Anybody planning a Ryzen build?


Recommended Posts

I just built my Ryzen server yesterday.  It crashed sometime when I was unable to get to it.  I have been watching the log and its spitting out this error a lot:

 

Jul 17 08:39:17 FatTony kernel: pcieport 0000:00:01.2: AER: Corrected error received: id=0008
Jul 17 08:39:17 FatTony kernel: pcieport 0000:00:01.2: PCIe Bus Error: severity=Corrected, type=Data Link Layer, id=000a(Transmitter ID)
Jul 17 08:39:17 FatTony kernel: pcieport 0000:00:01.2: device [1022:15d3] error status/mask=00001000/00006000

Which correpsonds to this:

IOMMU group 1: [1022:15d3] 00:01.2 PCI bridge: Advanced Micro Devices, Inc. [AMD] Device 15d3

 

Anyone else had this problem?  On a B350 mobo.

 

Edit:  After some googling and trying various things, I added pcie_aspm=off to my syslinux.cfg and the errors are not getting thrown around anymore.  I will let it run and see if my system is stable now.

Edited by choppyesq
Link to comment

So I ordered a 2700x, Gigabyte x470 Ultra Gaming and a whole bunch of other stuff. Excited to see how unRAID will run ?

I still have a 1080 Ti, will this work for pass through? I wonder what the power draw will be when not running the Windows VM and basically not using the GPU.

Edited by pushnoi
Link to comment
4 hours ago, pushnoi said:

So I ordered a 2700x, Gigabyte x470 Ultra Gaming and a whole bunch of other stuff. Excited to see how unRAID will run ?

I still have a 1080 Ti, will this work for pass through? I wonder what the power draw will be when not running the Windows VM and basically not using the GPU.

 

You won't be disappointed.

 

I have been running mine now for a couple of months with much the same hardware (exact spec in my signature) and it has been both pain and lockup-free.

 

There's no reason to expect any problem with the 1080Ti. I'm using a GTX 970 without hassle, but there plenty of reports of success with your model too. If unsure of the process then Gridrunner's guide is a handy reference.

 

The power draw will just be a trickle until it is in use. If you are curious then you can always invest a few <insert local currency> in a wall socket power-meter. It's always fun to actually see hard data as opposed to supposition. In my own case I'm idling at 95W with just a couple of minor container/VM running 24x7, and that spikes to 270W when gaming on a Windows VM.

 

 

 

Edited by Dev Null
Fleshing out power usage details.
  • Like 2
Link to comment
On 7/24/2018 at 4:22 AM, pushnoi said:

So I ordered a 2700x, Gigabyte x470 Ultra Gaming and a whole bunch of other stuff. Excited to see how unRAID will run ?

I still have a 1080 Ti, will this work for pass through? I wonder what the power draw will be when not running the Windows VM and basically not using the GPU.

 

Before you use Unraid extract the GPU BIOS using TechPowerUp GPU-Z and best to do it on a clean Win10 and then use the Hex editor to remove the headers from the BIOS file. Had to do that on my GTX 1060 because none of the online BIOS worked. 

 

On a side note, I finally got my RX Vega 64 to GPU passthrough but the Reset bug for AMD cards persists. I have to reboot the whole system to make it work again. 

 

 

I used this post and works with some success. Does anyone else have this struggle with the AMD reset bug? 

Link to comment
On 7/22/2018 at 3:37 PM, smashingtool said:

How should CPU pinning be handled on a ryzen chip? I assume the CCX makes it slightly more complicated. I guess the real question is, what is the layout of the CPU/Threads for 1-16?

 

They are actually numbered from 0 to 15 and they are paired even-odd. So threads 0 and 1 are physical core 0, threads 2 and 3 are physical core 1, etc. all the way to threads 14 and 15 being physical core 7. You'll see the arrangement if you switch to the Dashboard page of the webGUI. This is different from how an Intel processor is arranged, where in an i7, for example, threads 0 and 4 represent physical core 0, threads 1 and 5 are physical core 1, threads 2 and 6 are physical core 2 and threads 3 and 7 are physical core 3. I'm no sure why they are different but the AMD arrangement makes more sense to my simple mind.

Link to comment
21 minutes ago, John_M said:

 

They are actually numbered from 0 to 15 and they are paired even-odd. So threads 0 and 1 are physical core 0, threads 2 and 3 are physical core 1, etc. all the way to threads 14 and 15 being physical core 7. You'll see the arrangement if you switch to the Dashboard page of the webGUI. This is different from how an Intel processor is arranged, where in an i7, for example, threads 0 and 4 represent physical core 0, threads 1 and 5 are physical core 1, threads 2 and 6 are physical core 2 and threads 3 and 7 are physical core 3. I'm no sure why they are different but the AMD arrangement makes more sense to my simple mind.


The Intel numbering made sense when you had system code that didn't understand hyperthreading and you didn't used one processor with multiple cores but the servers (and also often the workstations) had multiple CPU slots where the motherboard often wasn't fully symmetric but required some hardware to be handled by a specific CPU.

 

It would have been very bad if code intended for the second CPU ended up running on the second thread of the first CPU.

