On Tue, Sep 26, 2000 at 12:14:18AM +0200, Andrea Arcangeli wrote:
> On Mon, Sep 25, 2000 at 10:52:08PM +0200, Peter Osterlund wrote:
> > Do you know why? Is it because the average seek distance becomes
>
> Good question. No I don't know why right now. I'll try again just to be 200%
> sure and I'll let you know the results.
These are the numbers produced by my current blkdev tree based on
test8-pre5 with only the spinlock-1 patch on it:
-----------------------------------------------------------------------------
2.4.0-test8-pre5 + blkdev-1 - IA32 2-way SMP LVM-stripe IDE
File Block Num Seq Read Rand Read Seq Write Rand Write
Dir Size Size Thr Rate (CPU%) Rate (CPU%) Rate (CPU%) Rate (CPU%)
------- ------ ------- --- ----------- ----------- ----------- -----------
. 254 4096 1 16.38 7.99% 0.647 1.16% 15.60 14.8% 1.330 5.53%
. 254 4096 2 16.34 10.9% 0.676 1.12% 15.70 17.2% 1.330 5.95%
. 254 4096 4 16.30 10.9% 0.690 1.07% 15.55 17.9% 1.324 6.24%
. 254 4096 8 15.71 12.1% 0.713 1.06% 15.11 17.8% 1.327 6.01%
File Block Num Seq Read Rand Read Seq Write Rand Write
Dir Size Size Thr Rate (CPU%) Rate (CPU%) Rate (CPU%) Rate (CPU%)
------- ------ ------- --- ----------- ----------- ----------- -----------
. 254 4096 1 16.41 7.82% 0.716 0.82% 15.91 14.6% 1.334 4.86%
. 254 4096 2 16.44 11.1% 0.715 0.91% 15.82 17.1% 1.316 4.67%
. 254 4096 4 16.39 10.9% 0.722 0.95% 15.52 17.9% 1.322 5.07%
. 254 4096 8 16.02 11.8% 0.742 0.99% 15.13 17.8% 1.329 5.06%
andrea@laser:/mnt/p > ~/dbench/dbench 40
40 clients started
.......................................................................................................................................................................................................................................................................................+.............................................................................................................................................................................................................................................................................+...................................+...+..+.........+............................................................................................................................................................+....................+..............................................................++...................................................................................+................+..........................................................+.................+
+...........................................................++....................................................................................+++...............................+................................................+....+....++..............+...........+...............................................................................+.........................+..........+.......................+........................+...............+.....+.+...+.+.....++.........................+****************************************
Throughput 10.7262 MB/sec (NB.4077 MB/sec 107.262 MBit/sec)
andrea@laser:/mnt/p > ~/dbench/dbench 40
40 clients started
...................................................................................................................................................................................................................................................................................................................................................................................................................+...............................+.+.+..........................................................+..........................................................................................+...........................................+.+...................................................................................................+.....................++...................++.............................................................................................................................................................++.......................................................................+............
..........................................................................................................................................................+.............................................................+...............+...........................+..+....+.........+.+............+....+............+............+.......+.......................+............+.............++...........................+.......+...........+...................................................+...+.+.....+****************************************
Throughput 11.7624 MB/sec (NB.703 MB/sec 117.624 MBit/sec)
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
200 7406 100.0 18626 16.0 8324 12.0 6897 88.0 18180 17.3 239.7 2.0
200 7405 99.9 18724 16.2 8153 11.9 6919 88.2 17827 18.3 231.2 2.1
-----------------------------------------------------------------------------
These are instead the numbers that I get by only applying your patch on
top of my tree:
-----------------------------------------------------------------------------
blkdev-1 (same as above) + elevator change proposed by Peter - IA32 2-way SMP LVM-stripe IDE
File Block Num Seq Read Rand Read Seq Write Rand Write
Dir Size Size Thr Rate (CPU%) Rate (CPU%) Rate (CPU%) Rate (CPU%)
------- ------ ------- --- ----------- ----------- ----------- -----------
. 254 4096 1 16.22 9.13% 0.707 1.13% 16.05 15.3% 1.384 5.05%
. 254 4096 2 11.04 7.99% 0.730 1.26% 16.16 17.2% 1.374 5.76%
. 254 4096 4 10.28 7.33% 0.749 1.11% 15.89 17.8% 1.358 5.67%
. 254 4096 8 9.726 7.36% 0.769 1.02% 15.64 18.2% 1.369 5.67%
andrea@laser:/mnt > tiobench.pl
No size specified, using 254 MB
Size is MB, BlkSz is Bytes, Read, Write, and Seeks are MB/sec . -T
File Block Num Seq Read Rand Read Seq Write Rand Write
Dir Size Size Thr Rate (CPU%) Rate (CPU%) Rate (CPU%) Rate (CPU%)
------- ------ ------- --- ----------- ----------- ----------- -----------
. 254 4096 1 16.50 9.22% 0.679 1.47% 16.23 15.0% 1.376 4.67%
. 254 4096 2 11.50 7.92% 0.682 1.26% 16.10 17.0% 1.395 5.67%
. 254 4096 4 10.23 6.92% 0.723 1.18% 15.91 17.8% 1.392 6.29%
. 254 4096 8 9.796 7.28% 0.751 1.28% 15.64 18.3% 1.368 5.49%
andrea@laser:/mnt/p > ~/dbench/dbench 40
40 clients started
...............................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................
....................................................................................................................................................................................................................................................................................................................................................................................................................................+.................+..+.+....+..........+++...++++++++++++++++++++++++++++++++****************************************
Throughput 5.78095 MB/sec (NB.22619 MB/sec 57.8095 MBit/sec)
andrea@laser:/mnt/p > ~/dbench/dbench 40
40 clients started
...............................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................
.......................................................................................................................................................................................................................................................................................................................................................................................................................................................................+++++.++++++++++++.+++++++++++++++++++++++****************************************
Throughput 6.14307 MB/sec (NB.67883 MB/sec 61.4307 MBit/sec)
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
200 7403 100.0 19425 16.7 8514 12.4 6693 85.1 17925 16.8 239.2 2.5
200 7409 99.9 19282 17.2 8487 11.2 6735 85.7 17930 17.1 238.4 2.5
-----------------------------------------------------------------------------
So the patch won't be included for now. The unfairness of the current
ordering algorithm is probably why it makes the numbers so better than
a perfect CSCAN that your patch implements.
I also want to make an update, the current elevator seems to perform
very well.
My current tree on scsi delivers 8mbyte/sec more for reads and a 30%
improvement in the seeks but that's because of further improvements not because
of final fixes or final reworks. The EXCLUSIVE wakeup thing seems not to make
difference in numbers.
Andrea
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Sep 30 2000 - 21:00:18 EST