May 2, 20251 yr 3 hours ago, ich777 said: Can you please post your Diagnostics? unraid-diagnostics-20250502-1611.zip
May 3, 20251 yr Author 15 hours ago, Plean said: Any help would be much appreciated First of all, what is this in your go file: # reload services after starting docker with 20 seconds grace period to allow starting up containers event=/usr/local/emhttp/webGui/event/docker_started mkdir -p $event cat <<- 'EOF' >$event/reload_services #!/bin/bash echo 'sleep 20;/usr/local/emhttp/webGui/scripts/reload_services' | at -M now 2>/dev/null EOF chmod +x $event/reload_services Are you sure you need this? Please remove /boot/config/modprobe.d/i915.conf and reboot. I don't know much about Battlemage but do you have a display or at least a HDMI dummy dongle installed? Are all the necessary power cables plugged in to the card?
May 3, 20251 yr Author 18 hours ago, carlks said: Updated to rc4 but no luck with that directory. Okay I see the issue now, you have to give me a bit since my time is really limited but I will definitely look into this further and let you know.
May 4, 20251 yr On 5/3/2025 at 5:25 AM, ich777 said: First of all, what is this in your go file: # reload services after starting docker with 20 seconds grace period to allow starting up containers event=/usr/local/emhttp/webGui/event/docker_started mkdir -p $event cat <<- 'EOF' >$event/reload_services #!/bin/bash echo 'sleep 20;/usr/local/emhttp/webGui/scripts/reload_services' | at -M now 2>/dev/null EOF chmod +x $event/reload_services Are you sure you need this? Please remove /boot/config/modprobe.d/i915.conf and reboot. I don't know much about Battlemage but do you have a display or at least a HDMI dummy dongle installed? Are all the necessary power cables plugged in to the card? I have a display plugged in an have tried re seating both the card and the power cables. I had a 1080ti running previously in that slot with that cable. I've removed the i1915.conf and rebooted to no avail On 5/3/2025 at 5:28 AM, ich777 said: Okay I see the issue now, you have to give me a bit since my time is really limited but I will definitely look into this further and let you know. I appreciated your help on this! No rush, I am already destined to buy you coffee for all your help
May 4, 20251 yr Author 1 hour ago, Plean said: I appreciated your help on this! No rush, I am already destined to buy you coffee for all your help This was not for your post... 1 hour ago, Plean said: I've removed the i1915.conf and rebooted to no avail Please create a post in the General subforums and create new Diagnostics. This is not related to intel_gpu_top since intel_gpu_top is just a application to show stats for your iGPU/GPU. However in your case I think it's a hardware related issue since it doesn't seem to load the driver properly.
May 4, 20251 yr Is this still the right topic for support with the Coral TPU drivers? I have a dual-edge TPU, meaning it contains two separate TPUs. This appears on the system as two PCIe devices: 05:00.0 System peripheral: Global Unichip Corp. Coral Edge TPU 06:00.0 System peripheral: Global Unichip Corp. Coral Edge TPU However, it seems like this plugin/driver is only using one of them? The Coral driver settings only shows "Coral TPU1", and I only have /dev/apex_0. /dev/apex_1 is missing. Any ideas?
May 5, 20251 yr Author 14 hours ago, Daniel15 said: Is this still the right topic for support with the Coral TPU drivers? Yes. 14 hours ago, Daniel15 said: I have a dual-edge TPU, meaning it contains two separate TPUs. This appears on the system as two PCIe devices Same over here. 14 hours ago, Daniel15 said: Any ideas? Please post your Diagnostics. EDIT: I did a quick test with my Dual Edge Coral TPU, I can see both Coral TPUs on the plugin page: and also on the Dashboard: and also listed under the devices: However this wouldn't be the first time that I see a defective Dual Edge Coral TPU.
May 5, 20251 yr @ich777 I've had a working dual coral TPU w/ pcie adapter frigate setup for some time. But I've been dealing with occasional machine crashes. After replacing way too much hardware, I've narrowed the problem down to the Frigate container, only when I enable the two TPUs as detectors. I've posted frigate related stuff here: https://github.com/blakeblackshear/frigate/discussions/18038 and Unraid related stuff here: The only somewhat interesting thing in the syslog is the write back errors: I saw you post earlier to "enable above 4G Decoding in your BIOS". I'm going to check that. Could that be causing this issue? Or something else?
