Intel X3100 - OpenGL support?

[sgRevan]

Hello world,

I'm going to buy a laptop from Novatux (a little french reseller) with a X3100 IGP and I would like to code a bit with pyOpenGL.

My issue is that I only saw OpenGL 1.5 to be supported. I wonder if newer Intel drivers implement OpenGL 2.0?

[LydianKnight]

Hello world,

I'm going to buy a laptop from Novatux (a little french reseller) with a X3100 IGP and I would like to code a bit with pyOpenGL.

My issue is that I only saw OpenGL 1.5 to be supported. I wonder if newer Intel drivers implement OpenGL 2.0?

If I have to follow in an exact way the specs available at WikiPedia (http://en.wikipedia.org/wiki/Intel_GMA), I'd say... NO, but I'm not sure, maybe there's a way for the driver to be able to let you use some of the OpenGL 2.0 featureset, but if I had to rely on the specs information... I wouldn't count on it :(

Julio

[crumja]

There's a glsl backend to 965. Mesa 7.1 should contain the bits to support opengl 2.0 on x3100+ hardware. Check out http://www.x.org/wiki/Events/XDS2007/Notes

[TechMage89]

I think the OpenGL 2 thing is just a driver issue. I'm fairly sure the hardware can do everything OpenGL 2 needs, but if it's not in the driver yet, it ought to be fairly soon.

[crumja]

Just waiting for Mesa 7.1. Fedora already bundles an rc version and the x3100 shows opengl 2.0 support in glxinfo.

[BlackStar]

Just to let you know that I installed Ubuntu Hardy to my business laptop (with X3100) and the result is... not pretty.

First impressions: Compiz works out of the box!

But that's about the only thing going for X3100.

Issues:
1. Graphics corruption in almost all Compiz effects. Windows leave trails behind, the screen stops refreshing (unless you "minimize all and show desktop"), windows stop responding etc etc.

2. Performance is horribly bad. So bad, that the card cannot display simple Compiz effects like "fold", "kick" or "glide1/2". The only thing that works, barely, is fade-in/fade-out. Moving large windows around is also rather slow, e.g. Firefox.

3. Driver only supports OpenGL 1.5.

Compared to my age old Ati 9700 Mobile, the X3100 is miles behind in drivers *and* performance, and that's saying a lot.

I'd stay as far away from this chip as possible, if I were you.

[Huenengrab]

You should work out your driver-issues then. I installed Linux on a 7 laptops with the Intel X3100 IGP and the configurations all differed, but I never had any of your problems.

The only thing I can think of was a font-corruption with the Fedora 9-LiveCD and that was fixed by temporarily disabling EXA in the xorg.conf and applying updates after the installation.

All of your problems remind me of the problems I had when I first got my GeForce 7900 GTX. Should I have reportet the card being useless too, even if it works fine, now that I know what I am doing?

[BlackStar]

So you've had OpenGL 2.0+ support, good performance and stable compiz with the X3100 chipset on 6/7 laptops? I don't know if I believe that, judging from my (admittedly limited) experience with this chip.

You should work out your driver-issues then.
That's what I will do, of course, but I was rather disappointed by the experience. "Intel chips work out of the box", or so they said - which is clearly an overstatement.

In fact, my ati 9700 Mobile did work out of the box on Hardy. So did my 9600 Pro. My father's 7600GS needed nvidia's binary blob, but was spotless afterwards. My X1950Pro was disappointing at first, but it's solid now.

And all of these are about many, many times faster than the X3100.

So, yes, I was disappointed and this shows in my post. I guess the moral of the story is to avoid the latest and greatest hardware - 12 months is a good baseline to allow devs time to work out the kinks.

Off now to find a fix for the atrocious performance. Hopefully there's some simple tweak to help here - any ideas?

[Svartalf]

So you've had OpenGL 2.0+ support, good performance and stable compiz with the X3100 chipset on 6/7 laptops? I don't know if I believe that, judging from my (admittedly limited) experience with this chip.


