Performance regressions with 256.25 driver?

[brucecadieux]

I personally get better performance with this driver, In all the Unigine benchmarks, glxgears (even though it really isn't a valid test), desktop is smoother, system and desktop responsiveness seems just a tad quicker, overall it is a slight improvement in performance.

No problems, no issues, just runs really well.

Have it installed on two openSuSE desktops and a Toshiba laptop, all are running great.

[wiresquire]

Originally Posted by Sean_W

Phoronix make too much drama out of this. Regression in a testing release? NO WAY!

Testing pre-release software and then claim there is some kind of regression or "disaster" (as in the 2.6.35 kernel tests) is just plain silly without identifying what it is.

I would agree that people should post their benchmarks on games with both drivers, specifically Openarea.

Bingo! It's only a beta release, Phoronix only tested one machine. RTFA. They didn't make any drama out of it at all. If you think they over-reacted, then quote from the article. They just reported what they got.

It's all the nvidia fanbois that make all the drama about Phoronix that makes me sick. "Oh, it's Phoronix, so it's BS". "Phoronix just want attention". Wrong. It's you.

FWIW, if you read the article, running a single benchy, especially Unigine:* would be stupid to try to prove there isn't a regression for you. If you wanted to make sure, you would be running Lightsmark2008 or OpenArena or one of the ones that showed a significant decrease in performance in the phoronix article. AFAICT, Licaon, you are the only one who did so.

That means everyone else is talking out their you know whats. That's what peeves me off...

[Sean_W]

Originally Posted by wiresquire

Bingo! It's only a beta release, Phoronix only tested one machine. RTFA. They didn't make any drama out of it at all. If you think they over-reacted, then quote from the article. They just reported what they got.

It's all the nvidia fanbois that make all the drama about Phoronix that makes me sick. "Oh, it's Phoronix, so it's BS". "Phoronix just want attention". Wrong. It's you.

FWIW, if you read the article, running a single benchy, especially Unigine:* would be stupid to try to prove there isn't a regression for you. If you wanted to make sure, you would be running Lightsmark2008 or OpenArena or one of the ones that showed a significant decrease in performance in the phoronix article. AFAICT, Licaon, you are the only one who did so.

That means everyone else is talking out their you know whats. That's what peeves me off...

Well, Phoronix do make dramas out of it. They test alpha software like it's meaningful.

[Lemmiwinks]

I also don't see a performance regression. Works really smooth for me. Many Phoronix forgot compiz's "--loose binding" option or something like that...

[xianthax]

Originally Posted by Licaon

that's the point: to test file systems at the DEFAULT settings ( provided by the distro ) because a generic user will use the defaults while the specialised ( server, workstation, etc ) user will know and want to optimise the settings as s/he sees fit anyway

Except that he concludes that Distro X is slower than Distro Y based on the default desktop installs using the default file system settings based on performance differences in server tests.

If you revisit his tests and look at only the desktop benchmarks they paint an entirely different picture of the differences between distros.

If he wants to run server benchmarks he should use distros designed for server use with file systems tuned for the particular workload. Otherwise, he's not addressing real world usage patterns and his benchmarks are useless.

Any moderately capable server admin will use the proper file system, file system configuration and kernel configuration for the role of that server. Finding the ideal configuration for various server workloads isn't always easy but is something that Phoronix could be extremely helpful with, if they would setup a reasonable test.

Especially useless are his benchmarks across various kernel versions. He again uses the default kernel/file system configuration options and proceeds to conclude kernel x is slower than kernel y not due to any actual performance differences in the kernel, but rather changes to the default kernel configuration. The resulting information is not useful at all. If he wants to test kernel performance lock to 1 kernel / file system configuration and run the tests so we can actually see if the kernel code paths have improved.

In general he is exceedingly bad at setting up anything resembling a controlled experiment. His tests almost always have multiple uncontrolled variables making the resulting data nearly useless in making any worthwhile conclusion.