May 5, 20251 yr Author 22 minutes ago, bmrowe said: TPUs Is this a Dual Edge TPU because you've got two? If it's a dual edge TPU, is it properly cooled so to speak do you have a Heatsink attached and proper Airflow over it?
May 5, 20251 yr 15 minutes ago, ich777 said: Is this a Dual Edge TPU because you've got two? If it's a dual edge TPU, is it properly cooled so to speak do you have a Heatsink attached and proper Airflow over it? I've got this card: https://coral.ai/products/m2-accelerator-dual-edgetpu/ plugged into this adapter https://www.makerfabs.com/dual-edge-tpu-adapter.html?srsltid=AfmBOorFfIb0tFJk7EvTSAjgGIHqytS6y9fDgoaJakAV_RjztfuqG_Zw with a thermal pad and this heatsink: https://www.mouser.com/ProductDetail/Advanced-Thermal-Solutions/ATS-CPX040040020-115-C2-R0?qs=MLItCLRbWswyZ7Q9sz7GLA%3D%3D&countryCode=US¤cyCode=USD I've never noticed the temps get high. Are you thinking it is heating to over 100C and shutting the whole unraid machine down?
May 5, 20251 yr Author 6 minutes ago, bmrowe said: I've never noticed the temps get high. Are you thinking it is heating to over 100C and shutting the whole unraid machine down? It would not be the first time that a Coral TPU took down the whole server but it could also be the case that it's defective. Do you possibly have Diagnostics when it crashed or is it just taking down the whole Server and prevents even pressing the power button?
May 5, 20251 yr 2 minutes ago, ich777 said: It would not be the first time that a Coral TPU took down the whole server but it could also be the case that it's defective. Do you possibly have Diagnostics when it crashed or is it just taking down the whole Server and prevents even pressing the power button? I have the syslog history through crashes. But it restarts instantly after crashing. In this example of a crash at 11:53am, this is what the syslog looked like around the crash time.
May 5, 20251 yr Author 48 minutes ago, bmrowe said: I have the syslog history through crashes. But it restarts instantly after crashing. So the server is restarting...? That is really weird because it should crash and not respond instead of just restarting. Are you sure that the power supply is up to the task?
May 5, 20251 yr 12 hours ago, ich777 said: EDIT: I did a quick test with my Dual Edge Coral TPU, I can see both Coral TPUs on the plugin page: Thanks - I rebooted and now I see both, and both are working fine. Very strange! 1 hour ago, ich777 said: If it's a dual edge TPU, is it properly cooled so to speak do you have a Heatsink attached and proper Airflow over it? For what it's worth, mine doesn't have a heat sink and it's been fine. Never gotten anywhere near the temperatures at which it thermal throttles. I used to pass it through to a Windows VM to use with Blue Iris and CodeProjectAI and that's worked fine for maybe 2 years now, but now I'm moving CodeProjectAI to be a Docker container instead, mostly to make maintenance easier. My security cameras have onboard motion detection using tripwires though (it only alerts if something crosses a line I've drawn in the config, rather than on any motion) so the Coral is only used when there's actually something to analyze. Edited May 5, 20251 yr by Daniel15
May 5, 20251 yr 1 hour ago, ich777 said: So the server is restarting...? That is really weird because it should crash and not respond instead of just restarting. Are you sure that the power supply is up to the task? Yes, the machine is hard powering off and then auto starting back up. I replaced the PSU last week thinking that it was a probable cause but the crashes continued. It's 850W for just: an intel i13500, LSI HBA card, 5 hard drives, 2 nvme drives, and the dual coral tpu pcie card. From my calculations, it's way overpowered for just that, right? I'm also under the impression that the coral is pretty low wattage - so the fact that not using the corals w/ frigate stops the crashes makes me think it's not PSU related. Also, I've shifted frigate to use the i13500 cpu for detection and the wattage use on the computer has increased significantly. Edited May 5, 20251 yr by bmrowe
May 5, 20251 yr Author 1 hour ago, Daniel15 said: For what it's worth, mine doesn't have a heat sink and it's been fine. Never gotten anywhere near the temperatures at which it thermal throttles. I used to pass it through to a Windows VM to use with Blue Iris and CodeProjectAI and that's worked fine for maybe 2 years now, but now I'm moving CodeProjectAI to be a Docker container instead, mostly to make maintenance easier. Back in the days they had a warning in the documentation that not adding a Heatsink these modules become a fire hazard, actually not the chips are the issue the power delivery seems to be the issue. Just saying, but maybe that isn't needed any more but that's just a wild guess.
