How to get more points?

[Aurum]

I noticed that GPUs would sit idle after finishing their WU since they didn't have a WU waiting in the wings. I've been using next-unit-percentage 98 on all my folders the last couple of days and now they'll have a WU waiting and will start folding it while the finished WU is ULing.

A small thing compared to using a passkey and getting through your first 10 WUs

[bruce]

... I've been using next-unit-percentage 98 on all my folders the last couple of days ...

A lot depends on the nature of your internet connection. If you're seeking maximum bonus, any time the new WU sits on your system BEFORE being started costs against your bonus, just as the time that the result sits waiting to be uploaded. For small WUs, my connection manages everything efficiently with next-unit-percentage=100. For large WUs, I set next-unit-percentage=99. with the goal of the download finishing just before the client is ready to run it. (Unfortunately I can't keep adjusting that setting based on whatever the next WU is going to be.)

As you suggested: It's (very) small difference.

[ComputerGenie]

...If you're seeking maximum bonus, any time the new WU sits on your system BEFORE being started costs against your bonus, just as the time that the result sits waiting to be uploaded...

Wish I knew that before, mine have been set to 95 to prevent ever idling.

[Aurum]

Thanks Bruce, I did not know that. I'm going to set all mine to 99.9 to avoid reducing bonus if WU sits on my HD idle for too long. Some of the long TPF WUs can leave a new WU sitting idle for well over 30 minutes using 98. I also just found out it can be done as a global option on the Expert tab.

[bruce]

It can be set to 100 in either the Expert tab (global) or in an individual slot's configuration.

Depending on the size of the WUs, my internet connection can usually download the new WU between the 100% message and the FINISHED_UNIT message so zero time is lost waiting for the new WU to arrive. This isn't always the case, though.

[kcanti]

So, I recently brought a GTX 1080, all ready to be blown away by the PPD paradine shift from my 960GTX ppd of 90-100k. And what should I get other then 80k. 80K!!!! I have no idea why my GTX 1080 STRIX edition is performing worse then a last gen budget card. I don't understand why, drivers are as up to date as I can get them.
This pc is specifically for folding and runs 24/7. Dare I say it, but until I resolve this issue ill be running Boinc, would much rather be running something useful like F@H.
Some things I have tried:
-Uninstalling and reinstalling my graphics drivers.
-Running FAH as admin in compatibility mode for Windows 7.
-Trying to run the visuliser (locks up on the loading screen and doesent update) which crashes my F@H app and will not restart unless I do a full system reboot.
-Running in SAFE MODE of all things.
-Disabling CPU folding to try eliminating any possible throttling.
Specs:
Windows 10
AMD FX 8350 OC 5.4GHZ
12GB DDR3 1833 MHz
Endless HDD's but my C drive is an M.2 500gb
GTX 1080 Strix Edition.

Thanks for your time,

Kcanti

[bruce]

Welcome to the FAH support forum, kcanti

Most likely you need to create a passkey and start using it.

(I'm merging your post with another topic on similar issues. Please read the topic from the top.)

[Aurum]

...drivers are as up to date as I can get them.

Is there a problem with the latest Nvidia graphics drivers
viewtopic.php?f=80&t=29276
My first GTX 980Ti is in the mail so I'm very curious which drivers are best for folding.

[Joe_H]

The last regular release of nVidia's drivers that worked was version 373.06. There is also a version 376.48 Hot Fix that has a workaround coded into it that has mostly been reported to work, but there is at least one failure report on one WU. The rest of the version 375 and 376 driver releases prior to the hot fix did not work with Core_21 projects.

[Aurum]

It can be set to 100 in either the Expert tab (global) or in an individual slot's configuration.

Depending on the size of the WUs, my internet connection can usually download the new WU between the 100% message and the FINISHED_UNIT message so zero time is lost waiting for the new WU to arrive. This isn't always the case, though.

This morning I checked FAHControl and just happened to see a GPU about to hit Progress 99% on the Status/Work Queue list. I've had next-unit-percentage=99.4 set globally on the Expert tab for days and watched to see if it worked. But the GPU did not poll the AS until Progress crossed 99.8%. But in the Log tab work progress is reported as integers:

Code: Select all

08:10:51:WU02:FS01:0x21:Completed 50000 out of 5000000 steps (1%)
08:42:23:WU02:FS01:0x21:Completed 100000 out of 5000000 steps (2%)
09:14:17:WU02:FS01:0x21:Completed 150000 out of 5000000 steps (3%)
09:45:52:WU02:FS01:0x21:Completed 200000 out of 5000000 steps (4%)

I bet the "next-unit-percentage" option uses a rounded up or off version of the "value" one assigns. If "value" can only be integers it'd be nice to know, e.g. reject a decimal entry when adding the option or a note in the Help file. It might be better to use ETA to trigger DLing the next WU.

The 99.9% bug: When a GPU is infected by the 99.9% bug it never seems to recover. The only way I know to cure it and get the GPU back to folding is to Quit F@H and restart it which wastes a lot of work when the other GPUs in the folding rig get set back to their checkpoints. I bet you're already working on a fix for 7.4.16 but since I see it daily just thought I'd mention it.

[SteveWillis]

On my Linux box I'm pretty sure that when I execute a reboot checkpoints are created at that point.

[Aurum]

No they're coded in by the WU creator. There's a discussion by those in the know on another thread here. Folders were asking for a Checkpoint-on-Pause button or a Pause-After-Checkpointing.

[bruce]

On my Linux box I'm pretty sure that when I execute a reboot checkpoints are created at that point.

If that's true (and I doubt it) it will depend on both the FAHCore that you're interrupting and on the Project being run.

Please document your suspicions by examining the log both before and after the reboot and showing at what point the WU was interrupted and at what point it was restarted. You may also note the timestamp on the checkpoint files in the /work directory if you can catch them before the FAH restart overwrites them.

[Joe_H]

This might happen for a WU using the new CPU core, Core_A7. From what I have seen it attempts to write a checkpoint when it receives a pause message. It doe not happen for Core_A4 or any of the GPU folding cores.

[SteveWillis]

Interesting thing. I decided to try and look at the timestamps on the checkpoint files. I found files checkpointState.xml and am guessing they are the checkpoint files since they are very large and update regularly and there is one for each WU. Strange though, I have checkpointing set to 30 minutes and these files are updating every 3 minutes. Which I guess explains why I never noticed any loss of work on a reboot.