Saturday, December 08, 2007

A comment on Nvidia

Recently a blog entry written on MyOpera back in August 07 has started being passed around again. The author dreams up a scenario where Nvidia introduced an Open-Source driver. I couldn't help but laugh while reading the article because Denilson F. de Sá had absolutely no clue about what he was talking about.

Now, I freely admit that I am probably biased towards ATi, for several reasons. First, ATi actually returned emails, and in some cases called me up point blank to talk about upcoming products. Chris Hook, Renee, and John S. all went out of their way to find me a spot during IDSA then ESA's E3's convention to go over upcoming products. Every time I've talked with ATi, even if it was blathering on about something far over my head, somebody took the time to listen or correct errors. Second, ATi was owning up to their previous problems. There was no denying that prior to the Catalyst driver set, ATi hardware was severely let down by the junk drivers. At the same time ATi was taking great strides to fix their drivers.

Now, I do have someone on record somebody at ATi saying they were not going to pursue performance drivers for Linux. ATi just wasn't interested in that market. That... actually was fine to me. I was not given an answer full of manure or misdirection. ATi stated they would improve the feature set of the Catalyst Linux drivers, and then completely blew by their rival with a streak of monthly driver releases going as far back as 2005.

Okay, yes, I'm quite happy that AMD has taken ATi further by opening up the Atom Bios, and by publishing the specifications for the graphics cards, and by funding Novell to develop an Open-Licensed driver. But what about before AMD came along?

Yet, what many people have to remember is that while ATi wasn't exactly Open-Source Friendly for several years... they weren't Open-Source hostile either. ATi funded driver development for an open-licensed 2D Linux driver for the Rage chip. ATi released specifications for the original Radeon and Radeon 8500 GPU's, so that driver developers were able to get working 3D drivers in place... before AMD came along.

ATi also got involved with package maintainers, and started a driver-preview program to insure that package maintainers were prepared for official releases, and that the ATi installer could be improved. ATi also had engineers around who would reportedly answer questions about the hardware to driver developers for the ATi driver and Radeon driver. ATi also listed links to an unofficial Wiki and Bug tracking system for Linux... on their home site. ATi's own engineers were known to trawl both the bug tracking system, and popular sites like Rage3D. ATi also stated at least as far back as 2005 that one of their goals was to build or release an Open-Licensed driver.

Again, all that before AMD ever entered the picture.


Now, lets compare that to Nvidia. A lot of Linux supports have said over the years, buy Nvidia, don't buy ATi. Yet, um... the only significant contribution Nvidia has made to Open-Licensed software is the 2D NV driver... which is little better than a widescreen patch on a VESA type driver. Nobody uses the NV driver to build off of because it's junk.

Nvidia has never released specifications for their cards. Nvidia refused to help or recognize engineers and developers working to build an Open-Licensed Nvidia driver. Nvidia never had or linked to a public accessible bug tracking system supported or listed a wiki for their cards. Their engineers have never seriously paid attention to any Nvidia fan site, nor answered questions.

From a technical standpoint, consider the blog I made looking for funding to go pick up an Nvidia card. Nvidia finally fixed memory handling problems in the Geforce4. That card was released in 2002, and 5 years later a fix is finally issued?

How about the the ATi vs. Nvidia 64bit comparison, where it's fairly obvious Nvidia has the broken driver set. It's sort of hard to ignore the problems with Nvidia-glx, and I've stated before I have absolutely no respect for Nvidia's driver team. I honestly don't think they know what they are doing.

From a personal standpoint, every time I've posed a question to Nvidia, I get an answer more crooked than a New York Senator seeking presidential office. Even questions asked to representatives faces during a convention were conveniently deferred to people who were not present. I've caught Nvidia reps outright lying to me on several occasions, and after a while, that gets a little old.


So, why would anybody believe, for a second, that Nvidia would honestly spearhead any charge for Open-Licensed drivers? A charge ATi started again, long before AMD ever entered the picture.

I can't answer that. Same way I can't answer why many Linux advocates think Nvidia products are good things.

Okay, yes, the situation in Linux Graphics has never been where many of us personally wanted it to be, and I have a laundry list of problems with ATi. I can't say I'm really happy about how long the average new product to Linux driver release has taken, or how long it took to get the new driver out the door.

However, looking back over the years, I can count a lot more Pro-Linux moments with ATi... than Nvidia will probably ever have.
Post a Comment