Need help selecting a Processor

[MeeLee]

Based on the relative performance comparison, if you got it folding it should meet all the deadlines and such, and I'd guess be worth 60-65K PPD. It's not much slower than the Vega 11 on my machine.

That being said, I don't mind letting this machine run 24/7 to fold. But if I had quick GPU's like you do, I think the energy would be better spend letting them fold more often. It should be even more power efficient in the long run, and why set up a slow rig to fold if it's not very efficient?

It could pave the way for ARM CPUs driving GPUs, basically the best of both worlds.
We're not there yet, but people like Jeff Beerling is slowly starting to get some GPUs work on the PCIE x1 slot on the Pi 4 (CM4).
It won't take long before all drivers will be updated, and a Pi 4 could run a lower end Nvidia GPU.
I don't think we'll see it drive RTX GPUs anytime soon (other than for bitcoin mining, or The Collatz Conjecture, that use very little CPU and PCIE bandwidth).

[Tuvar]

As promised I made a small spreadsheet and notes on what I came up with. Work and the holidays had me a bit distracted so it's a little late where I forgot to upload it. As it was stated you can only do a 1:1 ratio of GPU's to CPU threads. Unfortunately it does lock them out even though the real usage would accommodate for a higher GPU count if it did not. Since I had ordered the 6 cards before being aware (My fault for not researching it better) I went a head and tested it since some people seemed to allude to the more recent update changing this possibly. Overall as always I enjoyed the build none the less and when I build my NAS server I might even move them over to it. Next project is to see how well a Tesla M40 or K80 does. I don't expect great things from them but I'm going to use it to dink around with ML/AI so I figured I'd spin it up since it's going into my VM server anyhow. This may be helpful or maybe not, I'm also sure some of the notes are somewhat common sense for most people who have folded for awhile. It's habit and for anyone who may stumble across this down the line though.

Code: Select all

# of P106-090	Slot's Used (16x or 1x)	Power %	Core Clock Offset	Memory Clock Offset  	Est PPD in Thousands	      Average (PPD/# of Cards)	    Wattage (33 idle)			System Specs
2	                Two 16x	                        80	        50	                        200	                                415	                              207.5			                                                                Motherboard	           MSI z170 Gaming Plus
2	                Two 16x	                        80	        100	                        200	                                661	                              330.5			                                                                Processor	                   Intel i3-7100
2	                Two 16x	                        80	        125	                        200	                                589-680	                      295-340	                               156		                        Memory	                   2x4Gb RAM 
4	                Two 16x Two 1x	                80	        125	                        200	                                1200	                              300	                                       295		                        GPU	One thru Six         Geforce P106-090 Mining GPU
                                                                                                                                                                                                                                                                                        Power Supply	            850W EVGA G2 Gold
									                                                                                                                                                                                                                Wattage Meter	            Kill A Watt P3
                                                                                                                                                                                                                                                                                        OS	                            Window 10 Rev 20H2
									                                                                                                                                                                                                                FAHClient	                    7.6.21
									                                                                                                                                                                                                                Cause Preference	    Alzheimer's

Disclaimer: This is not a definitive guide, PPD will change depending on the WU, Cause Preference, Over/Under Clocking, and even room temperature if it doesn't cool your equipment properly. Heat is a natural enemy of electronics, keeping them cool increases efficiency and extends the life of your folding equipment.

        Note: After testing going beyond Core Clock offset of +125 no benefit noticed									
	Note: After testing going beyond Memory Clock of +200 no benefit was noticed									
	Note: 80% was the lowest Power Limit for these GPU's. Increasing it a long with Core or Memory Clock seemed to have no effect other than increased wattage and heat		
        Note: CPU Thread are a 1:1 ratio to GPU's, I tried 6, but the i3-7100 only has 4 threads therefore any more than 4 GPU's and the system overall took a hit and wasn't worth using so I pulled them and scaled it back to 4 total.

[Tuvar]

Ya know I formatted that into charts and it saves looking like crap

[toTOW]

You should try to use the

Code: Select all

 tags to keep formatting of your table ... it won't be perfect, but it will better than with nothing.