Many, many others seem to have had decent results with it. Keep in mind that many laptop and motherboard vendors do screwy things with the chips they implement- some of which would cause problems. I know for a fact that the driver works FINE on my eeePC, but the same GMA on my Dell Optiplex 745 at work is unstable with 3D operation and I had to swap in a 6200LE just to work with things on the machine. BIOS settings and timings can play a BIG part in things working or not.

I've seen things that way with pretty much all the Intel parts and any IGPs from ATI/AMD and NVidia (Just look at the X200M story to see proof of this...gad... It's largely NOT AMD's fault it's a mess, just FYI...).


That's what I will do, of course, but I was rather disappointed by the experience. "Intel chips work out of the box", or so they said - which is clearly an overstatement.


Actually, it's less of one than you're making it. YOU have found an issue with it. Having used computers for literally decades and dealt with the same sorts of garbage under Windows, I can safely say that I try to point fingers like that out of the gate. I try to see if it's a bad driver situation with the version of the OS I'm using, a bad BIOS or bad Mobo, etc.

Since you didn't share what brand machine you're trying to claim this on, I can't tell you what might be the beginning of the problem, but...


In fact, my ati 9700 Mobile did work out of the box on Hardy. So did my 9600 Pro. My father's 7600GS needed nvidia's binary blob, but was spotless afterwards. My X1950Pro was disappointing at first, but it's solid now.

And all of these are about many, many times faster than the X3100.


Yep. And for the notable exception of the 9700 Mobility, those parts are going to be in a completely different class of part- as they're all discrete devices. Discrete devices tend to paste IGP's out of the gate because they've got dedicated memory access and they've typically got higher clocks, more vertex (transform/lighting) and fragment (texturing) pipeline pieces so they can do more per clock. (And, comparing an X1950Pro to a GMA X3100?? C'mon...let's get realistic here. ;) ) I'm also not a bit surprised about the 9700 edging out the GMA right at the moment. If you're using the fglrx drivers for it, you're going to see an edge there as they've got a more optimized rendering path codeset compared to the current batch of Intel drivers- I could even see the same story with the R300 FOSS drivers since Intel's stuff is still a bit in it's infancy. The in-progress Gallium3D drivers may change that situation, but for now, heh...


So, yes, I was disappointed and this shows in my post. I guess the moral of the story is to avoid the latest and greatest hardware - 12 months is a good baseline to allow devs time to work out the kinks.


The other moral is to not presume, just because 3D "works" out of the box, that your system configuration will allow it to work or that it'll perform unless you're buying something like an Intel motherboard, and so forth.

[BlackStar]

That's what I will do, of course, but I was rather disappointed by the experience. "Intel chips work out of the box", or so they said - which is clearly an overstatement.
Actually, it's less of one than you're making it. YOU have found an issue with it. Having used computers for literally decades and dealt with the same sorts of garbage under Windows, I can safely say that I try to point fingers like that out of the gate. I try to see if it's a bad driver situation with the version of the OS I'm using, a bad BIOS or bad Mobo, etc.
You are missing the point. Intel is thought to be the leader in OSS video drivers as far as features, stability and user experience is concerned. As such, I was looking forward to (finally!) trying a chip that wouldn't need countless hours of xorg.conf tweaking just to work - but the expectation didn't play out.

I acknowledge that I'm just one person, testing a single hardware configuration (a Fujitsu-Siemens Esprimo Mobile to be exact). I'm not saying that everyone will encounter the same problems. On the other hand, if someone (like me) is thinking to go with an intel solution for driver stability and features, he might be wise to temper his expectations - there's simply no video card in the market today that doesn't suffer from quirky configuration or other random issues (e.g. OpenGL context construction on fglrx or 2d performance on nvidia's G80+).

I'm not saying this only occurs on Linux either. The very same X3100 chip would blue screen windows, until the July drivers fixed what looked like an invalid memory allocation (this certainly didn't help my opinion of this card).

Finally, I'm obviously not comparing ati's X1950 performance to intel's X3100 - but the latter should be able to take on ati's 5-year-old 9700M at least! Hopefully, things will change as X3100 drivers mature - they simply aren't there yet.

