LTT Official Folding Month VI

[GOTSpectrum]

1 minute ago, RollinLower said:

over halfway there tho!

yeah...

 

When I did the first event with @palespartan I didn't really consider that I would be doing it for this long. 

[GOTSpectrum]

Deciding between using the money I saved up for another GPU or for the intended purpose (steam deck)

[TylerD321]

31 minutes ago, GOTSpectrum said:

Deciding between using the money I saved up for another GPU or for the intended purpose (steam deck)

Is it possible to download folding @ home onto a steam deck and how many PPD would you get? 

[GOTSpectrum]

1 minute ago, TylerD321 said:

Is it possible to download folding @ home onto a steam deck and how many PPD would you get? 

no idea, possible to download for sure, you just pull from a repo, if it will run, CPU yes, GPU unknown 

[TylerD321]

Just now, GOTSpectrum said:

no idea, possible to download for sure, you just pull from a repo, if it will run, CPU yes, GPU unknown 

for science?

[IkeaGnome]

34 minutes ago, GOTSpectrum said:

no idea, possible to download for sure, you just pull from a repo, if it will run, CPU yes, GPU unknown 

 

33 minutes ago, TylerD321 said:

for science?

Remind me this weekend and I'll science.

[Birblover12]

Does it matter what I make my username as long as I use a passkey and join a team? What happens if someone uses the same name? 

[GOTSpectrum]

4 minutes ago, Birblover12 said:

Does it matter what I make my username as long as I use a passkey and join a team? What happens if someone uses the same name? 

the passkey is your unique identifier, usernames are none-unique 

[Birblover12]

7 minutes ago, GOTSpectrum said:

the passkey is your unique identifier, usernames are none-unique 

interesting. So how would giveaways work then? If there's two people with the name "bob", how would you know who won lol, since the sign up form only has the forum username and folding username.

[GOTSpectrum]

2 minutes ago, Birblover12 said:

interesting. So how would giveaways work then? If there's two people with the name "bob", how would you know who won lol, since the sign up form only has the forum username and folding username.

it isn't something that has come up as of yet, there are unique identifiers on the data so we would check this against the uid in the raw data 

 

generally if you make it the same as your forum name it makes it easier in most situations

[IgniVellex]

what is this?

[GOTSpectrum]

3 minutes ago, IgniVellex said:

what is this?

please read the OP then ask any questions you have 

[gravitywave]

There once was a Tech Tips folder

Whose GPU's had become older

He felt quite ill

With his electricity bill

And bought newer cards that ran colder.

[GOTSpectrum]

3 minutes ago, gravitywave said:

There once was a Tech Tips folder

Whose GPU's had become older

He felt quite ill

With his electricity bill

And bought newer cards that ran colder.

this is basically me 

[Alex Atkin UK]

3 hours ago, GOTSpectrum said:

no idea, possible to download for sure, you just pull from a repo, if it will run, CPU yes, GPU unknown 

CPU yes, GPU no, there are no OpenCL drivers installed and as I recall Valve don't include them in the Windows bundle either.  I think when I tried the CPU was too slow to hit the timeout though, but not 100% on that.

 

I did notice on Windows F@H tried to fold on my 7800X3D iGPU, but I don't think that was going to hit the deadline either.

[podkall]

1 hour ago, gravitywave said:

There once was a Tech Tips folder

Whose GPU's had become older

He felt quite ill

With his electricity bill

And bought newer cards that ran colder.

I've seen it firsthand, 1050 barely radiated heat on full throttle, but the 1080ti I'm probably gonna run it with 65°C power limit when folding

[BiotechBen]

1 hour ago, podkall said:

I've seen it firsthand, 1050 barely radiated heat on full throttle, but the 1080ti I'm probably gonna run it with 65°C power limit when folding

I've got a p600 (effectively a 1050 afaik) and at full 40w it's sitting at 72C on its single slot cooler with 150k PPD (before bonus), and the 1080's pull 1.2MPPD at 185W and 84C (blower cooler)

[Gorgon]

I've been really impressed with the 4070ti's efficiencies. Seeing as those 2-slot InnoColor3D GPUs aren't easily accessible in North America I'm running 2 rigs with 2 x Zotac Trinity OCs so there's only about 2-3mm clearance between the backplate of the lower GPU and the upper GPU but the temps are quite reasonable when clock-limited to just 2205MHz:

above showing Temperature, Power, Gclock & Fan%

[Alex Atkin UK]

6 hours ago, GOTSpectrum said:

this is basically me 

If I weren't restricted from going to CEX in person I would have traded my 3080 for a 4080 as it hurts to fold on something that gets half the PPD for the same power consumption.

