Folding Community Board

[RollinLower]

26 minutes ago, yaboistar said:

so i might've just pulled the trigger on a Vega64 Silver edition

we'll see how this one goes lmao

ooh, please do report core 22 performance on that thing!

[Metallus97]

Can anyone give me a tip on how to remove a stuck WU that won’t upload for days?

i tried removing and reading the slot -> didn’t help...

It’s a C22/11738

collection server is 128.252.203.4

[Metallus97]

9 minutes ago, yaboistar said:

have you restarted client to get it to make a new connection

yup... also the whole system and I removed the slots and re added em. Didn't help but made me loose some work  

[Gorgon]

7 hours ago, Metallus97 said:

yup... also the whole system and I removed the slots and re added em. Didn't help but made me loose some work  

Set the slots to finish then stop the FAHClient executable (using exit from the TaskBar item). Go into the work Directory:

C:\Users\<user_name>\AppData\Roaming\FAHClient\work

and delete any folder(s)

 

Restart FAHClient

[Favebook]

This 11737 (not sure which PRCG exactly) really is power hungry... With 1080 Ti, I get up to 1.7M from 1.1M on 14269.

However, 11737 is pinning my GPU up to 140% PL and is pushing it to 77C where 14269 uses maximum 105% PL and goes up to 45-50C.

 

Mathematically it is worth it but I think I will have to clock it down a bit so it doesn't hit a thermal limit especially when summer comes.

[Metallus97]

15 hours ago, Favebook said:

This 11737 (not sure which PRCG exactly) really is power hungry... With 1080 Ti, I get up to 1.7M from 1.1M on 14269.

However, 11737 is pinning my GPU up to 140% PL and is pushing it to 77C where 14269 uses maximum 105% PL and goes up to 45-50C.

 

Mathematically it is worth it but I think I will have to clock it down a bit so it doesn't hit a thermal limit especially when summer comes.

or build a big firiggn loop!  

[Favebook]

1 hour ago, Metallus97 said:

or build a big firiggn loop!  

Or that!

Tho, my custom hardline loop failed and spilled the liquid all over my GPU, GPU riser, PSU and motherboard. I had to take apart and clean the computer that took me about 36h working hours to make. Meanwhile, my GPU block leaked as well due to my mistake... However, I got really lucky that everything is still working full speed after this mess, but now I have EKWB parts that I cannot use as I got no more tubes...

[RollinLower]

4 hours ago, Favebook said:

Or that!

Tho, my custom hardline loop failed and spilled the liquid all over my GPU, GPU riser, PSU and motherboard. I had to take apart and clean the computer that took me about 36h working hours to make. Meanwhile, my GPU block leaked as well due to my mistake... However, I got really lucky that everything is still working full speed after this mess, but now I have EKWB parts that I cannot use as I got no more tubes...

to be honest if you are gonna do a work pc/server, like a F@H box, you should always go for flexible tubing anyway.

makes maintenance, upgrading/swapping parts and just general QOL a lot nicer.

[Favebook]

2 hours ago, RollinLower said:

to be honest if you are gonna do a work pc/server, like a F@H box, you should always go for flexible tubing anyway.

makes maintenance, upgrading/swapping parts and just general QOL a lot nicer.

Oh of course, but this was my main gaming/work and overall PC. I wanted it too look nice and to be made by me.

 

Other systems that I have are mostly striped mining rigs converted to F@H systems and heaters over winter.

[GOTSpectrum]

That time when you forget to buy gas... well at least GPU temps are looking fine on the 1080ti

 

[minibois]

F@H decided to give me a whole bunch of Core 21 WU's and it tanked my scores 

[minibois]

9 minutes ago, yaboistar said:

are you on the beta flag?

For sure!

 

I've just been given more and more C21 WU's these last couple days. Well, at least when I am looking at the F@H program.

Like right now I have a 0x21 WU running. Since I am not running my PC 24/7 it's difficult to see how many 0x22 vs. 0x21 WU's I have been getting, but I do know on 0x22 is get around 1.8m PPD vs. 1.2-1.3m PPD on 0x21

[GOTSpectrum]

Erm... you what mate?