[Tuvar]

Cleaned it up a little bit.

[shill_team_6]

For a Jetson to do GPU folding there would need to be a folding core coded to use it, and driver support to allow that usage for computing as opposed to video display.

The bulk of the engineering work that you suggest needs to be completed to run GPU folding on Jetson is already completed as of September. (There was F@H CUDA blog post, but I lack permission to include URLs.)

The OpenMM project (and by extension, the core22 folding core) builds and tests well on Jetson GPU via CUDA. There must be additional conditions imposed by FAHClient that restrict this folding core from running on Jetson GPU (e.g. GPUs.txt, perhaps?), as the fundamental core is functional on Jetson GPU. As far as I know, FAHClient remains closed-source, so this actual logic used to lock-out using core22/GPU/CUDA remains a mystery.

It would be nice for the Jetson users if whatever that final switch is could be enabled for Jetson GPU.

[PantherX]

Welcome to the F@H Forum shill_team_6,

...The OpenMM project (and by extension, the core22 folding core) builds and tests well on Jetson GPU via CUDA. There must be additional conditions imposed by FAHClient that restrict this folding core from running on Jetson GPU (e.g. GPUs.txt, perhaps?), as the fundamental core is functional on Jetson GPU. As far as I know, FAHClient remains closed-source, so this actual logic used to lock-out using core22/GPU/CUDA remains a mystery...

If OpenMM builds and tests well on Jetson GPU via CUDA, that's a good start. However, FahCore_22 does use a tweaked version so we would have to test that out. Moreover, the restriction of GPU comes from the Servers. We don't know what the PCIe Device ID is so we can't add it to the list. You can provide it here but it may not get added to the list just yet. Moreover, can you actually run FAHClient on it? That would be the first step.

Considering that it's a new hardware type, additional work would be involved to ensure that whatever data it processes, generates valid data. I don't think that the primary developer of FahCore_22 has access to Jetson.

Do note that there are plans to have a new version of FAHClient (and other components) that is open source which would allow the F@H Community to engage and contribute to it.

[shill_team_6]

Thank you PantherX for the warm welcome. I appreciate the insight shared in your reply.

We don't know what the PCIe Device ID is so we can't add it to the list. You can provide it here but it may not get added to the list just yet. Moreover, can you actually run FAHClient on it? That would be the first step.

FAHClient does run on Jetson. The CPU side of things is aarch64, so the recent ARM support in FAHClient helped a ton there. I have successfully ran/submitted some work units that run on the ARM CPU. The PCIe Device ID is a curious topic. As an embedded platform, the GPU is not identified by a PCIe link/ no link exists (at least, not identifiable by lspci). There are other tell-tell signs that a platform is Jetson, eg. availability of CUDA and Device 'Compute Capability' = 7.2 for Volta-generation Jetson products, 6.2 for Pascal-gen, 5.3 for Maxwell-gen.

Considering that it's a new hardware type, additional work would be involved to ensure that whatever data it processes, generates valid data. I don't think that the primary developer of FahCore_22 has access to Jetson.

Jetson is supported by nvidia as a CUDA platform, so I would expect that it goes through similar scrutiny to specific GPU models that may be similarly inaccessible to developers. (Except for the no-PCIe Device ID issue/workaround, which would definitely be easier with hardware access.)

[PantherX]

If FAHClient can start-up that's great! Would it be possible for you to share the log file which includes the first 100 lines for us to see what happens?

If FAHClient is unable to detect the GPU, which I suspect is the case, then it's not a trivial matter to modify the GPU detection.

[shill_team_6]

If FAHClient can start-up that's great! Would it be possible for you to share the log file which includes the first 100 lines for us to see what happens?

Here's the log from startup through the first WU.

