PhysX SDK Support Comes Back To Linux

[mirv]

Nvidia doesn't have a cpu last time I checked.

[deanjo]

Nvidia doesn't have a cpu last time I checked.

Well you would be wrong on that comment in many ways. They even have an x86 cpu.

http://www.nvidia.com/page/uli_m6117c.html

But your right, nVidia doesn't have a modern x86 cpu, so why the hell would they bother spending time and money on developing for SSE support. None of their products use it.

NVCC does the exact same thing that Overdrive does in regards to the processor. It makes the same calls but yet you can't even install Overdrive on a nvidia chipset based board despite it having the same capabilities and making the same calls. You can hack Overdrive once you remove the artificial restrictions in place to work on a nvidia chipset with a AMD processor.

[Joe Sixpack]

Lol do you live under a rock or something. Nvidia is the single most used GPU according to Phoronix Graphics Survey 07'-09' and most likely again this year.

Smolts hardware database shows pretty much the same results.

I'm late, but I read this a few weeks ago:

http://www.engadget.com/2010/07/30/ati-overtakes-nvidia-in-discrete-gpu-shipments/

[deanjo]

I'm late, but I read this a few weeks ago:

http://www.engadget.com/2010/07/30/a...gpu-shipments/

Shipments !=units currently in use

[energyman]

deanjo, you can stop with your fanboi bullshit. Nobody renamed old parts as Nvidia did. Nobody tried to rip off customers like they did.

When did AMD or Intel release drivers that made their hardware stop working if an nvidia part was used in the system?
When did AMD or Intel release the same cards with 4 different names?
When did AMD or Intel sold knowingly defective parts and blamed everybody else?

Oh - and the link you posted defending nvidia's choice of x87 instructions - those ghuys ignore the fact that x87 is DEAD for x86-64 processors.

But go on, defend nvidia.

[deanjo]

deanjo, you can stop with your fanboi bullshit. Nobody renamed old parts as Nvidia did. Nobody tried to rip off customers like they did.

LMFAO, ya uhhuh look at the AMD series between the 3000-4000-5000. They rename old parts to new series all the time especially on the low end.

When did AMD or Intel release drivers that made their hardware stop working if an nvidia part was used in the system?

The hardware still works. It just doesn't work in certian configs. No different then AMD chipset motherboards not allowing SLi and vice versa.

When did AMD or Intel release the same cards with 4 different names?

LMAO look at intels server chipsets. Same thing. AMD has rebranded their GPU's many times as well. Just look at any device inf file on their cards and look up their corresponding family and their advertised "series"

When did AMD or Intel sold knowingly defective parts and blamed everybody else?

Lets see, how about Granite Bay or AMD 750 chipset where AGP defects were supposed to be worked around in the manufacturers bios. Some did, others didn't.

Oh - and the link you posted defending nvidia's choice of x87 instructions - those ghuys ignore the fact that x87 is DEAD for x86-64 processors.

Well SSE has never even been incorporated on GPU's. Your point?

[blackshard]

This is pretty Off Topic, but so funny to read...

LMFAO, ya uhhuh look at the AMD series between the 3000-4000-5000. They rename old parts to new series all the time especially on the low end.

You kidding? They are different project with a shared base, but they are certainly different chips with progressively added features. There's just a new memory controller for GDDR5 memories (hd3000->hd4000), but hey, who cares? There's just support for dx11 features (hd4000->hd5000), but hey, who cares?

The hardware still works. It just doesn't work in certian configs. No different then AMD chipset motherboards not allowing SLi and vice versa.

Senseless response.

LMAO look at intels server chipsets. Same thing. AMD has rebranded their GPU's many times as well. Just look at any device inf file on their cards and look up their corresponding family and their advertised "series"

In your dreams maybe... Surely not in real word.
However I think you don't understand the meaning of rebranding: rebranding is the activity to take a chip and create a brand, then take the same chip and create a different brand, and so on...

Lets see, how about Granite Bay or AMD 750 chipset where AGP defects were supposed to be worked around in the manufacturers bios. Some did, others didn't.

I see no point here too. An errata is supposed to happen, a deliberate issue with construction material is not.

Well SSE has never even been incorporated on GPU's. Your point?

Senseless response again. Sincerely, I don't understand what's the meaning of this response...

[Joe Sixpack]

Shipments !=units currently in use

So... let me get this straight:

Nvidia's GPU card shipments drop 10%
Nvidia's integrated graphics solutions drop 10%
Apple drops Nvidia from Mac Pro's in favor of ATi.

None of this matters, but useless hardware polls from websites counts as valid support of your argument?

Interesting.

[Joe Sixpack]

Senseless response again. Sincerely, I don't understand what's the meaning of this response...

*backslap*

Don't you ever, ever, EVER argue with an Nvidia fanboy. They're the biggest waste of time on the internet. They can list 1,000 reasons why ATi sucks without actually using the product or even reading the reviews.

(p.s. The biggest re-branding ripoff of all time has to be the 9800GT)

[energyman]

deanjo, you are posting bullshit, as blackshard has shown.

There is a difference between an ERRATA and SEVERAL GENERATIONS OF CHIPSETS DYING.
There was no rebranding by AMD. The 3000 series as different from the 2000 series as the NVidia 7000 is different from the 6000. Same for 3000>4000 and 4000>5000. No rebranding.

But nvidia? 8XXX>9YYY rebranding 9YYYY> GTS2XY rebranding. And now 2XY>3ZX rebranding - to rip off customers.

There were perfectly fine systems incorporating an AMD product with an added Nvidia card to have physx.
What did Nvidia? Driver update that shut those cards down. People paid money, people got shafted. That is nothing but fraud.

And the last point: yes SSE is not on GPUs. But that is not the problem. The problem is that you have to jump to douzends of loops to not use SSE in binaries compiled for modern CPUs (CPUs not GPUs. You get it? ). And Nvidia did. Why? So nobody would realize how much Physx on GPUs sucks.