Got Any Questions For NVIDIA About Linux?

[phoronix]

Phoronix: Got Any Questions For NVIDIA About Linux?

If you have any (non-tech support) questions about NVIDIA and Linux, ask away! Phoronix will be hosting a question and answer session with NVIDIA regarding their Linux graphics driver. If you have any questions to ask, click on the "Comments and Discussion" button below and ask away in our forums...

http://www.phoronix.com/vr.php?view=NzQ5Nw

[macemoneta]

I typically think of Nvidia as a hardware company, yet you seem to want to be a software company. Nvidia's business competitors (AMD/ATI, Intel graphics) have left the software portion of the market, and support open source software driver development.

Which business segment is Nvidia in? If hardware, then why does Nvidia want to spend so much time developing software? Do you see Nvidia supporting the Nouveau drivers at some point over the proprietary drivers?

[eugene2k]

My question would be this: what are the reasons behind the choice to develop the linux drver as closed source?

P.S.
- Within the next 12 months, where do you hope the Catalyst Linux driver will be in regards to features, adoption, and support?
Isn't the Catalyst driver an AMD/ATI graphics driver?

[dandart]

I think it a good idea to release your driver code under a liberal license so it can easily be included in any operating system. It will surely make everyone's life easier. Are there any plans for this?

[deanjo]

I typically think of Nvidia as a hardware company, yet you seem to want to be a software company. Nvidia's business competitors (AMD/ATI, Intel graphics) have left the software portion of the market, and support open source software driver development.


Intel and AMD have left the software market? Since when? Intel in case you haven't noticed has been aquiring software companies lately (project offset and havoc for example) as well they still have their compiler suites. AMD like wise has been pushing out items like AMD Live. Their closed software is still alive and well.

[homerhomer]

Not really a Linux specific question. Is Nvidia working with Adobe to accelerate flash? And if so, when do you expect it to be released?

Thanks

[macemoneta]

Intel and AMD have left the software market? Since when? Intel in case you haven't noticed has been aquiring software companies lately (project offset and havoc for example) as well they still have their compiler suites. AMD like wise has been pushing out items like AMD Live. Their closed software is still alive and well.

I guess I need a disclaimer: This discussion is "Questions for Nvidia regarding Linux". The question is asked in that framework. Nvidia, AMD and Intel are large corporations with multiple lines of business. The questions refer only to the their relationship with software development of graphics hardware drivers for Linux.

[deanjo]

Michael I would drop this question since it's in the README for linux drivers.

Why don't you make install packages available as .DEB and .RPM's? http://us.download.nvidia.com/XFree86/Linux-x86/185.18.36/README/chapter-07.html

Why does NVIDIA not provide RPMs anymore?

Not every Linux distribution uses RPM, and NVIDIA wanted a single solution that would work across all Linux distributions. As indicated in the NVIDIA Software License, Linux distributions are welcome to repackage and redistribute the NVIDIA Linux driver in whatever package format they wish.I also think this:

Within the next 12 months, where do you hope the Catalyst Linux driver will be in regards to features, adoption, and support? is supposed to be this:

Within the next 12 months, where do you hope the Forceware Linux driver will be in regards to features, adoption, and support? Although in it's original form it would be interesting to see the reply. :p

The one question I would like to see is if we are going to see vdpau support in flashplayer since the windows version supports purevideo and they have been working with adobe on this.

[DDevine]

The questions I really want to see discussed are relating to Gallium 3D.

- Is NVIDIA considering developing an Gallium 3D driver?
- What is the engineers opinion on Gallium 3D?

[jander99]

I'm not sure whether or not this has been answered in any definitive way by nVidia, but here is my question for the nVidia guys.

The Linux kernel has progressed quite steadily recently on some technologies such as GPU memory management and kernel-based graphics mode setting. Does nVidia plan to utilize something like GEM and KMS in upcoming driver releases so features like flicker-free booting is possible?

I know it would probably take more person-power from nVidia in order to support some of the new technology coming along in the Linux/BSD space as far as kernel features and X.org features. I also know the Linux kernel modules is derived from the Windows driver, since registry keys can be used to change settings within xorg.conf. Isn't the OS X nVidia driver separate from the Windows driver? Since I don't have a mac I don't know how nVidia drivers within OS X.

[deanjo]

Hey here is a couple more, when are they going to include vdpau info and capabilites in their control panel? When is the openCL libraries and drivers going to be public? When are we going to get profile support in the control panel as well for SLi?

[greg]

deanjo, that's an interesting thought. Something like vdpauinfo should be included with the driver.

[deanjo]

Also, when are they going to be hooking up Phoronix with more nvidia cards to test? Surely they must know a few partners they can hook ya up with.

[deanjo]

deanjo, that's an interesting thought. Something like vdpauinfo should be included with the driver.

