Can't believe I missed this
_k
P-Town, Texas Icrontian
Surfing hardocp and reading their weekly shout out for people to join their folding team I saw they hyperlinked to something. What it linked to is something that seems very interesting and I am going to do a test deployment on a few rigs.
Check this out.
First issue I have found is that HFM.NET does not like this program and will not save the cfg if you point it to the work folders that are being generated. This makes using it a pain for farms but the client does have a GUI interface so checking on the clients ppd and progress is fast and easy. Single rig folders this would be a great idea because it mashes everything into a single program.
With just running a bigadv SMP I am loosing 11,980k of memory to FAH GPU Tracker. For its ppd calculations for P6053 it was spot on to HFM but P11020 it was off by about 30 points from HFM. I consider that to be reasonable despite that large of a variance across a whole farm could mean it looking like you have a wide spread problem if you are down a few hundred but all clients appear to be at standard production, might drive some people crazy.
HFM will start reporting ppd after a single frame but FAH GPU Tracker is waiting for a full 3 frames. There are a few cool things it does as allow for gaming pausing and heat control(based on WU not temp), there is a lot of little things that normally you would have to do manually or run a .bat to handle. Plus the program centralizes all of the clients. Check it out I am getting the feeling it is worth it.
The HFM config issue sucks but if you set it up and leave it running all the time you could have it create a website to a network location and view from where ever you want, assuming you put the HFM on the machine that stays live the most. I have already thought of several tricks to deal with this, just ask and we can talk.
The most important thing this program did that I saw was only download a3 WUs on machines that had been plagued by broken a4 WUs that were not running in SMP correctly, only 1 core instead of all 4 or 8. Also the one machine that wouldn't download a bigadv recent it fixed that issue right away. I am love'n it but will report back later today and this weekend on how everything is working out. Heck using this might give me the kick in the butt to finish the network GPU temp monitor I started.
Check this out.
First issue I have found is that HFM.NET does not like this program and will not save the cfg if you point it to the work folders that are being generated. This makes using it a pain for farms but the client does have a GUI interface so checking on the clients ppd and progress is fast and easy. Single rig folders this would be a great idea because it mashes everything into a single program.
With just running a bigadv SMP I am loosing 11,980k of memory to FAH GPU Tracker. For its ppd calculations for P6053 it was spot on to HFM but P11020 it was off by about 30 points from HFM. I consider that to be reasonable despite that large of a variance across a whole farm could mean it looking like you have a wide spread problem if you are down a few hundred but all clients appear to be at standard production, might drive some people crazy.
HFM will start reporting ppd after a single frame but FAH GPU Tracker is waiting for a full 3 frames. There are a few cool things it does as allow for gaming pausing and heat control(based on WU not temp), there is a lot of little things that normally you would have to do manually or run a .bat to handle. Plus the program centralizes all of the clients. Check it out I am getting the feeling it is worth it.
The HFM config issue sucks but if you set it up and leave it running all the time you could have it create a website to a network location and view from where ever you want, assuming you put the HFM on the machine that stays live the most. I have already thought of several tricks to deal with this, just ask and we can talk.
The most important thing this program did that I saw was only download a3 WUs on machines that had been plagued by broken a4 WUs that were not running in SMP correctly, only 1 core instead of all 4 or 8. Also the one machine that wouldn't download a bigadv recent it fixed that issue right away. I am love'n it but will report back later today and this weekend on how everything is working out. Heck using this might give me the kick in the butt to finish the network GPU temp monitor I started.
0
Comments
looks interesting ... too bad I don't have millions to spend on folding.