Posts by Palo M.
log in
1) Message boards : Number crunching : Linux 64 bit ATI application (Message 3194)
Posted 3008 days ago by Palo M.
I'm in!

Phenom II 940 X4, Ubuntu 9.04 x86_64, kernel 2.6.28 (custom)
HD4870 (1G), fglrx driver version 8.650 (aka Catalyst 9.9)
BOINC currently 6.4.5 (I can compile any version if needed)
2) Message boards : Number crunching : 0 credits (Message 2215)
Posted 3037 days ago by Palo M.
BTW: The machine 3568 has made trouble to many people (4 WUs for me). The results' stderr usually shows "needed 0 steps" and most probably it was overclocked:
Found 1 CAL device

Device 0: ATI Radeon HD 3800 (RV670) 512 MB local RAM (remote 1855 MB cached + 1855 MB uncached)
GPU core clock: 878 MHz, memory clock: 901 MHz

Meanwhile the owner seem to reduce core clock and add another card:
Found 2 CAL devices

Device 0: ATI Radeon HD 3800 (RV670) 512 MB local RAM (remote 1855 MB cached + 1855 MB uncached)
GPU core clock: 825 MHz, memory clock: 901 MHz
320 shader units organized in 4 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
supporting double precision

Device 1: ATI Radeon HD 3800 (RV670) 512 MB local RAM (remote 1855 MB cached + 1855 MB uncached)
GPU core clock: 825 MHz, memory clock: 901 MHz
320 shader units organized in 4 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
supporting double precision
And the results validate for the host...

But I'm curious about following possibility: My wingman will have overclocked GPU, so after we both return results, validation will be inconclusive nad WU will be sent to third host. And what happens, if that other host also has overclocked GPU, also returning wrong result... If there would be more-less random result returned (failed somewhere in the middle of computation), it's really not probable that 2 wrong results will be the same.
But looking at those wrong results mentioned above, I assume that the "needed 0 steps" could appear for 2 incorrect results and they could vote out the correct result. Then the "good guy" will lose his credit, but more important, the number will not be calculated at all and project could miss a chance.
My assumption is based only on stderr though, so maybe the 2 faulty results are not exactly same and therefore not validated... But I'm still curious - can validator discover such obviously faulty results?
3) Message boards : Science : The Collatz Conjecture (Message 2051)
Posted 3039 days ago by Palo M.
Slicker wrote:
This project's goal is to prove, or disprove, the Collatz Conjecture.

I'm just curious how this project can prove Collatz Conjecture.

I assume there is one way how the project can disprove the conjecture - if we find the loop of numbers which to not contain "black-hole" sequence 4-2-1, then the conjecture will be disproved...

And what about the other possibility which would disprove conjecture (that the numbers will increase without bound)? I also cannot imagine how project could achieve this...




Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.