• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Intel Arc series unveiled with the Alchemist dGPU to arrive in Q1 2022

Soldato
Joined
30 Aug 2014
Posts
5,963
Apparently Intel is dropping DX9 driver support for Arc and Xe. Intel will be using an open source software DX9 to DX12 API conversion for all DX9 games



Intel would probably do this for 10 and 11 as well if the software exists, they really don't want to make new drivers for anything other than DX12 games
DXVK does exist, is very good indeed and is in heavy use on Linux and makes up the basis of Valve's Proton translation layer (the Steam Deck relies on this). I wouldn't be surprised if Intel intends to use this; after all, people were fixing poor AMD DX11 performance in certain games on Windows by using DXVK dll's. Here is just one example: https://www.reddit.com/r/Amd/comments/ggx9ke/dxvk_on_windows_can_improve_the_performance_of/
 
Caporegime
Joined
22 Nov 2005
Posts
45,280
Intel would probably do this for 10 and 11 as well if the software exists, they really don't want to make new drivers for anything other than DX12 games
Microsofts claims the API is seamless and works great though... it's probably what the xbox uses seeing as Microsoft wrote it
 
Soldato
Joined
17 Aug 2009
Posts
10,719
That is extraordinarily convenient for Intel.

On the other hand, this opportunity didn't appear recently so why did they have reviews n stuff BEFORE this.
 
Soldato
Joined
26 May 2014
Posts
2,955
The problem is (and I’m sure there is a more technically accurate way to put it) that DX11 is an old API, which AMD and NVidia driver engineers hand optimise each DX11 game via drivers (often overwriting API commands) to improve performance and make it perform well (or as well as it possibly can on said hardware). This takes time, effort, experience and a lot of employees. AMD and Nvidia have been at it for a long time, Intel on the other hand are just getting started with discrete graphics, and it will take them a long time to catch up here. DX12 and Vulkan on the other hand does not require so much work from the driver teams and is therefore a lower hanging fruit and that would likely explain why ARC does better here.

A game developer spoke about this recently
On the Moores Law Is Dead Podcast, broken silicon, if you would like a more accurate explanation.
I know how these things work. It's not an excuse, and it's certainly not something that would make me any more inclined to buy a half-finished product either. An admittedly half-finished product, with Intel sending Tom Petersen out on the PR circuit to practically apologise for all its shortcomings. Intel are a huge company with near-unlimited resources. They're not some tiny startup who we have to accept can't compete with the big boys. They should have been working on the drivers from the moment final silicon was ready, and yet we know from leaks that they weren't. They didn't want to spend the money developing an entirely new driver stack and instead chose to repurpose and expand upon their iGPU drivers instead. The result of that choice is a burning garbage fire, and a ton of even the absolute biggest, most-played games on the planet running like absolute **** on Arc. A game like CS:GO is just plain broken. Performance doesn't change no matter what you do with the settings. It's just hard capped at ~140fps, which is below what even entry level cards like the RX 6400 deliver.

I think everybody was prepared to give Intel some leeway and accept that, sure, perhaps that obscure game from 2008 that you still play might have some issues initially. People would have been okay with that. What they don't get a pass for is literally anything that doesn't run on DX12 or Vulkan performing terribly, and all because Intel took a bunch of shortcuts and refused to put the work in. They didn't try and fail - they simply didn't try. Their leeway and good will ends right there. The state of the drivers currently is outright embarrassing. As the Gamers Nexus video pointed out, half of the features are just plain broken (hell, installing the things is just plain broken), before you even get into performance. I'm still rooting for Intel to get their **** together and provide some competition, but the desire to do so within the company appears limited outside of a small handful of people. It would be less of a surprise if Gelsinger pulled the plug on the whole thing than it would be if Intel got their drivers sorted any time soon.
 
Caporegime
Joined
22 Nov 2005
Posts
45,280
As the Gamers Nexus video pointed out, half of the features are just plain broken (hell, installing the things is just plain broken), before you even get into performance. I'm still rooting for Intel to get their **** together and provide some competition
someone keeps pointing out in youtube comments that there is a specific beta driver that everyone should be using, some other influencer type made a video about it apparently.

