JustJoshin

ESXI? I don't run VMware ESXI.

I put together a dual E5-2670 build earlier in the year and can comment on those cpus. Some of the other server chips mentioned don't turbo as high on all cores or when several cores are being used at the same time. Not a problem for server workloads but a bit more problematic for gaming if this is something you or any readers might want to be doing. So it really depends on what you want to do with the pc. I don't think you mentioned? Basically I'd say if it wasn't for gaming virtual machines then some of those are good / probably better options now. Especially the E5-2683 v3 I think. I myself run: Crashplan, Plex, Couch Potato, Sonarr, my primary windows 10 VM for software development and gaming and my wife's windows 10 vm. I have a Fury Nitro in my VM and a 280x in my wife's VM so she can do a bit of light gaming as well. DX 12 / vulkan games obviously make better use of the increased core count and lower core frequencies compared to DX 11 games. Having said that, DX 11 performance on my Fury is still fantastic albeit a little bit limited in DX 11 games due to the 3 GHZ turbo on all cores. I wouldn't want to go any lower on the clock speed if gaming was one of the tasks you would be expecting the the machine to do though. I just thought I'd throw this out there for anybody that might be reading this wondering about the E5-2670 build and does a bit of gaming. Does anybody know what the all core turbo speed of the E5-2683 v3 is?

Haven't made any progress on the pwm front. I installed the dynamix core temp plugin along with perl and re-ran sensors-detect but am stuck in regards to loading the required module / driver for pwm. Any other s2600cp motherboard users out there?

I'm not familiar with that particular drive but the peak theoretical bandwidth of pcie gen 2 8x should be about 4GB/s. I think some of the modern nvme drives are doing >3 so if I had to hazard a guess I would say probably not. I'd also like to hear from anyone else that knows a bit more on the subject. I haven't had time to really look into this but I'd like to add an nvme drive at some point.

My windows 10 VM runs fine until it has to reboot. Either through needing to reboot when uninstalling / re-installing new display drivers or forced restarts from Microsoft to apply windows updates. The funny thing is when it reboots I can see the bios screen and the windows 10 splash screen but just as it is about to load the windows login screen the screens blank and the fan goes to 100%. Re-setting the the physical pc at this point won't reset the graphics card. I have to do complete power cycle. I never experienced this on my 280x (this card is a Fury Nitro). Reading around the forum I heard some mention that some AMD cards don't 'reset' themselves very well. As it stands, I cannot leave my pc on and have to manually shutdown and start the machine every time I want to use it. It also means the other vm isn't available for my wife to use. Hibernate / sleep power settings are all turned off. In my research on this forum I also read something about supplying the videocard ROM file manually? Could this help this issue? If so, how do I go about doing this?

You don't really need to worry about the speed of these blue open ended 8x slots. For gaming it makes almost zero difference to frame rates. The bandwidth is more important when you consider things like hard drives.

I was about to create my own thread when I noticed this so thought I would come along and bump it in the hopes that someone will see it and we will both get an answer. Before running Unraid I ran FreeNAS and had iscsi targets setup which worked great. I am currently using shares in Unraid which works ok for some things and a lot of things just flat out refuse to install to a network drive. Off the top of my head some of the Windows folders cannot be easily moved to a network drive, Blizzard's Heroes of the Storm won't install and I also had a lot of problems with 3dMark. I'd really like to know how if Unraid has support for setting up iscsi as I don't have the option of physically passing through an entire disk.

Just thought I'd chime in as another person interested in adding nvme to their unraid server. I'm looking to purchase the Samsung 950 Pro nvme with a pcie slot card.

I'm curious if passing through an entire usb controller would have solved your issues with usb passthrough and the vive? Either passing through a controller on the motherboard, or a $20-$30 pcie usb hub that can be passed straight through to the vm. Then you get the benefit of hotplug working.

