====== Benchmark Linux vs. Solaris ======

Ce benchmark a été réalisé sur 2 UltraSparc 1, cadencées à 166Mhz et disposant de 128Mo de mémoire et d'un disque SCSI 4Go.

La première fonctionnant sous Solaris 2.6 livré préinstallé.

La seconde fonctionnant sous [[http://ultra.linux.cz/|UltraPenguin 1.0.9]] une distribution Linux pour UltraSparc basé sur la [[http://www.redhat.com|RedHat5.0]], le noyau testé est l'image CVS du noyau ce trouvant sur vger.rutgers. edu du 28 Avril 1998 (Linux-2.1.98).

Le benchmark utilisé est //BYTE UNIX Benchmarks (Version 3.11)//

^ Tests ^^^ Linux-2.1.98 ^^^ Solaris 2.6 ^^
^ Dhrystone 2 without register variables ^^| 325220.2 lps ||| 304253.9 lps ||
^ Dhrystone 2 using register variables ^^| 323780.6 lps ||| 304414.8 lps ||
^ Arithmetic Test (type = arithoh) ^^| 1370507.6 lps ||| 1371915.9 lps ||
^ Arithmetic Test (type = register) ^^| 21957.1 lps ||| 19271.3 lps ||
^ Arithmetic Test (type = short) ^^| 21129.5 lps ||| 18915.1 lps ||
^ Arithmetic Test (type = int) ^^| 21818.2 lps ||| 19280.6 lps ||
^ Arithmetic Test (type = long) ^^| 21980.7 lps ||| 19316.0 lps ||
^ Arithmetic Test (type = float) ^^| 51759.1 lps ||| 51460.4 lps ||
^ Arithmetic Test (type = double) ^^| 40460.7 lps ||| 40229.8 lps ||
^ System Call Overhead Test ^^| 59465.6 lps ||| 53604.2 lps ||
^ Pipe Throughput Test ^^| 59768.2 lps ||| 38769.7 lps ||
^ Pipe-based Context Switching Test ^^| 31745.6 lps ||| 16801.4 lps ||
^ Process Creation Test ^^| 2374.0 lps ||| 343.7 lps ||
^ Execl Throughput Test ^^| 1557.8 lps ||| 163.4 lps ||
^ File Read (10 seconds) ^^| 269467.0 KBps ||| 115662.0 KBps ||
^ File Write (10 seconds) ^^| 57573.0 KBps ||| 8699/0 KBps ||
^ File Copy (10 seconds) ^^| 7912.0 KBps ||| 8855.0 KBps ||
^ File Read (30 seconds) ^^| 269704.0 KBps ||| 118590.0 KBps ||
^ File Write (30 seconds) ^^| 56604.0 KBps ||| 8589.0 KBps ||
^ File Copy (30 seconds) ^^| 6985.0 KBps ||| 8780.0 KBps ||
^ C Compiler Test ^^| 210.5 lpm ||| NA ||
^ Shell scripts (1 concurrent) ^^| 329.4 lpm ||| 158.3 lpm ||
^ Shell scripts (2 concurrent) ^^| 175.7 lpm ||| 87.0 lpm ||
^ Shell scripts (4 concurrent) ^^| 91.0 lpm ||| 47.0 lpm ||
^ Shell scripts (8 concurrent) ^^| 46.0 lpm ||| 24.0 lpm ||
^ Dc: sqrt(2) to 99 decimal places ^^| 7741.6 lpm ||| 5954.8 lpm ||
^ Recursion Test--Tower of Hanoi ^^| 5988.7 lps ||| 5937.5 lpm ||
^ INDEX VALUES ^^^^^^
^ TEST ^ BASELINE ^ RESULT ^ INDEX ^ RESULT ^ INDEX ^
^ Arithmetic Test (type = double) | 2541.7 | 40460.7 | 15.9 | 40229.8 | 15.8 |
^ Dhrystone 2 without register variables | 22366.3 | 325220.2 | 14.5 | 304253.9 | 13.6 |
^ Execl Throughput Test | 16.5 | 1557.8 | 94.4 | 163.4 | 9.9 |
^ File Copy (30 seconds) | 179.0 | 6985.0 | 39.0 | 8780.0 | 49.1 |
^ Pipe-based Context Switching Test | 1318.5 | 31745.6 | 24.1 | 16801.4 | 12.7 |
^ Shell scripts (8 concurrent) | 4.0 | 46.0 | 11.5 | 24.0 | 6.0 |
^ SUM of 6 items | 199.5 |||| 107.1 ||
^ AVERAGE | 33.2 |||| 17.9 ||

Les résultats parlent d'eux même je pense, comme je suis pas un expert je laisse l'interpretation à d'autres ! ;)

----
 Stéphane Casset / [[mailto:casset@erm6.u-strasbg.fr| casset@erm6.u-strasbg.fr ]] 

Last Modified : \\