Leaderboard

Popular Content

Showing content with the highest reputation on 07/17/19 in all areas

  1. Hey everyone! After months of testing and learning, I finally managed to have 2 of my 4 1080TIs run in SLI. The information on how to do this has actually been online for a while but a bit scattered about (at least that was my experience). Overview of Steps 1) Achieve GPU pass-through 2) Mod the Nvidia Drivers to allow SLI in our VM 3) Use Nvidia Profile Inspector to get much better performance Edit Update After utilizing a few other VM optimizations, specifically CPU pinning, my SLI performance DRASTICALLY improved. My FPS in SLI went from the mid 40s to 70+ (I used a few different benchmarks such as the unigine benchmarks and also from personal experience playing ESO). When I started trying to get SLI to work in Unraid, I noticed that just passing through 2 GPUs to a single VM already resulted in very noticeable gain in performance. I am still tinkering with Nvidia Profile Inspector, so things might change. If they do, I will post an update. GPU pass-through My VM options: Bios: OVMF Machine: Q35.1 Sata for ISO drivers and VirtIO for Primary vDisk Follow the instructions in this Spaceinvader One video. Afterwards, pass through your 2 GPUs and they should appear in windows. Nvidia Drivers Mod Note: If you have any difficulties with this next part, you are better off asking for help here on the DifferentSLIAuto forum thread. So to my understanding motherboard manufactures must license the right to allow SLI on their boards from Nvidia. The reason we haven't been able to achieve SLI in Unraid is due to the fact that our VM's "motherboard" info simply not qualifying as a Nvidia approved motherboard for SLI. Luckily there has been a hack available for a while that allows SLI to be enabled for not only any motherboard but also any GPUs (aka, the GPUs don't even need to be the same model). This is what worked for me: I used Nvidia Driver version 430.86. If you use the same version, then these instructions SHOULD work for you. Install Nvidia Drivers The original method/program used by DifferentSLIAuto is no longer working for the latest versions of Nvidia drivers (driver versions 4xx and on). We have two choices, we can go with the old method and use an older driver or we can mod newer drivers manually. This is what I did, and what I'll be describing: Download DifferentSLIAuto version 1.7.1 Download a Hex Editor (I used HxD) Copy nvlddmkm.sys from C:\Windows\System32\DriverStore\FileRepository\nv_dispi.inf_amd64_b49751b9038af669 to your DifferentSLIAuto folder (NOTE: if you are not using driver version 430.86, then the nvlddmkm.sys file you must modify will be located some where else and you must find it yourself by going to Device Manager > Display adapters > YOUR CARD). Mod the copied nvlddmkm.sys file by opening it in a Hex Editor. Here are the changes for driver 430.86: Address: [OLD VALUE] [NEW CHANGED VALUE] 000000000027E86D: 84 C7 000000000027E86E: C0 43 000000000027E86F: 75 24 000000000027E870: 05 00 000000000027E871: 0F 00 000000000027E872: BA 00 000000000027E873: 6B 00 Save and exit Hex Editor In your DifferentSLIAuto folder, right click and edit install.cmd. Replace all instances of "nv_dispi.inf_amd64_7209bde3180ef5f7" with the location of where our original nvlddmkm.sys file was in our case this is "nv_dispi.inf_amd64_b49751b9038af669". The install.cmd file will modify the copy we added to the folder and replace the original one found at the location we specify here. Use this video for reference but please note that in the video the driver version is different then ours and they replace nv_dispi.inf_amd64_7209bde3180ef5f7, with nv_dispi.inf_amd64_9ab613610b40aa98 instead of nv_dispi.inf_amd64_b49751b9038af669. Move your DifferentSLIAuto Folder to the root of your C:\ drive. Set UAC to the lowest setting (OFF) in Control Panel\All Control Panel Items\Security and Maintenance. Run cmd.exe as admin and enter: bcdedit.exe /set loadoptions DISABLE_INTEGRITY_CHECKS bcdedit.exe /set NOINTEGRITYCHECKS ON bcdedit.exe /set TESTSIGNING ON Restart you computer into safe mode + network enabled (Video showing how to do it quickly using Shift + Restart) Within the DifferentSLIAuto folder located at "C:\", run install.cmd as admin. After only a few seconds the CMD window text should all be green indicating that all is well! Open up your Nvidia Control Panel and under 3d Settings it should now say "Configure SLI, Surround, PhysX". Click that option, and under SLI Configuration Select Maximize 3d performance and that's it! Nvidia Profile Inspector The default settings in the Nvidia Control Panel really suck. After FINALLY getting SLI to work I was getting only 40 FPS in SLI when before I was getting 100+ FPS prior to enabling SLI. I was about ready to give up when I came across Nvidia Profile Inspector! By