I am planning to build a multipurpose home server. It will be a NAS, virtualization host, and have the typical selfhosted services. I want all of these services to have high uptime and be protected from power surges/balckouts, so I will put my server on a UPS.
I also want to run an LLM server on this machine, so I plan to add one or more GPUs and pass them through to a VM. I do not care about high uptime on the LLM server. However, this of course means that I will need a more powerful UPS, which I do not have the space for.
My plan is to get a second power supply to power only the GPUs. I do not want to put this PSU on the UPS. I will turn on the second PSU via an Add2PSU.
In the event of a blackout, this means that the base system will get full power and the GPUs will get power via the PCIe slot, but they will lose the power from the dedicated power plug.
Obviously this will slow down or kill the LLM server, but will this have an effect on the rest of the system?
You must log in or register to comment.
The amount of absolutely wrong answers in here is astounding.
NO. PCIE is not plug and play. Moreover, having a dead PCIE device that was previously accepting information, and then suddenly stops, is almost guaranteed to cause a kernel panic on any OS because of an overflowing bus of tons of data that can’t just sit there waiting. It’s a house of cards at that point. It’s also going to possibly harm the physical machine when the power comes back on due to a sudden influx of power from an outside PSU powering up a device not meant for such things.
Why wouldn’t instead think of maybe NOT running an insane workload on such a machine with insanely power hungry GPUs, and maybe go for an AMD APU instead? Then you’ll get all the things you want.
PCIe is absolutely plug and play. Cards have been PnP since the ISA era. You probably meant hot-plug, but it’s hot-pluggable too: https://lwn.net/Articles/767885/
Any buffered data will sit in the buffer, and eventually be dropped. Any data sent to the buffer while the buffer is full will be dropped. I’m not intimately familiar with communicating with GPUs, but I imagine the only buffers are in the GPU driver (which would either handle the removal or crash) or in the application (which would probably not handle the removal and just crash). Buffering is not really where I would expect to see a problem.
That said, a GPU disappearing unexpectedly will probably crash your program, if not your whole OS. Physical damage is unlikely, though I definitely wouldn’t recommend connecting two PSUs to one system due to the potential for unexpected… well, potential. Inrush current wouldn’t really be my concern, since it would be pulling from the external PSU which should have plenty of capacity (and over-current protection too, I would hope). And it’s mostly a concern for AC systems, rarely for DC.
What’s wrong with 2 PSUs if both of them are connected to the same ground? I thought multiple PSUs is common in the server space too.
Server PSUs are designed to be identical and work on parallel (though depending on platform, they can be configured as primary/hot spare, too). I’d be concerned about potential difference in power, especially with two non-matching PSUs. It would probably be fine, but not probably enough for me to trust my stuff to it. They’re just not designed or tested to operate like that, so they may behave unexpectedly.
deleted by creator
You are mistaking “plug and play” with “hot swap/plug CAPABLE”. The spec allows for specifically designed hardware to come and go, like Express card, Thunderbolt, or USB4 lane-assigned devices, for example. That’s a feature built for a specific type of hardware to tolerate things like accepting current, or having a carrier chip at least communicating with the PCIE bridge that designates it’s current status. Almost all of these types of devices are not only designed for this, they are powered by the hardware they are plugged into, allowing that power to be negotiated and controlled by the bridge.
NOT like a giant GPU that requires it’s own power supply current and ground.
But hey, you read it on the Internet and seem to think it’s possible. Go ahead and try it out with your hardware and see what happens.
Right, it requires device support. And most GPUs won’t support it. But it’s by no means impossible.
I’ve got some junk hardware at work, I’ll try next time I’m in and let you know.
Wow. Okay. Pull that card, buddy. Pedantic on the Internet. What a shocker. 🙄
You have multiple accounts, and are sadly so consumed with Internet points, you used both of them to downvote when you’re won’t. You’re pathetic. Get a hobby. Maybe learning about hardware!