Code: Select all

*********************** Log Started 2020-12-16T05:16:24Z ***********************
05:16:24:******************************* libFAH ********************************
05:16:24:         Date: Oct 20 2020
05:16:24:         Time: 20:36:48
05:16:24:     Revision: 5ca109d295a6245e2a2f590b3d0085ad5e567aeb
05:16:24:       Branch: master
05:16:24:     Compiler: GNU 8.3.0
05:16:24:      Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
05:16:24:               -fdata-sections -O3 -funroll-loops -fno-pie
05:16:24:     Platform: linux2 4.19.0-9-arm64
05:16:24:         Bits: 64
05:16:24:         Mode: Release
05:16:24:****************************** FAHClient ******************************
05:16:24:      Version: 7.6.21
05:16:24:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
05:16:24:    Copyright: 2020 foldingathome.org
05:16:24:     Homepage: https://foldingathome.org/
05:16:24:         Date: Oct 20 2020
05:16:24:         Time: 20:39:10
05:16:24:     Revision: 6efbf0e138e22d3963e6a291f78dcb9c6422a278
05:16:24:       Branch: master
05:16:24:     Compiler: GNU 8.3.0
05:16:24:      Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
05:16:24:               -fdata-sections -O3 -funroll-loops -fno-pie
05:16:24:     Platform: linux2 4.19.0-9-arm64
05:16:24:         Bits: 64
05:16:24:         Mode: Release
05:16:24:         Args: --child /etc/fahclient/config.xml --run-as fahclient
05:16:24:               --pid-file=/var/run/fahclient.pid --daemon
05:16:24:       Config: /etc/fahclient/config.xml
05:16:24:******************************** CBang ********************************
05:16:24:         Date: Oct 20 2020
05:16:24:         Time: 18:38:03
05:16:24:     Revision: 7e4ce85225d7eaeb775e87c31740181ca603de60
05:16:24:       Branch: master
05:16:24:     Compiler: GNU 8.3.0
05:16:24:      Options: -faligned-new -std=c++11 -fsigned-char -ffunction-sections
05:16:24:               -fdata-sections -O3 -funroll-loops -fno-pie -fPIC
05:16:24:     Platform: linux2 4.19.0-9-arm64
05:16:24:         Bits: 64
05:16:24:         Mode: Release
05:16:24:******************************* System ********************************
05:16:24:          CPU: Unknown
05:16:24:       CPU ID: 
05:16:24:         CPUs: 8
05:16:24:       Memory: 15.45GiB
05:16:24:  Free Memory: 14.87GiB
05:16:24:      Threads: POSIX_THREADS
05:16:24:   OS Version: 4.9
05:16:24:  Has Battery: false
05:16:24:   On Battery: false
05:16:24:   UTC Offset: 
05:16:24:          PID: 5970
05:16:24:          CWD: /var/lib/fahclient
05:16:24:           OS: Linux 4.9.140-tegra aarch64
05:16:24:      OS Arch: ARM64
05:16:24:         GPUs: 0
05:16:24:CUDA Device 0: Platform:0 Device:0 Bus:0 Slot:0 Compute:7.2 Driver:10.2
05:16:24:       OpenCL: Not detected: Failed to open dynamic library 'libOpenCL.so':
05:16:24:               libOpenCL.so: cannot open shared object file: No such file or
05:16:24:               directory
05:16:24:***********************************************************************
05:16:24:<config>
05:16:24:  <!-- Client Control -->
05:16:24:  <fold-anon v='true'/>
05:16:24:
05:16:24:  <!-- Folding Slot Configuration -->
05:16:24:  <gpu v='false'/>
05:16:24:
05:16:24:  <!-- Network -->
05:16:24:  <proxy v=':8080'/>
05:16:24:
05:16:24:  <!-- Slot Control -->
05:16:24:  <power v='full'/>
05:16:24:
05:16:24:  <!-- User Information -->
05:16:24:  <passkey v='*****'/>
05:16:24:  <team v=''/>
05:16:24:  <user v=''/>
05:16:24:
05:16:24:  <!-- Folding Slots -->
05:16:24:  <slot id='0' type='CPU'/>
