Lyfesaver 133 Posted January 4 Posted January 4 Server <-> Emby (Best Practices) So, I am just looking for advice on some best practices as my "Media Server" just got flipped upside-down and turbo-charged for other, non-media purposes. Gone is the 12th Gen i7 w/32GB 4800 DDR5 & Arc A380. Swapped out the CPU, GPU and RAM to a 14th Gen i9 w/96GB 6400 DDR5 & RTX 5090 (with a 2nd GPU, like a 5070 Ti, being added with a month or so). Why? My boss said if I provide the server (not knowing I had one and meeting his reqs would be a simple CPU swap for me) told me he would hook me up with the RAM (this was back in early OCT, thank god) and the 5090 so we could move our AI development off the cloud and only run the much cheaper production use on Azure. About 6k a month in Azure bills for AI and easily 4K was development, so it made sense as we are a small co. We use AI for real time fraud detection and SQL analytics so production usage will always be pretty cheap. Anyway, knowing what my hardware I am using is important, but I tell you all this so you wont think me an idiot that is spending all this on a "media server". I mean, don't get me wrong. I am 100% an idiot... just not in this particular scenario. Altho, I bet if we put in even minimal effort we can find ways I am an idiot here too, lol. Okay, on to the meat and potatoes. Since my step in the upgrade (about 2 weeks ago I finally dropped in the 14th Gen with the better IMC and was able to bump RAM from 5600 to 6400) I have noticed issues. Now some of them I dont think have anything to do Emby in any way what-so-ever and almost all of them are on the "more a nuisance" scale than an actual roadblock/showstopper. However, I do feel that if I was properly schooled in a few best practices Anything will be isolated enough that it wont matter to Emby. So, here are a few questions I have: Is it smart, in my use case, to force Emby to use the onboard UHD 770, freeing up all 5090 resources. Heck, is that even possible with on-board GPUs? I just checked and only the 5090 is listed under Advanced in transcoders Is there a particular way I should be doing the tone-mapping with multiple GPUs? Would it make sense to bring the Arc A380 back into the server and not use the on-board? My main fear (outside of the bugs I am trying to squash) is that when Vanna is doing it's thing and going to town hard, a transcode will kick in and for the next hour or so Vanna is slowed down. Now, in my defense of being stupid, maybe this is all for nothing and a transcode will have little to zero effect on the CUDA. I don't know. I am quite ignorant about thing when we get below the surface/end-use level of CPUs & GPUs. Anywho, sorry to drag this on for so long but any guidance on this will greatly help my confidence that at least I am moving in the right direction. I really dont want to have two servers
Neminem 1518 Posted January 4 Posted January 4 Since you post in General / windows and you transcoding path looks like windows. Does the iGPU show up in windows? If not you need to enable it in your BIOS refire to your MB manual, for a how to. 1
Lyfesaver 133 Posted January 4 Author Posted January 4 2 minutes ago, Neminem said: Since you post in General / windows and you transcoding path looks like windows. Does the iGPU show up in windows? If not you need to enable it in your BIOS refire to your MB manual, for a how to.
Neminem 1518 Posted January 4 Posted January 4 Hmmm thats strange, what does Hardware diagnostics show you. They should be in your logs dir.
Lyfesaver 133 Posted January 4 Author Posted January 4 And yes, the fact that I am still in Windows is one of the ways I am currently an idiot Seriously tho, WSL2 is a good friend of mine but I have never had the confidence to be a fully Linux server guy... tho now that I am being forced into local development I have decided to ditch Windows but it will have be later on when I can afford the downtime.