Link to comment
26 minutes ago, pwm said:


The Intel numbering made sense when you had system code that didn't understand hyperthreading and you didn't used one processor with multiple cores but the servers (and also often the workstations) had multiple CPU slots where the motherboard often wasn't fully symmetric but required some hardware to be handled by a specific CPU.

 

It would have been very bad if code intended for the second CPU ended up running on the second thread of the first CPU.

 

From a historical viewpoint I can see that made sense. Thanks for the explanation.

Link to comment
On 7/25/2018 at 7:14 PM, unraidrocks said:

 

Before you use Unraid extract the GPU BIOS using TechPowerUp GPU-Z and best to do it on a clean Win10 and then use the Hex editor to remove the headers from the BIOS file. Had to do that on my GTX 1060 because none of the online BIOS worked. 

 

What's wrong with reading the bios in unRAID as explained in that shared video tutorial? Does that not work all the time?

Also - is it possible to use that same GPU for both say OSX and Windows? Obviously not at the same time, but say - shutting one VM down and loading another?

Link to comment

Most ATX-size X370 and X470 motherboards have three PCIe x16 sockets, the top two being wired via a switch that allows either single PCIe v 3.0 x16 or dual PCIe v 3.0 x8 operation, while the bottom one is wired with four PCIe v 2.0 lanes from the chipset. There are usually two or three PCIe v 2.0 x1 slots (and often a second M.2 socket) that take lanes from the bottom x16 slot if they are used. With that in mind, here's what I'd like to do:

 

Put a decent graphics card in the top slot and a SAS HBA in the second long slot so they would get x8 each. Put a cheaper, single slot graphics card in the bottom slot. Now, is there a way of making the primary graphics card be the one on the bottom slot, so that it is used for the unRAID console, leaving the more powerful card free to be passed through to a VM? I'm thinking there would need to be a BIOS option to select which slot has the primary GPU. Does anyone know of an X470 or X370 board that fits the bill? The two that interest me are the Asus PRIME X470-PRO and the Gigabyte AX370 Gaming K5. They both have the slots but I don't know if they have the BIOS option.

Link to comment
  • 2 weeks later...
On 7/29/2018 at 4:31 AM, John_M said:

 

They are actually numbered from 0 to 15 and they are paired even-odd. So threads 0 and 1 are physical core 0, threads 2 and 3 are physical core 1, etc. all the way to threads 14 and 15 being physical core 7. You'll see the arrangement if you switch to the Dashboard page of the webGUI. This is different from how an Intel processor is arranged, where in an i7, for example, threads 0 and 4 represent physical core 0, threads 1 and 5 are physical core 1, threads 2 and 6 are physical core 2 and threads 3 and 7 are physical core 3. I'm no sure why they are different but the AMD arrangement makes more sense to my simple mind.

 

So that would mean 0-7 are the first CCX, and 8-15 are the second, right? 

I assume that mixing and matching between the 2 CCXs would be bad. I'll likely give my VM sole access to a whole CCX, so 8-15, unless thats a bad idea for some reason.

Link to comment

Hi I read little in this post if you need a motherboard for 24 HDD   got this one now link  it are whit 2 sas controllers in the pci-e  and if I also like to get my GTX 970 in it.
there are we only talk abut a X470 right ?  I find only this of the B450 but the price and all are there any benefit then B450
also to be sure I like to use  it for my VR also so the VM must also have the USB of the host that will not be a problem or ? 

sorry for any misspelling 

Link to comment
10 hours ago, tola5 said:

Hi I read little in this post if you need a motherboard for 24 HDD   got this one now link  it are whit 2 sas controllers in the pci-e  and if I also like to get my GTX 970 in it.
there are we only talk abut a X470 right ?  I find only this of the B450 but the price and all are there any benefit then B450
also to be sure I like to use  it for my VR also so the VM must also have the USB of the host that will not be a problem or ? 

sorry for any misspelling 

keep in mind,  X8ST3-F is very old board with SAS1068E SAS controller - it wont support HDDs > 2TB..

Link to comment
2 hours ago, tola5 said:

So I take the jump and order it all come before time that my main pc motherboard is dining so hope later to get more rams
ASUS ROG STRIX X470-F GAMING
Ryzen 7 2700X
HyperX Predator DDR4 3000MHz 16GB
 
 wish me luck

 

Set your BIOS 'Power Supply Idle Mode' to 'Typical Current Idle' (or whatever the Asus equivalent wording is) and all should be golden.

 

Have fun :)

  • Like 1
Link to comment

Hi so sorry to post this but after reading it sound like it a ryzen thing but this thread are so long and after reading I can see some thing have changed so hope have some status what need when the newest unraid to get this to work do zenstate still need or ? I post down under my post  from a question hope it good enough now that it look like it truly are a ryzen probleme ?