05:16:24:</config>
05:16:24:Trying to access database...
05:16:24:Successfully acquired database lock
05:16:24:FS00:Initialized folding slot 00: cpu:8
[91m05:16:25:ERROR:WU00:FS00:Exception: Could not get IP address for assign1.foldingathome.org: Name or service not known[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Could not get IP address for assign2.foldingathome.org: Name or service not known[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Could not get IP address for assign3.foldingathome.org: Name or service not known[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Could not get IP address for assign4.foldingathome.org: Name or service not known[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Failed to find any IP addresses for assignment servers[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Could not get IP address for assign1.foldingathome.org: Name or service not known[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Could not get IP address for assign2.foldingathome.org: Name or service not known[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Could not get IP address for assign3.foldingathome.org: Name or service not known[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Could not get IP address for assign4.foldingathome.org: Name or service not known[0m
[91m05:16:25:ERROR:WU00:FS00:Exception: Failed to find any IP addresses for assignment servers[0m
******************************* Date: 2021-01-06 *******************************
00:53:42:WU00:FS00:Connecting to assign1.foldingathome.org:80
00:53:42:WU00:FS00:Assigned to work server 129.32.209.203
00:53:42:WU00:FS00:Requesting new work unit for slot 00: cpu:4 from 129.32.209.203
00:53:42:WU00:FS00:Connecting to 129.32.209.203:8080
00:53:42:WU00:FS00:Downloading 753.00KiB
00:54:04:WU00:FS00:Download 100.00%
00:54:04:WU00:FS00:Download complete
00:54:04:WU00:FS00:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:16929 run:35 clone:63 gen:68 core:0xa8 unit:0x0000003f000000440000422100000023
00:54:04:WU00:FS00:Starting
00:54:04:WU00:FS00:Running FahCore: /usr/bin/FAHCoreWrapper /var/lib/fahclient/cores/cores.foldingathome.org/lin/64bit-aarch64/a8-0.0.9/Core_a8.fah/FahCore_a8 -dir 00 -suffix 01 -version 706 -lifeline 5970 -checkpoint 15 -np 4
00:54:04:WU00:FS00:Started FahCore on PID 8072
00:54:04:WU00:FS00:Core PID:8085
00:54:04:WU00:FS00:FahCore 0xa8 started
00:54:04:WU00:FS00:0xa8:*********************** Log Started 2021-01-06T00:54:04Z ***********************
00:54:04:WU00:FS00:0xa8:************************** Gromacs Folding@home Core ***************************
00:54:04:WU00:FS00:0xa8:       Core: Gromacs
00:54:04:WU00:FS00:0xa8:       Type: 0xa8
00:54:04:WU00:FS00:0xa8:    Version: 0.0.9
00:54:04:WU00:FS00:0xa8:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
00:54:04:WU00:FS00:0xa8:  Copyright: 2020 foldingathome.org
00:54:04:WU00:FS00:0xa8:   Homepage: https://foldingathome.org/
00:54:04:WU00:FS00:0xa8:       Date: Oct 28 2020
00:54:04:WU00:FS00:0xa8:       Time: 22:19:53
00:54:04:WU00:FS00:0xa8:   Compiler: GNU 8.3.0
00:54:04:WU00:FS00:0xa8:    Options: -faligned-new -std=c++14 -fsigned-char -ffunction-sections
00:54:04:WU00:FS00:0xa8:             -fdata-sections -O3 -funroll-loops -fno-pie
00:54:04:WU00:FS00:0xa8:   Platform: linux2 4.15.0-108-generic
00:54:04:WU00:FS00:0xa8:       Bits: 64
00:54:04:WU00:FS00:0xa8:       Mode: Release
00:54:04:WU00:FS00:0xa8:       SIMD: arm_neon_asimd
00:54:04:WU00:FS00:0xa8:     OpenMP: ON
