I know should ask it on Intel forums, but then again you get only fanboy's opinions.
and yea i searched and searched on youtube no facts only the price range (which i did found on the step side)

a few thing's they not talk about is the following

1.
Drivers You hear nothing of it do it support Linux? is telementry baked in like Nvdia? do it support DX9 or upwards?.
etc

2.
DLSS it have the technolgie but it not say it's it backwards compatible to other DLSS? for example mech warrior 5?

1. The product has not been released yet, so nobody knows but driver support, even for Windows, will most likely be the weakest point of this GPU.

2. DLSS is proprietary to nVidia and will only be available on nVidia products. Intel will offer a similar technology called XeSS that will also be available on nVidia and AMD GPUs but it will not be compatible with DLSS.

Intel has already started committing code to their open source Linux drivers in anticipation of the new dGPUs. They will be supported by Mesa, much like their iGPUs currently are as well. All cards released these days provide support for anything up to DX12, including previous DX versions.

not entire true my current GPU do not support windows 10 LTSB so Windows 7 is out of the picture
so it's not complete crazy that DX10 or DX9 is no longer supported

"Developed under the previous codename "DG2", the first generation of Intel Arc GPUs (codenamed "Alchemist") is slated for release by the first quarter of 2022. Alchemist uses the Intel Xe GPU architecture, or more specifically, the Xe-HPG variant. Alchemist will support hardware-based ray tracing, XeSS or supersampling based on neural networks (similar to Nvidia DLSS), and DirectX 12 Ultimate"
https://en.wikipedia.org/wiki/Intel_Arc

^ They're generally backwards compatible, so it should play DX9 games. Look at how the Intel iGPU's like the UHD 630 that have DirectX12 feature set can play older DX7-11, even DirectDraw titles, and I don't see why it should be any different to that.

Try waiting for some actual data.

The Intel cards are likely to be hot, slow and unreliable.

No particular insult there as they are first effort (well, excluding the historic Larrabee) and while there are plenty of jokes to be made at Raja Koduri's expense it won't be his fault except perhaps any excess hype applied. They may be competitively priced as a loss leader, but probably not- as they're using an external foundry (TSMC 7nm), which also means they will not even be increasing supply but competing with AMD (and likely nVidia again too, next gen) for space. Them not being fabbed by Intel is unsurprising, as Intel 7 (formerly '10nm') is still pretty broken and the yield for GPU sized chips (often ~8x the size of a CPU chip) would be miniscule, while trying to get a competitive 14nm GPU working would result in nuclear fusion level temps.

Intel aren't going to be saying that they'll be hot and slow, but they're going from a standing start so it's pretty much inevitable. Probably the biggest elephant in the room is patents and other IP, Intel will have to work around AMD and nVidia patents etc and do a lot of reinventing the wheel to not fall afoul of them. It should also be noted that while AMD and Intel have a permanent cross licence agreement it's only for x86/x64, not graphics.

(The one Intel DG2 card already on the market is pretty much a complete joke, as a discrete card too, but it isn't representative either)

Actually, given past history, I would expect driver support on Linux to be *better* than driver support on Windows.

Consider, for example, that my Intel HD 4600 even has partial Vulkan support on Linux. According to a 5 year old reddit comment, it's supported on Linux but not Windows (at least at the time of the post).


Anyway, one question is whether one of these GPUs would make a good companion to the AMD 5600G (or 5700G). Of course, for this to happen:
* The Intel GPU needs to play well with the AMD CPU.
* The Intel GPU must be better at at least one task than the AMD CPU's integrated graphics. (Although one could just use it for pci passthrough, though the question is whether it's worth getting just for that; or, with Vulkan, one can do graphics on one GPU and compute on the other.)

Now I know a lot of people are saying their Windows drivers are crap (and I can't comment on that point to be honest, since I haven't used Windows in quite some time now), but their Linux drivers work surprisingly well with their iGPUs. I have no reason to think it will be any different with their dGPUs - initial jank excluded.

Agreed about the rest of the things you've said - their first dGPU effort is probably not going to impress anyone in terms of performance or architectural capability.

this discussion is so premature, it's not even funny. Intel hasn't announced anything, design is not finalized, I doubt Intel's own management or engineers know what will the final product be. Sure, they made a lot of promises but there's a huge gulf between stating goals and actually achieving them.

If you are using Intel graphics on Windows, it hepls to download and install a new driver.

ah they said they release it Q1 of 2022 that's less then 5 months tops
that means they already producing them in factory (that means finalized)

so no not premature

I haven't used their windows drivers for anything apart from Office type tasks for 10 years plus, and at that they've been absolutely fine. Their gaming support before that was pretty awful though, and at times was actively dishonest.

But, it's unlikely to be just driver issues contributing to unreliability. End of the day the best extant Intel iGPU is about on par feature and performance wise with a 1030. Jumping to the rumoured 3070ish level leads to a bunch of potential issues: temperature/ power/ performance management, hardware testing, management software etc that will largely have to be designed from the ground up.

Longer term they'll probably have decent drivers, but the initial ones are likely to be borderline unusable and that may well extend for months. AMD and nVidia very often have troubles* even when introducing iterative updates, and Intel wants to introduce a bunch of revolutionary ones (for them) at the same time. Nobody is that surprised or disappointed if an iGPU can't perform well or isn't stable with gaming, but if you're paying 400USD you expect a bit more.

*On the positive side, it's likely that a lot of those troubles are from legacy code and from having to support multiple generations and 5 year old cards. There are some advantages to be had from dealing with a clean slate as well as the disadvantages.

Things are now in motion that cannot be undone :P. It started in July this year.

I am somewhat impressed by what their Xe-LP iGPUs can do. Of course, they're nowhere near Vega cores from AMD, but nobody was seriously expecting them to be anyway. What they are, however, is a big leap in tech compared to their previous iGPU generations. If Intel can keep that momentum going, we may be heading for a triumvirate of GPU manufacturers, which is, IMHO, a good thing for everybody.