Re: [PATCH] proposed scheduler enhancements and fixes

From: Lawrence Manning (lawrence@aslak.demon.co.uk)
Date: Thu Mar 02 2000 - 19:21:55 EST


On Thu, 2 Mar 2000, Andrea Arcangeli wrote:

> This backouts my change and it fix dbench numbers, please Linus apply to
> 2.3.49pre2:
>
> --- 2.3.49pre2/mm/page_alloc.c Sun Feb 27 06:19:45 2000
> +++ /tmp/page_alloc.c Thu Mar 2 03:57:21 2000
> @@ -337,7 +337,7 @@
> zone_t *zone;
> int i;
>
> - sum = nr_lru_pages - atomic_read(&page_cache_size);
> + sum = nr_lru_pages;
> for (i = 0; i < NUMNODES; i++)
> for (zone = NODE_DATA(i)->node_zones; zone <= NODE_DATA(i)->node_zones+ZONE_NORMAL; zone++)
> sum += zone->free_pages;
 
> Now it's interesting to know the numbers between 2.3.45 + above patch and
> 2.3.46 + above patch, to see how much the new elevator harmed the dbench
> numbers.

Great! Now getting a nice 16-17meg from 2.3.47 + that fix; I haven't
looked at .48 yet.

Because you asked ;) Here's the numbers for 2.3.45+fix and 2.3.46+fix.
All tests were done in the order given, with no activity between.

2.3.45 + patch

Throughput 15.1465 MB/sec (NB=18.9332 MB/sec 151.465 MBit/sec)
Throughput 17.4895 MB/sec (NB=21.8619 MB/sec 174.895 MBit/sec)
Throughput 18.6803 MB/sec (NB=23.3504 MB/sec 186.803 MBit/sec)
Throughput 17.4141 MB/sec (NB=21.7676 MB/sec 174.141 MBit/sec)

2.3.46 + patch

Throughput 16.4924 MB/sec (NB=20.6155 MB/sec 164.924 MBit/sec)
Throughput 18.0024 MB/sec (NB=22.503 MB/sec 180.024 MBit/sec)
Throughput 17.6604 MB/sec (NB=22.0755 MB/sec 176.604 MBit/sec)

(interesting how the numbers increase for each run... probably not
relevatnt to real life though)

Well, the elevator dosn't seem to do any harm to dbench! At least on my
machine, which makes me very happy to see.

> Certainly I'll keep doing my best to make the elevator better to improve
> global performances too.)

Fabulous :)

> [stuff I don't really understand skipped... one day...]

I will get 2.3.48 ASAP and start hammering on it, but it's looking very
solid here. Thank you all for your great work!

Lawrence

-
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/



This archive was generated by hypermail 2b29 : Tue Mar 07 2000 - 21:00:13 EST