On my Athlon XP (SSE, 1733 MHz, 256 kB L2 cache) times are down from ~21,900 to ~12,950 seconds.
On my dual-core Opteron 165 (SSE3, 1800 MHz, 1MB L2 cache per core) times went down from ~18,600 to ~6,240.
What have you been using before? The old 0.40 FreeBSD Beta or a Linux App?
On my Athlon XP (SSE, 1733 MHz, 256 kB L2 cache) times are down from ~21,900 to ~12,950 seconds.
On my dual-core Opteron 165 (SSE3, 1800 MHz, 1MB L2 cache per core) times went down from ~18,600 to ~6,240.
What have you been using before? The old 0.40 FreeBSD Beta or a Linux App?
That's a comparision against old 0.40 FreeBSD core.
I've updated the FreeBSD App on the Beta Test Page, keeping the same version number. The md5sum of the new package is d53fe3cb16bbf00a9e8ce5d824fe27f9 . It was built as a static binary on a FreeBSD 5.5-PRERELEASE system and also tested on a FreeBSD 6 machine. Have a look.
Segfaults on amd64 architecture. The static binary advice I gave you turned out bad, I forgot that pthread is not portable between i386 and amd64 in this manner. My apology for that.
Could I, perhaps, ask you for dynamic binary built on that 5.5-PRERELEASE box?
It just so happened that I expected this - the page has been updated again, new md5sum for the dynamic binary package is d21630ac6abe657837db8e8709980869.
RE: On my Athlon XP (SSE,
)
What have you been using before? The old 0.40 FreeBSD Beta or a Linux App?
BM
BM
RE: RE: On my Athlon XP
)
That's a comparision against old 0.40 FreeBSD core.
Now I have four results which
)
Now I have four results which validated and credit was granted.
I've updated the FreeBSD App
)
I've updated the FreeBSD App on the Beta Test Page, keeping the same version number. The md5sum of the new package is d53fe3cb16bbf00a9e8ce5d824fe27f9 . It was built as a static binary on a FreeBSD 5.5-PRERELEASE system and also tested on a FreeBSD 6 machine. Have a look.
BM
BM
Seems to run on i386
)
Seems to run on i386 architecture.
Segfaults on amd64 architecture. The static binary advice I gave you turned out bad, I forgot that pthread is not portable between i386 and amd64 in this manner. My apology for that.
Could I, perhaps, ask you for dynamic binary built on that 5.5-PRERELEASE box?
It just so happened that I
)
It just so happened that I expected this - the page has been updated again, new md5sum for the dynamic binary package is d21630ac6abe657837db8e8709980869.
BM
BM
RE: It just so happened
)
Hehe, good work with Precog.
The dynamic binary works fine on i386, on amd64 with the usual pthread hack. I'll update the official FreeBSD port with it.
Thanks a lot again for supporting us!
Do you have any idea whether
)
Do you have any idea whether this will work on FreeBSD prior to 5.5-Pre? Probably on 5.4, but what about 5.3 or 5.2? Anyone still using these systems?
BM
BM
RE: Do you have any idea
)
It should just work on 5.3 and above, the 5.3+ line of code is ABI compatible.
It will not run on 5.2.1. But hardly anyone is using 5.2.1 today, it was AWOLed some time ago.
BTW: Thanks to Jürgen
)
BTW: Thanks to Jürgen Nickelsen in Berlin and Gerrit Kühn in Hannover for providing machines and support
BM
BM