 

[minibois]

3 hours ago, yaboistar said:

okay yeah you're running beta, i'm getting this too on my 2080ti system, it looks like they've introduced some more c21 units and they may be putting them in for initial testing just to make sure there's nothing wrong with them before moving them over to either the general pool or advanced. happens from time to time

 

https://apps.foldingathome.org/psummary

take a look at this if you want an idea of what projects they've got going on at any one time

Must be it.

The WU's were strangely tiny as well, to the point of it only taking 30 - 40 minutes to complete  

 

Thankfully I got a 0x22 WU now, so my PPD 'skyrocketed' closer to 2M PPD anyways. But that is one of those rare high value WU's/

[TVwazhere]

Anyone else have a small spike Sunday? ITNOS has been on a steady 1.5M PPD trend line since the end of the last Foldathon, so this is a bit of a departure from the norm. (Not using beta WU's)

[TVwazhere]

14 hours ago, yaboistar said:

Are you on advanced? You might’ve got a chain of a few more c22’s than usual

You mean the Advanced client? Yeah I'm using the advanced client. I could check the logs if they go back far enough.

[GOTSpectrum]

Do we think the 2060 KO will have a boosted Folding performance like in blender as reported by GN? 

 

I've got a limited window in which I can nab one for a decent price so if anyone can give hard data please HMU as soon as you can. 

 

 

[TVwazhere]

Still on track for the Platinum badge by the end of March!!!

(Red and blue are @Favebook and @yaboistar)

[GOTSpectrum]

6 minutes ago, yaboistar said:

i saw that too and wondered the same thing. i think it might be worth trying.

 

it's an interesting case nonetheless, because there's obviously something inside that chip which has been left enabled and is accelerating certain workloads. it seems to be the same core cut-down (not like the 1060 6gb/3gb) with 1920 cores, 48 ROPs, identical clocks and identical 6gb of GDDR6 on an identical 192 bit bus.

i'm wondering whether certain tiers of the turing chip have different microcode in them for managing the cores or memory, and perhaps this has been done to create an artificial performance defecit on the lower end cards, as here we see a cut down high end card clearly outperforming the stock card it's replicating in compute intensive workloads.

 

at the end of the day, if it doesn't perform any better or worse you could send it back

The TU-104 has a higher ratio of raster engines to CUDA Cores than the 106 due to the fact that the GPCs are laid out differently. each GPC has one raster engine, but on the 106 it has 12 SMs (384 CUDA Cores) per GPC, while on the 104 it has 8 (256 CUDA Cores). 

This may also mean that the pipelines on the 104 die are less congested than that of the 106, but this is just speculation as the level of my knowledge doesn't go this deep unfortunately. 

[GOTSpectrum]

Just now, yaboistar said:

i would suspect that may make a difference, less congestion would make for faster running in highly intensive workloads like compute.

That was exactly the logic I was going by with that comment. But, as I said I don't know if congestion is even a problem for team green like it has been for team red with their Stream processors stalling due to being memory starved and having a highly congested pipeline on high CU GCN parts. Not looked into RDNA as of yet in detail but it seems they have some of the same problems there too, just that they can have more Stream Processors before they start to feel it when compared to GCN.  

[Favebook]

1 hour ago, TVwazhere said:

Still on track for the Platinum badge by the end of March!!!

(Red and blue are @Favebook and @yaboistar)

Yeah, for a moment I saw @yaboistar on my threat list and I went HELL NO, and decided to pump up the production otherwise mine line would probably be closer to yours.

[Favebook]

59 minutes ago, yaboistar said:

i don't run 24h/day, i could theoretically push ~7.7mppd if i left things running constantly, but i'm happy with making about 3mppd. if i have a long weekend that figure does drop quite a lot as i'm not at home to run the machines and it's just my partner's 1070Ti running with a couple work units from my 1060m laptop

 

if i brought my v64 machine in and also added my old 970, i could hit roughly 9.3mppd

That's insane, how are you still below me? I am only running 2xR9 380, 1xRX580,1x1070 and 1 1080 Ti 24/7 and one 1080 Ti jumps in from time to time when I am not using my main computer.

[Favebook]

3 hours ago, yaboistar said:

 

@Favebook

It's time to RUN!

[RollinLower]

ah yes, i too like getting high PPD numbers.

300K on a Titan X and a 1070. noice

[Favebook]

2 hours ago, yaboistar said:

ah i remember the days when my redline was above 4500

Mine still is on Golf 4 2001