Well, I was lucky to find my GTX 660 Ti finishing a WU in 10 minutes so I edited the configuration from FAHControl to add <max-units v='1'/> and once it finished the WU, it stopped automatically, I didn't have to pause it or do anything. Thus, I do think that it will work out for you needs which I think is great news. Here is a very short version of the log since I can't seem to find the code tags:
 
20:47:00:WU00:FS01:0x17:Completed 2375000 out of 2500000 steps (95%)
20:47:04:Removing old file 'configs/config-20130425-133509.xml'
20:47:04:Saving configuration to config.xml
20:47:04:<config>
20:47:04:  <slot id='1' type='GPU'>
20:47:04:    <client-type v='beta'/>
20:47:04:    <max-slot-errors v='1'/>
20:47:04:    <max-unit-errors v='1'/>
20:47:04:    <max-units v='1'/>
20:47:04:    <next-unit-percentage v='100'/>
20:47:04:    <pause-on-start v='true'/>
20:47:04:  </slot>
20:47:04:</config>
20:49:04:WU00:FS01:0x17:Completed 2400000 out of 2500000 steps (96%)
20:57:43:WU00:FS01:FahCore returned: FINISHED_UNIT (100 = 0x64)
20:57:44:WU00:FS01:Sending unit results: id:00 state:SEND error:NO_ERROR project:7663 run:4 clone:8 gen:9 core:0x17 unit:0x0000000d01f98459518715d546a2439c
20:57:44:WU00:FS01:Uploading 5.82MiB to 171.64.65.68
20:59:44:WU00:FS01:Upload complete
20:59:44:WU00:FS01:Server responded WORK_ACK (400)
20:59:44:WU00:FS01:Final credit estimate, 7392.00 points
20:59:44:WU00:FS01:Cleaning up

I got my first 7663 looks like the PPD is going to be amazing. 

jinihammerer

I am folding beta linux GPU on my 690's ATM. its actually about 6 seconds slower per fold then in windows 8. i am about 7% into it with the latest nvidia drivers installed in linux mint 14. I have to double check the nvidia drivers may have me in sli mode.

Jini hope your setup works out better than mine the system I set this up on crashed hard and lost everything.
Just started with another HDD, completely wiped fresh install of Mint 14.
At least I was lucky enough for my last 8105 to finished and uploaded before the crash.
I was no longer GPU folding when this happened, but there was signs of problems were
the system would not come out of sleep mode when moving mouse or using keyb.
Never seen any problems like this before the nvidia driver and FAH V7 install.

OK, I have installed V7 this evening onto one of my Linux rigs running Ubuntu 12.04 LTS. 
 
That setting in V7's Slot config for "Max Units" is:
(example for if you want the max-units to be a value of 2, but you can set your own number)
In the Edit Slot Options area.
Name:  max-units
Value:  2
 
I have the first WU running now and will see (tomorrow morning) if V7 indeed stops processing additional WUs after it has finished 2 WUs in a row.

Came here to say this, you beat me to it :)
This option works fine, I use it all the time.

<edit>
the usage as quoted by panther-x is correct
max-units
X - V7 will complete X number of units on that slot, then stop folding on that slot.
 
 

Panther-X

Hi Guys,

Please note that while there is a request for adding a scheduler in V7 (ticket #300 which I created, can't post link yet), for now you can use this lesser-known-feature:

<max-units v='0'/>

This can be set as a per slot basis and the default of 0 means that it will continue to download and process WUs until the donor stops. However, you can set it to 1, 2, 3, etc and that will tell the slot to stop after 1, 2 3, WUs automatically.

That's the theory about this feature and hopefully, it works as such. I haven't yet tried it so if you fellows do try it and it works, would be nice to that :)

BTW, not sure if this is a feature or bug, I typed in Quick Reply and hit "Open Full Version" but my message didn't appear. I quit the Full Version and my message was still in the Quick Reply. Using Cyberfox 20.0.1, a 64-bit variant of Firfox.

This works great! Stopped after completing 2 Wu's.
Now on my second go round with both GPU's folding.
One set to work 2 units the other set to work only one.
Lets see how this works out. 