It would certianly save alot of "Why can't I run vdpau on my GeForce 5/6/7 etc posts".

[greg]

- Is NVIDIA considering developing an Gallium 3D driver?
- What is the engineers opinion on Gallium 3D?

I doubt NVidia is interested in Gallium3D since their driver stack is superior or at least equal to it.

[sakuramboo]

Ever since AMD/ATI opened up their drivers, has NVIDIA gotten any technology from the AMD/ATI code base or anything that might have benefited NVIDIA? Can just be a yes/no answer since that might go into the realm of NDA's.

[deanjo]

Another, when is that damn PerfHud coming for linux?

http://developer.nvidia.com/forums/index.php?showtopic=81

[deanjo]

Are they going to add any type of monitoring of GPU usage capability to the drivers?

[deanjo]

Is there any plans for ESA support in linux?

[deanjo]

Are there any plans to contribute openCL and/or Cuda code to accelerate video encoding in linux along the lines of what they gave us with vdpau?

[Michael]

Isn't the Catalyst driver an AMD/ATI graphics driver?

Typo, some of these questions were recycled :P

[deanjo]

OK only two more I promise:

Physx support on GPU in linux, is it even on a roadmap?

Will they please finally officially put the damn GTX275 on the linux support list? (Some RPM's packagers are really anal retentive about adding the PCIid to their packages until this is done. *cough* Stefan Dirsch *cough*)

[deanjo]

Typo, some of these questions were recycled :P

lol Ask them both.

[DMJC]

Why is my video card repeatedly freezing and locking up? I've tried gentoo, ubuntu, many different versions of the proprietary driver. And I get graphical corruption, freezes and nvrm Xid messages. I've got a Gigabyte Geforce 260gtx core 216 896mb and I had an 8800gts 320mb the gtx260 is more unstable than the 8800 was. My motherboard is an asus p5q deluxe cpu is an e2220 and I have 4gb of ram. The system appears to pass memory/video memory tests and I have tried everything I can think of to fix it. I have been using linux for 10 years and I have never had this much trouble up until these last two cards. Previously I've owned a TNT2 that had bad ram. A geforce 2mx 400 and a geforce 4 ti 4800se. Oh and under maximum load the highest measured gpu temperature is 59-65C

[DanL]

Not really a Linux specific question. Is Nvidia working with Adobe to accelerate flash? And if so, when do you expect it to be released? Thanks
See: http://blogs.adobe.com/penguin.swf/2008/05/flash_uses_the_gpu.html

[deanjo]

Why is my video card repeatedly freezing and locking up? I've tried gentoo, ubuntu, many different versions of the proprietary driver. And I get graphical corruption, freezes and nvrm Xid messages.

Didn't read the first line of the article I guess.

If you have any (non-tech support) questions about NVIDIA and Linux, ask away!

[deanjo]

See: http://blogs.adobe.com/penguin.swf/2008/05/flash_uses_the_gpu.html

That doesn't answer the question. We are not wondering about slow ass OGL support (btw which truly sucks in 64-bit I might add in flashplayer 10), we are wondering about vdpau support since adobe supports purevideo in windows.

http://www.businesswire.com/portal/site/home/permalink/?ndmViewId=news_view&newsId=20090601006564&newsLang=en

[virtex]

What plans do you have regarding support for Linux on the Tegra platform? What assistance will you provide vendors who wish to produce Linux devices built on the Tegra platform?

[L33F3R]

Heres a simple one.

My dearest nvidia :)

What do you dislike most about linux and its users. (im serious ;)).

[deanjo]

Heres a simple one.

My dearest nvidia :)

What do you dislike most about linux and its users. (im serious ;)).

lol Man is that a loaded question. I feel a 100 page article coming up.

[Redeeman]

When will nvidia fix the severe stability issues which alot of people are suffering from? its been there for like a year now, and spans across LOTS of different hardware.. via email/forums they claim to want to fix it and such, but cant reproduce, which seems highly unlikely..

http://www.nvnews.net/vbulletin/showthread.php?t=123912
http://www.nvnews.net/vbulletin/showthread.php?t=137613
http://www.nvnews.net/vbulletin/showthread.php?t=136622
http://www.nvnews.net/vbulletin/showthread.php?t=134181
http://www.nvnews.net/vbulletin/showthread.php?t=137161
http://www.nvnews.net/vbulletin/showthread.php?t=137215
and theres plenty more threads about these issues..

will nvidia ever fix this?

[EnderWiggin]

Nvidia's TwinView was probably the first stable and consistent multiscreen setup that I could use, however, the it seems that the open-source world has caught up and in some cases leap-frogged TwinView with RandR 1.2/1.3. Specifically, TwinView's refresh-rate mechanisms aren't recognized by most window managers and so screen-spanning windows and backgrounds behave rather poorly. If Nvidia doesn't have a plan to integrate RandR 1.2/1.3 in the near term, are there any plans to update TwinView to provide better multi-screen support in the popular window managers?

