% Execl Throughput Test || 67.9 -> 85.3 +25.63% C Compiler Test || 48.7 -> 51.9 +6.57% Pipe-based Context Switching Test || 9863.0 -> 10493.0 +6.39% Process Creation Test || 347.7 -> 368.9 +6.10% Shell scripts (1 concurrent) || 96.4 -> 101.9 +5.71% Shell scripts (4 concurrent) || 25.5 -> 26.5 +3.92% Shell scripts (2 concurrent) || 50.8 -> 52.6 +3.54% System Call Overhead Test || 29945.4 -> 30523.5 +1.93% Dhrystone 2 using register variables || 49658.9 -> 49910.8 +0.51% Arithmetic Test (type = int) || 8365.0 -> 8377.5 +0.15% Arithmetic Test (type = double) || 5063.6 -> 5071.1 +0.15% Recursion Test--Tower of Hanoi || 726.1 -> 727.1 +0.14% Arithmetic Test (type = arithoh) || 127103.2 -> 127265.6 +0.13% Arithmetic Test (type = short) || 7508.8 -> 7515.5 +0.09% Arithmetic Test (type = register) || 8367.3 -> 8372.8 +0.07% Arithmetic Test (type = float) || 5066.8 -> 5069.3 +0.05% Arithmetic Test (type = long) || 8371.8 -> 8372.1 +0.00% Shell scripts (8 concurrent) || 13.0 -> 13.0 +0.00% Pipe Throughput Test || 17850.0 -> 17726.9 -0.69% Dhrystone 2 without register variables || 51001.9 -> 49772.1 -2.41% File Read (30 seconds) || 4613.0 -> 4404.0 -4.53% File Read (10 seconds) || 4648.0 -> 4390.0 -5.55% Dc: sqrt(2) to 99 decimal places || 11886.7 -> 10822.2 -8.96% --8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<-- From: * Linux pimpinel.fluido.org 1.3.50 #78 Mon Dec 25 11:18:00 MET 1995 i486 * Mon Dec 25 12:11:11 MET 1995 To: * Linux pimpinel.fluido.org 1.3.51 #79 Wed Dec 27 19:16:06 MET 1995 i486 * Wed Dec 27 22:41:23 MET 1995 ******* Results ******* File Read (30 seconds) || 4404.0 -> 4548.0 +3.27% File Read (10 seconds) || 4390.0 -> 4516.0 +2.87% Dc: sqrt(2) to 99 decimal places || 10822.2 -> 11034.2 +1.96% Process Creation Test || 368.9 -> 375.0 +1.65% File Copy (10 seconds) || 2645.0 -> 2683.0 +1.44% File Copy (30 seconds) || 2636.0 -> 2669.0 +1.25% Shell scripts (4 concurrent) || 26.5 -> 26.8 +1.13% C Compiler Test || 51.9 -> 52.4 +0.96% Dhrystone 2 without register variables || 49772.1 -> 49808.7 +0.07% Arithmetic Test (type = long) || 8372.1 -> 8376.6 +0.05% Arithmetic Test (type = register) || 8372.8 -> 8376.1 +0.04% Arithmetic Test (type = float) || 5069.3 -> 5069.9 +0.01% Arithmetic Test (type = short) || 7515.5 -> 7515.7 +0.00% Shell scripts (8 concurrent) || 13.0 -> 13.0 +0.00% Arithmetic Test (type = double) || 5071.1 -> 5067.4 -0.07% Arithmetic Test (type = arithoh) || 127265.6 -> 127172.0 -0.07% Dhrystone 2 using register variables || 49910.8 -> 49870.6 -0.08% Arithmetic Test (type = int) || 8377.5 -> 8370.4 -0.08% Recursion Test--Tower of Hanoi || 727.1 -> 724.3 -0.39% Shell scripts (2 concurrent) || 52.6 -> 52.3 -0.57% Shell scripts (1 concurrent) || 101.9 -> 101.2 -0.69% File Write (30 seconds) || 9460.0 -> 9356.0 -1.10% Pipe Throughput Test || 17726.9 -> 17227.1 -2.82% Pipe-based Context Switching Test || 10493.0 -> 10194.8 -2.84% System Call Overhead Test || 30523.5 -> 29476.8 -3.43% Execl Throughput Test || 85.3 -> 81.5 -4.45% File Write (10 seconds) || 8070.0 -> 7680.0 -4.83% --8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<-- From: * Linux pimpinel.fluido.org 1.2.10 #59 Tue Jun 13 09:46:17 MET DST 1995 i486 * Tue Jun 27 10:18:13 MET DST 1995 To: * Linux pimpinel.fluido.org 1.3.51 #79 Wed Dec 27 19:16:06 MET 1995 i486 * Wed Dec 27 22:41:23 MET 1995 ******* Results ******* Process Creation Test || 119.9 -> 375.0 +212.76% Pipe-based Context Switching Test || 3446.9 -> 10194.8 +195.77% Dc: sqrt(2) to 99 decimal places || 5906.2 -> 11034.2 +86.82% Execl Throughput Test || 51.6 -> 81.5 +57.95% Shell scripts (8 concurrent) || 10.0 -> 13.0 +30.00% Shell scripts (1 concurrent) || 78.3 -> 101.2 +29.25% Shell scripts (4 concurrent) || 21.0 -> 26.8 +27.62% Shell scripts (2 concurrent) || 41.0 -> 52.3 +27.56% C Compiler Test || 42.2 -> 52.4 +24.17% Dhrystone 2 without register variables || 46284.2 -> 49808.7 +7.61% File Write (30 seconds) || 8911.0 -> 9356.0 +4.99% Recursion Test--Tower of Hanoi || 709.2 -> 724.3 +2.13% System Call Overhead Test || 29238.2 -> 29476.8 +0.82% Arithmetic Test (type = float) || 5055.6 -> 5069.9 +0.28% Arithmetic Test (type = long) || 8353.5 -> 8376.6 +0.28% Arithmetic Test (type = register) || 8353.2 -> 8376.1 +0.27% Arithmetic Test (type = short) || 7496.0 -> 7515.7 +0.26% Arithmetic Test (type = double) || 5056.8 -> 5067.4 +0.21% Arithmetic Test (type = int) || 8353.4 -> 8370.4 +0.20% Arithmetic Test (type = arithoh) || 126914.4 -> 127172.0 +0.20% Dhrystone 2 using register variables || 49802.4 -> 49870.6 +0.14% File Write (10 seconds) || 8100.0 -> 7680.0 -5.19% Pipe Throughput Test || 18512.0 -> 17227.1 -6.94% --8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<-- Enjoy! Carlo -- * ...Ma appena fuori tutto e' gomma, * K * Carlo E. Prelz - fluido@marktest.pt tutto e' cicca impiastricciata... * (Marco Zappa-Niente cicca nella scuola) -- +----------------------------------+-----------------------------------------+ |Julian Assange | "if you think the United States has | |FAX: +61-3-9819-9066 | has stood still, who built the largest | |EMAIL: proff@suburbia.net | shopping centre in the world?" - Nixon | +----------------------------------+-----------------------------------------+