NVIDIA 304.37 Linux Driver Brings 41 Official Changes

[pingufunkybeat]

and for all of Intel's 'heavy lifting' ~ their GPUs still don't touch Nvidia, nor does the perfomance or features match Nvidia.

That's a hardware issue, their drivers are top-notch!

[johnc]

That's a hardware issue, their drivers are top-notch!

My understanding from the last Phoronix benchmark is that the open Intel Linux drivers got curb-stomped by the closed Windows drivers?

[ninez]

That's a hardware issue, their drivers are top-notch!

it's not just a hardware issue, Intel's drivers do NOT match Nvidia's and if you don't believe me ~ then go do a comparison of what they implement (as far as standards, like OpenGL) vs. Nvidia...

Intel's drivers are 'decent' NOT 'top-notch'.

[ninez]

My understanding from the last Phoronix benchmark is that the open Intel Linux drivers got curb-stomped by the closed Windows drivers?

that is my understanding too. While Nvidia's drivers are fairly similar on both platforms.

[pingufunkybeat]

It's my understanding that things have changed:

http://openbenchmarking.org/prospect...fc89be48b2e792

Also look at this thread: http://phoronix.com/forums/showthrea...nux-3-6-Kernel

[RealNC]

There's also the issue of trying to force manufacturers to disclose their hardware specs and force open source down their throats. If they don't want to, then you've got problems. And the source of those problems is you, not the manufacturer.

[ninez]

It's my understanding that things have changed:

http://openbenchmarking.org/prospect...fc89be48b2e792

Those are just Intel benchmarks, not Intel compared to Nvidia. Sure they have improved - that doesn't mean they are more performant and better than Nvidia, and doesn't change the fact that Nvidia produces drivers that support all of the latest GL specs, while Intel doesn't (and isn't even close to - none of the OSS drivers are).

Also look at this thread: http://phoronix.com/forums/showthrea...nux-3-6-Kernel

That thread clearly cites that they have improved and in *some cases* are catching up with Windows.

[pingufunkybeat]

Those are just Intel benchmarks, not Intel compared to Nvidia.

You were talking about Intel Linux vs. Intel Windows. If these are similar, then it is a hardware issue, and not the driver. Like I claimed.

and doesn't change the fact that Nvidia produces drivers that support all of the latest GL specs, while Intel doesn't (and isn't even close to - none of the OSS drivers are).

Yes, Mesa is lagging on some OpenGL stuff. OpenGL 4 is some ways off. But 3.3 is just around the corner. For me, a driver with VAAPI, OpenGL 3.3, excellent powersaving, blazingly fast 2d and perfect integration is a top-notch driver.

Nvidia driver is fast, but it will fuck up your system if you upgrade the kernel, doesn't support most mobile hardware currently being sold, and is not available out-of-the-box on pretty much anything, requiring jumping through hoops to get it working. Pick what you prefer.

That thread clearly cites that they have improved and in *some cases* are catching up with Windows.

That thread clearly says that the benchmarks are running 6 months-old drivers, and that huge improvements have landed in the meantime, which had not been benchmarked there.

[pingufunkybeat]

There's also the issue of trying to force manufacturers to disclose their hardware specs and force open source down their throats. If they don't want to, then you've got problems. And the source of those problems is you, not the manufacturer.

Nobody is shoving anything down your throat, you are free to use whatever you want -- including a blob.

It is you who are shoving closed-source down our throats, and into the kernel. Just open up the documentation, then OSS proponents can ALSO have a choice.

For you, forcing closed source blobs on everyone is great, but offering a CHOICE between an open driver and a closed one is "forcing open source down their throats". If you don't want to run a FLOSS system, there are many operating systems for you to choose from. Why are you trying to dictate Linux developers how to license their shit? You have BSD, you can shove any old shit in there, go use that.

[ninez]

You were talking about Intel Linux vs. Intel Windows. If these are similar, then it is a hardware issue, and not the driver. Like I claimed.

and the Linux driver serverly lacks behind what the windows driver has;http://www.phoronix.com/scan.php?pag...tem&px=MTA5NzM

Yes, Mesa is lagging on some OpenGL stuff. OpenGL 4 is some ways off. But 3.3 is just around the corner. For me, a driver with VAAPI, OpenGL 3.3, excellent powersaving, blazingly fast 2d and perfect integration is a top-notch driver.

top-notch would mean to me, being equal or surpasses all other drivers - this is not the case with Intel's linux drivers. and as far as VAAPI is concerned, i have been using VDPAU for a very long time, yet VAAPI is something you currently still can't use, so claiming your driver is 'top-notch' is off the mark.

Nvidia driver is fast, but it will fuck up your system if you upgrade the kernel, doesn't support most mobile hardware currently being sold, and is not available out-of-the-box on pretty much anything, requiring jumping through hoops to get it working. Pick what you prefer.

Actually, no it doesn't 'fuck up my system'. When i have upgraded my kernel - i run a single command to install nvidia ~ how is that screwing up my system, exactly? (hint: it isn't by any stretch of the imagination). the worst that could possibly happen is when i reboot X fails ~ big deal, boo hoo ~ very easy to fix (with one command, followed by loading the nvidia module and stating X. As far as Optimus is concerned in laptops, thier is lots of blame to go around and doesn't only fall on Nvidia (as already discussed) and i think you are exaggerating to begin with. As far as OOTB - it takes me no effort at all to install the nvidia drivers on a new system - ie: i don't 'jump through hoops' to get it working ~ that is a load of BS.

and yes, i have 'picked what i prefer' ~ which is Nvidia because it is better than AMD or Intel for GFX in Linux.

That thread clearly says that the benchmarks are running 6 months-old drivers, and that huge improvements have landed in the meantime, which had not been benchmarked there.

then why don't you benchmark your system with these new improvements, rather than just making (what i consider baseless claims) unless, you provide 'real world' data to support your claims.