[ethana2]

Which situation would be more pleasant for nVidia driver developers--
20% of the market owned by Apple, or 20% of the market powered by linux?

What is their time frame for the release of bridge chip documentation (for coreboot)?

Do they encourage companies like EVGA to include their drivers as .deb files on their install discs? If not, do they realize that Ubuntu has more interest (measured by Google), and probably market share, than all other linux based desktop operating systems combined?

[deanjo]

Do they encourage companies like EVGA to include their drivers as .deb files on their install discs?

Heh, and the first thing a good tech will tell you is to chuck that disk as far away as possible as they are so completely out of date by the time they hit the shelves.

That last statement is pretty grandiose too.

[BhaKi]

What about programming documentation for GPUs? Any plans on disclosing them? Or still want to continue to lock people into using windows and linux?

[dotancohen]

Why so many technical questions? My users have just one question:
When can we start considering nVidia hardware with the confidence that it will run our software (media center, games, composing desktop) with a popular desktop Linux-based OS out of the box?

[smitty3268]

When will the nv driver actually be usable?

[MakotoTheKnight]

Here's a question I'm curious about.

What plans does NVidia have for revolutionizing the Linux Games market?

[smitty3268]

How about asking their opinion on the KDE4 transition. I know there was some frustration on the KDE developers side when they ran into NVidia bugs and couldn't seem to get in contact with the correct people right away (but eventually this got handled, and nvidia was hardly alone in this). Does NVidia have any thoughts about how this type of thing should be handled in the future? Any permanent open-source contacts which developers can take their concerns to about future changes to their codebases? Are there any tools developers have access to that can help debug issues in the drivers, like performance (either bugs in how the software is using them, or bugs inside the drivers themselves), etc...

[unix_epoch]

Will NVIDIA do anything to improve the 2D performance of their drivers (especially when used with a compositing manager)? It's rather disappointing that the integrated i845 graphics in my old laptop can run wobbly windows smoother than my dual-GPU desktop. Granted, some of this is likely due to the advantage integrated graphics have when transferring data from/to GPU memory, but Windows 7 and MacOS seem to do compositing just fine on NVIDIA hardware.

Will any work be done on making VDPAU smooth when being composited with Compiz or kwin?

[MaestroMaus]

Ever since AMD/ATI opened up their drivers, has NVIDIA gotten any technology from the AMD/ATI code base or anything that might have benefited NVIDIA? Can just be a yes/no answer since that might go into the realm of NDA's.

Lol come on, the answer is always going to be no even if they did benefit from AMD going open source.

[rafi]

Here my questions:

Your program "The way it's meant to be played" is very popular on Windows platform. Does it exist for Linux too? If not, do you think it could be something interesting to do?

[axet]

why game work under linux slower?

[Babets]

My questions would be:
How about opening drivers at least for unsupported cards that used the old driver serie 89 (geforce 1,2) and maybe also the 96/173 series (geforce 3,4 and geforce 5)?
With a geforce 1 or 2 you can't use the newer version of xorg with 3d.
A driver for the nvidia-based motherboard's firewall?

[BlackStar]

Three non-tech-support questions to consider:

AMD and Intel offer significantly better driver support "out of the box". What is Nvidia planning to do in order to catch up? Are there any plans to help the Nouveau developers, either by providing documentation (either under NDAs or in the open) or by answering technical hardware-related questions?

How does Nvidia tackle the issue of security in their drivers? Is there any formal security audit process in place? This is especially important in the workstation space.

How does the driver development process work? In the recent months, there was a torrent of alpha and beta quality drivers (more like daily builds), but afterwards drivers releases started to slow down. Is this a reflection of the internal driver development process (introduce features -> work on stability -> release stable)? Who calls the shots on releasing or not releasing a specific driver build (is there a formal process of testing)?

[Licaon]

My question:
What about 3D Stereo on Linux on consumer cards ( 3D Vision (http://www.nvidia.com/object/3D_Vision_Main.html) and 3D Vision Discover (http://www.nvidia.com/object/3D_Vision_Discover_Main.html) )?
iZ3D (http://www.iz3d.com/t-dcdriver.aspx) drivers have a way to run OpenGL games in 3D in Windows ( emulated quad-buffer ) that works for anaglyph red/cyan even on a weak nVidia 6200, yet the official nVidia 3D drivers don't offer anything like this, not even the basic anaglyph.

[StringCheesian]

If the nVidia Linux driver devs had a wishlist for changes to Linux and accompanying libraries/software, what would the top few items be?

What package (Xorg, kernel, etc) has the most room for improvement to better complement video drivers such as nVidia's for 3d acceleration, desktop effects, video playback, etc and in what way?

