ATI driver features

[barkas]

I was aware about the problems with 8000 series. Mine was quite hot (~60? idle) but it never melted. But if they are encountering the same problems with 200 series I understand why we aren't ready to see nvidia laptop for a while....

Anyway, is there a topic or a website dedicated to xvba and va-api for radeonhd and fglrx ?

Thanks

It is said that nvidia did have problems with heat cycling and faulty bumps that would crack after a few hundreds or thousands of those cycles, which a laptop has more often. I believe that is fixed by now.

There is no xvba/va-api on open source drivers at the moment. But there is a xvba thread somewhere around here that goes into detail.

[Guilo]

Tearing on flash clips doesn't bother me, I nearly never watch those clips.

Powersaving is not really important at the moment.

My advice is - DO NOT BUY an ATI CARD.

Unfortunately I haven't got much choice

[cutterjohn]

Hi all !

My laptop is dead and I will have to buy a new one. But almost all laptops are using ati graphics. I have been using nvidia drivers for a long now and was really happy with what had been done with vdpau and ease of installation. So Ihave a few questions to determine if I can switch on ATI.

Wow! here for a while it seemed like they might be moving to more ATI based notebooks, which is part of the reason why I got one with a 4850 mobility, however not long after they ALL switched back to nVidia GPUs. Even the replacement model for my notebook went back to nVidia apparently as soon as MSI ran out of built stock of this one(GT725).

- I have read that XvBA what finally doing something. But is it already as efficient and easy to use as vdpau? Ismplayer supported. I mostly watch x264 1080p. Is this codec supported ?

- What about video sync and tearing. I had a desktop with ati and experienced a lot of problems of that kind. Is it fixed ?

Dont' know much about video tech specifics(and don't care really), but IME nVidia is MUCH better at video playback than fglrx/catalyst under linux. In windows they seem to be about the same, but I still give the edge to nVidia.

- I am planning to keep my kubuntu with KDE4. Are 3Deffects well supported by the driver ? I remember the time where these effects where hard to get working with the ATI driver.

Nah they seem to work fine now, but a few versions ago I had to disable compiz/beryl otherwise my machine would NOT successfully resume after suspending. Oh yes, they're also vvvvveeeeeerrrrrryyyyyyy ssssslllllloooooowwwww.

- And last question : what about the open-source driver ? Where is the developpement heading to and what can be achieved as of now with this driver ?

From what I understand 2d is supposedly better than fglrx already, but 3D is nowhere near finished. I'm really hoping for the OSS driver as catalyst/fglrx progress is pathetically slow and tends to have annoying regressions frequently. Decent 3D for the OSS driver is apparently about a year off for R600 GPUs and newer.

Thanks a lot for all your answers, I am anxious at the idea of leaving nvidia.

I'd stick with nVidia for a notebook and save the great ATI experimentation for desktops where it's easier/cheaper to swap out GPUs... I wish that I had done that now, but that said I'm mostly living with the 4850 mobility plus Ubuntu 9.10 x86-64 and catalyst 9.11 now, but it's just like I've got this nagging itch that I can't scratch unlike it is with nVidia GPU + driver.

[EDIT]
IMO that heat cycling problem that nVidia had probably really should've been TSMC's fault as they were the manufacturer contracted to produce the chips utilizing their own processes.
[/EDIT]

[EDIT2]
Seeing the Intel comments, if you want to do any serious 3D/opengl you've GOT to go either ATI or nVidia. Intel IGPs are too anemic to do much useful with, but they should be OK for video/2D.
[/EDIT2]

[Hoodlum]

[EDIT]
IMO that heat cycling problem that nVidia had probably really should've been TSMC's fault as they were the manufacturer contracted to produce the chips utilizing their own processes.
[/EDIT]

It's a similar problem intel had with the P4. Trying to get a massive die and hugely complex design (about 4-5x the ati one iirc) manufactuered without leakage / other issues. Wouldn't really have mattered too much who manufactered it the yields would be extremely low because of the design.

[Guilo]

Has the opensource driver a GUI like amdccle ?