[crumja]

The x3100 may be comparable in performance to a radeon 7500 or geforce 2 gts once the drivers mature. However, it is capable of handling more features. Comparing it to a 9700 is not valid. Integrated graphics are designed for low cost, low power usage, and decent features for output and video acceleration. Gamers are not the target audience.

[BlackStar]

The x3100 may be comparable in performance to a radeon 7500 or geforce 2 gts once the drivers mature. However, it is capable of handling more features. Comparing it to a 9700 is not valid. Integrated graphics are designed for low cost, low power usage, and decent features for output and video acceleration. Gamers are not the target audience.

Who is talking about games? I am talking about basic compiz functions (e.g. moving windows around, opening windows), features and stability - not games.

Both the 9700M and the X3100 are perfectly stable in my experience. However, the former wins on all other regards, features included, despite being 5 years old. The X3100 may be a DX10 compatible chip, but the drivers only expose GL1.5 - not good at all.

[Redeeman]

you realize that the 9700 drivers doesent expose more, right? in fact, the hardware doesent support opengl 2.0..

and btw, X3100 has EXCELLENT compiz support, in facts it works a whole lot better with aiglx than anything fglrx and nvidia can do.

[BlackStar]

you realize that the 9700 drivers doesent expose more, right? in fact, the hardware doesent support opengl 2.0..
You are kidding me, right? GL2.0 is supported just fine on R300+ and the drivers *do* expose this functionality. I know, I have two R300 parts running Linux :rolleyes:

and btw, X3100 has EXCELLENT compiz support, in facts it works a whole lot better with aiglx than anything fglrx and nvidia can do.
My experience is totally different. I currently run compiz on two R300, one R500 and a G70 part - DRI2 issues aside, all of these work very nicely (Ubuntu Gutsy and Hardy, x86 and x86_64).

My x3100 on the other hand exhibits artifacts and horrible compiz performance on Hardy, when using any effect other than the "fade" and "genie" transitions. Enabling the "blur" plugin (mip-maps) kills compiz. The water effect turns all windows into solid colored rectangles. The zoom effect draws at <1 fps. This is as far from EXCELLENT as it gets!

I don't doubt this are driver issues and will be resolved in time. In fact, it would be great if anyone has any pointers to share. Right now, I'm using the laptop without compiz (it works fine), but it's a shame to leave things like that.

[crumja]

I haven't seen these compiz bugs on my x3100 setup.

[Svartalf]

You are missing the point. Intel is thought to be the leader in OSS video drivers as far as features, stability and user experience is concerned. As such, I was looking forward to (finally!) trying a chip that wouldn't need countless hours of xorg.conf tweaking just to work - but the expectation didn't play out.


Had you read what I'd said to you:


Many, many others seem to have had decent results with it. Keep in mind that many laptop and motherboard vendors do screwy things with the chips they implement- some of which would cause problems. I know for a fact that the driver works FINE on my eeePC, but the same GMA on my Dell Optiplex 745 at work is unstable with 3D operation and I had to swap in a 6200LE just to work with things on the machine. BIOS settings and timings can play a BIG part in things working or not.


You'd have already gotten the idea that I didn't miss your point. Moreover, you indicate that this device has issues with WINDOWS- which indicates something a bit more problematic than a "driver" issue. In your case, your hardware vendor did something that made things complicated.

I quote your comments here:


I'm not saying this only occurs on Linux either. The very same X3100 chip would blue screen windows, until the July drivers fixed what looked like an invalid memory allocation (this certainly didn't help my opinion of this card).


This has more to do with the way Fujitsu deployed the chipset and the BIOS settings (Keep solidly in mind that other people don't have your problems, including under Linux...) than the drivers or Intel itself- start blaming people in order of the evidence (and not just yours...), in this case, very, very few people have issues with the X3100. That should point to a motherboard vendor screwing something up.

As for performance... My GMA950 on my eeePC works well as does my X3500 on my HTPC. The Dell Optiplex 745 GMA950 DOES NOT work stably and when it does run, doesn't do 3D as well as it ought to. I strongly suspect you've got a problem with your laptop that puts the GMA in a suboptimal configuration.

[Kano]

If you have got stability problems check your ram or replace it for testing.