AMD was able to open source and/or document a lot by separating out the parts they couldn't legally disclose. Similar problems have been cited as preventing nVidia from open sourcing their driver (licensed 3rd parts code, etc). Could nVidia use the same strategy?

Can you speculate on the future of the 2D only open source nv driver vs the nouveau driver? Will be dropped, beefed up, stay the same?

[b15hop]

Phoronix: Got Any Questions For NVIDIA About Linux?

If you have any (non-tech support) questions about NVIDIA and Linux, ask away! Phoronix will be hosting a question and answer session with NVIDIA regarding their Linux graphics driver. If you have any questions to ask, click on the "Comments and Discussion" button below and ask away in our forums...

http://www.phoronix.com/vr.php?view=NzQ5Nw


I like all those questions. =)

What tools are used to create, debug, compile code.

[bugmenot]

Are there any plans to provide GEM / KMS interfaces (interfaces, because AFAIK nvidia driver has own memory manager.) to NVIDIA hardware (and submit patches to kernel if it's not possible to provide interfaces) ? It is important because it allows to speed up developments like http://groups.google.com/group/wayland-display-server , or maybe google's (now secret) windowing system, and eventually abolish old problematic X.

[FunkyM]

- There had been talks about focusing on supporting RandR, however nothing happened since in that direction. Is RandR 1.2/1.3 integration planned? TwinView is nice, but RandR is an interface used by basically the whole graphics tools/driver stack on the desktop except NVIDIA's.

[eugene2k]

Another question I have. If the reasons for having closed-source drivers are just third-party components, isn't it possible to open specifications for the video cards, and work closer with nouveau developers to produce what would later become THE nVidia graphics driver for Linux?

[superfoor]

I have been using nvidia products since my first card, a Geforce 4. I have a very simple question, why not open source the drivers. I mean what do you have to loose ATI is so far behind you. I would think that if nothing else the publicity of releasing nvidia drivers would get you more customers. Remember the Linux community are the guys/girls generally making the IT Decisions at jobs.

[Kjella]

Does nVidia plan to utilize something like GEM and KMS in upcoming driver releases so features like flicker-free booting is possible?

I think Bridgman answered this one for both of them, to use KMS for output you must give up output control including crap like HDCP. So even if it were useful, which it mostly isn't since Catalyst/nVidia implement their own cross-platform version of things, they couldn't.

[nanonyme]

I think Bridgman answered this one for both of them, to use KMS for output you must give up output control including crap like HDCP. So even if it were useful, which it mostly isn't since Catalyst/nVidia implement their own cross-platform version of things, they couldn't.I'm not sure this is completely true nor whether he exactly said that. I do know though that rewriting nvidia or fglrx drivers for KMS would require entire rewrite in both userspace and kernelspace drivers for them. They would need to write userspace stuff to be compatible with libdrm and then kernelspace stuff so that they'd have a module that taps into DRM and plays with it like the opensource modules. It would not, however, possibly need to be opensource itself. "Just" implement the KMS DRM API instead of directly dealing with the kernel API. We're probably talking of years of rewrite work here.
Correction: Well, they could always redesign their own KMS though... Wouldn't probably take any less time than the rewrite.

[yoshi314]

what are the biggest drawbacks/annoyances of developing a closed source linux/bsd video driver?

which part of X is most troublesome to work with?

[Kjella]

I'm not sure this is completely true nor whether he exactly said that.

In the thread about KMS for Catalyst:
"Modesetting includes output control. Output control includes HDCP, Macrovision, and a bunch of other things that prefer the dark."

http://www.phoronix.com/forums/showthread.php?p=76108&highlight=%22prefer+dark%22#post76108

Of course, they should just stop wasting everyone's time and forget DRM, just like music sales didn't die from removing DRM from CDs and the iTunes store neither would movies. CSS is broken, HDCP is broken, AACS/BD+ is in practice broken and if they weren't there'd be hardhacks and the analog hole.

Anyway, I guess I'm taking this thread way off topic.

Question for nVidia:
More and more focus is being put on universal shaders, rather than fixed-function hardware. Do you see the possbility of graphics drivers separating into two parts, one implementing basic access to shaders and the other on implementing DirectX/OpenGL/Cuda/OpenCL etc. in a manufacturer-independent way?

[Ant P.]

Ooh, I've gone one!

Now that Intel, AMD and even VIA are actively participating in the open source community, what are nVidia's plans for staying relevant in the Linux graphics market in future?

(or if you want a different one)

Why has nVidia chosen to release proprietary drivers for even non-graphics hardware like their motherboard chipsets, an area where every other manufacturer has been much less hostile to their customers?

[phred14]

Can you comment on commonality and separation between the driver codebases for Windows and Linux, and how they are brought to functional equivalence?

