Tritech

Members
  • Posts

    57
  • Joined

  • Last visited

Everything posted by Tritech

  1. Take a look here: https://github.com/guydavis/machinaris/wiki/Generic
  2. It's been great over the last year using Unraid, from spinning up linux vms in seconds, hosting game servers for friends, hosting my family home movies and picture,to backing up all the pcs on my network. The single feature I've wanted was a way to "snapshot" vm xmls. When customizing xmls for my threadripper build it took MANY iterations to get it right from switch numa nodes, pinning cores, and many custom edits. Every time an edit was made and I need to change the form view, all of these were reverted. It would be nice to be able to take a snapshot of a known working xml, maybe even have a dropdown menu of several saved templates. This would save so much time when making custom edits, and save a lot of unorganized copy-pasting to text files on my PC.
  3. 0 issues on RC4. Also to note, I've never had any of the DB corruption issues either.
  4. I just get the simple recaptcha checkbox, but I haven't done anything special. EVGA 1080ti
  5. With the way xmls get modified when you alter any setting on the form view can we get a way to clone or make a copy of an xml before making any changes? I've got over a hundred xml variants trying to track down better latency and just being able to have a known good working xml, make a copy/clone of it, and the work off that would be nice. Maybe even being able to set a template for your common settings as well.
  6. Geforce experience works fine on my end. Did you just try later, maybe they were having an outage?
  7. Can anyone point me in the right direction for enabling hugepages? Will 32gb be enough with only 1 vm allocating 10gb and a handful of dockers?
  8. https://pastebin.com/7CG7ntKw Take a look at mine, running a similar system (Asrock x399fatality, 1950x, 1080ti, passed through NVME controller and a few usb controllers) Specifically, take a look at the <hyperv> and <cpu> section differences. I've got latency down to reasonable levels. Still have audio pops and cracks every now and then though, but it brought down my latency from over 2000 to 100ish average. You could also experiment with the "<topology sockets='1' cores='8' threads='1'/>" section and try <topology sockets='1' cores='4' threads='2'/>... well however many cores you allocated.
  9. Change any instances of "pcie-root" to "pci-root".
  10. @jbartlett Off topic, but do you have a single controller card you can recommend, preferably 4 port that plays nice with usb3? Looking for something in the sub $50 usd range.
  11. I'm on Pro. I have all windows updates off, and usually update all machines in the house monthly. Like I said, this last instance the machine went down and the screen went black while in use. It's just odd that it's pegging the cores when it crashes.
  12. Hey, bastl. We've exchanged a few times here on the "at my wits end..." thread dealing with latency... I have the same board as you. I haven't changed anything in over a month or so. Software wise in the VM, I haven't really installed anything new, just steam/browser updates. The windows logs looked fine, no weird errors other than the one above when trying to restart it. Maybe it was just a fluke. I keep my vm running 24/7. Should I just bring it down every few days as a preventative measure?
  13. Currently on 6.7 RC5. My VM has crashed twice this week. Once when I was asleep, and woke up to a functioning server. All dockers and shares were accessible, but the main win 10 VM was down, and all the allocated cores were pegged to 100%. Stopping the vm fixed the CPU usage, but it was unable to restart. The only thing I could see was: host-cpu char device redirected to /dev/pts/0 (label charserial0) in the vm log, and then stating that it crashed. Only a hard reboot solved the issue. Second time was just now, while I was just browsing, screen just went black, and same maxxed cpu cores, just the ones pinned to the vm. Any ideas of what to look for the next time it happens?
  14. I've converted to vm and back a few times to test with the same windows install. I'm not sure what you mean by this, I'm assuming you mean to config the vm. This is my current vm, the xml is heavily customized but you should get the idea from here. There is no vdisk location as I've passed through the NVME controller that you see on the bottom. As far as extra vdisk images, just check when they were last modified, and if there are any existing vms that point to that .img.
  15. <qemu:commandline> <qemu:arg value='-global'/> <qemu:arg value='pcie-root-port.speed=8'/> <qemu:arg value='-global'/> <qemu:arg value='pcie-root-port.width=16'/> </qemu:commandline> Tested the new RC4 and looks like it's working!
  16. Just chiming in to say I am waiting for this and any other threadripper performance related changes as well.
  17. Step one would be keeping this on the first page and visible here as well 😁 Devs, you pickin' up what we're putting down?
  18. Yea I was just wondering if its driver related on the host side/vfio.
  19. Actually I let it run a bit longer and both of the highest execution are network related. ndis.sys and adf.sys. Come to think of it, you're using a different ethernet port than I am. I wonder if that may have some issue. I'm using the 10G port, which I don't really have a use for right now, the rest of my network is gigabit.
  20. I get what your saying, I think its saying that they should be in the included range. You know how you left out cores 8/24? Well I think they have to be on the same "domain" to be used at all, well to at least get the most out of them. At least that's the way I interpret it. I've tweaked my config for now just so they're all on the same domain. I'll fix it later when I change my isolcpus at reboot. Here's some updates as well, seems that storport.sys is whats giving me the highest execution time. Gonna see if I can track down any gains there.
  21. I cross-referenced that several times. Really helpful stuff. I reapplied the Epyc "hack" and that further brought down my latency, to ~300u, as low as 125ish. https://pastebin.com/dLWncwhV
  22. Evidently it does start with 1. I think the "0-15" refers to the vcpupin'd cpus, not physical.
  23. You're right... mine seems to be grabbing almost 1.5 gb from node0.