Nouveau Benchmarks On Ubuntu 12.10

[phoronix]

Phoronix: Nouveau Benchmarks On Ubuntu 12.10

For those curious about the open-source Nouveau graphics driver performance on Ubuntu 12.10, here's a few benchmarks from a recent test release of the Quantal Quetzal with the reverse-engineered NVIDIA driver code...

http://www.phoronix.com/vr.php?view=MTE4NzA

[TobiSGD]

For some of the Linux OpenGL benchmarks, the Nouveau performance is down on Ubuntu 12.10 over Ubuntu 12.04 LTS, but this may come down to Unity/Compiz being slower in Ubuntu 12.10.

And this sentence is what makes the whole article totally useless, since no one can know if this is a regression in the driver or a problem with Unity/compiz without actually testing that with a different WM.
Another Phoronix article that only exists to have the buzzwords "Benchmark" and "Ubuntu 12.10" in the title to get page hits from Ubuntards, without actually delivering any usable information in the article.

[willmore]

There are three major variants of the 9600GSO. One is a 96 core part with a 192 bit memory buss. One is a 48 core part with a 128 bit memory buss. I think there's a 32 core one with a 256 bit memory buss, even. I own the first two variants and I think you did a review of one that was the latter of the three.

Any idea what's really under the hood?

[Temar]

And this sentence is what makes the whole article totally useless, since no one can know if this is a regression in the driver or a problem with Unity/compiz.

What difference would that make for the normal user? None!

The developers can easily do their own benchmarks if they care to fix it? They have to do it anyway, if they want to find the commit which caused the regression.

[uid313]

It is too bad that Wayland does not work on Nouveau.
Some bugs in Nouveau.

[gens]

i tried the almost latest nouveau/mesa on my 8800GT (same G92 chip) and it works fairly well, almost as good as the propreitary (taking into account that the memory cant be reclocked nearly as far as it is in the propreitary due to some memory scheduler missing or something)
il start using it when it starts supporting opengl 3.3, and that might take a while due to geometry shaders

PS it dosent render a few things in xonotic, thus the benchmark is unbiased (unless its just on my custom setup)

[TobiSGD]

What difference would that make for the normal user? None!

The developers can easily do their own benchmarks if they care to fix it? They have to do it anyway, if they want to find the commit which caused the regression.

The developers are not Michael's target audience, they know how fast their drivers are. At least their should.
So what use has a benchmark when the used testbed admittedly may or may not be the cause of some random slowdowns?
And what use has it to write an article about such a benchmark?

The answer to the first question: None, it is totally pointless.
The answer to the second question: None, except you use some buzzwords in the title to generate page hits to get money from an article that delivers exactly this amount of real information: Zero.

When I subscribed to the RSS feed I did this because I thought this is a professional news site. Looking at the ever declining quality of the articles, the joke that is called forum moderation and the sheer amount of trolls in this forum I realize that there is only one thing in this site/forum that is professional: the ability to generate page hits without delivering any information.
Unsubscribed and out of here, searching for a site/forum that is really professional.

[schmidtbag]

The developers are not Michael's target audience, they know how fast their drivers are. At least their should.
So what use has a benchmark when the used testbed admittedly may or may not be the cause of some random slowdowns?
And what use has it to write an article about such a benchmark?

The answer to the first question: None, it is totally pointless.
The answer to the second question: None, except you use some buzzwords in the title to generate page hits to get money from an article that delivers exactly this amount of real information: Zero.

When I subscribed to the RSS feed I did this because I thought this is a professional news site. Looking at the ever declining quality of the articles, the joke that is called forum moderation and the sheer amount of trolls in this forum I realize that there is only one thing in this site/forum that is professional: the ability to generate page hits without delivering any information.
Unsubscribed and out of here, searching for a site/forum that is really professional.

I rarely call out people by using a direct insult, but you are just simply an idiot. I don't see how you don't get the point of this article, as it is pretty obvious. What other info are you looking for in a software related benchmark? Is, to you, benchmarking about seeing what software or hardware is the best? If so, since when is that so practical? If you don't have the hardware and don't intend to, it is completely irrelevant. If you do have the hardware, well you should already know how good it is. The point of these benchmarks is to compare the progression/regression of drivers, desktop environments, kernels, or any other kind of software. While this can be helpful to devs who might not have tried something on a different DE, kernel, or hardware platform, it's also helpful to the users who may want to make a switch or upgrade to something.

As for professionalism, why do you even care? If you get the results you want, does it matter how it's worded? Well, I guess considering your nonsensical concept of what a benchmark is supposed to be, I'm sure you'll be just as irrational about these questions too. As long as an article gets to the point, shows and analyzes all relevant data, and is generally unbiased then that's all that matters. While some articles on phoronix may seem sloppy or offensive in some way, they do what they're supposed to.

[gens]

And what use has it to write an article about such a benchmark?
The answer to the question: None, except you use some buzzwords in the title to generate page hits to get money from an article that delivers exactly this amount of real information: Zero.

i play xonotic on a G92, would like to play on nouveau
its nice to know how far along is it without recompiling git mesa to see

its almost as good to play as the blob, at least in my min fps

[smitty3268]

What difference would that make for the normal user? None!

The developers can easily do their own benchmarks if they care to fix it? They have to do it anyway, if they want to find the commit which caused the regression.

It does make a difference for anyone who doesn't run Ubuntu. We don't know whether or not the same slowdown will be present for us or not.

That said, i don't think it's necessarily bad that Michael published this info. It just would have been nice if he'd gone the extra mile to make it more useful for everyone.