Mining Rig

[foldy]

I know the big RAM usage projects are a mistake and John Chodera wants to reduce RAM usage again.
See thread viewtopic.php?f=38&t=35052

@bruce: system thrashing does it mean PPD goes down (how much) or it does not change PPD?

[MeeLee]

Yes it needs one CPU thread per GPU to feed it. RAM usage should be 0.5 GB per GPU + 2 GB for OS. So for your 6 GPUs it is 8 GB.

Your numbers are obsolete. The 0.5 GB per WU is on the low side. Yes, there still are some WUs that fit in 0.5 GB but I've also seen some that need 1.5 GB. Notice that we're now seeing projects that have 400K atoms PLUS. I had two of them at the same time. Even with my paging file on SSD, the system was thrashing.

But those WUs, are for fast GPUs, no?
They probably won't be assigned to a GTX 1060.
It is best for them to be assigned to RTX 2080 and higher.

Not sure if the OP was planning on putting 6x RTX GPUs in his system, as they will be PCIE bottlenecked (unless he has a server CPU and motherboard, allowing of at least 4PCIE lanes per GPU).

[foldy]

Only thing OP can do to optimize PPD is switching to Linux

[Joe_H]

But those WUs, are for fast GPUs, no?
They probably won't be assigned to a GTX 1060.
It is best for them to be assigned to RTX 2080 and higher.

By preference those WUs may get assigned to faster, higher end GPUs, but that is only a preference and not the only GPUs they will get assigned to

[Darth_Peter_dualxeon]

Hi! I use my old mining rig for folding, it has 6 P106 GPUs and I can do 1,7M PPD

Did you register a passkey for yourself? 1.7M is a bit small, my RTX2070 + 2 old xeon processors almost reach that level.
If you have passkey, and folded ahandful of workunits successfully, then you may get quick return bonus points with passkey, that is a factor of ~ 7x

You have 6 GPUs of Pascal architecture, it is not that old. I suppose it can do more.
(just check the bandwidth usage of pcie slots somehow, maybe pcie x1 risers are not that good for folding)

[MeeLee]

Hi! I use my old mining rig for folding, it has 6 P106 GPUs and I can do 1,7M PPD

Did you register a passkey for yourself? 1.7M is a bit small, my RTX2070 + 2 old xeon processors almost reach that level.
If you have passkey, and folded ahandful of workunits successfully, then you may get quick return bonus points with passkey, that is a factor of ~ 7x

You have 6 GPUs of Pascal architecture, it is not that old. I suppose it can do more.
(just check the bandwidth usage of pcie slots somehow, maybe pcie x1 risers are not that good for folding)

They're sufficient for GTX GPUs under Linux.

[bruce]

I think you're looking at the delay where a page that has been written to external storage cannot be deallocated instantaneously and the resource monitor has to list it as belonging to something. If you look at the log, FAHCore_xx is terminated and passes a return code to FAHClient before the new copy of FAHCore_xx starts. Both copies are not putting demands for memory at the same time. Certainly the new copy of FAHCore_22 is allocating memory pages rapidly even if the OS's garbage collection routines have not deallocated all of the pages used by the previous program yet. The allocation process has a higher priority than the deallocation processing.

[ajm]

Yes, you are right. A new request is sent to the server when the WU reaches 99%. And then it can happen that a new WU is downloaded and ready before the first one is even finished. But, as you say, the new one actually starts only afterwards. Not sure whether it would work well if there was not enough RAM available for both, though.

