Q 6600 to xeon
13 Mar 2015 3:14:42 UTC
Topic 198012
(moderation:
I ordered a Intel Xeon X5460 SLBBA 3.16GHZ 12MB 1333MHZ to replace my Q6600 . I run Einstein ; Rosetta ; poem and milkyway . which ones will benefit the greatest from the upgrade I am going to try the 775 to 771 conversion . my graphics card is a amd 7800 version with a 1x pcie
Q 6600 to xeon
)
I took a quick at your tasks list on the website and I see that you are just crunching BRP6 GPU tasks with the standard (non-beta) version of the app. I can't comment on other projects (nil or too little experience) but you could have got a very significant boost for your current host just by switching from the standard (non-optimised) BRP6 app to the latest (optimised) beta test version. I presume you meant 16x PCIe rather than 1x?? I'd recommend you accept beta test tasks in your preferences and see the benefit for yourself.
The 7800 series GPUs are excellent for BRP5/BRP6 - I run 4 concurrent tasks on my 7850s - and the ones in older hosts (Q8400 rather than Q6600, but that's not going to make any big difference) are showing a tremendous improvement now that the PCIe bus bandwidth bottleneck seems to have been solved.
No doubt, the beta test app will become the standard app at some point, but you may as well enjoy the performance boost right now. Over my range of hosts, the CPU used (dual or quad, current or 5 year old) makes very little difference to the crunch times of GPU tasks on a HD7850.
Cheers,
Gary.
thanks for the input yes
)
thanks for the input yes pcie 16x typo ;-] . I hope the swap will go good
To run beta test
)
To run beta test applications, you need to enable that setting in your project preferences. Go to your account page on the website and click on the Einstein@Home preferences link. Find the setting for beta test apps and edit it to "yes".
If you want to run multiple GPU tasks concurrently, there is also a setting for that, two below the beta test setting. It's called "GPU utilization factor of BRP apps" and you set it to the fraction of a GPU that each task will 'use'. For two concurrent tasks, set 0.5, for three set 0.33, for four set 0.25, etc. I run four on all my HD7850s, but I haven't yet done the experiments to see if that is really appropriate for the new app. If you change that setting, the new value comes into play when you next download new tasks. The warnings for that setting relate to extra heat from working the GPU harder so you need to be sure that your cooling is good.
Cheers,
Gary.
Listen to Gary, what he says
)
Listen to Gary, what he says will improve your Einstein throughput by far more than any CPU swap.
MrS
Scanning for our furry friends since Jan 2002