Common Misconceptions of Intel Integrated Graphics

If you have any questions regarding this article, Intel kindly asks that you direct them to the Intel® Graphics Developer Community.

Of course, general comments and feedback are still welcome here.

Such a shameless commercial from Intel... Though not surprising from their side, I find it very disappointing for a respectable developer site like GameDev.net to publish such articles. At least it should be marked as [Sponsored] then. Very disappointing.


Ivan

I agree. This is obviously little more than promotional material, it should be labeled as such.

As for content though, I must admit that I find it hard to understand what they're trying to accomplish. They state on the first page that their wiz-bang X3000 architecture is "just as good as those other guys!" which is great and all (I'm all for an improved graphical baseline) but...

1) Most people, even if they are using integrated graphics, DON'T have the X3000 architecture. It's only been out for a year or so as near as I can tell, which means that a vast majority of integrated graphics users are still stuck in the stone age. You simply can't code to the level of graphics they profess if you want to reach the integrated market.

2) Even for people who do have the new architecture, it doesn't really matter to developers. We code to DirectX and OpenGL, not NVIDIA, ATI, or Intel. The only thing this really means is that we have yet another platform to QA. Fact is, if Intel did everything as wonderfully as they claim then developers are free to pretty much completely ignore them since it should "just work."

So, in a nutshell: Sorry Intel, but this doesn't change things at all. If we want to target high-end graphics we're going to target dedicated chipsets, and if we want to target low-end we still have to deal with your legacy products. Maybe two or three years from now this might matter, but for now the net worth is near zero.

"consumer experts agree that it is possible to achieve a playable experience with most games on the market using the Intel X3000 series"

I'm glad it's possible to finally achieve a playable experience with Intel integrated chipsets. ;)

I feel very strongly against Intel integrated graphics. I used to have to do tech support for a video game company, and I always hated explaining to customers that their "new" computer wouldn't run a game because of their stupid Intel graphics chipset.

Intel integrated chipsets have been utter and complete garbage for playing games. It seems like that may be changing with their newest technology, (which is definitely a plus considering the sheer number of computers out there with them), but I'm still not that impressed.

When programming for Intel GPUs, does one still have to use a completely round-about way of determining maximum primitive count?

That's just silly.

Quote:

Original post by Tojiro67445 As for content though, I must admit that I find it hard to understand what they're trying to accomplish.

Ditto. I could understand an article like this on TomsHardware or something, aimed at consumers, but I'm not sure what developers are supposed to take away. If we don't specifically target Intel GPUs right now, it's because we prefer to target the hardware which our target demographic is actually using, and which is industry standard. I'd certainly be willing to program to improve performance on the X3000 too, assuming the tradeoffs weren't too severe, but what useful information does the article present on how to do this? "Use a profiler." Gosh, thanks, Mr. DeSylva.

Like others said, this is no article, this is a publicity and the link to it should belong on a banner. It should be taken down, moved or at least have a disclaimer at the start.

It's the same stupid reason they have been giving developers for a few years now.

For example they say it can you recent games, but they don't say at what resolution or graphic quality.

They say they have 50% of the integrated market, just like they said before, true.
But what they don't say is that they own about 0% of the gamer demographic.
You don't lower your min spec(and graphic quality) so that people who don't play games can play your game.

They should work on accelerating flash games and desktop games, thats the only gaming those who own these will do anyway. Ho and web page display should also be accelerated!

The whole point of integrated graphic is to save cost... and so Intel can make more money selling you a package. They should concentrate on rendering Aero glass and stop there disguised publicity.

If Intel didn't pay for this, I can only assume that Gamedev is getting desperate for article submissions...

"Myth 5: Intel GPUs have abysmal fillrate"

Oh wait, that's still true. And pretty much cripples their use for anything non trivial. Looking at their "Developer community" forum is just depressing, nothing but bugs, conflicts and complaints about incompatible games.

I'm glad to see that I'm not the only one who thinks that article wasn't objective. I didn't need to go to the final page to realize that the one who wrote this article works at Intel.

It's good to see that 50% of the integrated chipsets will at some point support SM 3.0

Integrated video cards weren't ever fast. I know this because I've seen:
SiS (620, 630 and 740), nForce (2, 4, and 6100), VIA Unichromes models (can't recall versions).
They were good, but not capable to compite against dedicated video cards. They're new, you can play the latest games, and after one year they just can't catch up with newer games.

Quote:

The Intel X3000 graphics core is an integrated solution that has demonstrated the capability to run some of the mainstream games at playable frame rates

Well they were at the _minimum_ playable rate. I'm sure they won't be able to stand against the upcoming games.

Like it's been said:
*What % of the video card global-market do they own? Not only integrated ones..
*The top-games they ran: What quality settings did they use?
*The benchmark was made with completely different PCs. Integrated video cards depend a lot on the motherboard they are running on, and the speed and quality of the system RAM, even if they use the same CPU. And if you go to the source of the benchmark, there is a little very INTERESTING note on the bottom regarding the test.

Another interesting note is that Intel didn't say (I don't know) if they drivers are still updates. All integrated cards I've seen (excluding NVIDIA nForces) have pretty old drivers, because they're abandoned. Does Intel do the same?

After all this, I conclude, integrated video card still are what they always were, cheap alternatives.

