Folding@Home bigadv EOL 3rd Anniversary Challenge: [H] vs TAAT

biodoc

Weaksauce
Joined
Sep 23, 2019
Messages
74
I'd like to thank Gilthanis and pututu for their help in organizing this challenge series. :)

The challenge will start at 12:00 UTC January 23rd (time 0 stats) and ends 12:00 UTC on January 30th.

Here's the link to the thread in the Anandtech forum.

Last year's race thread on the [H] forum: Link
Last year's race thread on the Anandtech forum: Link

Details:
1) [H] total team Folding@home points vs. the points of Team Anandtech members who sign up for the challenge.
2) Team Anandtech members must sign up for the challenge before the start date. I will not add racers once the challenge starts.
3) To clarify, there's no need for [H] members to sign up. I will take the total output of [H] on FAH during the race.
4) I'll collect stats from EOC and will post updates every 12 hours.
 
Last edited:
biodoc, thanks for setting this up.

To all [H] folders. If you are in US, challenge starts on Jan 23rd 7am/4am PT/ET time and ends on Jan 30th 7am/4am PT/ET.

Some tips:
  • gpu is better than cpu in folding for most people unless you have gazillion of cores.
  • average daily PPD is higher when running in linux than windows (up to 10%)
  • have one full thread per gpu during folding. I usually leave one full core.
If you are looking for gpu and cpu performance, visit folding.lar.systems. You can also install a chrome client to monitor your real time gpu/cpu performance or just use the standard FAH client from FAH. Remember to have a passkey to claim extra credits (bonus). If you need to install Folding in Windows 10, visit Gilthanis how-to video here. For other flavors of OS, see official FAH website.

Our current daily production is just over 100M PPD. Last year during the week long challenge, we average about 210M PPD. Hope to see more folks firing up their gpus during this upcoming challenge :).

Also this is for a good cause, with the most recent one relating to patent-free covid anti-viral development.

For nostalgic reason, see video below for the good old days when [H] was the top team during the BigAdv days. Ok, the video does show the rise (2013/2014) and fall of [H] team, nothing to cheer about. I think it reminds us of what we can achieve.





2021 Result (2nd BigAdv Anniversary Challenge)
1641224537007.png
 
Last edited:
Please note you'll need nvidia drivers that support cuda 11.x to use FAHcore 22 v0.0.18. If you have older drivers, the fahclient will use opencl on nvidia cards and your PPD will drop significantly.
If you are on linux, you will also need glibc 2.17 or later to use FAHcore 22 v0.0.18.

More info at the folding forum is here.
 
Please note you'll need nvidia drivers that support cuda 11.x to use FAHcore 22 v0.0.18. If you have older drivers, the fahclient will use opencl on nvidia cards and your PPD will drop significantly.
If you are on linux, you will also need glibc 2.17 or later to use FAHcore 22 v0.0.18.

More info at the folding forum is here.
And I noticed a boost on my GTX1060 last week when I updated a year old driver to newer one with Cuda 11.4 vs the older 11.1 . YMMV
 
If 1 of you all join our ETF Event (can still fold for Team 33), I might join in on this and help with 115 million PPD. Ask firedfly for details as his ETF Team needs a folder to fold 1 piece of hardware 24/7. Edit: spot filled
 
Last edited:
Who is Jac_Sac? Please drop a line here. We appreciate your FAH contribution. Keep it up ;)
 
I'll be in, in some capacity depending what else is going on in the crunching world. Why you ask? Because pututu is the man!!
cuarc001 too!
(y)
 
I think I'll join in, I'll have to reconfigure my mining cards to participate, spread them out so they run more efficiently instead of hammering a poor old Athlon 435 triple core!
 
I think I'll join in, I'll have to reconfigure my mining cards to participate, spread them out so they run more efficiently instead of hammering a poor old Athlon 435 triple core!
Time to mow down more slackers (include me, lol) ;)
 
Keep those gpus locked and loaded!

I'll try to publish daily top 10 or top 20 contributors when the challenge starts. Hopefully to give some motivations to the rest of the [H]ordes.
 
Keep those gpus locked and loaded!

I'll try to publish daily top 10 or top 20 contributors when the challenge starts. Hopefully to give some motivations to the rest of the [H]ordes.
After some experimenting, I will have to put the 3080Ti in my main PC with the 3060 in the second x16 slot to fully utilize them! 😲

Then I will have to put 3060Ti in primary x16 slot with 1070Ti in second x16 slot of the triple core Athlon to get moar PPD!
 