changing a few settings with Nvidia Profile Inspector, I was able to finally get great SLI results (70 FPS). Keep in mind that I only changing settings in profile inspector for only a few hours, so I'm sure there are many optimizations to be made, so hopefully we can figure it out as a community. Run Nvidia Profile Inspector. I recommend the following settings for now for the _GLOBAL_DRIVER_PROFILE (Base Profile). Nvidia Profile Inspector Settings: 1 - Compatibility - SLI compatibility bits: 0x02C00005 SLI compatibility bits (DX10 + DX11): 0x080000F5 5 - Common - Power management mode: Prefer maximum performance Thread optimization: On 6 - SLI - NVIDIA predefined number of GPUs to use on SLI rendering mode: 0x00000002 SLI_PREDEFINED_GPU_COUNT_TWO NVIDIA predefined number of GPUs to use on SLI rendering mode (on DirectX 10): 0x00000002 SLI_PREDEFINED_GPU_COUNT_TWO NVIDIA predefined SLI mode: PLAY WITH BOTH: 0x00000002 SLI_PREDEFINED_MODE_FORCE_AFR & 0x00000003 SLI_PREDEFINED_MODE_FORCE_AFR2 NVIDIA predefined SLI mode on DirectX 10: PLAY WITH BOTH: 0x00000002 SLI_PREDEFINED_MODE_FORCE_AFR & 0x00000003 SLI_PREDEFINED_MODE_FORCE_AFR2 SLI rendering mode: Try: 0x00000000 SLI_RENDERING_MODE_AUTOSELECT, 0x00000002 SLI_RENDERING_MODE_FORCE_AFR. 0x00000003 SLI_RENDERING_MODE_FORCE_AFR2 MAKE SURE TO HIT APPLY CHANGES ON TOP RIGHT HAND CORNER Next we will make some changes in Control Panel > Nvidia Control Panel Nvidia Control Panel Manage 3D Settings > Global Settings Power management mode: Prefer maximum performance SLI rendering mode: start by leaving this alone, and then make it match your Nvidia Profile Inspector settings (so if you are trying 0x00000002 AFR, set this to Force alternate frame rendering 1, and if your are trying 0x00000003 AFR2, set this to alternate frame rendering 2) And that's it! Now keep in mind the settings above are far from the best, and are only a starting point for us. It is probably best to find individual game profiles for each title and go from there. I will be googling "Nvidia Profile Inspector <insert game here>" for a while and trying different settings out. Make sure you change the "NVIDIA predined number of GPUs" settings to TWO if you change profiles because in my experience it was defaulting to FOUR (this may be because I do have 4 physical cards installed on the motherboard, so if someone else gets different results please let me know). SOME CLOSING THOUGHTS I did some additional research which lead me to open up my motherboard manual. I discovered that in my case, my mother board PCIe slots change speed depending on a wide number of factors (for example, if I have a 28 lane CPU, some of my PCIe 3.0 slots (PCIE1/PCIE3/PCI5) STOP functioning in x16 and instead run at x16/x8/x4. If that wasn't a big enough kick in the nuts, since I have an m.2 SSD in my m.2 slot, my PCIE5 slot doesn't function at all). All in all, this was fun adventure for me, and I really hope this information helps people who are interested in trying SLI via VMs!
    1 point
  2. The plugin didnt take it out. Guarantee that. Something else happened. If it happens again, post your diagnostics before attempting to fix Sent from my NSA monitored device
    1 point
  3. Extra parameters line (switch to advanced view when editing)
    1 point
  4. I would like to also mention that UNRAID was a 1 day affair. Freenas was over a week. for a newbie for a particular system/OS/Enviorment, unraid felt like a breeze. i Day includes other activities. LOL. So far UNRAID is KEWL.
    1 point
  5. Question 6. Wanting to install a Cache drive. Want to go with an M.2 Form Factor. question is NVME or SATA. Also incase there is some data on the cache, and if (for whatever reason the machine goes off), what happens to the data in the cache drive. Can I make sure somehow that its written to disk? Can we have some system as in (sorry) freenas that there is "cop", hope term is (correct)? Also for the first 3-6 months can I backup to a normal external USB drive every night for the whole system (data) No music No videos, just pure data, Say aboout 500GB MAX MAX zipped. The files can be zipped and copied off. Once all is well then okay. Even Incremental will do. Thanks in advance.
    1 point
  6. 1 point
  7. My guess would be that it does not update automatically. And why should it? The data does not usually change that often. If you really want/need an update of the page, you can simply refresh it. By the way, even the auto updating of the Main page was was a big debate a few years back as I recall. The updating process consumes a fair number of CPU cycles and this impacted the speed of the parity check with the (very) slow CPU's that many of us were using back in those days. The impact was considerable--hours in additional time. I seem to recall that when first implemented in a stable release, you had to turn it 'on' via a setting!
