Untitled

wonder how this will work on linux, since we use vulkan anyways

There will be very limited run of A770 and A750. Most silicon will go into mobile Alchemist, OEMs will do A380

Not yet, the hardware should be there, even in Xe iGPUs back to Tiger Lake, you should get something similar to the following if you run lspci -vvvv. I doubt the dGPUs have removed it.

00:02.0 VGA compatible controller: Intel Corporation TigerLake-LP GT2 [Iris Xe Graphics] (rev 01) (prog-if 00 [VGA controller])
Subsystem: Dell Device 0a20
Flags: bus master, fast devsel, latency 0, IRQ 142
Memory at 6052000000 (64-bit, non-prefetchable) [size=16M]
Memory at 4000000000 (64-bit, prefetchable) [size=256M]
I/O ports at 3000 [size=64]
Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
Capabilities: [40] Vendor Specific Information: Len=0c
Capabilities: [70] Express Root Complex Integrated Endpoint, MSI 00
Capabilities: [ac] MSI: Enable+ Count=1/1 Maskable+ 64bit-
Capabilities: [d0] Power Management version 2
Capabilities: [100] Process Address Space ID (PASID)
Capabilities: [200] Address Translation Service (ATS)
Capabilities: [300] Page Request Interface (PRI)
Capabilities: [320] Single Root I/O Virtualization (SR-IOV)
Kernel driver in use: i915
Kernel modules: i915

You still need drivers though and they won't be upstreamed until Q4 next year apparently. You can try using the linux-intel-lts branch to try it but I doubt it's in any state ready where you don't need internal Intel code that isn't public yet in that Github.

github.com/intel/linux-intel-lts/issues/33

>I doubt the dGPUs have removed it.
github.com/intel/gvt-linux/issues/160
Internally we have some prototype but there is no formal plan to support GVT-g on discrete GPU like DG1. In a client platform which has had integrated GPU, are there still needs to share DG1 among multiple virtual machines? Thanks!

The rebar shit kills it for me.

GVT-g isn't hardware which is what I meant when I said it wasn't removed, it's software and it's understandable why Intel wants to discontinue it for Xe and above because SR-IOV is supported in hardware and it will get rid of limitations with GVT-g like synchronization issues especially with multiple VMs because of needing to schedule GPU tasks internally which SR-IOV lowers the overhead of and should eliminate.

Considering games on linux often rely on wine/proton i suspect the terrible dx9/11 performance will impact linux gaming as well. Vulkan and dx12 shouldn't be a problem, OpenGL as the legacy api to vulkan probably has comparable performance problems. Intel Marketing guy said on ltt they would price according to "tier 3" (meaning the games where alchemist sucks) performance, but there's no way in hell the a770 will be comparable to an 6600xt or even a 3060 in those titles, so i do hope Intel is ready to go down significantly with their pricing.
>There will be very limited run of A770 and A750
I guess for the purpose of testing their products that's probably the best idea. They should still drop those prices to make a more favorable first impression. Just look at OnePlus, first Phone was a banger with limited numbers for a good price, every phone after that just got marginally better with significantly higher prices. I really hope they can establish themselves in the gpu market, but I'm not convinced Intel has the balls to do what's necessary, even though they claim they do. Anyway, it's going to be an interesting year nonetheless.

Intel is claiming they will price the card according to the performance on the worst case of gaming.
Which for non-gaming purposes may give it a killer price

Who would buy such toy cards? I bet they'll struggle even at 1440p and I see no chance they will be able to maintain even shitty 60fps in most games.

>i suspect the terrible dx9/11 performance will impact linux gaming as well.
but nowadays we use dxvk for dx9/11 and vkd3d for dx12 to turn that shit into vulkan

I was under the impression that the api implementation cannot fundamentally fix the shortcomings of the api layer it mimics, but maybe I'm wrong. If that were possible then amd/intel/nvidia could simply write their own dx9todx12 for windows (or modify dxvk) and easily increase performance. Especially intel would have done this in a heartbeat to fix their abysmal performance on dx9/11. The fact that they haven't is imo. a clear indication that even though it can help in some cases, it is not a suitable fix for all of the shortcomings of the older apis. But this is of course all just speculation on my part.

If they're priced well, many people

Why didn't the ape developers from these popular old games port them to Vulkan or low-level dx by now? I mean GTA 5 still earns rockstar big money and yet..

Let me guess, 1080p is not enough for you.