After some experimenting, I will have to put the 3080Ti in my main PC with the 3060 in the second x16 slot to fully utilize them! 😲

Then I will have to put 3060Ti in primary x16 slot with 1070Ti in second x16 slot of the triple core Athlon to get moar PPD!
Talking about effect of pcie lanes on folding, Nathan_P shared his study/result here. Feel free to share your result. I suspect the ampere cards might take a bit more hit with PCIE 3.0 x1.
 
wow, thats dragging up an old thread. I'd forgotten i'd done most of that work!
 
Talking about effect of pcie lanes on folding, Nathan_P shared his study/result here. Feel free to share your result. I suspect the ampere cards might take a bit more hit with PCIE 3.0 x1.
I know for sure that having a powerful enough CPU to feed GPUs information and mobo IO bandwidth seems to play a part.
The 2080 Super is on the x1 lane and still was able to reach 2.7 million PPD, so it doesn't seem restricted. The 3070 was not doing well in the riser card so I put it on the mobo and it gained about 1 million PPD. the proccessor is an ancient 6600K, 4 cores/8threads.
The classic Athlon 435 has PCI 2.0 slots, so it may limit in an x1 lane scenario. However, this PC is rock solid for stability, which is why it gets used primarily for mining. The Z170 system has some stability issue, intermittent, but it is there.
 
I will say one thing, for those complaining about mining damaging GPUs, I think those claims are unfounded because F@H gets the 3070 hot, but when mining it gets slightly warm on the backplate.
 
biodoc, thanks for setting this up.

To all [H] folders. If you are in US, challenge starts on Jan 23rd 7am/4am PT/ET time and ends on Jan 30th 7am/4am PT/ET.

Some tips:
  • gpu is better than cpu in folding for most people unless you have gazillion of cores.
  • average daily PPD is higher when running in linux than windows (up to 10%)
  • have one full thread per gpu during folding. I usually leave one full core.
If you are looking for gpu and cpu performance, visit folding.lar.systems. You can also install a chrome client to monitor your real time gpu/cpu performance or just use the standard FAH client from FAH. Remember to have a passkey to claim extra credits (bonus). If you need to install Folding in Windows 10, visit Gilthanis how-to video here. For other flavors of OS, see official FAH website.

Our current daily production is just over 100M PPD. Last year during the week long challenge, we average about 210M PPD. Hope to see more folks firing up their gpus during this upcoming challenge :).

Also this is for a good cause, with the most recent one relating to patent-free covid anti-viral development.

For nostalgic reason, see video below for the good old days when [H] was the top team during the BigAdv days. Ok, the video does show the rise (2013/2014) and fall of [H] team, nothing to cheer about. I think it reminds us of what we can achieve.





2021 Result (2nd BigAdv Anniversary Challenge)
View attachment 427898

The [H] got passed by EVGA, then caught up to and passed them, then passed the Default group; that is [H]ard! Then comes Curecoin from out of nowhere and passes everyone...
 
I know for sure that having a powerful enough CPU to feed GPUs information and mobo IO bandwidth seems to play a part.
The 2080 Super is on the x1 lane and still was able to reach 2.7 million PPD, so it doesn't seem restricted. The 3070 was not doing well in the riser card so I put it on the mobo and it gained about 1 million PPD. the proccessor is an ancient 6600K, 4 cores/8threads.
The classic Athlon 435 has PCI 2.0 slots, so it may limit in an x1 lane scenario. However, this PC is rock solid for stability, which is why it gets used primarily for mining. The Z170 system has some stability issue, intermittent, but it is there.
Yeah, i'm resolving all the bandwidth and PCIe bandwidth issues when the next folding rig gets upgraded. It will be my current threadripper system, I'm going to get something a bit newer for day to day use.
 
