Start Folding the Easy Way!!

Happy this hit the front page or I would have missed it. Just started folding for [H] afew days ago.
 
Welcome to all you new folks on the team. You guys picked a very good time to join us as we are in the battle of battles right now.
 
thanks for this post. I always forget to restart my client after gaming. The gaming mode on this works perfect for me
 
Thank you jedi95.

I have been waiting a long time for someone to come up with a easy way of folding that actually worked for me.
 
Just saw the post about reaching 4 Billion, well done guys! Thats amazing work, with any luck i'll be able to join you soon.
 
Just installed this and I'm not folding the for [H]. Running with SMP but its going very slow and is 0.0 PPD while my GPU is green and 5,482.9. Is this normal?
 
Nice work Jedi95, installed this last night, works great!!! Maybe it's just me, but I think my GPU PPD is a bit higher than with just the 6.23 client, PLUS I'm now using my CPU to fold also :)



 
fishoil said:
Just installed this and I'm not folding the for [H]. Running with SMP but its going very slow and is 0.0 PPD while my GPU is green and 5,482.9. Is this normal?
Click to expand...

Same for me. SMP seems like it just sits there even though the Task Manager shows all the CPU's at 100%. GPU works fine.
 
The SMP client will take much longer to complete WUs compared to the GPU clients. (unless you have some odd configuration, like an i7 980x with a 9600GT) The 0.0 PPD is normal when just starting. It takes a few frames to get an accurate TPF average in order to calculate PPD. The Tracker needs 3 frames to calculate TPF (meaning the WU must progress at least 3% without being stopped/restarted before it will calculate PPD)

Also, to clarify on why you will sometimes see a nonzero yellow PPD, the yellow status means that whatever is shown for PPD is based on benchmarks, and not the current TPF. If no benchmarks exist for that WU, then you will see 0.0 PPD. If a benchmark exists for that WU, then the PPD will be calculated using the benchmark TPF.

The green PPD status means that the PPD shown is based on the TPF calculated from the last 3 frames.
 
Ok I will let it go for 24 hours and see what happens. Realize I was running the normal client before trying this software, but my business went under so now I will come back and fold for [H].

 
Is it better to just run CPU instead of SMP? I have a dual core AMD X2 Dual Core 5600+ (yes, I know I need to upgrade).
 
No, SMP would still be worthwhile to run. It will give you more points compared to the normal CPU client.
 
SMP is CPU......are you asking if in your case it would be better to run just the regular CPU Client? If so, possibly. I dont know how fast a 5600+ can do a SMP work unit, the uni processor(regular CPU) will gain less points for you(no bonus system), though there is a back log of those units ans they need to get done.....I would say wait for osmeone who knows if your CPU can do SMP ontime......but i'm thinking it would be best running uni processor for now, from heat/power perspective.....

edit: nevermind it appears i missed jebo's post.....would be up to you really.
 
I just joined...have 5 rigs folding at the moment:

E6600 @ 3.6GHZ./GTX 460
E6600 @ 3.4GHZ/HD4850
PS3
i5 750/ 2 x 5770
1 laptop T9600/ GTX 260

Do SMP calculations take a lot longer than the GPU calculations?
 
mekkanic said:
I just joined...have 5 rigs folding at the moment:

E6600 @ 3.6GHZ./GTX 460
E6600 @ 3.4GHZ/HD4850
PS3
i5 750/ 2 x 5770
1 laptop T9600/ GTX 260

Do SMP calculations take a lot longer than the GPU calculations?
Click to expand...
nice little farm. Welcome to the team!0
Posted via [H] Mobile Device
 
jedi95 I would like to buy you a beer or two for all your hard work. Are you taking donations?
 
First of all I would like to thank Kyle once again for getting me a 5830 to test with the Tracker.

I have figured out ATI + Nvidia folding on the same computer. It's kind of interesting what needs to be done to make it work correctly.

The test system I am using has an ATI 5830 and an Nvidia 8800 GTS 512. In order to get the 2 cards to fold at the same time BOTH cards need the -gpu 0 flag. For the 5830 you can use GPU2 or GPU3, but you need -forcegpu ati_r700 either way.

Second, if FahCore_11.exe already exists in the client folder, you may have to delete it. There are 2 versions of FahCore_11.exe, one for Nvidia and one for ATI. As a result, you need to delete the core (which causes the correct core to be downloaded) when switching between the 2.

Third, I had to install the ATI drivers after the Nvidia drivers. I couldn't get the 5830 to fold without doing this. This step also applies if you are doing other distributed computing projects through BOINC like Milkyway@Home, since BOINC wasn't recognizing the GPU.

That being said, if you want to use the Tracker for ATI + Nvidia folding this is what you need to do:

1. Change the -gpu x flags:
For ATI + Nvidia folding, the -gpu x flags are independent of eachother. So the first ATI card needs -gpu 0, the second -gpu 1, and so on. The Nvidia cards in the same system also start with -gpu 0 for the first card, then -gpu 1 for the second card, and so on. You will need to edit these flags in client.cfg in the Tracker's GPU client folders. You can use whatever clients for whatever GPUs, it doesn't matter. The -gpu x flags don't have to match the name of the client.

2. Remove FahCore_11.exe from all the client folders.
The client will NOT automatically re-download the correct ATI or Nvidia core if the wrong one is present. If you use the wrong one it will fold on that GPU type regardless of the -forcegpu setting. So if you have a client with -gpu 0 and -forcegpu ati_r700, but you have the Nvidia FahCore_11.exe, it will actually try to fold the WU on Nvidia GPU 0.