Here is a log showing this kind of situation with a 1050 ti:

Code: Select all

05:22:33:WU01:FS02:0x22:Completed 490000 out of 500000 steps (98%)
05:28:52:WU01:FS02:0x22:Completed 495000 out of 500000 steps (99%)
05:28:53:WU00:FS02:Connecting to assign1.foldingathome.org:80
05:28:53:WU00:FS02:Assigned to work server 206.223.170.146
05:28:53:WU00:FS02:Requesting new work unit for slot 02: RUNNING gpu:1:GP107 [GeForce GTX 1050 Ti]  2138 from 206.223.170.146
05:28:53:WU00:FS02:Connecting to 206.223.170.146:8080
05:29:05:WU00:FS02:Downloading 46.04MiB
05:29:11:WU00:FS02:Download 24.03%
05:29:17:WU00:FS02:Download 46.70%
05:29:23:WU00:FS02:Download 70.32%
05:29:29:WU00:FS02:Download 97.48%
05:29:30:WU00:FS02:Download complete
05:29:30:WU00:FS02:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:14201 run:130 clone:0 gen:30 core:0x22 unit:0x00000023cedfaa925eb99cbe4b110c77
05:35:12:WU01:FS02:0x22:Completed 500000 out of 500000 steps (100%)
05:35:27:WU01:FS02:0x22:Saving result file ..\logfile_01.txt
05:35:27:WU01:FS02:0x22:Saving result file checkpointState.xml
05:35:33:WU01:FS02:0x22:Saving result file checkpt.crc
05:35:33:WU01:FS02:0x22:Saving result file positions.xtc
05:35:33:WU01:FS02:0x22:Saving result file science.log
05:35:33:WU01:FS02:0x22:Folding@home Core Shutdown: FINISHED_UNIT
05:35:35:WU01:FS02:FahCore returned: FINISHED_UNIT (100 = 0x64)
05:35:35:WU01:FS02:Sending unit results: id:01 state:SEND error:NO_ERROR project:14201 run:826 clone:0 gen:22 core:0x22 unit:0x00000019cedfaa925eb99bb9e4ff9f23
05:35:35:WU01:FS02:Uploading 40.92MiB to 206.223.170.146
05:35:35:WU00:FS02:Starting
05:35:35:WU01:FS02:Connecting to 206.223.170.146:8080
05:35:35:WU00:FS02:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:\Users\ajmch\AppData\Roaming\FAHClient\cores/cores.foldingathome.org/v7/win/64bit/Core_22.fah/FahCore_22.exe -dir 00 -suffix 01 -version 706 -lifeline 13948 -checkpoint 15 -gpu-vendor nvidia -opencl-platform 0 -opencl-device 1 -cuda-device 1 -gpu 1
05:35:35:WU00:FS02:Started FahCore on PID 18204
05:35:35:WU00:FS02:Core PID:18380
05:35:35:WU00:FS02:FahCore 0x22 started
05:35:35:WU00:FS02:0x22:*********************** Log Started 2020-05-26T05:35:35Z ***********************
05:35:35:WU00:FS02:0x22:*************************** Core22 Folding@home Core ***************************
05:35:35:WU00:FS02:0x22:       Type: 0x22
05:35:35:WU00:FS02:0x22:       Core: Core22
05:35:35:WU00:FS02:0x22:    Website: https://foldingathome.org/
05:35:35:WU00:FS02:0x22:  Copyright: (c) 2009-2018 foldingathome.org
05:35:35:WU00:FS02:0x22:     Author: John Chodera <john.chodera@choderalab.org> and Rafal Wiewiora
05:35:35:WU00:FS02:0x22:             <rafal.wiewiora@choderalab.org>
05:35:35:WU00:FS02:0x22:       Args: -dir 00 -suffix 01 -version 706 -lifeline 18204 -checkpoint 15
05:35:35:WU00:FS02:0x22:             -gpu-vendor nvidia -opencl-platform 0 -opencl-device 1 -cuda-device
05:35:35:WU00:FS02:0x22:             1 -gpu 1