00:54:04:WU00:FS00:0xa8:       CUDA: OFF
00:54:04:WU00:FS00:0xa8:       Args: -dir 00 -suffix 01 -version 706 -lifeline 8072 -checkpoint 15 -np 4
00:54:04:WU00:FS00:0xa8:************************************ libFAH ************************************
00:54:04:WU00:FS00:0xa8:       Date: Oct 28 2020
00:54:04:WU00:FS00:0xa8:       Time: 22:19:24
00:54:04:WU00:FS00:0xa8:   Compiler: GNU 8.3.0
00:54:04:WU00:FS00:0xa8:    Options: -faligned-new -std=c++14 -fsigned-char -ffunction-sections
00:54:04:WU00:FS00:0xa8:             -fdata-sections -O3 -funroll-loops -fno-pie
00:54:04:WU00:FS00:0xa8:   Platform: linux2 4.15.0-108-generic
00:54:04:WU00:FS00:0xa8:       Bits: 64
00:54:04:WU00:FS00:0xa8:       Mode: Release
00:54:04:WU00:FS00:0xa8:************************************ CBang *************************************
00:54:04:WU00:FS00:0xa8:       Date: Oct 28 2020
00:54:04:WU00:FS00:0xa8:       Time: 22:19:08
00:54:04:WU00:FS00:0xa8:   Compiler: GNU 8.3.0
00:54:04:WU00:FS00:0xa8:    Options: -faligned-new -std=c++14 -fsigned-char -ffunction-sections
00:54:04:WU00:FS00:0xa8:             -fdata-sections -O3 -funroll-loops -fno-pie -fPIC
00:54:04:WU00:FS00:0xa8:   Platform: linux2 4.15.0-108-generic
00:54:04:WU00:FS00:0xa8:       Bits: 64
00:54:04:WU00:FS00:0xa8:       Mode: Release
00:54:04:WU00:FS00:0xa8:************************************ System ************************************
00:54:04:WU00:FS00:0xa8:        CPU: Carmel
00:54:04:WU00:FS00:0xa8:     CPU ID: NVIDIA Family 8 Model 0 Stepping 0
00:54:04:WU00:FS00:0xa8:       CPUs: 4
00:54:04:WU00:FS00:0xa8:     Memory: 15.45GiB
00:54:04:WU00:FS00:0xa8:Free Memory: 14.23GiB
00:54:04:WU00:FS00:0xa8:    Threads: POSIX_THREADS
00:54:04:WU00:FS00:0xa8: OS Version: 4.9
00:54:04:WU00:FS00:0xa8:Has Battery: false
00:54:04:WU00:FS00:0xa8: On Battery: false
00:54:04:WU00:FS00:0xa8: UTC Offset: -7
00:54:04:WU00:FS00:0xa8:        PID: 8085
00:54:04:WU00:FS00:0xa8:        CWD: /var/lib/fahclient/work
00:54:04:WU00:FS00:0xa8:********************************************************************************
00:54:04:WU00:FS00:0xa8:Project: 16929 (Run 35, Clone 63, Gen 68)
00:54:04:WU00:FS00:0xa8:Unit: 0x00000000000000000000000000000000
00:54:04:WU00:FS00:0xa8:Reading tar file core.xml
00:54:04:WU00:FS00:0xa8:Reading tar file frame68.tpr
00:54:04:WU00:FS00:0xa8:Digital signatures verified
00:54:04:WU00:FS00:0xa8:Calling: mdrun -c frame68.gro -s frame68.tpr -x frame68.xtc -cpt 15 -nt 4 -ntmpi 1
00:54:06:WU00:FS00:0xa8:Steps: first=34000000 total=34500000
00:54:39:WU00:FS00:0xa8:Completed 1 out of 500000 steps (0%)
00:57:05:WU00:FS00:0xa8:Completed 5000 out of 500000 steps (1%)
00:59:02:WU00:FS00:0xa8:Completed 10000 out of 500000 steps (2%)
01:00:52:WU00:FS00:0xa8:Completed 15000 out of 500000 steps (3%)
01:02:47:WU00:FS00:0xa8:Completed 20000 out of 500000 steps (4%)
01:04:34:WU00:FS00:0xa8:Completed 25000 out of 500000 steps (5%)
01:06:20:WU00:FS00:0xa8:Completed 30000 out of 500000 steps (6%)
01:08:05:WU00:FS00:0xa8:Completed 35000 out of 500000 steps (7%)
01:09:51:WU00:FS00:0xa8:Completed 40000 out of 500000 steps (8%)
01:11:37:WU00:FS00:0xa8:Completed 45000 out of 500000 steps (9%)
01:13:22:WU00:FS00:0xa8:Completed 50000 out of 500000 steps (10%)
01:15:07:WU00:FS00:0xa8:Completed 55000 out of 500000 steps (11%)
01:16:56:WU00:FS00:0xa8:Completed 60000 out of 500000 steps (12%)
01:18:41:WU00:FS00:0xa8:Completed 65000 out of 500000 steps (13%)
01:20:27:WU00:FS00:0xa8:Completed 70000 out of 500000 steps (14%)
