![]() |
[QUOTE=ewmayer;543980]Say I'm remotely logged in and wish to start or check up on multiple gpuowl runs. How do I manage that? Is that that what the Linux screen utility is useful for?[/QUOTE]
Screen is very handy for such situations. You can also do it raw (anywhere; nothing to install): [code] ./executable </dev/null >./exec.log 2>./exec.err & [/code] ...and then exit the shell (after everything's launched). You can then come back in and "tail -f exec.*" without risking the executable being HUP'ed if the SSH connection drops (network link failure, or your workstation being rebooted). |
I've had very good luck doing the initial Ubuntu install with the GPU active.
I do use screen. It is very effective. I have crontab call startup_reboot at boot time. startup_reboot for a 3 GPU system running 6 gpuowls looks like this: /etc/init.d/screen-cleanup sleep 10 screen -d -m -S mprime /home/george/Prime95/mprime -d /home/george/startup1 /home/george/startup2 /home/george/startup3 screen -d -m -S owl7 /home/george/gpuowl/tools/primenet.py -u gw_2 -p XXXX --dirs /home/george/gpuowl1 /home/george/gpuowl2 /home/george/gpuowl3 -w PRP --tasks 3 screen -d -m -S owl7a /home/george/gpuowl/tools/primenet.py -u gw_2 -p XXXX --dirs /home/george/gpuowl1a /home/george/gpuowl2a /home/george/gpuowl3a -w PRP --tasks 3 startup1 looks like this: echo "mypassword" | sudo -S /home/george/gpu-settings1 screen -d -m -S owl1 /home/george/gpuowl1/gpuowl -dir /home/george/gpuowl1 sleep 5 screen -d -m -S owl1a /home/george/gpuowl1a/gpuowl -dir /home/george/gpuowl1a gpu-settings1 looks like this: #!/bin/bash if [ "$EUID" -ne 0 ]; then echo "Radeon VII init script needs to be executed as root" && exit; fi #Sleep a few seconds to give the OS time to set up the GPU sleep 4 #Allow manual control echo "manual" >/sys/class/drm/card0/device/power_dpm_force_performance_level #Undervolt by setting max voltage # V Set this to 50mV less than the max stock voltage of your card (which varies from card to card), then optionally tune it down echo "vc 2 1801 1030" >/sys/class/drm/card0/device/pp_od_clk_voltage #Overclock mclk up to 1200 echo "m 1 1200" >/sys/class/drm/card0/device/pp_od_clk_voltage #Push a dummy sclk change for the undervolt to stick echo "s 1 1801" >/sys/class/drm/card0/device/pp_od_clk_voltage #Push everything to the card echo "c" >/sys/class/drm/card0/device/pp_od_clk_voltage #Put card into desired performance level /opt/rocm/bin/rocm-smi -d 0 --setsclk 4 --setfan 170 # 1050(881mV) mem 1200 working for 2 days # 1030(875mV) mem 1200 working for 10 days At any time I can look at the 6 gpuowls via "screen -r owl1" or "screen -r owl2a", etc. |
Some other useful commands:
* nohup, e.g. nuhup ./gpuowl & "no hang up", keep the process running on shell exit. * multitail, e.g. multitail dir1/gpuowl.log dir2/gpuowl.log To see live output from multiple processes * kill -INT <pid> Simulates a CTRL-C on the process, allows to clean shutdown gpuowl instances |
[QUOTE=preda;544002]Some other useful commands:
* nohup, e.g. nuhup ./gpuowl & "no hang up", keep the process running on shell exit. * multitail, e.g. multitail dir1/gpuowl.log dir2/gpuowl.log To see live output from multiple processes * kill -INT <pid> Simulates a CTRL-C on the process, allows to clean shutdown gpuowl instances[/QUOTE] Yes, absent the screen-utility option my plans was to use nohup to start the jobs, tail-log to monitor them and kill for my daily kill-and-restart-to-reduce-kworker-task-parasitism cycle. ====================== Houston, we have failure to launch ... 2 sticks [8GB each] ddr4 ram arrived today, mobo manual suggets using slots 1 and 2 [color-coded red] for 2-stticks, but I plugged into slots 3 and 4 [black] because those require shoving the main power cable to the side a smidge less. Took George's advice, plugged in one of the Radeon VIIs cards and hooked up the two 6+2 pin power connectors coming from the PSU. Finally carefully hooked up the 2-pin pwr and reset mini-switches and the same-bundle red [hdd activity] and green [power] LEDs. Hooked up to monitor via HDMI, keyboard and mouse via USB and inserted the Ubuntu boot-image thumb drive into an unoccupied USB port. Switched power switch on PSU on, no activity anywhere. Clicked pwr mini-switch, lots of LEDs came on on the mobo, GPU LEDs and fans also kicked in, as did PSU and water-cooler fans. The green pwr-on LED lit immediately; the red hdd-activity one was off at first then started blinking at regular intervals a little while later. But ... no activity on the monitor, no BIOS menu, nothing. Again powered off, tried switching SSD connector from the ASATA0 to the SATA0 one, no change on power-up. Powered off, unplugged the GPU and set aside, switched RAM modules to the recommded slots 1 and 2, again clicked power switch, same deal, just sans GPU, obviously. Double-checkd my hookups against the mobo-manual's Internal Connectors section, only thing that cuahgt my eye was that the guy who sold me the mobo, who said he'd applied fresh thermal grease and reseated the water cooler prior t shipment, had its 4-pin connector plugegd into the CPU_FAN one on the mobo, not the CPU_OPT (water cooling CPU fan header) one next to it. But if that were an issue, I wouldn't expect the cooler LEDs and fan to come on. Clues as to what-to-try-next welcome. |
Do you have the 8-pin CPU plug inserted to the motherboard?