I can't rmember what name the comment said something like wizard, I don't pay attention to most youtubers so no idea who it was. but apparently the guy did some test with different drivers
 
Caporegime
Joined
17 Mar 2012
Posts
47,662
Location
ARC-L1, Stanton System
I know how these things work. It's not an excuse, and it's certainly not something that would make me any more inclined to buy a half-finished product either. An admittedly half-finished product, with Intel sending Tom Petersen out on the PR circuit to practically apologise for all its shortcomings. Intel are a huge company with near-unlimited resources. They're not some tiny startup who we have to accept can't compete with the big boys. They should have been working on the drivers from the moment final silicon was ready, and yet we know from leaks that they weren't. They didn't want to spend the money developing an entirely new driver stack and instead chose to repurpose and expand upon their iGPU drivers instead. The result of that choice is a burning garbage fire, and a ton of even the absolute biggest, most-played games on the planet running like absolute **** on Arc. A game like CS:GO is just plain broken. Performance doesn't change no matter what you do with the settings. It's just hard capped at ~140fps, which is below what even entry level cards like the RX 6400 deliver.

I think everybody was prepared to give Intel some leeway and accept that, sure, perhaps that obscure game from 2008 that you still play might have some issues initially. People would have been okay with that. What they don't get a pass for is literally anything that doesn't run on DX12 or Vulkan performing terribly, and all because Intel took a bunch of shortcuts and refused to put the work in. They didn't try and fail - they simply didn't try. Their leeway and good will ends right there. The state of the drivers currently is outright embarrassing. As the Gamers Nexus video pointed out, half of the features are just plain broken (hell, installing the things is just plain broken), before you even get into performance. I'm still rooting for Intel to get their **** together and provide some competition, but the desire to do so within the company appears limited outside of a small handful of people. It would be less of a surprise if Gelsinger pulled the plug on the whole thing than it would be if Intel got their drivers sorted any time soon.

Right exactly.

On the public marketing side they like to boast about how "They are the biggest manufacture of graphics and have been doing for decades" but when reviewers get a hold of the only GPU they actually have on the market and show it for how bad it is Intel change tune and complain "Don't be so hard on us we are new to this"

One thing that keeps coming up time and time again, for years now, is the state of Intel's internal politics, infighting between departments and even individuals, it may well be one reason for the incompetence but also a lot of the shameless and cynical public facing BS that is responsible for Intel increasingly looking like a joke.

Intel are a massive and rich company, so the only reason for this is sheer incompetence. As AIB's have apparently told MLID "We don't think Intel know what they are doing"
 
Caporegime
Joined
17 Mar 2012
Posts
47,662
Location
ARC-L1, Stanton System
I think one problem is Intel are used to winning, they are used to getting their way, they have done some really cynical BS in their time and keep getting away with it.

Now ARM, Nvidia and AMD are literally stealing their lunch like the biggest school yard bully, they don't know how to handle the situation, "AMD are now back in the rear view mirror" really???? are you sure Pat? What about soon to be released Genoa? and Bergamo? What about AMD's Zen 5 road map? how can you say AMD are back in the rear view mirror when you still haven't got a response to end of life Zen 3 in data-centre???? <- That was pretty much the literal response from Intel's investors to that quib!

A bit of humble pie, Intel, get off your high horse. Before Nvidia and AMD pull you off it and beat you to death.
 
Last edited:
Caporegime
Joined
17 Mar 2012
Posts
47,662
Location
ARC-L1, Stanton System
MLID makes a few good points, there is no evidence that ARC is something "designed for modern API's" even with DX12 performance is all over the place, and the reasoning given for needing ReBar was again claimed to be because "its a modern Memory Controller"
Its obvious nonsense, its the audacity to sit there with a perfectly straight face like this with this poo.

This sort of thing is a pet-pev of mine, it comes from a lack of respect of their audience, like they think they are talking to stupid people, its indicative of the attitude of these people and when one has been watching them for many years of the same sort of nonsense the rumours that they have internal politics problems rings absolutely true, if they can behave like this in their marketing i have no doubt they rub eachother up the wrong way constantly.