01:22:12:WU00:FS00:0xa8:Completed 75000 out of 500000 steps (15%)
01:23:58:WU00:FS00:0xa8:Completed 80000 out of 500000 steps (16%)
01:25:43:WU00:FS00:0xa8:Completed 85000 out of 500000 steps (17%)
01:27:28:WU00:FS00:0xa8:Completed 90000 out of 500000 steps (18%)
01:29:13:WU00:FS00:0xa8:Completed 95000 out of 500000 steps (19%)
01:30:58:WU00:FS00:0xa8:Completed 100000 out of 500000 steps (20%)
01:32:43:WU00:FS00:0xa8:Completed 105000 out of 500000 steps (21%)
01:34:27:WU00:FS00:0xa8:Completed 110000 out of 500000 steps (22%)
01:36:13:WU00:FS00:0xa8:Completed 115000 out of 500000 steps (23%)
01:37:58:WU00:FS00:0xa8:Completed 120000 out of 500000 steps (24%)
01:39:43:WU00:FS00:0xa8:Completed 125000 out of 500000 steps (25%)
01:41:28:WU00:FS00:0xa8:Completed 130000 out of 500000 steps (26%)
01:43:13:WU00:FS00:0xa8:Completed 135000 out of 500000 steps (27%)
01:45:01:WU00:FS00:0xa8:Completed 140000 out of 500000 steps (28%)
01:46:53:WU00:FS00:0xa8:Completed 145000 out of 500000 steps (29%)
01:48:39:WU00:FS00:0xa8:Completed 150000 out of 500000 steps (30%)
01:50:25:WU00:FS00:0xa8:Completed 155000 out of 500000 steps (31%)
01:52:11:WU00:FS00:0xa8:Completed 160000 out of 500000 steps (32%)
01:53:57:WU00:FS00:0xa8:Completed 165000 out of 500000 steps (33%)
01:58:16:WU00:FS00:0xa8:Completed 170000 out of 500000 steps (34%)
02:01:57:WU00:FS00:0xa8:Completed 175000 out of 500000 steps (35%)
02:06:04:WU00:FS00:0xa8:Completed 180000 out of 500000 steps (36%)
02:11:15:WU00:FS00:0xa8:Completed 185000 out of 500000 steps (37%)
02:14:55:WU00:FS00:0xa8:Completed 190000 out of 500000 steps (38%)
02:18:34:WU00:FS00:0xa8:Completed 195000 out of 500000 steps (39%)
02:22:14:WU00:FS00:0xa8:Completed 200000 out of 500000 steps (40%)
02:25:48:WU00:FS00:0xa8:Completed 205000 out of 500000 steps (41%)
02:29:27:WU00:FS00:0xa8:Completed 210000 out of 500000 steps (42%)
02:33:07:WU00:FS00:0xa8:Completed 215000 out of 500000 steps (43%)
02:36:49:WU00:FS00:0xa8:Completed 220000 out of 500000 steps (44%)
02:40:31:WU00:FS00:0xa8:Completed 225000 out of 500000 steps (45%)
02:44:12:WU00:FS00:0xa8:Completed 230000 out of 500000 steps (46%)
02:47:54:WU00:FS00:0xa8:Completed 235000 out of 500000 steps (47%)
02:51:36:WU00:FS00:0xa8:Completed 240000 out of 500000 steps (48%)
02:55:17:WU00:FS00:0xa8:Completed 245000 out of 500000 steps (49%)
02:58:58:WU00:FS00:0xa8:Completed 250000 out of 500000 steps (50%)
03:02:37:WU00:FS00:0xa8:Completed 255000 out of 500000 steps (51%)
03:06:24:WU00:FS00:0xa8:Completed 260000 out of 500000 steps (52%)
03:09:55:WU00:FS00:0xa8:Completed 265000 out of 500000 steps (53%)
03:13:27:WU00:FS00:0xa8:Completed 270000 out of 500000 steps (54%)
03:16:28:WU00:FS00:0xa8:Completed 275000 out of 500000 steps (55%)
03:19:52:WU00:FS00:0xa8:Completed 280000 out of 500000 steps (56%)
03:21:36:WU00:FS00:0xa8:Completed 285000 out of 500000 steps (57%)
03:23:22:WU00:FS00:0xa8:Completed 290000 out of 500000 steps (58%)
03:25:07:WU00:FS00:0xa8:Completed 295000 out of 500000 steps (59%)
03:26:52:WU00:FS00:0xa8:Completed 300000 out of 500000 steps (60%)
03:28:38:WU00:FS00:0xa8:Completed 305000 out of 500000 steps (61%)
03:30:23:WU00:FS00:0xa8:Completed 310000 out of 500000 steps (62%)
03:32:08:WU00:FS00:0xa8:Completed 315000 out of 500000 steps (63%)
03:33:53:WU00:FS00:0xa8:Completed 320000 out of 500000 steps (64%)
03:35:38:WU00:FS00:0xa8:Completed 325000 out of 500000 steps (65%)
03:37:23:WU00:FS00:0xa8:Completed 330000 out of 500000 steps (66%)
03:39:20:WU00:FS00:0xa8:Completed 335000 out of 500000 steps (67%)