Razen 2700k 16 GB ram and a Nvida 970 
when run the vm whit vnc it work  but when try run it whit  gpu passthrough it not I have  give it the bios file , I have try 2 different but that don't help I think it not that ?   when look at the log I can see   non essential  but it sound like it give me problem anyway ? 
hope some help from your smart people ;-) and sorry for the post setup  whit the xml hope you get it ? also sorry for any misspelling if you have any problem just ask and will try to rephrase it 

IOMMU group 19:	[1022:1455] 0b:00.0 Non-Essential Instrumentation [1300]: Advanced Micro Devices, Inc. [AMD] Device 1455
2018-08-18 07:45:18.259+0000: starting up libvirt version: 4.0.0, qemu version: 2.11.1, hostname: Backup
LC_ALL=C PATH=/bin:/sbin:/usr/bin:/usr/sbin HOME=/ QEMU_AUDIO_DRV=none /usr/local/sbin/qemu -name 'guest=Windows 10,debug-threads=on' -S -object 'secret,id=masterKey0,format=raw,file=/var/lib/libvirt/qemu/domain-10-Windows 10/master-key.aes' -machine pc-i440fx-2.11,accel=kvm,usb=off,dump-guest-core=off,mem-merge=off -cpu host,hv_time,hv_relaxed,hv_vapic,hv_spinlocks=0x1fff,hv_vendor_id=none -drive file=/usr/share/qemu/ovmf-x64/OVMF_CODE-pure-efi.fd,if=pflash,format=raw,unit=0,readonly=on -drive file=/etc/libvirt/qemu/nvram/a0321852-5996-c06c-f365-f29de45eb6b8_VARS-pure-efi.fd,if=pflash,format=raw,unit=1 -m 12800 -realtime mlock=off -smp 14,sockets=1,cores=14,threads=1 -uuid a0321852-5996-c06c-f365-f29de45eb6b8 -display none -no-user-config -nodefaults -chardev 'socket,id=charmonitor,path=/var/lib/libvirt/qemu/domain-10-Windows 10/monitor.sock,server,nowait' -mon chardev=charmonitor,id=monitor,mode=control -rtc base=localtime -no-hpet -no-shutdown -boot strict=on -devic2:54:00:1b:58:6e,bus=pci.0,addr=0x2 -chardev pty,id=charserial0 -device isa-serial,chardev=charserial0,id=serial0 -chardev 'socket,id=charchannel0,path=/var/lib/libvirt/qemu/channel/target/domain-10-Windows 10/org.qemu.guest_agent.0,server,nowait' -device virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=org.qemu.guest_agent.0 -device vfio-pci,host=08:00.0,id=hostdev0,bus=pci.0,addr=0x5,romfile=/mnt/user/domains/vbios/Gigabyte.GTX970.4096.141224.rom -device vfio-pci,host=0b:00.3,id=hostdev1,bus=pci.0,addr=0x6 -device vfio-pci,host=08:00.1,id=hostdev2,bus=pci.0,addr=0x8 -device usb-host,hostbus=1,hostaddr=2,id=hostdev3,bus=usb.0,port=1 -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x9 -msg timestamp=on
2018-08-18 07:45:18.259+0000: Domain id=10 is tainted: high-privileges
2018-08-18 07:45:18.259+0000: Domain id=10 is tainted: host-cpu
2018-08-18T07:45:18.299967Z qemu-system-x86_64: -chardev pty,id=charserial0: char device redirected to /dev/pts/0 (label charserial0)
2018-08-18T07:45:20.132183Z qemu-system-x86_64: vfio: Cannot reset device 0000:0b:00.3, depends on group 19 which is not owned.
2018-08-18T07:45:21.161893Z qemu-system-x86_64: vfio: Cannot reset device 0000:0b:00.3, depends on group 19 which is not owned.
Link to comment
1 hour ago, tola5 said:

Hi so sorry to post this but after reading it sound like it a ryzen thing but this thread are so long and after reading I can see some thing have changed so hope have some status what need when the newest unraid to get this to work do zenstate still need or ? I post down under my post  from a question hope it good enough now that it look like it truly are a ryzen probleme ?

Razen 2700k 16 GB ram and a Nvida 970 
when run the vm whit vnc it work  but when try run it whit  gpu passthrough it not I have  give it the bios file , I have try 2 different but that don't help I think it not that ?   when look at the log I can see   non essential  but it sound like it give me problem anyway ? s on group 19 which is not owned.

 

Hey Tola5,

 

On the zenstates, no you don't need to play around with those any more. It turned out that the lockup problem was to do with most power supplies idling at too l;ow a voltage for the Ryzen CPU, which resulted in that lockup. To avoid that problem the motherboard manufacturers introduced a #Power Supply Idle' mode. Find that in the BIOS and set it to 'Typical Current Idle'.

 

On the GPU issue, take a look at Gridrunner's youtube video:

 

 

 

Hopefully that will help.

Edited by Dev Null
  • Like 2
Link to comment

Sorry one more question 

Need to encoding lot off stuff so like to be sure my temp . So I google and read there was no driver whit dynamic temp plugin but you cold use a orther did that . But can only see 3 and one of them are to hot for my liking so are it right or ? (The top hit 80) 

Edit 

Yes them temp was the same in Windows the stock cooler just get it like that new on the way ?

 

Screenshot_20180819-105914_Chrome.jpg

Edited by tola5
Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.