With my current problem, I bought an nVidia 8400GS-based card specifically for it's rich capabilities for displaying on my "legacy" (Sony CRT) TV. After getting the card, I found that it had excessive overscan, and that the "overscan" option in xorg.conf no longer works for any 8000+ GPU. I've since heard that the "overscan" option has been reenabled for Windows, but is still not available in Linux.

Once upon a time, I thought I heard that the driver codebases had been merged, but this appears to not be the case. By the way, many blame the NTSC spec and old TVs, and try to place the burden there. This particular TV has worked correctly with EVERY other thing I've thrown at it as a signal source, including a Mac laptop and a PC laptop with an ATI graphics adapter. Only nVidia has overscan problems.

[King InuYasha]

What does nVidia think of what AMD/ATi has done to support open source drivers with 3D capabilities?

Has the work AMD/ATi done to support open source graphics card drivers caused people within the company to think following a similar path would be beneficial to the company's goals of selling people nVidia graphics cards?

Are there any future plans within nVidia to follow steps similar to AMD/ATi and release specifications and help with development of the open source driver (nouveau)?

[lbcoder]

What possible reason(s) do I or anybody else have for buying/owning nvidia hardware given the history of poor manufacturing quality (I am, of course, referring to this kind of stuff: http://www.tmworld.com/blog/640000064/post/990031899.html ) and that we would be forced to use an unreliable binary driver of questionable security in contrast with AMD's superior manufacturing quality, lower prices, and reliable open sourced drivers given roughly equivalent overall performance?

[gildea]

NVIDIA: I have two monitors, and I would like to be able to rotate one of them. RANDR 1.2 would let me do this; when will your driver support it?

[Licaon]

Another question:
What about PhysX support on Linux? If CUDA and OpenCL are here already why not port it too?
Ageia had a Linux SDK at http://devsupport.ageia.com/ since 2005.

[birdie]

Are they ever going to release even the smallest part of specs required for implementation of open source drivers, like ATI is already doing.

There's no need to release parts of specs covered by patents - I just want basic stuff like 2D acceleration and basic OpenGL 2.0 acceleration to work under NVIDIA card.

[ostromark]

Will NVIDIA do anything to improve the 2D performance of their drivers...

I would like to add another strong vote of support for this question. It is clear to me that there are performance deficiencies in NVIDIA's Linux drivers for those of us who are not running games.

I have performed some tests recently where a brand new quad-core desktop can be downright unusable with an app workload consisting entirely of 2D business applications on RHEL5. Profiling reveals that the nvidia driver code causes the single threaded Xorg process to pin one of the CPU cores at 100%, at which point the user interface becomes sluggish. The same system running the same applications on an ATI card runs just fine with low CPU utilization. The recent driver updates from NVIDIA have proven to be disappointing. Are there not enough of us 2D "power users" for this to be a priority?

[Max Spain]

As a Linux customer and Nvidia 7000 series user, I prefer graphics cards that feature low power consumption, a lack of DRM, open drivers/specs, and of course great performance ;) While there are lots of good questions here, I don't think these have been asked:

How difficult would it be for Nvidia or their partners to release cards without DRM considering the demand for such DRM (http://www.tomsguide.com/us/Blu-ray-drive-DVD-RW-PC,news-4504.html) remains low?

Seeing graphics cards require 2 PCI-E power connectors scares me. Does Nvidia plan on releasing modern video cards (current gen) that only require one PCI-E power connector?

[Goga777]

my questions are follow -

- does support vdpau cards correctly interlace output ? is there any messing or missing field in that interlaced mode ? how is it possible to adjust the corrected interlaced mode ?
have a look please on that discussion
http://www.linuxtv.org/pipermail/vdr/2009-August/021148.html


the main reason why nvidia chooses to deinterlace always even if you use an
interlaced video timing is not the scaling problem you mention. This could
be eventually solved (albeit not perfectly) by scaling both fields
independently.

The main reason is: even with VDPAU there still exists no synchronization
between stream and video timing.

That means there will be an arbitrary amount of lost/doubled fields with
current VDPAU implementation.

Nvidias workaround now is to deinterlace all content in any case. As a
result these field losses are (mostly) not directly visible to the human eye.

Ceasing from deinterlacing here would reveal field sequence problems
(caused by missing stream-synchronization) immediately.

Cheers
Thomas

[CrystalCowboy]

"Stereo" 3D on the Samsung 120 Hz LCD monitor:

When will there be a driver for Linux?

What hardware will and will not work for this?
- What graphics cards?
- Is it necessary to buy the "GeForce 3D" emitter and glasses? I have heard the CrystalEyes equip is not compatible.

I have $$$$$, and I am ready to buy! Give me a product that works, clear information on what I need, and I'll buy half a dozen of these subsystems the next day. I need them for scientific visualization.