[Redeeman]

#14:
so i must have missed the commit providing GLSL support? i guess i did..

andwith the issues you describe, its blatantly obvious that you have some sort of giant issue, which people normally dont have..

my experience with X3100 spans 5 laptops, all working excellently, and then my brother has a non-X 3100(htpc), which also runs excellently..

I also know of people running GMA950 with compiz perfectly fine.

by now you should realize that either there is something extremely wrong with your hardware, or your install.

[BlackStar]

I haven't seen these compiz bugs on my x3100 setup.
This is a good sign. Which distro? Which xorg version?

Many, many others seem to have had decent results with it. Keep in mind that many laptop and motherboard vendors do screwy things with the chips they implement- some of which would cause problems. I know for a fact that the driver works FINE on my eeePC, but the same GMA on my Dell Optiplex 745 at work is unstable with 3D operation and I had to swap in a 6200LE just to work with things on the machine. BIOS settings and timings can play a BIG part in things working or not.

You'd have already gotten the idea that I didn't miss your point. Moreover, you indicate that this device has issues with WINDOWS- which indicates something a bit more problematic than a "driver" issue. In your case, your hardware vendor did something that made things complicated.
The windows crashes were 99% driver-related. They started happening after an update to the intel video drivers. The stopped happening with another video driver update. (Note that these drivers came from intel, not fujitsu.)

I'm not saying that the BIOS or vendor implementation can't break things - I've worked around my share of ACPI bugs. However, there's no indication of BIOS problems here (standby/hibernate work perfectly on both platforms) - given the aforementioned experience on windows, it's only natural to assume the drive.rs are at fault

If you have got stability problems check your ram or replace it for testing.
Thanks for the suggestion, but I haven't had any stability problems under linux (or windows after the driver update) :D Besides, I've already replaced my RAM (moved from 2x1GB to 2x2GB - memtest succeeds in both cases).

so i must have missed the commit providing GLSL support? i guess i did..
Just checked the mobility 9700: the vendor/driver strings return "ATI Technologies Inc. ATI MOBILITY RADEON 9600-9700 Series 2.1.7412 Release". A quick test with OpenTK shows that 1288 out of 1528 OpenGL functions return non-null entry points.

Maybe the open drivers haven't caught up yet, but soon enough they will. The point is, R300 can and does support GL2.0+.


Anyway, thank you, your replies have helped me create a clearer picture regarding intel video hardware: it is solid (unless you fall into some specific hardware mis-configuration, which could happen anyway). Stability is great, 2d and video works fine. 3d is being worked upon (GL2.0 support is coming and let's hope GL3 support won't be too far off). If integrated graphics cover your needs, it's a great choice.

Now on to find some solution to the issues I'm seeing...

[crumja]

I'm running hardy 64 bit on a thinkpad t61. Xorg = 7.3

Are you using fglrx to get opengl 2.0 support on r300? I was under the impression that the OSS driver doesn't have the bits for it included yet.

I would like to add that the intel driver is on the forefront of adding new features. DRI2, TTM/GEM, xvmc, aiglx support, accelerated EXA, and new product support all arrive faster than other vendors' drivers.

[BlackStar]

I'm runing hardy 64 bit on a thinkpad t61. Xorg = 7.3
Thanks, that's the exact same configuration as here.

Are you using fglrx to get opengl 2.0 support on r300? I was under the impression that the OSS driver doesn't have the bits for it included yet.
Yes, using fglrx for now. I think you are right regarding GL2 in the OSS driver, but I'm too lazy to uninstall fglrx to check.

The only reason for using fglrx (on this laptop) was that I knew off-hand how to setup a vertical big desktop - I just didn't have the time at that point to learn how to configure the OSS driver. Btw, the intel chip was a breeze to configure - just typed the mode ("1440x900+1280x1024") and xorg picked it up at once. Great!

Anyway, for now I'm content using the x3100 for 2d work and mesa indirect for 3d development on the go. Back home, I test on a R500 and a G70 - with the target hardware being a quadro 3-thousand-something. A strange combination, but you get good compatibility (and become familiar with all shorts of driver quirks...)

Once the x3100 gets stable GL2 support now, I'll be happy. :)