As I know I have seen a few S2600CP builds in this forum I was wondering if anybody has gotten cpu fan speed throttling working? My cpu fan seems to always run at 100% which is quite noisy on my Cooler Master 212 Evo and it's driving me nuts. Can anybody point me in the direction of how I can get the temperature based variable fan speed working? root@Tower:~# sensors coretemp-isa-0000 Adapter: ISA adapter Physical id 0: +34.0°C (high = +90.0°C, crit = +100.0°C) Core 0: +33.0°C (high = +90.0°C, crit = +100.0°C) Core 1: +33.0°C (high = +90.0°C, crit = +100.0°C) Core 2: +31.0°C (high = +90.0°C, crit = +100.0°C) Core 3: +29.0°C (high = +90.0°C, crit = +100.0°C) Core 4: +34.0°C (high = +90.0°C, crit = +100.0°C) Core 5: +33.0°C (high = +90.0°C, crit = +100.0°C) Core 6: +33.0°C (high = +90.0°C, crit = +100.0°C) Core 7: +31.0°C (high = +90.0°C, crit = +100.0°C) i350bb-pci-0700 Adapter: PCI adapter loc1: +52.0°C (high = +120.0°C, crit = +110.0°C) Additionally, pwmconfig doesn't detect / show anything. root@Tower:~# pwmconfig # pwmconfig revision 6243 (2014-03-20) This program will search your sensors for pulse width modulation (pwm) controls, and test each one to see if it controls a fan on your motherboard. Note that many motherboards do not have pwm circuitry installed, even if your sensor chip supports pwm. We will attempt to briefly stop each fan using the pwm controls. The program will attempt to restore each fan to full speed after testing. However, it is ** very important ** that you physically verify that the fans have been to full speed after the program has completed. /usr/sbin/pwmconfig: There are no pwm-capable sensor modules installed

There's two major ways to deal with it, one is to allow the guest OS to request and control sleep features, which isn't fully supported and may or may not work. The other is to use the KVM virsh to suspend and resume, which I believe Limetech is working to implement. I don't know that the virsh method is foolproof, but it should be easier to support. I may be misunderstanding but virsh requires you have another pc from which to ssh in and suspend/resume/restart etc.? I think this probably isn't all that useful for the average user that may be running say a nas/gaming setup (which is one of the ways LT choose to market unRaid). They just want the vm to go to sleep like it normally would come back some time later and wiggle the mouse / tap a key on the keyboard to wake it up again. They don't necessarily have another computer with a teminal, putty or similar in which to ssh into the box to initiate kvm virsh. Please feel free to correct me if I got the wrong idea though.

Sorry to dig up an old thread but I have noticed that my cpu cores are able to power down much lower when the vm isn't running. E5-2670 likes to sit at around 2600mhz just idling at the desktop for some strange reason. Suspending the vm will allow most / all cores pinned to the vm to clock down to around 1300mhz. This is a fairly significant saving when you have two cpus (which I do, just waiting for another 8pin power cable to arrive to plug the other one in). I notice that when you sleep the vm unraid correctly recognises the vm has been suspended. There just doesn't seem to be a way to wake the vm from sleep. I really hope this can be fixed in future.

Hi GridRunner. For the benefit of myself and others not familiar with cpupinning, can you explain this part. I thought you could basically just put unraid on core 0,6 (physical core 1) and then dedicate any remaining cores to virtual machines. What is the function of having a dedicated core and hyper thread for emulatorpin?

[quote name="dmwilson1990" post="474563" timestamp="1464736799"] I received dual GTX 1080s this morning and ran Fire Strike 1.1 to see how it performed. The results are in the link below: [url=http://www.3dmark.com/fs/8643594]http://www.3dmark.com/fs/8643594[/url] The GPU performed great in the benchmark and Overwatch is running at ~100fps. However, I was rather surprised with how poorly my CPU performed. I have an Intel 5820k with 6 cores assigned to each VM and no overclocking. Is there something I'm doing wrong? ... [/quote] For comparison to throw a data point into the ring: http://www.3dmark.com/compare/fs/8643594/fs/8638558 I'm currently running a single xeon e5-2670 @ 3Ghz with 7 cores and hyper threading passed through to my windows 10 vm (1 core+HT left to Unraid). My Physics score was ~50% higher which would suggest your cpu score is a bit low? Not sure how much one extra core of mine would make. Doubt it would be 50%.

Yeah I finally figured out what the problem was (which seems obvious in hind sight). Server motherboards do a lot of additional checks and diagnostics on post then consumer boards and I think it was freaking out that it was missing the rest of it's installation environment from wherever it was pulled from. After tracking down the location of the bios reset jump in the manual, all is good. I've got it posting and loading unraid now. Had a busy weekend so just haven't had time to sit down and start porting over my existing unraid environment from another machine. TBC...