May 5, 20251 yr Author 1 hour ago, bmrowe said: From my calculations, it's way overpowered for just that, right? Yes, seems overpowered to me. 1 hour ago, bmrowe said: Also, I've shifted frigate to use the i13500 cpu for detection and the wattage use on the computer has increased significantly. Yes, that's expected. 1 hour ago, bmrowe said: I'm also under the impression that the coral is pretty low wattage - so the fact that not using the corals w/ frigate stops the crashes makes me think it's not PSU related. Can you maybe try to use just one TPU and see if maybe one is the issue or if the whole module crashes, which I can't imagine because they each have their own PCIe lane.
May 5, 20251 yr 1 hour ago, bmrowe said: I'm also under the impression that the coral is pretty low wattage Yes, it's only 2 watts maximum. 4 watts if you have the dual TPU.
May 5, 20251 yr 22 minutes ago, ich777 said: Yes, seems overpowered to me. Yes, that's expected. Can you maybe try to use just one TPU and see if maybe one is the issue or if the whole module crashes, which I can't imagine because they each have their own PCIe lane. This is what I was thinking too. I'll enable one, run it for a bit, and see if it crashes.
May 5, 20251 yr Author 18 minutes ago, Daniel15 said: Yes, it's only 2 watts maximum. 4 watts if you have the dual TPU. 14 minutes ago, bmrowe said: This is what I was thinking too. I'll enable one, run it for a bit, and see if it crashes. Oh see, it's still in there but they worded it differently: Source: https://coral.ai/static/files/Coral-M2-Dual-EdgeTPU-datasheet.pdf And there is the note about the power ICs: ...just to let you know that it would be good to mount something since nothing is monitoring the Power ICs @Daniel15.
May 5, 20251 yr Author @carlks please update to 7.1.0 and see if anything is changing in terms of the NPU and also post your Diagnostics if you are still having issues.
May 5, 20251 yr 1 minute ago, ich777 said: Oh see, it's still in there but they worded it differently: Good catch! I've never seen mine above 55C (especially since it's used infrequently for ~100ms at a time, rather than it having constant load) but I'll definitely keep it in mind. I'm surprised they don't factory-install a heatsink, but I guess for $25 for the single TPU / $40 for the dual TPU we can't expect everything
May 5, 20251 yr Author Just now, Daniel15 said: Good catch! I've never seen mine above 55C (especially since it's used infrequently for ~100ms at a time, rather than it having constant load) but I'll definitely keep it in mind. Please look at the docs: I would always recommend to have some kind of cooling in place because even the single chip Coral TPUs have some kind of heat sink on them (metal cage on top <- you don't see them in most screenshots). Just saying, the documentation is here.
May 5, 20251 yr 37 minutes ago, ich777 said: @carlks please update to 7.1.0 and see if anything is changing in terms of the NPU and also post your Diagnostics if you are still having issues. success! thank you for the help. I will test the passthrough to docker.
May 6, 20251 yr On 5/5/2025 at 2:05 PM, ich777 said: Can you maybe try to use just one TPU and see if maybe one is the issue or if the whole module crashes, which I can't imagine because they each have their own PCIe lane. I was getting multiple crashes per 24 hours. Running on one TPU was stable for an entire day. Now trying with just the second TPU. If both work independently, but not together, would that point to a driver or motherboard issue? I can't imagine the magic-blue-smoke pcie adapter is the problem.
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.