Like some in this forum, I don't see either the point of this article for developers. I won't judge GameDev.Net for that, but it's risky that non-objective material ends up in a page like GD.Net
They're not even offering an own tool to debug their cards (except for the well-known VTune Analyzer)

Dark Sylinc

PS: However, it is impressive how Intel improved in the graphic field, at this rate, pherhaps in a future, they may be able to compite aginst dedicated GPUs

Quote:

Original post by Matias Goldberg PS: However, it is impressive how Intel improved in the graphic field, at this rate, pherhaps in a future, they may be able to compite aginst dedicated GPUs

Yea they have made progress, they at least try and support up to date features now. They could also match the performance of the low end card some day, probably, but you have to ask yourself why is the same chip slower on a MB compared to a GC? probably memory bandwidth? can they improve that... yes but the cost will go up.
Use a chip with more shader units? also increases cost.
If they want integrated graphics with real gaming performance the MB will at least cost 300$, and that won't give you a GF8800's performance.
So what is the point of having it on the MB then?
With a graphic card, when it's too old or break, at least you can change it.

Now that I look at it, this seems to be a direct copy of the article here. Doesn't GDNet have a general policy against posting articles already available elsewhere?

Quote:

Original post by Matias Goldberg It's good to see that 50% of the integrated chipsets will at some point support SM 3.0

yeah, and by the time that that happens, the rest of the world will be using Shader Model 5.0.

I'll start paying attention to Intel as a serious graphics hardware provider when they produce a product that doesn't have abysmal performance. Wasn't every game in their supposed benchmark now at least a year old? If it is playing a game that is already a year old at barely playable framerates (on the lowest quality settings, no doubt), that doesn't strike me as being very good performance.

At the very least this article should be declared as sponsored. (gosh I hope GDnet got paid to put this up!)

The most disappointing part of this article is the fact that there is nothing useful for me to take away as a graphics developer, e.g. favorable texture/data formats, cache mannerisms, etc.

The "optimizations" section of this article is laughable, search for optimization on the nvidia developer site and you will be presented with in depth tips and techniques, stuff you can actually use.

I don't think this will have a major effect on my opinion of GDnet, everyone makes mistakes from time to time, and I'm willing to ignore this shameless commercial stunt (by intel).

I have a GMA X3100 in my Toshiba laptop. When I first got it, it would play a 2D online game at about 2-4fps. So I went to Intel to get the driver, downloaded it, and when I tried to install it, my laptop said "No! Not signed by Toshiba, ain't installing it". I was like crap, went to Toshiba website and there were no drivers yet. So I stuck to using the laptop for casual stuff.

Finally, I went back to Toshiba and there it was in all its glory, a driver! Downloaded it, installed smoothly. Now I can do just about anything with it. That 2D game runs about 150fps now and every sample app in the DirectX sample browser (excluding dx10) runs silky smooth, almost as good as my Geforce 7600. Of course the only comparison I can make is with cards I own,Geforce 5500,7600,8500,and 8800. I have no games that would stress test it. And I can run/program XNA games on it now (which is why I bought it in the first place).

All in all, its not a 8800GTS but its not bad.

Note: I think the only reason the driver has to be signed is because the optical drive is a HD-DVD drive.

Is this article any different than the sudden reviews of all those new $10,000 programs and libraries? I always wonder just who those things are targeted at. For the most part, we are a group of people working with milkshape, fragmotion, or other low cost (or even free) solutions.

Quote:

Original post by Vampyre_Dark Is this article any different than the sudden reviews of all those new $10,000 programs and libraries? I always wonder just who those things are targeted at. For the most part, we are a group of people working with milkshape, fragmotion, or other low cost (or even free) solutions.

Right but many of us also have jobs in the game industry proper, and may be looking for new tools for our company.

Quote:

This article sought to put to rest some misperceptions in the graphics community regarding Intel’s position in the graphics marketplace and the strength of its graphics technology.

It failed.

We all know that Intel graphics will run many games "good enough" for the average Joe consumer. We all know that most casual gamers don't have a fancy graphics card. We all know that if our code is too slow we can use a profiler... and the bit about buggy drivers? They are buggy, and telling us how to report the bugs to Intel doesn't help the immediate situation.

Fact is most DirectX games already do support Intel graphics, they just look prettier on a better card. Surprise, surprise! The question Intel should be answering: how can we make the graphics on an Intel card better than we have been in the past? I'm not impressed that they got decent framerates on a bunch of 3-5 year old games... now if they showed good benchmarks for Crysis and Oblivion I'd have learned something new...

On a scale of 1-10 I give this article a zero for effort and an eleventy-three for stating the obvious.

Wow, so negative.

Obviously a GMA950 won't run Crysis at max settings (what will?), but for less demanding and older games it works quite well.

What about battery life? Considering there are a lot of laptops nowadays, I'm sure the average Joe will appreciate getting more battery life to type up that essay, rather than a noisy fan and heat blowing on your lap just for the ability to game the latest games.

Everything has its time and place. Integrated graphics, included. IMO.

Quote:

Original post by shurcool Wow, so negative. ... Everything has its time and place. Integrated graphics, included. IMO.

Certainly. I just don't think that its "time and place" is an obvious advertisement masquerading as a programming article. It hurts the respectability of this site, since now I have to wonder with each article I read whether or not it's written objectively.

All I'm expecting for is GameDev.net to call an article an article, and an ad an ad.