Hey guys,
First time posting on the [H] forums ever, active EHW member and pretty dedicated to F@H for various reasons.
firedfly made me aware of this competition. He's been helping our team for some time, so I felt the need to donate the output from some of my hardware for this competition.
Planning to fire up at least a 3080Ti and a 2080Ti here, maybe more.. :rolleyes:
 
Hey guys,
First time posting on the [H] forums ever, active EHW member and pretty dedicated to F@H for various reasons.
firedfly made me aware of this competition. He's been helping our team for some time, so I felt the need to donate the output from some of my hardware for this competition.
Planning to fire up at least a 3080Ti and a 2080Ti here, maybe more.. :rolleyes:
Welcome! I've been folding for the [H] for years, though I stopped a couple of years ago, tight on money and running a CPU and GPU 24/7 was adding to electric bill. I'm retooling and testing so I can hopefully bring 10-15 million PPD when I switch my ETH mining GPUS over to F@H.
 
Agent_N on that last screenshot, I would try testing the PPD with less CPU threads on the CPU slot. In the last FAH competition we had on our side (I'm from TeAm Anandtech, the competitor here, so take what I say with a grain of salt :ROFLMAO: ) , we noticed the conventional wisdom of "number of threads minus 1" when setting up CPU slots wasn't necessarily still correct. I'm not sure if it's changes they've made to FAH or what, but some of us noticed if we didn't reserve more CPU than 1 thread it would lower the GPU PPD. I guess I don't know what CPU you have in there, so maybe it's not set as 'cpu threads minus 1' currently but that was my assumption with it saying 15 and your signature saying you have a 3800x. I would try something like 7 or 8 CPU threads instead and see where the PPD ends up if that's on your 3800x machine. I would imagine it varies based on video card too, so maybe a 3060 doesn't need as much CPU I'm not sure.
 
Welcome! I've been folding for the [H] for years, though I stopped a couple of years ago, tight on money and running a CPU and GPU 24/7 was adding to electric bill. I'm retooling and testing so I can hopefully bring 10-15 million PPD when I switch my ETH mining GPUS over to F@H.
Thanks! I had some downtime too, used to fold for OCN. Started again over there 2 years ago, but switched over to EHW and continued folding over there. Slowly making my way up the ranks :)
10-15 mil is a good amount!

Agent_N on that last screenshot, I would try testing the PPD with less CPU threads on the CPU slot. In the last FAH competition we had on our side (I'm from TeAm Anandtech, the competitor here, so take what I say with a grain of salt :ROFLMAO: ) , we noticed the conventional wisdom of "number of threads minus 1" when setting up CPU slots wasn't necessarily still correct. I'm not sure if it's changes they've made to FAH or what, but some of us noticed if we didn't reserve more CPU than 1 thread it would lower the GPU PPD. I guess I don't know what CPU you have in there, so maybe it's not set as 'cpu threads minus 1' currently but that was my assumption with it saying 15 and your signature saying you have a 3800x. I would try something like 7 or 8 CPU threads instead and see where the PPD ends up if that's on your 3800x machine. I would imagine it varies based on video card too, so maybe a 3060 doesn't need as much CPU I'm not sure.
It's good to test a GPU without folding on the CPU first. For example, if I fold on my CPU even at 4 threads(16 core) it will reduce my GPU ppd more than what I gain from the CPU. So I fold GPU only, but they are all high end cards anyways.
With mid range cards it might have less of an impact on the GPU.
 
Last edited:
Hey guys,
First time posting on the [H] forums ever, active EHW member and pretty dedicated to F@H for various reasons.
firedfly made me aware of this competition. He's been helping our team for some time, so I felt the need to donate the output from some of my hardware for this competition.
Planning to fire up at least a 3080Ti and a 2080Ti here, maybe more.. :rolleyes:
Welcome to [H] forum. Thanks for lending a hand to this good cause. (y)

In case you are not aware, this is to commemorate the ending of BigAdv expt on Jan 31st.
 
Agent_N on that last screenshot, I would try testing the PPD with less CPU threads on the CPU slot. In the last FAH competition we had on our side (I'm from TeAm Anandtech, the competitor here, so take what I say with a grain of salt :ROFLMAO: ) , we noticed the conventional wisdom of "number of threads minus 1" when setting up CPU slots wasn't necessarily still correct. I'm not sure if it's changes they've made to FAH or what, but some of us noticed if we didn't reserve more CPU than 1 thread it would lower the GPU PPD. I guess I don't know what CPU you have in there, so maybe it's not set as 'cpu threads minus 1' currently but that was my assumption with it saying 15 and your signature saying you have a 3800x. I would try something like 7 or 8 CPU threads instead and see where the PPD ends up if that's on your 3800x machine. I would imagine it varies based on video card too, so maybe a 3060 doesn't need as much CPU I'm not sure.
This might be worth exploring. My LHR 3060 gets over 3mil PPD on most projects on my R3600 with only WUProp running on CPU currently. It is OC'd with 120% pwr limit, +130 core and +1100 mem but folding doesn't get the clocks as high as gaming. OC'd or not you might be leaving PPD on the table.
 
