Love my job...

[jonp]

I've tested this on multiple hardware hosts with various GPUs / CPUs / Motherboards and have had great success repeating this. It won't work for everyone / every GPU, but I bet it works for a lot.The last testing I'm doing today is to pass through a blu-ray / DVD player device that I have through to the VM to see if I can get that working solid. 

 

The next test I want to perform at some point is getting a cable card and putting it into the server, pass it through to OpenELEC. 

 

Now you could use your unRAID server as your storage for your DVR/PVR for your cable TV.  You could then use XBMC as the interface for that...

 

What happens if a parity check is in progress?

 

I run all my VMs on the cache pool, so their performance is completely unaffected by the array / parity.  I can watch a movie on the VM while doing a parity check on the array and you wouldn't notice.

[piotrasd]

think I just did, didn't I (share the details)?  If you're looking for developer level stuff, I didn't actually compile OpenELEC myself, Eric did.

I mean maybe config - xml of VM Openelec (openelec i can compile/build myself with that options what you tell me, but i have small expirence with KVM

[jonp]

After beta 9, I will post something.

[Owel]

Will it be possible to do a GPU passthrough with the CPU integrated GPU for AMD? (on intel side a iGPU)

So I have an AMD 5350, will it be possible to use the GPU with this XBMC without an extra GPU Card?

[piotrasd]

After beta 9, I will post something.

 

Ok thanks

[jonp]

Will it be possible to do a GPU passthrough with the CPU integrated GPU for AMD? (on intel side a iGPU)

So I have an AMD 5350, will it be possible to use the GPU with this XBMC without an extra GPU Card?

Probably not, but to be fair, we haven't tested AMD integrated graphics pass through at all.  Intel just does not work.  We keep testing from time to time, but no change.

[Owel]

Ok, thanks for this info.

 

Do you have the equipment to test AMD integrated GPU? Or should I volunteer to test it with your help?

 

Regards

Owel

[jonp]

Ok, thanks for this info.

 

Do you have the equipment to test AMD integrated GPU? Or should I volunteer to test it with your help?

 

Regards

Owel

 

I do not have an AMD systems for testing at this time (just many variants of Intel).  This isn't because we're AMD-haters or anything that, its just that the majority of folks run on Intel and that is somewhat of the standard in our industry.  That said, if you're willing to test for us, we'd really appreciate it!

[jumperalex]

[shrug] for what its worth, as best i can tell, $/iommu availability ratio (and ecc $/availability) says AMD might be something you want to invest in at least a single box for testing.  I know Intel is the "standard" for the "industry" but we aren't always talking industry for your non-commerical clients.  I'd also guess it is those non-commercial clients who are most likely to want to pass-thru GPU for things like HTPC or gaming.

 

Just my two cents and back to the topic: Where's my case badge?!?!?! 

[jonp]

[shrug] for what its worth, as best i can tell, $/iommu availability ratio (and ecc $/availability) says AMD might be something you want to invest in at least a single box for testing.  I know Intel is the "standard" for the "industry" but we aren't always talking industry for your non-commerical clients.  I'd also guess it is those non-commercial clients who are most likely to want to pass-thru GPU for things like HTPC or gaming.

 

Just my two cents and back to the topic: Where's my case badge?!?!?! 

 

FYI, when I say standard for the industry, I mean server and desktop.  All my personal unRAID servers are running on desktop grade Intel systems (no Xeon's here).  The only Xeon I have is in the AVS 10/4 test system from Lime Tech.  All my friends that run gaming rigs all run Intel as well.  Intel is the standard in the industry for processors...period.  AMD has been encroaching on that territory though and we will take another look at them, but there's only so much we can tackle at one time in development.

 

In theory, the IOMMU technology should work the same for both vendors so long as you have your kernel flags for them correctly assigned.