[Alex Atkin UK]

20 hours ago, BiotechBen said:

I know it's a rather small sample size, but apparently not? (P620 is a slightly higher clocked p600 with 4gb instead of 2gb)

https://folding.lar.systems/gpu_ppd/brands/nvidia/folding_profile/gp107gl_quadro_p620

 

That's interesting as certainly on newer cards its quite a difference:

 

The bottom being Linux.

 

I admit I haven't checked the TDP on those two cards to see if there is any differences, but usually 40x0 series perform pretty much the same as they hit the voltage limit not TDP.

A quick check shows the card on Linux is pulling 186W on this job but for some reason Windows is hitting 170W max and fluctuates a lot more.  Even if I stop the CPU job to make sure its not using up too much CPU it ping-pongs 172-176W.

 

I think it stalls for longer when hitting a checkpoint too, I wonder if thats the Windows PC using SATA SSD vs Linux on NVME?  Though that doesn't explain the lower utilisation in general.

My understanding is Windows leaves more spare GPU power to prevent the UI from stuttering whereas Linux gives no Fs.    So Windows its 90-95% GPU utilisation whereas Linux sits at 98% and doesn't really move.

 

Even more interesting when you consider I run voltage curve optimisation on Windows but you can't on Linux.

[Alex Atkin UK]

55 minutes ago, Gorgon said:

I've been really impressed with the 4070ti's efficiencies. Seeing as those 2-slot InnoColor3D GPUs aren't easily accessible in North America I'm running 2 rigs with 2 x Zotac Trinity OCs so there's only about 2-3mm clearance between the backplate of the lower GPU and the upper GPU but the temps are quite reasonable when clock-limited to just 2205MHz:

above showing Temperature, Power, Gclock & Fan%

I'd imagine it helps also that modern GPUs have that extra bit on the end that effectively allows air to flow through from one card to the other.

 

The Inno3D card is not exactly readily available in the UK either, I think I only found one retailer who now are out of stock.  Plus some cheeky so and so trying to flog the card on Amazon for £272 markup over RRP (almost 4080 price).

[BiotechBen]

38 minutes ago, Alex Atkin UK said:

That's interesting as certainly on newer cards its quite a difference:

 

The bottom being Linux.

 

I admit I haven't checked the TDP on those two cards to see if there is any differences, but usually 40x0 series perform pretty much the same as they hit the voltage limit not TDP.

A quick check shows the card on Linux is pulling 186W on this job but for some reason Windows is hitting 170W max and fluctuates a lot more.  Even if I stop the CPU job to make sure its not using up too much CPU it ping-pongs 172-176W.

 

I think it stalls for longer when hitting a checkpoint too, I wonder if thats the Windows PC using SATA SSD vs Linux on NVME?  Though that doesn't explain the lower utilisation in general.

My understanding is Windows leaves more spare GPU power to prevent the UI from stuttering whereas Linux gives no Fs.    So Windows its 90-95% GPU utilisation whereas Linux sits at 98% and doesn't really move.

 

Even more interesting when you consider I run voltage curve optimisation on Windows but you can't on Linux.

I feel like with the p600, it's a low enough power card that it's like doing a cold air intake on your '92 Honda Civic DX

[Alex Atkin UK]

2 minutes ago, BiotechBen said:

I feel like with the p600, it's a low enough power card that it's like doing a cold air intake on your '92 Honda Civic DX

It may also be Windows or the NVIDIA drivers behave different based on consumer vs professional cards.  Or only the card connected to a display might be throttled, I've never really tested that as modern motherboards always put the bottom x16 too far down the board to fit a second GPU and the second x16 (if you have it) gets obstructed by your main GPU.

[BiotechBen]

1 hour ago, Alex Atkin UK said:

It may also be Windows or the NVIDIA drivers behave different based on consumer vs professional cards.  Or only the card connected to a display might be throttled, I've never really tested that as modern motherboards always put the bottom x16 too far down the board to fit a second GPU and the second x16 (if you have it) gets obstructed by your main GPU.

At 40w max, and with seemingly locked down performance (nothing applies with afterburner) it seems like the "it's a professional card, it go brr at prescribed amount" seems to be the case.

(Currently in HP business desktop with integrated graphics doing display)

[Alex Atkin UK]

18 minutes ago, BiotechBen said:

At 40w max, and with seemingly locked down performance (nothing applies with afterburner) it seems like the "it's a professional card, it go brr at prescribed amount" seems to be the case.

(Currently in HP business desktop with integrated graphics doing display)

Now you've got me wanting to re-enable iGPU on my desktop to see if not using the dGPU for display is a factor.  I probably wont though, too much hassle.