Messages in this thread | | | From | cd_smith@ou ... | Date | Sun, 1 Aug 1999 11:11:59 -0500 (CDT) | Subject | Results from Artur Skawini's SCHED_IDLE patch |
| |
Hi,
Here's some test results from Artur Skawini's SCHED_IDLE patch on 2.3.13-pre1. What I did was run both bonnie and lmbench on a stock 2.3.13-pre1 kernel to get a reference point, and then run each on 2.3.13-pre1 + SCHED_IDLE, with the other test running in the background as an idle task.
The lmbench results came out just about the same... Based on my observations from running lmbench a lot yesterday, I think all the variations fit well within the variation I got running lmbench multiple times on the same kernel. Bonnie performance dropped considerably, though, when lmbench was running in the background. This is most likely because compared to lmbench, bonnie tends to use a lot less CPU time. So lmbench had a chance to get in there and pollute cache for the bonnie task. This is an expected result, and the only way to fix it reasonably is to not run idle tasks.
Now for the full results:
Bonnie on straight 2.3.13-pre1, no other active tasks:
-------Sequential Output-------- ---Sequential Input-- --Random-- -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks--- MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU 500 1853 86.4 4125 14.2 2581 14.4 1882 86.7 8053 32.5 51.5 3.7
Bonnie on 2.3.13-pre1 + SCHED_IDLE, with lmbench in the background:
-------Sequential Output-------- ---Sequential Input-- --Random-- -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks--- MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU 500 1785 84.0 4094 14.7 2193 13.4 1857 85.4 7571 31.0 45.9 3.7
lmbench results, first two runs are on straight 2.3.13-pre1 with no other active tasks. Second two rows are on 2.3.13-pre1 + SCHED_IDLE, with bonnie in the background.
L M B E N C H 1 . 9 S U M M A R Y ------------------------------------ (Alpha software, do not distribute)
Processor, Processes - times in microseconds - smaller is better ---------------------------------------------------------------- Host OS Mhz null null open selct sig sig fork exec sh call I/O stat clos inst hndl proc proc proc --------- ------------- ---- ---- ---- ---- ---- ----- ---- ---- ---- ---- ---- i586-linu Linux 2.3.13 166 0.9 1.5 15 20 0.11K 4.4 6 1.4K 6K 31K i586-linu Linux 2.3.13 166 0.9 1.5 15 22 0.11K 4.3 6 1.6K 6K 31K i586-linu Linux 2.3.13 166 0.9 1.6 18 28 0.11K 4.4 6 1.5K 6K 32K i586-linu Linux 2.3.13 166 0.9 1.6 15 21 0.11K 4.4 6 1.5K 6K 32K
Context switching - times in microseconds - smaller is better ------------------------------------------------------------- Host OS 2p/0K 2p/16K 2p/64K 8p/16K 8p/64K 16p/16K 16p/64K ctxsw ctxsw ctxsw ctxsw ctxsw ctxsw ctxsw --------- ------------- ----- ------ ------ ------ ------ ------- ------- i586-linu Linux 2.3.13 7 82 348 127 568 151 614 i586-linu Linux 2.3.13 6 83 384 128 575 148 607 i586-linu Linux 2.3.13 9 76 490 109 573 142 597 i586-linu Linux 2.3.13 8 82 355 123 548 149 607
*Local* Communication latencies in microseconds - smaller is better ------------------------------------------------------------------- Host OS 2p/0K Pipe AF UDP RPC/ TCP RPC/ TCP ctxsw UNIX UDP TCP conn --------- ------------- ----- ----- ---- ----- ----- ----- ----- ---- i586-linu Linux 2.3.13 7 24 65 130 213 677 i586-linu Linux 2.3.13 6 25 72 129 216 662 i586-linu Linux 2.3.13 9 29 67 148 201 672 i586-linu Linux 2.3.13 8 27 75 131 198 663
File & VM system latencies in microseconds - smaller is better -------------------------------------------------------------- Host OS 0K File 10K File Mmap Prot Page Create Delete Create Delete Latency Fault Fault --------- ------------- ------ ------ ------ ------ ------- ----- ----- i586-linu Linux 2.3.13 36 6 67 12 1271 2 0.0K i586-linu Linux 2.3.13 36 6 67 12 1334 2 0.0K i586-linu Linux 2.3.13 36 6 67 12 1300 2 1.4K i586-linu Linux 2.3.13 36 6 66 12 1306 2 0.0K
*Local* Communication bandwidths in MB/s - bigger is better ----------------------------------------------------------- Host OS Pipe AF TCP File Mmap Bcopy Bcopy Mem Mem UNIX reread reread (libc) (hand) read write --------- ------------- ---- ---- ---- ------ ------ ------ ------ ---- ----- i586-linu Linux 2.3.13 38 24 16 31 84 43 42 84 86 i586-linu Linux 2.3.13 34 24 16 33 84 42 42 84 86 i586-linu Linux 2.3.13 37 23 16 34 84 42 42 84 86 i586-linu Linux 2.3.13 35 23 17 33 84 42 42 84 86
Memory latencies in nanoseconds - smaller is better (WARNING - may not be correct, check graphs) --------------------------------------------------- Host OS Mhz L1 $ L2 $ Main mem Guesses --------- ------------- --- ---- ---- -------- ------- i586-linu Linux 2.3.13 166 12 96 146 i586-linu Linux 2.3.13 166 12 102 146 i586-linu Linux 2.3.13 166 12 98 146 i586-linu Linux 2.3.13 166 12 93 146
Chris Smith <cd_smith@ou.edu>
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |