• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Boinc pentathalon race

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
A reminder, it's now 40 minutes until the end of the Einstein@home-part, so if you've not already done so, upload & report any finished work ASAP.


Hmm, just finished one Einstein-wu, and is left with a single one. Unfortunately, it's 1h35m from finish, so won't make it. 🙁

I've stock-piled Rosetta-work on the non-GPU-computers, and hopefully they won't start erroring-out...

Also, for anyone that isn't passing on FreeHal (like I'm doing), should start running it now if not already running.
 
How come you're passing on FreeHAL? I thought about it, but it's almost crazy not to run it - unless you're low on RAM. 25 WUs can run per machine, with virtually no CPU, each takes 90 minutes or so and gives you 15 cobblestones. I'll also try one or two virtual machines - I may be able to do up to 60 WUs at a time on this one machine!
 
Update: FreeHAL seems to like to take more memory than it claims it will take: 90MB/process on 64-bit Linux, when it claimed 30. So now I'm running it exclusively in virtual machines on this machine.

That leaves me free to possibly switch to making a Rosetta stockpile. Can you tell me, how much memory does each Rosetta process take?
 
Ken g6 said:
Update: FreeHAL seems to like to take more memory than it claims it will take: 90MB/process on 64-bit Linux, when it claimed 30. So now I'm running it exclusively in virtual machines on this machine.

That leaves me free to possibly switch to making a Rosetta stockpile. Can you tell me, how much memory does each Rosetta process take?
Click to expand...
Quick look, of the Rosetta-tasks currently running, their peak is anything from 260 MB to 390 MB.

Hmm, 90 MB/process * 25 processes is over 2 GB, none of the computers running Rosetta has so much memory that can keep FreeHal satisfied... Even a single instance would likely lead to trashing on one of the computers, since it's only got 1 GB memory...

But in any case, FreeHal doesn't seem to do any useful work at all, so don't see much point to waste resources on this project. Also, they've got a message plastered all over their forums and many of their pages about "don't use that browser, instead 'upgrade' to a so-called 'better' browser instead" 😡 D:
 
Last edited:
You don't *have* to do 25 processes at a time. This laptop I'm on is set for 10. Plus the 32-bit version I'm running in two VMs seems to require less. 30mb free per process plus a decent swapfile, all in a VM, seems to work.

As for not doing any useful work, well that's why I almost didn't participate. And one of a few reasons why I won't when this race is over. 🙂
 
We've dropped to 14th place.

Final Einstein stats: Nice job all! 🙂

stats_day5.png


MilkyWay stats: Great Crunching!! 🙂

MW_day1.png
 
We are getting totally smoked on FreeHal. 🙄

Too bad they didn't place a decent CPU project like Rosetta from now until the finish of the race. :\
 
Well, see, that's why I'm running FreeHAL in a VM. (Actually, two.) I'm going to switch my main system to Rosetta in a couple of hours and disable its networking, but the two VMs can still connect. 🙂

Ubuntu 9.04 32-bit with 894MB RAM is running 25 FreeHALs just fine.
 
Rattledagger said:
Quick look, of the Rosetta-tasks currently running, their peak is anything from 260 KB to 390 KB.
Click to expand...
Yeah...when you said KB you meant MB, didn't you? Each one I'm running has over 300 MB virtual memory, and at least 100MB physical.

Had to shut down one VM - though it wasn't doing much anyway. FreeHAL doesn't seem to be keeping up with demand. 🙁
 
Ken g6 said:
Yeah...when you said KB you meant MB, didn't you? Each one I'm running has over 300 MB virtual memory, and at least 100MB physical.

Had to shut down one VM - though it wasn't doing much anyway. FreeHAL doesn't seem to be keeping up with demand. 🙁
Click to expand...

😳

Yes, I'm meaning MB.

With only 1 GB memory, when you runs 2 Rosetta-instances, even a single FreeHal is likely to interfere with Rosetta if it's using 90 MB memory.

Hmm, one of the Rosetta-wu's terminated after only 4 hours, hopefully the rest won't be so fast, since in that case I'll run-out of work before the Rosetta-race-start...
 
Last edited:
Those milkyway optimised apps are amazing, my times dropped from 35k secs/wu to 9.5K secs/wu. with the win64 sse3 app. Too bad there is no sse4 win app. It might be time to get my vmware reinstalled and get ubuntu back up.


EDIT: to 9.5K/sec (those pesky K's)
 
Last edited:
waffleironhead said:
Those milkyway optimised apps are amazing, my times dropped from 35k secs/wu to 9.5 secs/wu. with the win64 sse3 app. Too bad there is no sse4 win app. It might be time to get my vmware reinstalled and get ubuntu back up.
Click to expand...

I only glanced at these will have to check out further. You saw a gain with gpu as well?
 
Last edited:
The SSE 4.1 x86-64 Linux app consistently crashed on me. 🙁

I'm now buffering Rosetta work, like RD.
 
A reminder, it's now less than 1 hour till the start of Collatz, so for anyone having GPU's that can't run Milkyway, now is the time to start running Collatz. 🙂

Collatz has a quorum of 2, so validation won't normally be instantaneous. Also, it's limited to 150 active tasks at once, so can't cache-up a large amount like it was possible in Einstein@home and Rosetta@home.

For anyone running Milkyway on GPU, it's best to continue running for the 2 remaining days. Still, in case Milkyway runs-out of work beforehand, it's an idea to enable Collatz, setting a resource-share of zero, and you won't download any Collatz-work except if idle GPU. You'll AFAIK need a v6.10.xx-client to correctly handle zero resource-share...
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top