16th Annual Folding@Home Holiday Season Race: The race is over and the Moonshots win!

[Pokey]

Must've been end of Q3 or beginning of Q4 2020. That's were I see a significant increase in my stats.

That's what I was thinking too.

 

[biodoc]

The new cuda app was released into beta testing in late September 2020.

 

[Ken g6]

Depends on how long you plan to pause (is this a consecutive 4h break you're planning on, or are these 4 peak hours distributed somehow over the day?). Because you'd lose the quick return bonus if the pause is too long.

Four consecutive hours. 4pm to 8pm. It's actually quite annoying. And part of the bill is based on the highest usage of any 60 minute period within those times, so I can't just cheat for a day or a week, or even an hour.

 

[StefanR5R]

While FAHClient doesn't feature a daily schedule like BOINC, you can implement it yourself with a control script. The simplest way would obviously be to either pause/unpause or abort/unpause right on time. Perhaps more elegant would be to finish ahead of time/ unpause on time. The trick would be to trigger the 'finish' at a time when a reliable estimation of the time remaining of the running workunit is available.

There was a brief thread about fahclient control here.

 

[Ken g6]

The real trick would to estimate the running time of the next WU before I fetch it.

Or maybe I should just turn everything down to a minimal clock speed and see how much power that saves.

 

[voodoo5_6k]

Four consecutive hours. 4pm to 8pm. It's actually quite annoying. And part of the bill is based on the highest usage of any 60 minute period within those times, so I can't just cheat for a day or a week, or even an hour.

I've made a quick estimation based on the total points equation and a WU that's running right now on my GPU.

With:
k=0.75 (baseline value, doesn't matter for the relative loss)
base points=67770
deadline length=2880 minutes
elapsed time (extrapolation)=365 minutes
elapsed time (your worst case, i.e. slap 4h on top for the pause)=605 minutes

Now I get the following:
total points=164,861
total points (worst case)=128,052

That's a loss of ~22% due to the 4h break. Not too bad. So, you should be able to pause/unpause over the 4h annoyance without too much impact to PPD. Combined with what @StefanR5R said and linked above, a quite comfortable solution seems possible

 

[salvorhardin]

Count me in.
salvorhardin - salvorhardin - 500k ppd - full month
Nothing new added since last year, so should be around the same as last year.

 

[biodoc]

I've chosen the Captains of the 2 teams. Is it ok to recycle last years team names? If not, the captains should choose another team name.


@Pokey is the captain of the Moonshot Gang team
@Markfw is the captain of Mark's Marauders team

 

[Markfw]

I've chosen the Captains of the 2 teams. Is it ok to recycle last years team names? If not, the captains should choose another team name.


@Pokey is the captain of the Moonshot Gang team
@Markfw is the captain of Mark's Marauders team

BTW back in the hospital but should be home for the race start. Check etc stats for my current average I think its only 20 mill. It will go up a few mill before the race start.

 

[Pokey]

Wow, OK……………… I’m game. Moonshot Gang it is.

Thanks for heading up the event @biodoc.

 

[crashtech]

Update, looks like my output should be ~23 Mppd.

 

[Orange Kid]

Is there a particular project that is best to run?
Any tips on the advanced settings?

Getting things set up. Should have a PPD by the 30th.

 

[Pokey]

Is there a particular project that is best to run?
Any tips on the advanced settings?

Getting things set up. Should have a PPD by the 30th.

I don't know if one is better than the other. I have always run Covid-19 on all my rigs except for one which is set to All, but it has lesser GPUs, so no help.

 

[cellarnoise]

Love the team captains!

Team names are o.k., but would fun to have team members contribute to banter and maybe new names. Though not as necessary as team contribution points!

I just like some change from year to year?

Though I will try to stay quite some. Ass I think I contribute too mucho...sometimes.

 

[VirtualLarry]

How about the two teams:

!) I've got Bills
2) I've got Money.

See if the money or the bills come out on top.

Edit: On second thought, that while kind of cute, it's nearly political.


PS. Not participating this year. I do have bills, and I need the money. My PG participation recently is all you're going to get for a while.

 

[Orange Kid]

I have one machine that runs winblows and three that are on Mint.
The mint boxes I have upgraded the NVIDIA drivers to the 495 version. All run great and are getting what I expected for F@H.

On the winblows box I was running 465 drivers and got 3m ppd.
I updated to the latest 496 drivers and ppd declined to less that 1m.
I have since reverted back to the 465 version and ppd is climbing.
So I was wondering if anyone else is experiencing the same problem?

 

[Markfw]

@biodoc , While Pokeys may be 40m, mine is 20m now, and not sure when I can get more (still in the hospital)

 

[biodoc]

Here's a first attempt at assigning participants to teams. Please be patient because a few have unknown PPD and some will be diverting resources to the upcoming PG challenge.

 

[biodoc]

On the winblows box I was running 465 drivers and got 3m ppd.
I updated to the latest 496 drivers and ppd declined to less that 1m.
I have since reverted back to the 465 version and ppd is climbing.
So I was wondering if anyone else is experiencing the same problem?

I looked briefly over at the folding forum and didn't find anything that was an obvious solution. Did you install the 496 drivers directly from Nvidia? I've heard that updating drivers via microsoft (windows) can sometimes lead to issues with F@H.

 

[Orange Kid]

The drivers were from NVIDIA.
With the 496 version I noticed the clock speed was held at 780MHz and with the 465 drivers, MHz is at 1770. It is a definite throttling by the drivers. Noticed this problem first when running some moo wrapper. Thought it was due to the open cl version, so that is when I first changed driver version from 495 to 465. Speed effectively doubled then. After firing up F@H, I upgraded the driver and it went to poop again.
Most certainly a driver issue with winblows.
Going to maybe make a separate Linux disk for this machine and see what kind of ppd I get.

 

[salvorhardin]

Looks like my ppd is going to be higher than I thought, it's showing around 1m ppd.

 

[Markfw]

OK, I am home, and my PPD is at its max, so I am ready for Wednesday !

 

[Markfw]

Pokey is already at 47m ! Might have to change the teams before we start !

 

[cellarnoise]

Hey, before I am not speaking with half the TeAm for the next month...

I am running 1 GPU with no CPU slots allocated to folding@home on one Windows 10 computer that most of the time will just be crunching. Any suggestions on how to set up all remaining available CPU % for running BOINC, besides a few % to support the GPU running Folding@Home? And at this point any participants not on my future 1/2 of TeAm will be suspect?

Too many variables to really have a method or direct answer? Just observations from Task Manager or HWiNFO64 that I use?

I don't want any CPU cycles going to waste or my waist as the Galumpkis is "on that" task already...

 

[Pokey]

Pokey is already at 47m ! Might have to change the teams before we start !

Well, let’s see. I averaged about 41mil last year and I am running the exact same rigs this year……………..
I have no idea why my production would be higher.
Either I will fall back to the norm, or everyone will see an increase.
Anyway, that's my line of thought right now.