    1 point
  8. The rebuild of Disk 8 was successful while connected to the motherboard SATA port. I've now stopped the array, added the new 10TB Ironwolf as the 2nd parity drive, and restarted the array. The parity sync/build for Parity 2 is underway.... 20 hrs to go for dual failure protection to be active. LSI got back to me re: the pictures and although the card looks good physically, the labelling is good except for the serial number. The number is not in their manufactured database and is in a format that's wrong for this series of card. They've given me some instructions and tools that should work from a terminal session under unRAID to do some diagnostics. I will wait for the 2nd parity disk to finish syncing before doing anything else. I also got shipping confirmation on the replacement miniSAS cables and they could be here by the end of this week. Once they arrive I'll shutdown, swap out the cables, disconnect all unRAID disks and do some testing with the controller and the tools that LSI supplied. I have two spare drives that I can use for testing purposes, one 10TB and one 8TB which I can move between bays of the enclosure to run the tests on each of the 4 SAS/SATA connections per backplane. The only ones I'll not bother testing are the ones that are connected to motherboard SATA ports as none of them have ever reported errors. I may use them initially, since they have been reliable and would be a good indicator if they failed while attached to the LSI controller vs the MB SATA. When attached to the MB SATA ports, those drive bays have been error-free - that would almost definitely point to the LSI controller as the faulting unit. When I get my next disability check, I'm going to go ahead and order a new LSI card from the eBay vendor that guarantees a new-in-box unit from LSI. It won't hurt to have a spare even if the existing controller proves to be OK. More to come when I get some of the testing done.
    1 point
  9. @drkCrix No, I haven't sorted this out. According to the following link, it appears there will be a fix in the Linux 5.3 kernel. I'm not Linux savvy, so I'm not sure how that will play out for unRAID. https://forum.proxmox.com/threads/only-one-of-two-nvme-detected-in-linux-duplicate-subnqn.54480/ I thought the NQN was something configurable by the manufacturer. So, I would blame on ADATA and not Realtek. I could be completely wrong about that though. I contacted ADATA "customer service" (in quotes since it is a joke). Their response email looks completely automated as it starts with "Dear Customer" and doesn't even have the name I entered when filling out the online form to contact them. The email starts out acknowledging I have an issue and then goes right into stating they are here to assist with the return of the product. The email did provide a few generic troubleshooting steps, but had absolutely no mention of the issue I explicitly detailed for them (again, since this is an automated, impersonal email.) I'll try replying to the email to see if that actually gets anywhere. I think I'll probably just return the drives and spent a bit more for another brand.
    1 point
  10. @Alabaster I found this today https://lkml.org/lkml/2019/7/15/57 Looks like it is for drives with the realtek controller like we have.
    1 point
  11. Tested with an AMD R9 Nano ( GPU + Audio Passthrough ) and the new Unraid 6.6.2 version. So far everything looks good, but I will test this bug with an AMD RX 56 Nano this weekend.
    1 point
  12. Yes, works fine. I have an unRAID box using the 2-port version of the same card, no problems. The 2-port is identical to the 4-port, apart from having half the socket parts missing.
    1 point
  13. My Win10 laptop started doing this a lot more recently. I am assuming some Windows update has prompted the change. I can resolve it by telling windows to log into the share with username of "\" and leave the password box blank. I hope that works for you.
    1 point
  14. Would it be possible to have a nice GUI to make child and parent shares in unRAID? I feel that is missing... Instead of having to map a lot of network drives, I can mount a few ones, and then browse down into them. Instead of having 4 documents shares (one for each person), I can just have 1 share, with some child shares under it. Then I can set read only on the documents share, and read/write on the child shares. Example: Instead of having: Documents - Me Documents - Brother Documents - Mom Documents - Dad Recipts As individual shares, I can istead have all my documents under one network share, then each person as a child share. More like this: Documents - Documents - Me - Documents - Brother - Documents - Mom - Documents - Dad - Recipts That looks a loot cleaner when you set it up on the clients machines, since I would not have to map so many network drives as I have to do now.
    1 point
  15. For completeness, then, after adding the three users and creating the "documents" share, the command line work goes something like this: mkdir /mnt/user/documents/brother chown brother:users /mnt/user/documents/brother chmod 700 /mnt/user/documents/brother mkdir /mnt/user/documents/mother chown mother:users /mnt/user/documents/mother chmod 700 /mnt/user/documents/mother mkdir /mnt/user/documents/me chown me:users /mnt/user/documents/me chmod 700 /mnt/user/documents/me Caveat: I haven't really tested this with SMB.
    1 point