I do something similar to op, however, running llms is what finally convinced me to switch over to kubernetes for these exact reasons, I needed the ability to have gpus running on separate nodes that then I could toggle on or off. Power concerns here are real, the only real solution is to separate your storage and your compute nodes.
What OP is suggesting is not only not going to work, and cause damage probably to the motherboard and gpus, but I would assume is also a pretty large fire hazard. One GPU takes in an insane amount of power, two gpus is not something to sneeze at. It’s worth the investment of getting a very good power supply and not cheaping out on any components.
You’re forgetting that the card would still be receiving it’s 75W of power from the PCIe bus. This is what powers cards that don’t have extra power connectors.
This is a terrible idea, no really.
Any system that shares power and grounds (i.e. on the same bus), keep on the same power supply/domain.
Even, if!!!, it doesn’t fry your computer when one power system goes off but the other stays on - the system will absolutely not be stable, and will behave in unexpected ways.
DO NOT DO THIS.
There computer:
I think the safe option would be to use a smart UPS and Network UPS Tools to shutdown the LLM virtual machine when it’s running on battery. I do something similar with my NAS as it’s running on an older dell R510 so when the UPS goes onto battery it’ll safely shut down that whole machine to extend how long my networking gear will stay powered.
What you want are two servers, one for each purpose. What you are proposing is very janky and will compromise the reliability of your services.
It looks like regular PSUs are isolated from the mains ground with a transformer. That means that two PSUs’ DC grounds will not be connected. That will likely cause problems for you, as they’ll have to back flow current in places that do NOT expect back flow current to account for the voltage differences between the two ground potentials. Hence it might damage the GPU which is going be the mediator between these two PSUs - and maybe the mobo if everything goes to shit.
Now I am not saying this will be safe, but you may avoid that issue by tying the grounds of the two PSUs together. You still have the issue where if, say, PSU1’s 12V voltage plane meets PSU2’s 12V voltage plane and they’re inevitably not the same exact voltage, you’ll have back flowing current again which is bad because again nothing is designed for that situation. Kind of like if you pair lithium batteries in parallel that aren’t matched, the higher voltage one will back charge the other and they’ll explode.
It sounds like a fire hazard. Do not mix power supplies especially if only one is plugged into a UPS
I imagine this would be up to the application. What you’re describing would been seen by the OS as the device becoming unavailable. That won’t really affect the OS. But, it could cause problems with the drivers and/or applications that are expecting the device to be available. The effect could range from “hm, the GPU isn’t responding, oh well” to a kernel panic.
Nope. I actually did that unintentionally on a PC I built. I only used one power wire when the GPU needed 2 so it couldn’t use all the power it needed when running 100%. My understanding was PCI doesn’t support disconnecting devices so the system expects all components it starts up with to be available all the time. Lose one and the system goes down.
PCIe absolutely does support disconnecting devices. It is a hot swap bus, that’s how ExpressCard works. But it doesn’t mean that the board/uefi implements it correctly.
in other words: OP either needs to get a thunderbolt dock or straight up have 2 computers. The latter should not even consume that much more power if the PC gets shut down in the evening and woken up using wakeonlan in the morning.
Oh nice! I knew hot swapping was supported on many other devices but not PCIe itself. Feels wrong to rip a card out while the system is powered up.
Also some GPUs support running without the external power connectors/not all of them. My old GTX 1080 ran for about 3 months off of just the PCIe slots power because I forgot to plug them in. Newer GPUs are FAR more power hungry though and not all newer cards support that. Plus I’ve never tried yoinking the power cables while it’s on. That can’t be good.
Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I’ve seen in this thread:
Fewer Letters More Letters NAS Network-Attached Storage PCIe Peripheral Component Interconnect Express PSU Power Supply Unit PoE Power over Ethernet SSL Secure Sockets Layer, for transparent encryption
5 acronyms in this thread; the most compressed thread commented on today has 4 acronyms.
[Thread #806 for this sub, first seen 15th Jun 2024, 14:15] [FAQ] [Full list] [Contact] [Source code]
