Originally Posted by Bouf0010
So is there any tweaking/setting up i should do with my sig rig?
So far ive installed the latest client and i see that my cpu and gpus are running at 100% - is there more to it than that? Whats the -bigadv (or whatever) ppl keep mentioning?
Any help/pointers will be greatly appreciated - i just want to contribute as much as i can for this event
Well I'm going to post a guide that will help you get the most out of your hardware, but if you want the simplest route then just sticking with client V7 is what I'd recommend. If you don't mind doing just a little extra work though then the Linux Virtual Machine instructions will get you about 10% more out of your CPU PPD. I personally really like the benefits of a VM but if you're only looking to fold just for the CC then whatever is simplest might work best for you, but if you'd like to fold more regularly then the little extra effort of setting up a VM will pay off, actually it will pay off even over as little time as 10 days. The Linux VM will give you about 10% better PPD from the CPU because the folding work has a more direct access to your CPU because Linux is a less bloated and better just generally OS for folding, but from your perspective it runs like any other program so no ill effects of native Linux.
First what you need to do is finish the work units that you are currently folding in v7 on your CPU, do this by right clicking the CPU slot and then selecting finish from the right click menu and then verify that your CPU slot now says finishing instead of running. This will finish your current work unit (WU) but not download a next WU. We do this because we will be changing clients and we want to finish your current work rather than make a change in the middle. If we had moved to a different client while you were working you would not finish your current WU and it would have been lost. That is always something to avoid because when you are assigned a WU but either delete it or do not finish it, that WU is not reassigned to someone else for a very long time so it slows down and hurts the research project as a whole. So just remember to always try to avoid doing anything that will cause you to delete or otherwise lose a WU.This is the guide
for a Linux VM that I use and recommend, it's very easy to follow. First you will want to go into your BIOS and enable virtualization, then you just follow the guide as normal until you get to the web configuration part of the guide. At this point we want to make a few changes from the "base set of options" shown on the screenshot in the guide. Right click and open my screenshot at the bottom of this post in a new tab to see the full size settings appropriate for you. You want those exact settings except for putting in your own folding name and passkey (go here if you don't have one yet
), leave the primary DNS I blacked out to it's default. Once you have those settings and hit submit your VM will reboot and then start folding. To monitor it's progress the monitoring program HFM
is recommended, to set it up to monitor your VM follow step 7 of the VM guide
but in the latest version of HFM you will want to go to "clients, add a new legacy client (v6)" instead of what it used to be called just "add a new client" and then just fill in the log folder address with the one shown in your VM window.
Once you've verified that your new VM is up and running successfully (let me know if you need more help) then you can delete the CPU slot that you should have 'finished' so it will no longer be folding, by going to configure, slots, then select the SMP slot and pressing delete. You can also add the two 580s you still have folding in v7 (which is a fine client for them) to HFM monitoring by following this
guide if you'd like to be able to see all your folding in one place, or you don't have to do that if you don't want to.
Oh to answer your bigadv question, bigadv work units are CPU WUs that take a longer time to fold but give better PPD. Right now they are incredibly rare for 8 core/thread CPUs such as ours (about once a month in my experience), but with my VM instructions you can get them once in a while. Some people can increase their chances of getting bigadv WUs by tricking the client in thinking they have more cores/threads than they do which will make it assign different WUs by doing what's called a core hack. While the core hack isn't a bad thing right now IMO I don't have much experience with it and the days of being able to complete those WUs are coming to an end so I personally don't recommend it right now, not to the point where I recommend against it yet but just that I'm not the person to help you with it.
Originally Posted by Pentium4 531 overclocker
10 days away!!!
AMD and SMP's? hmmm might reconsider... ill check it out later... CPU is at 3.8GHz stable, but 3 Cypress Cores... thats like 60% peak, most of the time it draws about 45% CPU power
So im stuck at SMP 4 instead of SMP 8... You guys think it would give me more points if i fold only CPU?
or maybe only 1 GPU? i dunno i have to play with it... but im running BOINC right now, can't play around with F@H
No way, with 3 58XX series GPUs and only a 870 for a CPU you will be getting the majority of your points from your GPUs. So yes, 4 out of 8 threads of CPU with 3 GPUs is much better than just all 8 threads of CPU and no GPUs (in every metric other than energy efficiency). If you wanted to try something different then I would recommend trying 6 threads of CPU and both GPUs of the 5970, that should be pretty close to the PPD of 4 CPU threads and 3 GPUs, but will save the energy of the third GPU.