arla-0.38 on FreeBSD/amd64

[Love]

Garrett Wollman <wollman at khavrinen.lcs.mit.edu> writes:

> <<On Thu, 6 Jan 2005 19:24:51 -0500 (EST), I wrote:
>
>> There still remain a number of LP64 and other type bugs in the 0.38
>> release.
>
> One other thing I noticed: utimes() seems to be broken.  Files
> extracted from an archive (for example) had their mtimes set to
> (time_t)0xffffffff, indicating some sort of sign-extension and/or
> endianness bug.

Indeed, I'll check the other BSD if you patch is ok and if there is a
better way of solving it.

> Performance seems to be a bit lacking, even compared to OpenAFS.  I'm
> not sure how much of that is due to the server, and how much is due to
> the cache manager not trying hard enough to hold on to cached data.
> I'm running some benchmarks while watching with "amon" and never see
> more than two workers running simultaneously.

What types of benchmarks are you running ? We have the last few years
concentrated on maintenance, bugfixes and functionallity. I have no doubt
there could be improvements done to performance.

You'll see several threads being used at the same time if you run multiple
instances of the benchmark at the same time, preferably in diffrent
directories so they don't case lock-stepping.

> Any advice as to whether lwp or pthreads is preferred, going forward?
> pthreads is at least SMP-capable.

The pthread code in arla is to emulate lwp-threads, so it have have the
same problem as LWP threads, even when using it on a smp machine.

Love


-------------- next part --------------
A non-text attachment was scrubbed...
Name: pgp00002.pgp
Type: application/octet-stream
Size: 823 bytes
Desc: "PGP signature"
Url : http://lists.stacken.kth.se/pipermail/arla-drinkers/attachments/00000000/939a427a/pgp00002.obj