Well, There Is No i8xx Fix For Ubuntu 10.10

[Craig73]

You obviously did not take the point...

I did not say that just because hardware is old, it should have functionality removed.

I said, that if you are still using such hardware, your use case certainly doesn't include anything which requires its limited gpu accelaretion...


What part of it some of you do not understand?

Well, as it was already stated - those who bought hardware with this chip (series) had (and still have) use cases that it's limited GPU acceleration (2d and 3d) was sufficient for.

Perhaps at the time their use case didn't justify GPU acceleration in the order of magnitude of 3d FPS games or Cuda based simulations, but that does not mean they didn't want nor benefit from GPU acceleration, which is what you stated.

[droidhacker]

You obviously did not take the point...

I did not say that just because hardware is old, it should have functionality removed.

I said, that if you are still using such hardware, your use case certainly doesn't include anything which requires its limited gpu accelaretion...


What part of it some of you do not understand?

And what has been stated back to you more than once now is that you ARE WRONG. If your use EVER required its "limited gpu acceleration", then why should that have changed just because time passes? Fact is that it HASN'T changed. If you used it THEN, then you most likely STILL use it NOW. This isn't an all-or-nothing situation! If it did what it can do then, then it should STILL be able to do exactly the same thing NOW as it did back THEN. And if it was enough for your use THEN, then it is STILL sufficient for doing the SAME THING NOW.

The situation that is being run into NOW though, is that it will NOT be able to do the same thing NOW as it USED to do back THEN.

It is WRONG to take functionality away from a particular piece of hardware for just the reason that time passes. Intel decided to change their driver architecture. That's nice, but they screwed up! They had two reasonable options: they could STOP supporting the old hardware, which means BRANCHING and leaving the old hardware support to the old driver, OR they could CONTINUE supporting the old hardware, which means developing new drivers to support them using the new driver architecture. They tried to do the second option, but weren't smart enough, so they gave up and are now dropping everything into a trash bin.

The CORRECT SOLUTION (given that they lack the intelligence to make the new driver work with the old hardware) is to have BOTH drivers: the old one to support the old hardware -- give it up to the community to continue supporting it if you don't like to, and the new driver for supporting new hardware. You don't just retract everything for the old hardware, because then suddenly the old hardware can no longer do what it has previously been able to do, which means that people DEPENDING on that functionality will no longer be able to use the hardware.

[Craig73]

My frustration is that I have this hardware - and perhaps an fortunate enough to have common enough hardware that it runs sort of OK on.

Perhaps this is a naive question -- but how many people are using this hardware out there? Even if it's only 1% ... if there are there even 10,000 machines still in use such that if each user was willing to pay $1 or $5 that we couldn't get a dedicated developer to fix the outstanding bugs/quirks for blocks of users with common hardware?

Ubuntu is putting considerable effort to work around a driver with bugs - why can't they put in place an infrastructure to identify and request people pool resources to address common issues. Not all of us have coding skills or time to learn driver development - but we have some cash to close off this issue for once and for all

[sturmflut]

Not all of us have coding skills or time to learn driver development - but we have some cash to close off this issue for once and for all

The problem is that is issue won't be solved "once and for all". Development on X.Org and the Intel driver continues, this is an on-going effort. Remember that i8xx chips worked for all those years, if we chip in now to fix the problem who guarantees us that there will not be regressions again in the next release? Intel doesn't seem to spend much time on quality control, otherwise this problem wouldn't even exist. As said I'm already running into problems with the next chipset generation, i915.

Intel should just stop experimenting so much with their Linux drivers. How many times have they delivered new and untested features with their "stable" drivers? How many technologies has this driver seen now? UMS, KMS, EXA, UXA, GEM, ...

Compared to AMD and NVIDIA Intel has very few different chipsets out in the wild, and although they seem to have the largest open source team of those three manufacturers they just can't seem to get it right.

[xeros]

We are talking about a decade old hardware here... The few people who are still using this hardware do not care for gpu accelaration anyway...

Please talk for yourself.
These integrated chipsets are still very popular and are used with both P4 and CoreDuos.
And people expect 3D, too.

[Craig73]

The problem is that is issue won't be solved "once and for all". Development on X.Org and the Intel driver continues, this is an on-going effort.

I guess I just assumed (perhaps naively) that the key issue was getting the quirks programmed into the new architecture of the current driver. It might not continue to get enhancement but the 2d and basic 3d would stay stable enough for a few years (no?)

[sturmflut]

I guess I just assumed (perhaps naively) that the key issue was getting the quirks programmed into the new architecture of the current driver. It might not continue to get enhancement but the 2d and basic 3d would stay stable enough for a few years (no?)

If Intels stops screwing around and starts focusing on stability: Yes. But I don't see that happening anytime soon. They've changed the architecture at least three times in five years already.

They should've just split the driver before changing the architecture so much. I don't care for KMS and the other stuff, but I care for working suspend/resume and video playback. Why not keep a stable and working "legacy" driver for i8xx/i915 and move the fancy new code for more recent chipsets into a new driver?

[deanjo]

Why not keep a stable and working "legacy" driver for i8xx/i915 and move the fancy new code for more recent chipsets into a new driver?

Because that would be logical.

[danboid]

Thats it Intel!

First Poulsbo and now this!

I've got a Dell Mini 10 and my dad has a 2.4Ghz P4 laptop w/ firewire, 5 USB2 ports and upgraded to 2GB RAM etc but that is now crippled (as far as modern Linux is concerned) by its i8xx gfx. That laptop, despite being over five years old now, is still perfectly good for many uses IF its gfx hardware is properly supported ie web browsing and video playback.

While the intel IGPs have never been known for their performance, at least you used to be able to count on them working under Linux. You'd think this was some strapped for cash company we were talking about here!

Seriously not impressed Intel!

[mat69]

I am not impressed either.
The machin of my parents also has an Intel i8XX chip because at that time it was the only chip I knew of with good FOSS Linux support.

Guess I'll opt for an old ATI card. Do you know if powersaving is supported well for R500 on Ubuntu? Because if I buy one it will be passively cooled and running it on max all the time would be bad in that case.