LTT "Summer" Folding Sprint

[IkeaGnome]

51 minutes ago, Gorgon said:

I’m assuming these are in the Asus ASUS Rog Maximus Code XI.

Correct.

51 minutes ago, Gorgon said:

Try changing the PCIe16_3 slot from “auto” to x4 in BIOS. It shares 2 lanes with SATA6G_56 so this will disable those two ports.

This system is normally my Unraid server, but it's not doing that right now. I have the Unraid cache SSD in SATA6G_4, and Windows boot is in _2. I just shut down and swapped cards around. 1070 and 1080 swapped. 1070 is now in a 4x mining riser working just fine, one 1080 is in the second x16 slot working away seemingly fine and the second 1080 is in a x16 riser in the very bottom x16 slot and still isn't recognized. If I was 1080s around, which ever is in the top slot works just fine. It's something with lanes and that one slot. It's possible that slot it's self has been dead since I got the board and I just never noticed. This board hasn't been used with this many PCIE devices. Usually just a HBA in the top 1x slot, then 2070 then 1070 in the x16s. At this point, it's either dead slot or bad configuration in F@H advanced viewer's settings. It does seem like something is weird in my GPU settings. pci-slot and pci-bus getting reused. 

 

Spoiler

 

But even in that, there should be 5 GPUs. The 4 dedicated and UHD 630, which I'm not using for anything in F@H. GPUz and MSI afterburner aren't seeing that bottom slot either.

Spoiler

 

 

 

[dogwitch]

i am following said post. due to a mining mobo i have. that i run multi gpu on.

still a learning curve for me on that.

 

onto a current issue.

seems thermal past went or something. seeing i have really bad temps on 1 side of cpu die

near 10 c difference. so 1 of the main folding machine will be down for a few hours...

 

[LazyDev]

I'm beaten, but not defeated. My 3080 now has a 2080 Super floating in its face.

 

 

[Kilrah]

That has to be the jankiest card / setup in existence 

[dogwitch]

1 hour ago, Kilrah said:

That has to be the jankiest card / setup in existence 

you not seen mine yet.

[RollinLower]

7 hours ago, LazyDev said:

I'm beaten, but not defeated. My 3080 now has a 2080 Super floating in its face.

 

those caps look jank-tastic aswell! was this for power stability or is this a repaired card back from the dead?

[Gorgon]

18 hours ago, IkeaGnome said:

Correct.

This system is normally my Unraid server, but it's not doing that right now. I have the Unraid cache SSD in SATA6G_4, and Windows boot is in _2. I just shut down and swapped cards around. 1070 and 1080 swapped. 1070 is now in a 4x mining riser working just fine, one 1080 is in the second x16 slot working away seemingly fine and the second 1080 is in a x16 riser in the very bottom x16 slot and still isn't recognized. If I was 1080s around, which ever is in the top slot works just fine. It's something with lanes and that one slot. It's possible that slot it's self has been dead since I got the board and I just never noticed. This board hasn't been used with this many PCIE devices. Usually just a HBA in the top 1x slot, then 2070 then 1070 in the x16s. At this point, it's either dead slot or bad configuration in F@H advanced viewer's settings. It does seem like something is weird in my GPU settings. pci-slot and pci-bus getting reused. 

If GPUz isn’t seeing the GPU in the bottom slot then the FAHClient never will.

 

what happens when you plug a GPU only into the bottom slot without a PCIe riser? Does it show up in the BIOS? Does it show up in GPUz?
 

Could be a bad riser or something to do with the IOMMU groups on that motherboard conflicting.

 

Ive run 6 GPUs on z470/570 boards with PCIe bifurcation and m.2 to PCIe adaptors under Linux and Windows. There is a limit in the FAHClient apparently of 10 GPUs in a system.

[cbigfoot]

@Pezui Why you always breaking the script...For shame...

[GOTSpectrum]

6 minutes ago, cbigfoot said:

Pezui Why you always breaking the script...For shame...

@Pezui

 

[GOTSpectrum]

Bloggy Bloggy

 

 

[cbigfoot]

@GOTSpectrum @Kilrah @mindstormer7 excuse me Freight train coming through....

 

[dogwitch]

and while i was a sleep. power went out. took 75% of  my folding ability offline for 10 hours...

wonderful

[Kilrah]

Getting up this morning to see a 2Mpt dip from the previous days, looks like F@H had issues serving WUs for about 4 hours. Others have a dip too so seems it was widespread...

[cbigfoot]

1 hour ago, Kilrah said:

Getting up this morning to see a 2Mpt dip from the previous days, looks like F@H had issues serving WUs for about 4 hours. Others have a dip too so seems it was widespread...

Checked the logs on my server babysitter actually got activated for the first time in a long time. She got triggered two times on my 3070 and two times on my 2060 Overnight. 

[TVwazhere]

12 hours ago, cbigfoot said:

At this rate I may catch @marknd59, but just barely

[Cester]

Welp, had a bug where the GPU workload refused to go beyond 99.9%, even after a preserving reinstall of F@H, so I had to wipe it all and give up those points. Oh well. At least we're running along nicely again. Also, to the person who lost a 2060, damn, that sucks. Hope things go well beyond that!

[dogwitch]

i was hoping to hit 20 mill mark. but stupid power out.  well i wont be able to get to that now.... .

[marknd59]

1 hour ago, TVwazhere said:

At this rate I may catch @marknd59, but just barely

  I should have kept hold of those 2x RTX2060 a bit longer.

[TVwazhere]

39 minutes ago, marknd59 said:

  I should have kept hold of those 2x RTX2060 a bit longer.

I just have Pascal's Army

[RollinLower]

1 minute ago, TVwazhere said:

I just have Pascal's Army

well, seeing as some 1080Ti's are getting the same PPD as my 6900xt, you're far better off price to performance wise.

[TVwazhere]

18 minutes ago, RollinLower said:

well, seeing as some 1080Ti's are getting the same PPD as my 6900xt, you're far better off price to performance wise.

According to @LAR_Systems's database, 1080ti's average 1.8-2M PPD, 6900XT's are more like 2.6m (someones getting 3.9M PPD on linux???)

https://folding.lar.systems/gpu_ppd/brands/amd/folding_profile/navi_21_radeon_rx_6900_xt

[RollinLower]

17 minutes ago, TVwazhere said:

According to @LAR_Systems's database, 1080ti's average 1.8-2M PPD, 6900XT's are more like 2.6m (someones getting 3.9M PPD on linux???)

https://folding.lar.systems/gpu_ppd/brands/amd/folding_profile/navi_21_radeon_rx_6900_xt

Mines doing  about 2.1m currently

[TVwazhere]

2 minutes ago, RollinLower said:

Mines doing  about 2.1m currently

F

[RollinLower]

13 minutes ago, TVwazhere said:

F

scratch that, 1.9M actually

seems like my card is the opposite of a golden sample when it comes to folding. 

[Pezui]

20 hours ago, cbigfoot said:

@Pezui Why you always breaking the script...For shame...

I am back on my bullshit helping validate the code and enforcing sanitization of data inputs. Just kidding but after working validation for years I feel you on having something break and zero easy ways to fix it.