Panther-X 

Here is a very short version of the log since I can't seem to find the code tags:

I think this is the link you were looking for:
Folding@home Client Control

Been doing some experimenting, it seems that Core17 is locked to use a maximum of 1 CPU core (when folding with nVidia, not sure about AMD).
 
When overclocking you GPU, there comes a point where it becomes CPU limited, and actually ends up slowing down with increased clock rates - may be an effect of the boost-clock-rate bouncing up and down a bit???
 
Anyway, on my system 2600K @ 4.5 the 660ti hits the limit at 1137 and the 560Ti at 945.  Anything above that and the tfp start to increase, not decrease.
 
Anyone else noticed this ?
 
btw running Driver version 314.07
 

Simba123

Been doing some experimenting, it seems that Core17 is locked to use a maximum of 1 CPU core (when folding with nVidia, not sure about AMD).

When overclocking you GPU, there comes a point where it becomes CPU limited, and actually ends up slowing down with increased clock rates - may be an effect of the boost-clock-rate bouncing up and down a bit???

Anyway, on my system 2600K @ 4.5 the 660ti hits the limit at 1137 and the 560Ti at 945.  Anything above that and the tfp start to increase, not decrease.

Anyone else noticed this ?

btw running Driver version 314.07

the tpf increase could be an instability in the OC, I've pushed both my 7970's to 1050 (+50mhz over base) and my 680's are in the 1260 range. all lowed tpf with each increase in their OC. I haven't pushed any into crashing yet. I can say that HFM ppd/tpf does jump up and down a lot, it maybe only 1-2sec per 1% either way but it affects the ppd with a 1-3k swing.

wrinvert

any jump in ppd with Linux? If so I might through an extra ssd in the wifes tower since it just folds anymore.

 
 
Not enough data to say yes at the moment. the GTX690 installed in SLI mode by default so when i was testing it i was actually 5 seconds slower per fold then windows. that's not saying its slower or faster, that's saying it was not optimized for folding when i started. tonight i will do the same experiment on my OCed 8770 k with the gtx 580 and gtx 670 ftw and also again on the pair of gtx 690 once i take it out of SLI.
 
I want to drive in a warning and reminder.
 
Evga Precision does not have a linux port over and from my understanding, never will. so if you are OCIng or rely heavily on the fan curve provided by the tool we will need to find an alternitive or come up with a WINE configuration that will allow it to run in Linux.
 
Till then, watch your heat levels on your cards, the basic way to do that is to open the Nvidia control panel it will tell you your GPU core temps. 
 
If you cant find the control panel, do a search in apps for Nvidia and you should find it. 
 
 

I was mentioning the fact that on the EVGA Forum, I can't find the Code Tags that are present in FoldingForum. However, I experimented in the Full Editor and it was available under the "Source Code". Sorry for not being clear in the first place.
 

TheWolf

Panther-X 

Here is a very short version of the log since I can't seem to find the code tags:

I think this is the link you were looking for:

Things I love about the new Core 17
*GPU runs quietier (very little to no coil whine noise)
*GPU seems to run cooler as well (thou I don't monitor this much anymore with the change to Kepler vs Fermi)
*Higher PPD
*Allows more and different WU projects to be used
 
Things I don't like about it
*Uses up a whole thread or core

johnerz

I just opened a ticked and used XZ's references, I hope I get a decent reply :)

All we can do is try and hope we do not get an answer like Simba123 got. So far I am on my second reply and by the name of person involved it looks like they are outsourced.

My Nvidia driver answer: “NVIDIA is currently working on updating the support for Open CL1.2"
No eta and watch for updates.

TheWolf

Xavier Zepherious

when you get results post the PPD improvement of using linux over Win7

I'll probably switch over for the gain

Its been a while since I ran the 660 in W7, not sure I still have any stats.
In Linux Mint14 TPF 2:04 PPD 60K GPU folding only.
I need to find a way to lock cores in Linux with this bigadv+GPU rig.
Other wise looks like it will be a fight for core usage and both may suffer.

For reference, my 660 Ti FTW is running 2:04 with a PPD of around 60K in windows.  I have an i7-860 (SMP:6) folding as well. 
 
Also - I notice that my GPU load (as reported in GPU-Z) is around 93%.  I may have missed something in the tread (or other thread), but are others seeing this as well?  Most, if not all, other WU's I have in recent memory have put my GPU's to 100% utilization.