[djtm]

1. Will NVIDIA bring out support so VDPAU can be used in editing software as well (copy the decoded video data back into main memory).

2. Will NVIDIA facilitate CUDA support e.g. in x264, xvid, ffmpeg or any other linux projects? If so, then how?

[volkris]

Any chance of getting any, even partial, VDPAU support for first gen PureVideo cards (GeForce 6-8)?

[Kano]

When will 71 series support current xserver? It was announed first that this will happen, but it did never happen.

[greg]

1. Will NVIDIA bring out support so VDPAU can be used in editing software as well (copy the decoded video data back into main memory).

I know this thread is mostly for asking questions, not commenting the questions, but...

That's already possible (http://http.download.nvidia.com/XFree86/vdpau/doxygen/html/group___vdp_video_surface.html#ge3550dc62e4e4c390c 1874b37f96e7c0), and has been for quite some time.

[greg]

I do know though that rewriting nvidia or fglrx drivers for KMS would require entire rewrite in both userspace and kernelspace drivers for them.

According to NVidia developers (http://www.nvnews.net/vbulletin/showpost.php?p=1946400&postcount=5), the missing piece mostly is a framebuffer driver. Modesetting is already done in the kernel. I doubt implementing something similar to KMS has a high priority, though.

[greg]

If the nVidia Linux driver devs had a wishlist for changes to Linux and accompanying libraries/software, what would the top few items be?

What package (Xorg, kernel, etc) has the most room for improvement to better complement video drivers such as nVidia's for 3d acceleration, desktop effects, video playback, etc and in what way?

AMD was able to open source and/or document a lot by separating out the parts they couldn't legally disclose. Similar problems have been cited as preventing nVidia from open sourcing their driver (licensed 3rd parts code, etc). Could nVidia use the same strategy?

Can you speculate on the future of the 2D only open source nv driver vs the nouveau driver? Will be dropped, beefed up, stay the same?

I think these questions are particularly interesting.

[L33F3R]

I have $$$$$

can L33F3R has some? :D

[L33F3R]

What does the great nvidia predict will be the future of OpenGL.

[deanjo]

What does the great nvidia predict will be the future of OpenGL.

Given that nvidia's Neil Trevett is the President of the Kronos Group and he created the openGL ES group I would say that they have a pretty good opinion of it's future.

[oscar]

Some interesting questions for a developer using Linux:

*Physx GPU support? Since Windows supports it and it's based on CUDA which is supported on Linux too, there will be PhysX GPU on linux?
*Since release 185 windows drivers provide cuvenc.dll which support GPU accelerated H.264 encoding.. Will this library supporting CUDA H.264 ported to Linux?
The VDPAU seems a similar to cuvid.dll so what about cuvenc.dll..
*This is somewhere not yet released but as in:
http://www.youtube.com/watch?v=FLQuqXhlx40
"NVIDIA NEXUS: Visual Studio-based GPU Computing and Graphics Development"
there's something similar planned for Linux users i.e. a addin for a IDE for GPU debugging (Netbeans,Eclipse,etc..) using the CUDA gdb debugger..
*3D Vision for Linux when?.. I mean support for OpenGL QB stereo on Quadro Cards..
If not I have heard that are going to launch 3D vision Pro with API support
for glasses with accelerometer, etc.. Are this going to be supported

So briefly are:
1. Physx GPU support?
2. GPU accelerated H.264 encoding API? or extending VDPAU..
3. GPU debbugging integrated in some IDE as Nexus for Visual Studio?
4. 3D Vision for Linux and OpenGL QB Stereo on Quadro cards? when?

[oscar]

Some interesting questions for a developer using Linux:

*Physx GPU support? Since Windows supports it and it's based on CUDA which is supported on Linux too, there will be PhysX GPU on linux?
*Since release 185 windows drivers provide cuvenc.dll which support GPU accelerated H.264 encoding.. Will this library supporting CUDA H.264 ported to Linux?
The VDPAU seems a similar to cuvid.dll so what about cuvenc.dll..
*IDE with GPU based debugging:
This is somewhere not yet released but search youtube for:
"NVIDIA NEXUS: Visual Studio-based GPU Computing and Graphics Development"
there's something similar planned for Linux users i.e. a addin for a IDE for GPU debugging (Netbeans,Eclipse,etc..) using the CUDA gdb debugger?..
*3D Vision for Linux when?.. I mean support for OpenGL QB stereo on Quadro Cards..
If not I have heard that are going to launch 3D vision Pro with API support
for glasses with accelerometer, etc.. Are this going to be supported

So briefly are:
1. Physx GPU support?
2. GPU accelerated H.264 encoding API? or extending VDPAU..
3. GPU debbugging integrated in some IDE as Nexus for Visual Studio?
4. 3D Vision for Linux and OpenGL QB Stereo on Quadro cards? when?