03:41:07:WU00:FS00:0xa8:Completed 340000 out of 500000 steps (68%)
03:42:52:WU00:FS00:0xa8:Completed 345000 out of 500000 steps (69%)
03:44:38:WU00:FS00:0xa8:Completed 350000 out of 500000 steps (70%)
03:46:29:WU00:FS00:0xa8:Completed 355000 out of 500000 steps (71%)
03:48:16:WU00:FS00:0xa8:Completed 360000 out of 500000 steps (72%)
03:50:06:WU00:FS00:0xa8:Completed 365000 out of 500000 steps (73%)
03:51:51:WU00:FS00:0xa8:Completed 370000 out of 500000 steps (74%)
03:53:36:WU00:FS00:0xa8:Completed 375000 out of 500000 steps (75%)
03:55:20:WU00:FS00:0xa8:Completed 380000 out of 500000 steps (76%)
03:57:06:WU00:FS00:0xa8:Completed 385000 out of 500000 steps (77%)
03:58:51:WU00:FS00:0xa8:Completed 390000 out of 500000 steps (78%)
04:00:36:WU00:FS00:0xa8:Completed 395000 out of 500000 steps (79%)
04:02:20:WU00:FS00:0xa8:Completed 400000 out of 500000 steps (80%)
04:04:04:WU00:FS00:0xa8:Completed 405000 out of 500000 steps (81%)
04:05:50:WU00:FS00:0xa8:Completed 410000 out of 500000 steps (82%)
04:07:37:WU00:FS00:0xa8:Completed 415000 out of 500000 steps (83%)
04:09:22:WU00:FS00:0xa8:Completed 420000 out of 500000 steps (84%)
04:11:08:WU00:FS00:0xa8:Completed 425000 out of 500000 steps (85%)
04:12:53:WU00:FS00:0xa8:Completed 430000 out of 500000 steps (86%)
04:14:39:WU00:FS00:0xa8:Completed 435000 out of 500000 steps (87%)
04:16:24:WU00:FS00:0xa8:Completed 440000 out of 500000 steps (88%)
04:18:09:WU00:FS00:0xa8:Completed 445000 out of 500000 steps (89%)
04:19:56:WU00:FS00:0xa8:Completed 450000 out of 500000 steps (90%)
04:21:41:WU00:FS00:0xa8:Completed 455000 out of 500000 steps (91%)
04:23:26:WU00:FS00:0xa8:Completed 460000 out of 500000 steps (92%)
04:25:10:WU00:FS00:0xa8:Completed 465000 out of 500000 steps (93%)
04:26:55:WU00:FS00:0xa8:Completed 470000 out of 500000 steps (94%)
04:28:41:WU00:FS00:0xa8:Completed 475000 out of 500000 steps (95%)
04:30:26:WU00:FS00:0xa8:Completed 480000 out of 500000 steps (96%)
04:32:11:WU00:FS00:0xa8:Completed 485000 out of 500000 steps (97%)
04:33:56:WU00:FS00:0xa8:Completed 490000 out of 500000 steps (98%)
04:35:41:WU00:FS00:0xa8:Completed 495000 out of 500000 steps (99%)
04:35:42:WU01:FS00:Connecting to assign1.foldingathome.org:80
04:35:42:WU01:FS00:Assigned to work server 129.32.209.203
04:35:42:WU01:FS00:Requesting new work unit for slot 00: cpu:4 from 129.32.209.203
04:35:42:WU01:FS00:Connecting to 129.32.209.203:8080
04:35:43:WU01:FS00:Downloading 759.00KiB
04:35:43:WU01:FS00:Download complete
04:35:43:WU01:FS00:Received Unit: id:01 state:DOWNLOAD error:NO_ERROR project:16931 run:34 clone:11 gen:76 core:0xa8 unit:0x0000000b0000004c0000422300000022
04:37:26:WU00:FS00:0xa8:Completed 500000 out of 500000 steps (100%)
04:37:26:WU00:FS00:0xa8:Saving result file ../logfile_01.txt
04:37:26:WU00:FS00:0xa8:Saving result file dhdl.xvg
04:37:26:WU00:FS00:0xa8:Saving result file frame68.gro
04:37:26:WU00:FS00:0xa8:Saving result file frame68.xtc
04:37:26:WU00:FS00:0xa8:Saving result file md.log
04:37:26:WU00:FS00:0xa8:Saving result file science.log
04:37:26:WU00:FS00:0xa8:Saving result file state.cpt
04:37:26:WU00:FS00:0xa8:Folding@home Core Shutdown: FINISHED_UNIT
04:37:27:WU00:FS00:FahCore returned: FINISHED_UNIT (100 = 0x64)
04:37:27:WU00:FS00:Sending unit results: id:00 state:SEND error:NO_ERROR project:16929 run:35 clone:63 gen:68 core:0xa8 unit:0x0000003f000000440000422100000023
04:37:27:WU00:FS00:Uploading 3.46MiB to 129.32.209.203
04:37:27:WU00:FS00:Connecting to 129.32.209.203:8080