In this particular example he tested 1 single system with a mobile GPU, notorious for variations in implementation, and concluded that the entire driver was slower in general. Failed logic.

Phoronix is in a great position to provide some real value in linux performance testing. However, as long as they care more about sensational headlines than they do about setting up controlled tests with real value, they do more harm than good.

[Arup]

Testing on a mobile GPU and passing judgement about the performance of the driver was a bad move indeed, I have been running this driver on few PCs and one laptop and see no such regressions, in fact desktop draw performance and vdpau performance on full screen with mplayer has improved, also CPU spikes are far less when doing graphic intensive activities.

[Licaon]

guys/gals do post test results as well to backup your claims either for slowdowns or speedups or else simple 'impressions' might not be anything else but placebo

get the latest Unigine Sanctuary / Tropics / Heaven or LightsMark and do test them with each driver version on your system at this moment

if you don't bother to test why bother to post?

Thanks

[damentz]

Why are you guys so pissed off at Phoronix? They use the same software and hardware configuration. They install the new beta driver and they experience major drops in FPS.

If I was in nvidia's shoes, that would be really nice to know early too.

[wiresquire]

So does anyone have benchmarks using the newer 256.29 beta that refute the same tests from Phoronix with the 256.25 drivers? Or is the performance regressions still applicable?

[Licaon]

Originally Posted by wiresquire

So does anyone have benchmarks using the newer 256.29 beta that refute the same tests from Phoronix with the 256.25 drivers?

it won't matter what everyone else gets since only Phoronix got the performance issue anyway

Originally Posted by wiresquire

Or is the performance regressions still applicable?

since only Phoronix saw the performance issues (presuming that the fixed regression was actually at fault for the lower Phoronix results) we won't know until they publish some results with the same laptop test system that they used for the initial 256.25 beta testing

my 256.29 results mirror the 256.25 results mentioned earlier

[Sean_W]

Originally Posted by xianthax

Except that he concludes that Distro X is slower than Distro Y based on the default desktop installs using the default file system settings based on performance differences in server tests.

If you revisit his tests and look at only the desktop benchmarks they paint an entirely different picture of the differences between distros.

If he wants to run server benchmarks he should use distros designed for server use with file systems tuned for the particular workload. Otherwise, he's not addressing real world usage patterns and his benchmarks are useless.

Any moderately capable server admin will use the proper file system, file system configuration and kernel configuration for the role of that server. Finding the ideal configuration for various server workloads isn't always easy but is something that Phoronix could be extremely helpful with, if they would setup a reasonable test.

Especially useless are his benchmarks across various kernel versions. He again uses the default kernel/file system configuration options and proceeds to conclude kernel x is slower than kernel y not due to any actual performance differences in the kernel, but rather changes to the default kernel configuration. The resulting information is not useful at all. If he wants to test kernel performance lock to 1 kernel / file system configuration and run the tests so we can actually see if the kernel code paths have improved.

In general he is exceedingly bad at setting up anything resembling a controlled experiment. His tests almost always have multiple uncontrolled variables making the resulting data nearly useless in making any worthwhile conclusion.

In this particular example he tested 1 single system with a mobile GPU, notorious for variations in implementation, and concluded that the entire driver was slower in general. Failed logic.

Phoronix is in a great position to provide some real value in linux performance testing. However, as long as they care more about sensational headlines than they do about setting up controlled tests with real value, they do more harm than good.

Agreed. Just look at the kernel test they did recently. They made a panic doomsday headline and within a week or so, the problem was fixed. This is 'testing' and welcome to the world were testing releases get broken and regress and testing it all is about catching these things, no making headlines about it.

[bistory]

At least, one thing is sure, and I can say that even without Phoronix :
2D is fracking slow !

My netbook with a slower Intel graphic chip runs smoother Compiz and Docky than my notebook (with an Nvidia 8400M GT), it's a pitty...

Ok, a friend said me that he updated recently the BIOS of his Asus notebook and the sluggish 2D disappeared... But Sony doesn't update their BIOS...
Even that, the drivers have a problem I think.