Are two pci-e plugs inserted snugly into the graphics card? Is the graphics card plugged into slot 1, again snugly -- you said it was. Is the first port of the graphics card used for the hdmi plug? |
You can try to plug in a tiny speaker on the motherboard (these usually come from a computer case, or another motherboard) -- then the MB makes a series of beeps on boot-up which may help diagnose the problem. (Some MBs have a 2-digit LED that show a diagnosis code).
I would attempt in order: - re-seat the RAM (It's common to insert the RAM partially). - check you plugged the CPU power in the motherboard (this is a separate cable from the MB power) - [maybe] re-seat the CPU (?). This usually requires re-applying thermal paste. If you have another CPU known-good you can try that, or try a known-good MB with the CPU. Probably a good idea to get the MB speaker, and a known-good MB for comparison. |
When the system is (hopefully) "booting", does the caps lock light on the keyboard go on and off with the caps lock key? If so, the motherboard is probably working and the problem is likely in the video sub-system.
Use the jumper on the motherboard to clear the CMOS. If it doesn't have one or you can't find it, remove the battery for a minute or two then try again. |
[QUOTE=paulunderwood;544018]Do you have the 8-pin CPU plug inserted to the motherboard?
Are two pci-e plugs inserted snugly into the graphics card? Is the graphics card plugged into slot 1, again snugly -- you said it was. Is the first port of the graphics card used for the hdmi plug?[/QUOTE] Both the 2x12 and 2x4-pin power connectors are firmly seated. By "graphics card" you mean the Radeon 7? In my first try yesterday that was firmly seated, the pci-e plastic locking tab was engaged, and a pair of auxiliary 6+2-pin connectors from the PSU inserted - as I said, on power-up the Radeon 7 lit up as expected. I had it plugged into the pci-e slot nearest the CPU, which on this mobo is labeled PCIEX16; the other 2 full-length slots are labeled PCIEX4 and PCIEX1_3 in order of increasing distance from the CPU. The hdmi adapter (my monitor uses VGA cable, so I have adapters for my Intel NUC and odroid, which lack VGA-outs) plug was always inserted into the hdmi port on the mobo back panel, since I intend to use the onboard gfx for the display and the Radeons for crunching. Since my first power-up failure yesterday, the Radeon has been yanked, there is nothing in any of the PCI slots. [QUOTE=preda;544019]You can try to plug in a tiny speaker on the motherboard (these usually come from a computer case, or another motherboard) -- then the MB makes a series of beeps on boot-up which may help diagnose the problem. (Some MBs have a 2-digit LED that show a diagnosis code). I would attempt in order: - re-seat the RAM (It's common to insert the RAM partially). - check you plugged the CPU power in the motherboard (this is a separate cable from the MB power) - [maybe] re-seat the CPU (?). This usually requires re-applying thermal paste. If you have another CPU known-good you can try that, or try a known-good MB with the CPU. Probably a good idea to get the MB speaker, and a known-good MB for comparison.[/QUOTE] I'll need to see if I can roust up a suitable speaker - maybe one of my mini-headsets. Already reseated the RAM yesterday after first power-up failure, at which point I powered off, yanked the Radeon card, reseated RAM and double-checked all the other connectors. I always make sure to push the RAM in until the locking tabs engage on their own, but just tried it again to make sure, no change on power-up. Don't have a spare mobo lying around, that and reseating CPU will be last-resort flailings. [QUOTE=PhilF;544031]When the system is (hopefully) "booting", does the caps lock light on the keyboard go on and off with the caps lock key? If so, the motherboard is probably working and the problem is likely in the video sub-system. Use the jumper on the motherboard to clear the CMOS. If it doesn't have one or you can't find it, remove the battery for a minute or two then try again.[/QUOTE] Nothing lights up on the keyboard on power-up, I watched for ~60 seconds. Don't have a spare 2-pin jumper anywhere in my PC toolkit, so just pulled the button-cell battery for ~5 mins, then popped that back in and tried another power-up ... no change. Is there a low-tech way to bridge the jumper, maybe with a pair of needlenose pliers? And I want to bridge it, then leave it bridged during power-up? Or bridge it while powered off? |
[QUOTE=ewmayer;544097] plug was always inserted into the hdmi port on the mobo back panel, since I intend to use the onboard gfx for the display and the Radeons for crunching.
[/QUOTE] I think once you plug in the graphics card it takes over as the main video -- try plugging in the card and the hdmi cable into the port nearest the board. HTH :smile: |
[QUOTE=paulunderwood;544104]I think once you plug in the graphics card it takes over [/QUOTE]Some motherboard/BIOS combinations do, some don't. I have a 6-slot mobo that does not, with up to 5 gpus in it. Some switch to a gpu automatically and it can not be disabled on some. Even if there's no monitor connected or driver working.
|
[QUOTE=paulunderwood;544104]I think once you plug in the graphics card it takes over as the main video -- try plugging in the card and the hdmi cable into the port nearest the board. HTH :smile:[/QUOTE]
Good thought - tried that, the board-nearest hdmi port on the R7 was uncovered, i.e. the previous owner had likely used it the same for driving the gfx. Tiny bit of progress - now the num-lock key lights up a few seconds after power-up, but still nothing on the monitor, and I double-checked that hdmi-to-vga-to-monitor connector chain. One other thing occurs to me - I made no attempt to format the new SSD I bought, just hooked it up to power and sata, figuring any needed formatting would get done after the bootloader read the Ubuntu image on the USB I plugged in. Is that correct? And, how do I clear the CMOS via the jumper - jump it during power-up? |
| All times are UTC. The time now is 12:59. |
Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.