3. Set forcegpu for all clients.
This step may or may not be required depending on the system, but to be sure it will work this is necessary. Otherwise there may be problems since you are using multiple clients with the same -gpu x flag.

Hopefully this will be helpful to anyone with ATI + Nvidia in the same system. Getting this to work is complicated right now, but I will be adding proper support for this type of configuration in a future version of the Tracker.
 
I wasn't able to get my 5870 folding with an nVidia GPU in the same system before, but I'm giving those instructions a shot. Hopefully it'll work out. Thanks for figuring that out.
 
BTW, Jedi95... When I restarted the FAH Tracker a few minutes ago, it informed me that there was a new version available (3.22). When I let auto-update try to install the update, it would fail with an error about the wrong MD5 checksum. I was able to manually download the new client and install it that way though.
 
Fifth Horseman said:
BTW, Jedi95... When I restarted the FAH Tracker a few minutes ago, it informed me that there was a new version available (3.22). When I let auto-update try to install the update, it would fail with an error about the wrong MD5 checksum. I was able to manually download the new client and install it that way though.
Click to expand...

This has been fixed along with another bug in 3.23. In 3.22 the deadline warning would appear when it shouldn't be doing so. The MD5 checksum was probably wrong in the config file (the Tracker checks an update config file that contains information about the update, including the MD5 checksum of the downloaded file) If the MD5 is wrong in the config file then the update will fail with that error every time.
 
thanks for the latest update jedi95 . out of curiosity, any plans to update the monitoring to be something like hfm.net with the bonus points?
 
firas said:
thanks for the latest update jedi95 . out of curiosity, any plans to update the monitoring to be something like hfm.net with the bonus points?
Click to expand...

What exactly do you mean? The Tracker already takes bonus points into account when recording stats and calculating PPD.
 
well I’m new to the whole scene but I heard people saying it’s better to use hfm.net with it and that’s what I’ve been doing. but if it’s not really necessary that would be great. I remember now, I think it was you saying “current version doesn’t calculate bonus points” in some other site but I guess that should be an old post
 
jedi95 said:
What exactly do you mean? The Tracker already takes bonus points into account when recording stats and calculating PPD.
Click to expand...
It's also a full blown production monitoring program?
 
well tried out the program.. not to bad.. only problem i had was not being able to set the -advmethods flag. and that for some reason even setting my 8800GS's to the gpu3 client in the program i only received GPU2 WU's.. after about 4 WU's i finally gave up went back to my gpu3 clients i had installed and the last 4 WU's have been GPU3 WU's.. so im not sure if its something with the program or what.. other then that its a pretty slick program...


and yes it does show PPD and all that apollo..
 
Geeze, you guys weren't kidding when you say that if you can run the GPU folding client over the SMP client, it's almost not even worth running the SMP client...

I just upgraded my system to a Phenom II X6 1090T today. Even overclocked to 3.8Ghz, as I type this my two SSC GTX260 Core216's have completed 75% of a work unit each. The CPU (with all six cores going)? 4%.

[Cleveland Voice]DAAAAAYMM![/Cleveland Voice]

I mean; that's still a heck of a lot faster than my old Opteron 185, but compared to the GPU's it's pathetic. :D

Next week I get my third SSC GTX260 Core216 (triple SLI)... hey, I got them cheap on eBay! :D
 
Fifth Horseman said:
Geeze, you guys weren't kidding when you say that if you can run the GPU folding client over the SMP client, it's almost not even worth running the SMP client...

I just upgraded my system to a Phenom II X6 1090T today. Even overclocked to 3.8Ghz, as I type this my two SSC GTX260 Core216's have completed 75% of a work unit each. The CPU (with all six cores going)? 4%.

[Cleveland Voice]DAAAAAYMM![/Cleveland Voice]

I mean; that's still a heck of a lot faster than my old Opteron 185, but compared to the GPU's it's pathetic. :D

Next week I get my third SSC GTX260 Core216 (triple SLI)... hey, I got them cheap on eBay! :D
Click to expand...

The time to complete a WU is not the only thing that matters. SMP WUs may take longer, but they are usually worth a LOT more points than GPU WUs. The measurement you should be using is PPD (points per day) which is shown in the right column on the Tracker's main window.
 
Fifth Horseman said:
Geeze, you guys weren't kidding when you say that if you can run the GPU folding client over the SMP client, it's almost not even worth running the SMP client...

I just upgraded my system to a Phenom II X6 1090T today. Even overclocked to 3.8Ghz, as I type this my two SSC GTX260 Core216's have completed 75% of a work unit each. The CPU (with all six cores going)? 4%.

[Cleveland Voice]DAAAAAYMM![/Cleveland Voice]

I mean; that's still a heck of a lot faster than my old Opteron 185, but compared to the GPU's it's pathetic. :D

Next week I get my third SSC GTX260 Core216 (triple SLI)... hey, I got them cheap on eBay! :D
Click to expand...


your looking at the small picture.. time means nothing.. on that processor your looking at about 10-14k PPD compared to the gtx 260's which are doing about 6-7k PPD each and consuming twice the amount of power compared to that x6 1090T..
 
Is it possible to set up some sort of calendar/timer? I'm happy to do this, but I don't want to have to deal with starting/stopping the clients, and I can't have em running while i"m working...
 
if you have the time you can tell windows to turn it on as an event and turn it off as an event.....though it would be a bit of work to do....
 
You must log in or register to reply here.
Back
Top