Lyfesaver 133 Posted January 4 Author Posted January 4 3 minutes ago, Neminem said: Hmmm thats strange, what does Hardware diagnostics show you. They should be in your logs dir. I do not know why there are three hardware_detection-63903112047.txthardware_detection-63903112109.txthardware_detection-63903112106.txt
Lyfesaver 133 Posted January 4 Author Posted January 4 Sidebar: Thanks for looking out and trying to help. I have always enjoyed our interactions, Neminem 1
Q-Droid 989 Posted January 4 Posted January 4 The Intel iGPU is more than adequate and in most cases preferable for media over discrete GPUs. Gaming and computational prowess doesn't translate directly to media decode/encode capacity. Also, the path to getting the newest integrated and discrete GPUs working in Linux is more challenging than in Windows. In your case you might need to get the latest Windows drivers directly from Intel (can't remember the name of their utility) for your system and Emby to detect properly.
Lyfesaver 133 Posted January 4 Author Posted January 4 4 minutes ago, Neminem said: Ohh are you using VSL2 to run Emby? No, sorry. Was just clarifying why I am not using a full Linux server. My Emby is 100% running as a Windows service on the server.
Lyfesaver 133 Posted January 4 Author Posted January 4 4 minutes ago, Q-Droid said: The Intel iGPU is more than adequate and in most cases preferable for media over discrete GPUs. Gaming and computational prowess doesn't translate directly to media decode/encode capacity. Also, the path to getting the newest integrated and discrete GPUs working in Linux is more challenging than in Windows. In your case you might need to get the latest Windows drivers directly from Intel (can't remember the name of their utility) for your system and Emby to detect properly. Thanks for the feedback, and yes because recently I was running Arc, I have both their GFX Utility and their Driver Updater for chipsets and stuff running.
Neminem 1518 Posted January 4 Posted January 4 If you pull the Nvidia card does it work then? And only use the iGPU. If that works. Then you should look at your BIOS. Another thing to try. Where is your monitor connected? Nvidea or MB?
Lyfesaver 133 Posted January 4 Author Posted January 4 Nvidea. I cant pull the GPU atm, so I will have to do that debug step at another time.
Neminem 1518 Posted January 4 Posted January 4 Try plugging in the iGPU to your monitor to see what happens after a reboot. 1
Neminem 1518 Posted January 4 Posted January 4 To explain further, if you don't see boot screen on your monitor. Then you should look in BIOS. Some MB's disable onboard gpu's if it detects a discrete gpu. You can then enable the igpu in bios. 1
Lyfesaver 133 Posted January 4 Author Posted January 4 2 minutes ago, Neminem said: To explain further, if you don't see boot screen on your monitor. Then you should look in BIOS. Some MB's disable onboard gpu's if it detects a discrete gpu. You can then enable the igpu in bios. Early this morning when debugging some issues (before I took those screenshot) I went into the BIOS and set the iGPU to enabled but what I did incorrectly is set the primary to be PEG. When I can, I plan to go back into the BIOS and set the primary to iGPU so that Windows will move most of what it is currently using in the 5090's VRAM over to the iGPU's shared system memory. It could be in hour or tomorrow morning but that should give us some data to add to it. I wont have time to actually remove the card until tomorrow morn.
Neminem 1518 Posted January 4 Posted January 4 I would start looking in the bios and plugging in the monitor to that. Before taking out the nvidia card 2
Lessaj 467 Posted January 4 Posted January 4 Windows generally only makes display adapters available if they have a monitor connected to them, you can achieve this with a dummy plug or pretty much every single monitor that has more than one input on it still presents the EDID even for inactive inputs, so you can just connect another cable to an unused input and it should enable the iGPU. You don't need to do that on Linux. 3
Lyfesaver 133 Posted January 4 Author Posted January 4 Damn server crashed again (non Emby reasons I think) so I was able to jump into BIOS real quick and switch if from PEG to CPU and plug monitor into the on-board hdmi but if I am not mistaken "we" *the collective that was helping me in the forums... maybe you were part of that, it was a long time ago) could not get rid of the dupes and I just disabled it on the BIOS.
Q-Droid 989 Posted January 4 Posted January 4 All you need now is a motherboard to drop in the the 12th Gen i7 w/32GB 4800 DDR5 & Arc A380 and you can have a dedicated Emby server. 1 1
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now