Agent_N on that last screenshot, I would try testing the PPD with less CPU threads on the CPU slot. In the last FAH competition we had on our side (I'm from TeAm Anandtech, the competitor here, so take what I say with a grain of salt :ROFLMAO: ) , we noticed the conventional wisdom of "number of threads minus 1" when setting up CPU slots wasn't necessarily still correct. I'm not sure if it's changes they've made to FAH or what, but some of us noticed if we didn't reserve more CPU than 1 thread it would lower the GPU PPD. I guess I don't know what CPU you have in there, so maybe it's not set as 'cpu threads minus 1' currently but that was my assumption with it saying 15 and your signature saying you have a 3800x. I would try something like 7 or 8 CPU threads instead and see where the PPD ends up if that's on your 3800x machine. I would imagine it varies based on video card too, so maybe a 3060 doesn't need as much CPU I'm not sure.
I’m on top of that stuff, I have been tweaking and experimenting and didn’t find that the PPD was reduced any appreciable amount. I found other issues that were factors in the reduced PPD for my rigs.
 
This might be worth exploring. My LHR 3060 gets over 3mil PPD on most projects on my R3600 with only WUProp running on CPU currently. It is OC'd with 120% pwr limit, +130 core and +1100 mem but folding doesn't get the clocks as high as gaming. OC'd or not you might be leaving PPD on the table.
I haven’t started OCing too heavily, I have found +100-150 core and +1100 or so memory and increasing PL increases performance. I do have enough cards that I have to watch A/C circuit draw… also, various projects produce various PPD and I haven’t ran F@H for a bit so I may not be getting the best PPD WUs. I’ll look into the CPU threads a bit more when I have a moment to tinker some more.
 
Last edited:
He means 40 million since he just bought another 3080 ti. or 30 million it is ;)
Might be less. I (foolishly?) ordered a water cooled 3080ti and have never setup a water cooled system before (don't have the parts or knowledge). So, it's doubtful that card will be up before the competition. Meanwhile, a friend is buying one of my existing 3080ti's. So, I'll be down a 3080ti.

However, I just got a notification that I can buy the new 12gb EVGA 3080TI FTW3 card that came out this week. So, that one should be running in time!

I feel like I'm on a roller coaster right now.
 
Might be less. I (foolishly?) ordered a water cooled 3080ti and have never setup a water cooled system before (don't have the parts or knowledge). So, it's doubtful that card will be up before the competition. Meanwhile, a friend is buying one of my existing 3080ti's. So, I'll be down a 3080ti.

However, I just got a notification that I can buy the new 12gb EVGA 3080TI FTW3 card that came out this week. So, that one should be running in time!

I feel like I'm on a roller coaster right now.
I just put in for a step up to the new 3080 12GB, but I highly doubt that I will get it before the competition
 
Recruit your friend to fold with their new to them 3080ti.
 
  • Like
Reactions: BWG
like this
Might be less. I (foolishly?) ordered a water cooled 3080ti and have never setup a water cooled system before (don't have the parts or knowledge). So, it's doubtful that card will be up before the competition. Meanwhile, a friend is buying one of my existing 3080ti's. So, I'll be down a 3080ti.

However, I just got a notification that I can buy the new 12gb EVGA 3080TI FTW3 card that came out this week. So, that one should be running in time!

I feel like I'm on a roller coaster right now.

I will trade toy my 3080ti ftw3 for your hydrocopper
 
I will trade toy my 3080ti ftw3 for your hydrocopper

For now, I plan to dip my toes into water cooling. It will just take longer than a few weeks before I get things setup. If I change my mind, I'll let you know :)

Recruit your friend to fold with their new to them 3080ti.

I'll see if I can convince him. Meanwhile, I think I may have someone else willing to devote a card towards the competition.
 
Me :) Just let me know if there's anything else I can do to help, but otherwise, I'll keep folding as long as I can
Woo hoo. Thanks for dropping by. Just keep crunching and see how we fare against our good friend from TAAT. ;)
We lost to them twice in the first two races.
 
Back
Top