If FAHClient is unable to detect the GPU, which I suspect is the case, then it's not a trivial matter to modify the GPU detection.

It appears the GPU is detected as a CUDA device, but it does not get included as a "GPU".

[PantherX]

Thanks for the log file. It seems that using the GPU would require FahCore_22 running on ARM platform. That's a massive undertaking and one that I am not sure would be feasible since there's other stuff in the pipeline for the future of FahCore_22. However, this could be revisited in future

[shill_team_6]

Thank you for the discussion. I'm hopeful for the future, but definitely understand the reservation given the required investment.

I tried to do some searching, but now figure its easier to just ask: Is FahCore_22 open source? if not, is there potential/discussion for it (or the 'next' core) to become open similar to / with future FAHClient?

[Joe_H]

OpenMM on which Core_22 is based is open source, the core itself is not. There is discussion and work being done on making the client fully open source, some parts already are.

The code the CPU folding cores are based on is also open source, Gromacs.

i don't know if there is any discussion on making the folding cores open source.

[xnupanic]

I was wondering if there were any other people here with an Xavier. Nice to see I am not alone and that I came to the same conclusions as here. Since FAHClient detects the CUDA device, does it matter that it cannot detect the CPU on PCI-E? Why isn't identifying the CUDA device enough? I'm genuinely curious about this.

Of course, then there is the problem of FahCore_22 not having an aarch64 binary...

[bruce]

I doubt the Xavier is going to work with FAH but I don't understand exactly what it's intended to do. FAH does not use AI processing capabilities. It uses the compute capabilities of the geometry engine in a GPU. It's designed specifically to use the Floating Point resources of a CPU and/or GPUs. Those GPUs that are intended to do AI processing have resources that FAH does not need or use.

The Xavier is not a CPU nor a supported GPU.