I've said it before, IMO Ryan Shrout is cancer, however, he probably fits in quite well with what's already there.

I want a third competitor, but Intel keep reminding me why i like seeing them suffer, do the decent thing Intel, for once in your 40+ years of existence, sell ARC to someone who knows what they are doing, please.

 
Soldato
Joined
1 Feb 2006
Posts
3,397
MLID makes a few good points, there is no evidence that ARC is something "designed for modern API's" even with DX12 performance is all over the place, and the reasoning given for needing ReBar was again claimed to be because "its a modern Memory Controller"
Its obvious nonsense, its the audacity to sit there with a perfectly straight face like this with this poo.

This sort of thing is a pet-pev of mine, it comes from a lack of respect of their audience, like they think they are talking to stupid people, its indicative of the attitude of these people and when one has been watching them for many years of the same sort of nonsense the rumours that they have internal politics problems rings absolutely true, if they can behave like this in their marketing i have no doubt they rub eachother up the wrong way constantly.

I've said it before, IMO Ryan Shrout is cancer, however, he probably fits in quite well with what's already there.

I want a third competitor, but Intel keep reminding me why i like seeing them suffer, do the decent thing Intel, for once in your 40+ years of existence, sell ARC to someone who knows what they are doing, please.

His comments about Resizable BAR/SAM are not correct as Resizable BAR is in the PCIe spec (think PCIe 2) so AMD did not invent it, they just started using it first and changed its name like it was something new. All it does is allow bigger transfers so N*256MB vs 1(Max VRAM size).
 
Caporegime
Joined
17 Mar 2012
Posts
47,662
Location
ARC-L1, Stanton System
His comments about Resizable BAR/SAM are not correct as Resizable BAR is in the PCIe spec (think PCIe 2) so AMD did not invent it, they just started using it first and changed its name like it was something new. All it does is allow bigger transfers so N*256MB vs 1(Max VRAM size).

Yeah he's not too bright himself. Doesn't change it, Intel didn't create an IMC that only works with ReBar. At least not deliberately. He didn't need to over think the reasoning, its just obvious.
 
Caporegime
Joined
17 Mar 2012
Posts
47,662
Location
ARC-L1, Stanton System

There is some really odd journalism around this GPU, firstly bang on about how badly Nvidia / AMD react to criticism for about 5 minutes, just making tripple sure everyone knows Intel good, everyone else bad, and then add the disclaimer at the end almost in passing to say actually they haven't been like that for a couple of years, well why the #### make that a huge talking point as if its critically relevant?

Then he talks about how good it is of Intel not to over egg the performance, as if everyone else does, and then play the clip which Intel's marketing guy talks about how they are going to "kick ass" i had to rewind that bit 3 times looking for the intended irony humour but no, its not there. He completes that segment by talking about how trustworthy we should all view Intel's performance slides.

In mean ffs Steve...
 
Caporegime
Joined
17 Mar 2012
Posts
47,662
Location
ARC-L1, Stanton System
"ReBar on Intel GPU's doesn't cause problems....... Sometime games with certain API's don't work well on AMD and Nvidia"

Oh come on Steve, that is one hell of a way to gloss over the fact that you have to enable ReBar on Intel GPU's or its flat out broken.

Is Ryan Shrout writing this script for you? He is isn't he? Yes he is.
 
Associate
Joined
31 Dec 2010
Posts
2,441
Location
Sussex
Umm.. no. Intel specifically say you need to enable ReBar. Their drivers are designed for it to be on.
Forgot that part already when I just replied in the 1630 thread. So for older systems, Intel is also out?
Despite the performance the 6400 or 6500 XT loose without PCIe 4.0, it still came out ahead of the 1630 for old system. A380 I dismissed because of its inconsistent performance, having totally forgotten how dependent it is on ReBar.

Presenting Navi24, the very best of a bunch of very bad options. Not exactly a ringing endorsement.
 
Back
Top Bottom