Greetings,
I posted this also on the FedoraForums in case this problem is specific to Fedora.

I'm seeing a very unusual drop in performance with sql-bench, in particular the "create" benchmark with MySQL between 5.1.47 and 5.1.55. Anyone else seeing this or is this a known difference between the releases?

In my case, the main differences between the two systems is one if Fedora 12 and the other is Fedora 13. I can also verify the issue in Fedora 14. All being 64-bit distros.

Any tips, insights or suggestions appreciated! More importantly, if you have an older/newer kernel with MySQL 5.1.55 that doesn't have the horrible "create" times for mysql-bench, I'd love to hear from you.

Primary results of run-all-test in the mysql-bench benchmark (full results available if desired):

Fedora 12 system:
=================
# dd if=/dev/zero of=/var/lib/mysql/bench.raw oflag=direct bs=64k count=10k
10240+0 records in
10240+0 records out
671088640 bytes (671 MB) copied, 10.9853 s, 61.1 MB/s

Benchmark DBD suite: 2.15
Date of test: 2011-02-23 9:17:38
Running tests on: Linux 2.6.32.26-175.fc12.x86_64 x86_64
Arguments:
Comments:
Limits from:
Server version: MySQL 5.1.47
Optimization: None
Hardware:

alter-table: Total time: 6 wallclock secs ( 0.03 usr 0.00 sys + 0.00 cusr 0.00 csys = 0.03 CPU)
ATIS: Total time: 7 wallclock secs ( 2.92 usr 0.11 sys + 0.00 cusr 0.00 csys = 3.03 CPU)
big-tables: Total time: 5 wallclock secs ( 1.40 usr 0.24 sys + 0.00 cusr 0.00 csys = 1.64 CPU)
connect: Total time: 47 wallclock secs (14.99 usr 8.02 sys + 0.00 cusr 0.00 csys = 23.01 CPU)
create: Total time: 43 wallclock secs ( 1.37 usr 0.72 sys + 0.00 cusr 0.00 csys = 2.09 CPU)
insert: Total time: 416 wallclock secs (155.83 usr 30.66 sys + 0.00 cusr 0.00 csys = 186.49 CPU)
select: Total time: 156 wallclock secs (16.01 usr 1.09 sys + 0.00 cusr 0.00 csys = 17.10 CPU)
transactions: Test skipped because the database doesn't support transactions
wisconsin: Total time: 4 wallclock secs ( 1.24 usr 0.38 sys + 0.00 cusr 0.00 csys = 1.62 CPU)

Fedora 13 System:
===============
# dd if=/dev/zero of=/var/lib/mysql/bench.raw oflag=direct bs=64k count=10k
10240+0 records in
10240+0 records out
671088640 bytes (671 MB) copied, 10.68 s, 62.8 MB/s

Benchmark DBD suite: 2.15
Date of test: 2011-02-23 13:06:02
Running tests on: Linux 2.6.34.7-56.fc13.x86_64 x86_64
Arguments:
Comments:
Limits from:
Server version: MySQL 5.1.55
Optimization: None
Hardware:

alter-table: Total time: 21 wallclock secs ( 0.04 usr 0.02 sys + 0.00 cusr 0.00 csys = 0.06 CPU)
ATIS: Total time: 13 wallclock secs ( 3.90 usr 0.22 sys + 0.00 cusr 0.00 csys = 4.12 CPU)
big-tables: Total time: 10 wallclock secs ( 2.66 usr 0.42 sys + 0.00 cusr 0.00 csys = 3.08 CPU)
connect: Total time: 77 wallclock secs (26.81 usr 11.55 sys + 0.00 cusr 0.00 csys = 38.36 CPU)
create: Total time: 1968 wallclock secs ( 3.04 usr 1.10 sys + 0.00 cusr 0.00 csys = 4.14 CPU)
insert: Total time: 602 wallclock secs (237.14 usr 37.98 sys + 0.00 cusr 0.00 csys = 275.12 CPU)
select: Total time: 205 wallclock secs (22.00 usr 2.29 sys + 0.00 cusr 0.00 csys = 24.29 CPU)
transactions: Test skipped because the database doesn't support transactions
wisconsin: Total time: 4 wallclock secs ( 1.34 usr 0.32 sys + 0.00 cusr 0.00 csys = 1.66 CPU)

Both systems have identical my.cnf, both use EXT4 filesystem and both are 2xSATA drives with Software RAID1 (2 partitions, 200MB /boot and remainder LVM)