distributed.net Pre-Release clients v2.8012.465 now available

[Fardringle]

Well, it would appear that this new client is optimized for Intel CPUs, and not for AMD CPU's, as my AMD took a big hit, but my CeleronII 700 jumped from ~4.7Mnodes/sec on the old client to over 5.5Mnodes/sec with the beta and my Celeron 466 jumped from 3.2Mnodes/sec to ~3.6Mnodes/sec!! I'm definitely going to have to do some more beta testing with these machines!

 

[sciencewhiz]

I haven't had a chance to check it out yet, but I have a question for you people reporting OGR speed increases/decreases.

Are the speed changes you are reporting (ie my client increases from 3.3 mnodes/s to 4.0 mnodes/s) based on comparisons with the 463 client or the 464 client. The 464 client used a new OGR core that increased speed for PIIIs and decreased the speed on K6 series processors. Are these changes on top of the 464 changes or are they just a consequence of that core being included in the new client.

 

[Viztech]

As far as some of the earlier problems with the 464 beta versions, Engineer and I experienced troubles from clients that were behind an M$ Proxy2 firewall that we were the first to report. Cyp worked quite hard to resolve these issues and sent out some alpha versions to only us to test the fixes. That all worked out nicely for us and Dnet.

Like Moose said, not all of us are in the position to beta test clients, but those of us that do are doing everyone a favor.

I am quite interested to see what the differences are as well in the OGR speeds. Many of us have been using the now discontinued 464g Beta that had a ~15% OGR speed increase over the 463b release client, but only with Intel P2/P3/Celeron CPUs. AMD Athlon/Duron/K6 saw a slight drop with that Beta, so we were hopefull of some improvements there.

viz

 

[imported_Thunder]

I don't have the time right now to try and mess with the pre-release, but I am keenly interested in the OGR speeds on P6 class processors. I'm still using the 8010.463b clients because I wasn't quite sure if I was in a position to deal with the potential headaches that came with the .464 series of pre-release clients.

Most of my machines are on a corporate network and need to run totally unattended for long periods of time reliably. That means that for me, pre-release = (to paraphrase the crocodile hunter) danger, danger, danger.

To those daring souls that are trying this stuff out; thanks! and keep us informed of what you learn.

-Brian

 

[crYnOid]

my K6-3 450 seems slower in OGR. With normal use it is normally about 2,900,000nodes/sec (463b), but with 465 it is down to 2,700,000nodes/sec


Edit: i just benchmarked this cpu with the rc5 core, and it looks like it is about 20,000-30,000 keys/sec faster

 

[Fardringle]

The changes in speed I have reported are in comparison to the 463 version of the client...I haven't tried any other beta client, so I don't have a comparison between those and this newest client...