[L33F3R]

Given that nvidia's Neil Trevett is the President of the Kronos Group and he created the openGL ES group I would say that they have a pretty good opinion of it's future.

true.

Dearest nvidia.

Carmack is switching to Direct X over openGL. Why do you have to suck so much.

[deanjo]

true.

Dearest nvidia.

Carmack is switching to Direct X over openGL. Why do you have to suck so much.

Heh, it was when Neil took over that the Khronos Group finally pulled their head out of their asses and started getting some long over due stuff done.

What still has to happen is a complete Khronos solution to compete with all of DirectX facets and put together the tools and SDK's to support them. Such as input, sound, graphics, etc etc.

[L33F3R]

Heh, it was when Neil took over that the Khronos Group finally pulled their head out of their asses and started getting some long over due stuff done.

What still has to happen is a complete Khronos solution to compete with all of DirectX facets and put together the tools and SDK's to support them. Such as input, sound, graphics, etc etc.

god damn you deanjo i'll never have a question :p

[deanjo]

god damn you deanjo i'll never have a question :p

You can ask when will the leafs win?:p

[L33F3R]

Dear my loving and caring nvidia.

Please tell deanjo that the leafs will win the cup next :D

[b15hop]

I just remembered a question that I feel is very important.

Mobile phones will soon all be GPU powered. NVidia is spending up big in this area. It seems there is a race to get the foot into the door with this technology. Smart phones will become Smarter phones. Since *nix type operating systems have been used on embedded devices for such a long time. Mainly due to the efficiency, cost, performance vs ease of development time. Will NVidia Tegra development be available on the linux scene?

http://www.youtube.com/watch?v=rQa9nP4yyms

Google has the Android platform:
http://code.google.com/android/
http://www.youtube.com/watch?v=CO7Yxyux1_k

Notice android uses virtual screens much like a Linux desktop. (great fun)

I can develop for Android on linux:
http://developer.android.com/sdk/1.5_r3/index.html

Google seems to be steam rolling ahead. Providing universal support rather than selfishly cornering the market. The user (me) wants a phone that can do it all. I don't want an iPhone or Microsoft phone. I want a phone where I choose the operating platform, the hardware and allow the community input it's own software.

My Sony z610i isn't bad. Though it seems far too Sony focused rather than innovating solutions for the customer.

Regards,

Phillip Kilby.

[rah_]

Question:

Both Intel and AMD supply open, NDA-less documentation for their graphics chips. The buying decisions within the free software and GNU/Linux communities are very strongly focussed on these manufacturers precisely because of this. The documentation enables an open, bazaar-style development model of drivers which has economic advantages over a closed, cathedral-style development model. It also enables increased synergy with the rest of the free software desktop, particularly the X windowing system.

Last year there was a 61% increase in the market share of GNU/Linux on the desktop, which is quite phenomenal. These increases will no doubt continue. The future upper limit of this market share cannot be known but it seems clear that the GNU/Linux desktop must factor into any long-term strategy in the IT industry.

With the advantages that open documentation brings to Intel and AMD, NVIDIA seems in danger of being left behind, if it hasn't been already. How will NVIDIA mitigate the disadvantages of a closed product?

[mysticalzero]

Here's a question regarding VDPAU:

I remembered hearing VDPAU is not supported by G80-based GPU due to the absence of a true dedicated video decoder. However, seeing that PureVideo works to some extent in Windows as to offload video decoding from the CPU, is it possible to extend the functionality of VDPAU to include G80-based GPU? Probably through other means like the use of CUDA?

[BlueJayofEvil]

Some of my questions are Linux/BSD specific and a few others are general nVidia-related.

1) Does nVidia have plans to upgrade the memory on its cards to match/surpass that of AMD/ATI? AMD/ATI currently uses GDDR5 while nVidia's highest-end current-gen cards still use GDDR3. Is there any plan to change that in the next upcoming line-up?

2) With AMD/ATI releasing unencumbered hardware documentation to allow free and open-source drivers to be made separate from their proprietary driver, will nVidia consider doing something similar?

3) With FreeBSD 8.0 coming within a month (hopefully) and most (if not all) of nVidia's requests implemented, will there be a 64-bit nVidia FreeBSD driver soon?

4) Is the Tegra platform going to be supported by proprietary-only drivers or will there be some openness in that area?

5) Any plans for GEM/KMS support?

[deanjo]

1) Does nVidia have plans to upgrade the memory on its cards to match/surpass that of AMD/ATI? AMD/ATI currently uses GDDR5 while nVidia's highest-end current-gen cards still use GDDR3. Is there any plan to change that in the next upcoming line-up?