05:35:35:WU00:FS02:0x22:     Config: <none>
05:35:35:WU00:FS02:0x22:************************************ Build *************************************
05:35:35:WU00:FS02:0x22:    Version: 0.0.5
05:35:35:WU00:FS02:0x22:       Date: Apr 22 2020
05:35:35:WU00:FS02:0x22:       Time: 04:42:59
05:35:35:WU00:FS02:0x22: Repository: Git
05:35:35:WU00:FS02:0x22:   Revision: 2d69202c898bd9bb3e093f51cd32bf411c2a0388
05:35:35:WU00:FS02:0x22:     Branch: HEAD
05:35:35:WU00:FS02:0x22:   Compiler: Visual C++ 2008
05:35:35:WU00:FS02:0x22:    Options: /TP /nologo /EHa /wd4297 /wd4103 /Ox /MT
05:35:35:WU00:FS02:0x22:   Platform: win32 10
05:35:35:WU00:FS02:0x22:       Bits: 64
05:35:35:WU00:FS02:0x22:       Mode: Release
05:35:35:WU00:FS02:0x22:************************************ System ************************************
05:35:35:WU00:FS02:0x22:        CPU: Intel(R) Core(TM) i7-4930K CPU @ 3.40GHz
05:35:35:WU00:FS02:0x22:     CPU ID: GenuineIntel Family 6 Model 62 Stepping 4
05:35:35:WU00:FS02:0x22:       CPUs: 12
05:35:35:WU00:FS02:0x22:     Memory: 31.94GiB
05:35:35:WU00:FS02:0x22:Free Memory: 26.69GiB
05:35:35:WU00:FS02:0x22:    Threads: WINDOWS_THREADS
05:35:35:WU00:FS02:0x22: OS Version: 6.2
05:35:35:WU00:FS02:0x22:Has Battery: false
05:35:35:WU00:FS02:0x22: On Battery: false
05:35:35:WU00:FS02:0x22: UTC Offset: 2
05:35:35:WU00:FS02:0x22:        PID: 18380
05:35:35:WU00:FS02:0x22:        CWD: C:\Users\ajmch\AppData\Roaming\FAHClient\work
05:35:35:WU00:FS02:0x22:         OS: Windows 10 Pro
05:35:35:WU00:FS02:0x22:    OS Arch: AMD64
05:35:35:WU00:FS02:0x22:********************************************************************************
05:35:35:WU00:FS02:0x22:Project: 14201 (Run 130, Clone 0, Gen 30)
05:35:35:WU00:FS02:0x22:Unit: 0x00000023cedfaa925eb99cbe4b110c77
05:35:35:WU00:FS02:0x22:Reading tar file core.xml
05:35:35:WU00:FS02:0x22:Reading tar file integrator.xml
05:35:35:WU00:FS02:0x22:Reading tar file state.xml
05:35:37:WU00:FS02:0x22:Reading tar file system.xml
05:35:38:WU00:FS02:0x22:Digital signatures verified
05:35:38:WU00:FS02:0x22:Folding@home GPU Core22 Folding@home Core
05:35:38:WU00:FS02:0x22:Version 0.0.5
05:35:41:WU01:FS02:Upload 0.46%
05:35:47:WU01:FS02:Upload 9.78%
05:35:53:WU01:FS02:Upload 19.71%
05:35:59:WU01:FS02:Upload 29.63%
05:36:05:WU01:FS02:Upload 39.56%
05:36:11:WU01:FS02:Upload 49.49%
05:36:17:WU01:FS02:Upload 59.42%
05:36:23:WU01:FS02:Upload 69.20%
05:36:23:WU00:FS02:0x22:Completed 0 out of 500000 steps (0%)
05:36:23:WU00:FS02:0x22:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900
05:36:29:WU01:FS02:Upload 70.88%
05:36:35:WU01:FS02:Upload 80.81%
05:36:41:WU01:FS02:Upload 90.74%
05:36:48:WU01:FS02:Upload complete
05:36:48:WU01:FS02:Server responded WORK_ACK (400)
05:36:48:WU01:FS02:Final credit estimate, 100875.00 points
05:36:48:WU01:FS02:Cleaning up
05:42:47:WU00:FS02:0x22:Completed 5000 out of 500000 steps (1%)
05:49:07:WU00:FS02:0x22:Completed 10000 out of 500000 steps (2%)