[bridgman]

I don't think TSMC does any of the packaging. They just make wafers AFAIK and other companies handle the rest of the mfg chain.

Has the opensource driver a GUI like amdccle ?

There's no driver-specific GUI; the open source drivers tend to expose standard APIs to desktop tools instead. There is driconf for 3D settings.

[cutterjohn]

It's a similar problem intel had with the P4. Trying to get a massive die and hugely complex design (about 4-5x the ati one iirc) manufactuered without leakage / other issues. Wouldn't really have mattered too much who manufactered it the yields would be extremely low because of the design.

Yield has ABSOLUTELY NOTHING to do with the heat cycling problem, nor should the size if TSMC was up to the task. It would seem more likely to me that TSMC was using a new lower lead solder IIRC and quite clearly had not gotten it quite right.

Intel: Missed their heat cycling problem entirely unless you're referring to yield.

In any event yield if simply a matter of fine tuning the manufacturing process which can be exacerbated by simultaneously moving to a new "size" process or other changes in the base manufacturing process.

[Kano]

Would really like to know if Intel has got 32 nm already working...

[Hoodlum]

Yield has ABSOLUTELY NOTHING to do with the heat cycling problem

I wasn't refering to that. Yield has everything to do with die size.

nor should the size if TSMC was up to the task.

Larger die size = greater chance of failure, always. Hence lower yield. It is inherant in any manufactuering.

It would seem more likely to me that TSMC was using a new lower lead solder IIRC and quite clearly had not gotten it quite right.

No, it is the high lead solder that is the issue "Given that Nvidia claims to be transitioning from high-lead to eutectic bumps, it is only a matter of time until the high-lead inventory is depleted, and the Macbooks are safe to buy."
This was in relation to the 9xxx series (aka 250) but it still applies to the other 200s. Here are pictures through an electron microscope http://www.theinquirer.net/inquirer/...-bump-material

Intel: Missed their heat cycling problem entirely unless you're referring to yield.

They had a bloated die size issue which caused low yields and high temperatures. I believe we're talking about 2 completely different hardware failures.

In any event yield if simply a matter of fine tuning the manufacturing process which can be exacerbated by simultaneously moving to a new "size" process or other changes in the base manufacturing process.

This was another issue intel had with the Prescott chips.


Nvidia quotes:
"Hara talked about how the original problem announced by Nvidia on July 2 was rectified. "A more robust underfill would have taken the stress off the bumps and kept that (original problem) from happening. What we did was, we just simply went to a more robust underfill. Stopped using that (previous) underfill, kept using high-lead bumps, but we changed the underfill. And now we don't see the problem."

"Intel has shipped hundreds of millions of chipsets that use the same material-set combo. We're using virtually the same materials that Intel uses in its chipsets," Hara said.

Hara also said Taiwan Semiconductor Manufacturing Company (TSMC) ships a "staggering" number of chips to many companies worldwide with high-lead bumps. TSMC is the world's largest contract chip manufacturer and makes chips for Nvidia, Advanced Micro Devices, and many other companies.".

Underfill details:

"Next up, we have the long shot scenario, that Nvidia packaging engineers, if they actually have them rather than outsourcing everything, simply missed an entire branch of science. They all took a class on semiconductor engineering, but they all slept through that day. And didn't read the book.

One last thing to toss into the mix, cost. The PI layer is expensive, it adds about $50 to the cost of a wafer. Wafers from TSMC on a high end process cost about $3,000 to $5,000 depending on a lot of details. Adding the PI layer increases the cost of silicon by a noticeable amount, and adds to the defect rate."

http://www.theinquirer.net/inquirer/news/1050052/nvidia-chips-underfill

Conclusion - It was an underfill issue. Nvidia cheaped out and it caused failures. This is what happens when you have to compete on price when you get significantly less dies per wafer (more cost). You have to make profit somewhere.

Don't get me wrong. I don't care for either company but faulty hardware is faulty hardware.

[alazyworkaholic]

System76 has nVidia laptops

System76.com has a range of good-looking laptops. If you don't care about games & just want video to work the X4500MHD from intel ought to do, & if you need something more there are a couple upper-end laptops with nVidia cards.