When it comes to memory speed they are pretty much on par. Nvidia's user of a wider memory bus offsets the faster but narrower bus card with GDDR5.

[Michael]

About 24 hours or so until I will end the call for questions.

[Huenengrab]

I have to admit that I didn't read all questions in this thread, so I'll just hope it hasn't been asked already:

Will Nvidia support the usage of .ICC-Color-Profiles in their drivers?

It'd be enough to just enable nvidia-settings to load a specified ICC-profile via a special parameter. It his extremely annyoing to reload my ICC-profile via xcalib after running the Nvidia-GUI or nvidia-settings which overrides any color-calibration. It would be even better if it supported ICC-profiles per plugged device, which is a feature xcalib is still lacking.

[curaga]

Why not release documentation freely for old / very old products? As mentioned, Geforce 1/2 can not be used with the latest Xorg.

[Kano]

gf 2 mx does work, gf 2 gts does not.

[tmpdir]

"newer NVIDIA GPUs now support decoding MPEG-4 Part 2, DivX 4, and DivX 5 formats. In order to use VDPAU with these formats, a NVIDIA GeForce GT 230M, GT 240M, G210M, GTS 250M, or GTS 260M is required. For these GPUs there is now also a higher quality video scaling algorithm that's used by the driver. Owners of other NVIDIA hardware are unaffected by these VDPAU changes."

will the new formats and higher quality algorithms become available for the other (newer) hardware as well?

[deanjo]

"newer NVIDIA GPUs now support decoding MPEG-4 Part 2, DivX 4, and DivX 5 formats. In order to use VDPAU with these formats, a NVIDIA GeForce GT 230M, GT 240M, G210M, GTS 250M, or GTS 260M is required. For these GPUs there is now also a higher quality video scaling algorithm that's used by the driver. Owners of other NVIDIA hardware are unaffected by these VDPAU changes."

will the new formats and higher quality algorithms become available for the other (newer) hardware as well?

http://www.nvnews.net/vbulletin/showthread.php?t=138336

No, the HW features required are specific to GPUs with VDPAU feature set C. -Stephan Warren

[Ex-Cyber]

Why not release documentation freely for old / very old products? As mentioned, Geforce 1/2 can not be used with the latest Xorg.My guess would be some combination of:

- Documentation doesn't exist in an organization/format suitable for release

- Legal concerns ("old" by PC standards is still a lot less time than a patent term)

- These GPUs may still have some things in common with newer ones, thus releasing docs may reveal information about newer GPUs

- From a business standpoint, why bother going to extra trouble/risk to support hardware that they haven't been selling for years?

[deanjo]

My guess would be some combination of:

- Documentation doesn't exist in an organization/format suitable for release

- Legal concerns ("old" by PC standards is still a lot less time than a patent term)

- These GPUs may still have some things in common with newer ones, thus releasing docs may reveal information about newer GPUs

- From a business standpoint, why bother going to extra trouble/risk to support hardware that they haven't been selling for years?


Not to mention that there are not that many of those cards still in use. Even the ones that are out there still in use, I doubt they are running cutting edge releases simply because the machines that those are running on would lead to a unsatisfactory desktop experience running the latest kde/gnome etc for example. That's quite a bit of work for such an extremely small group.

[tmpdir]

http://www.nvnews.net/vbulletin/showthread.php?t=138336

Thanks Deanjo.

---

The G210 and GT220 are missing in this list (ftp://download.nvidia.com/XFree86/Linux-x86_64/190.32/README/appendix-a.html). These cards are available as oem and will be in retail next month. Looking at the features my guess is these will be class c as well.

[Kano]

Ok, in that case it should be possible to get a test card. Those slow cards are sure on ebay as they are sold in lots of OEM boxes. Feel free to send me one ;)

[mibo]

Something in CUDA of the "newer than" 180.xx linux driver was changed so that they are no more compatible with the windows driver. That can be seen when trying to run the windows executable of GPU-Folding@Home and a dll-wrapper.
Is this a bug in the linux drivers that will be fixed, or is it an intended difference between the platforms?
http://forums.nvidia.com/index.php?showtopic=105878

Cheers,
mibo

[Gamester17]

Will NVIDIA developers assist FFmpeg with patches to get them supporting VDPAU's new MPEG-4 part 2 (DivX 4/5, etc.) decoder support?

This question is interesting since loads of open source multimedia software projects out there for Linux uses FFmpeg demuxer and codec suit.

In addition, I like to know why did NVIDIA create VDPAU instead of use/extend VAAPI nativly?

Also, does NVIDIA ever plan on bringing VDPAU to Mac OS X? and if so will it then have a public API so third-party software like XBMC Media Center could use it on Mac OS X?


Another thing already mentioned here that I too would be interesting to known if there an official Gallium 3D driver is planned? and if so, will that driver will have VDPAU support?