3dfx and Nvidia's public flame war

We've said it before, but ain't it great when you're reminded that businesses are run by real people with real prejudices and flaws? This one's a treat and was played out on Glide Underground.

Nvidia and 3dfx are getting increasingly uptight with each other. They were neck and neck once, but Nvidia is now storming ahead (nabbing M$'s X-box contract, for example).

3dfx is not a happy bunny. The announcement that it was to delay release of Voodoo5 because of field failure rates inspired a cocky email from Nvidia's Dianne Vanasse - cc'ed to a range of hardware sites.

Senior PR man at 3dfx, Brian Burke (admittedly egged on by HardOCP), responded publicly. Just a daft spat you might think but there was much obviously a lot of paddling going on underwater since soon afterwards Voodoo 5 is suddenly fine and 3dfx had been "overly cautious" - it'll be shipped according to schedule. And hang on, but Nvidia's 5.22 drivers are suddenly non-beta. As our man John Davis from Glide Underground said: "This industry is something else."

The emails are reproduced below:

"From: Diane Vanasse
Sent: Wednesday, May 24, 2000 3:46 PM
Subject: 3dfx Delays Voodoo5 Schedule

Hence our argument about shipping product versus non-shipping products. I believe we were on the shelves on April 30th?"

Kyle Bennett from HardOCP then asked Brian Burke (the man who wrote the original delay press release):

"I have been thinking more and more about this 'shipping Vs. non-shipping' product and wondered what you thought about how it pertained to leaked drivers as well as product?"

And in response:

"From: Brian Burke
Subject: RE: 3dfx Delays Voodoo5 Schedule

Fair question, I'll bite. I think that as long as the comparison is representative of what customers will be using, then it is a fair comparison. We are proud of the products we give out and have no problem with them being scrutinized. If one FSAA method is technically superior to another, then the tests should reflect that. We think that their max FSAA setting is comparable to our 2X setting and can back it up scientifically. We think the tests results should reflect that. On the other hand, we have no on board T & L, so test our products that way and we will stand behind the results. Real world is the criteria that should be used and testing should be as apples-to-apples as possible. Words like "leaks" and "Beta" should not be used as a smoke screen to relieve accountability or dodge results. If we do not want you to test something, we do not give it to you to test.

I'm a little confused by Diane's message:

"Hence our argument about shipping product versus non-shipping products." -- Diane Vanasse from Nvidia.

I was not privy to that argument. Has Nvidia discouraged the testing of the shipping GEforce 2 FSAA vs the non-shipping 3dfx FSAA? I am a little confused, the FSAA drivers from Nvidia are BETA and considered non-shipping, right? Even when they shipped in the ELSA box on the CD on April 30th they were BETA. The last official release on Nvidia's site is 3.68, with no FSAA, right? Well, on the WHQL site there is a 3.84 WHQL driver from Nvidia, but it has no FSAA either. So you are comparing two non shipping products when you compare the FSAA quality the two companies offer. True for now, correct? So what is the problem?

Take it a step further. Maybe you should test as she asks. To support the request made by Diane (who is an official spokesperson for Nvidia), editors should only test shipping product vs. shipping product, non shipping vs. non shipping.

Since no "shipping driver" from Nvidia supports FSAA, this lifts any level of support to customers and leaves them out in the cold. It also conveniently lifts accountability for the quality or lack of quality in the driver from Nvidia. Consider this, when the Voodoo5 ships, Nvidia will have no FSAA of any kind to compare to the Voodoo5 FSAA, per Diane's wishes in an official company communication to the top online editors as shown below. When the Voodoo5 is released, testing of all BETA drivers from Nvidia should stop immediately when making comparisons to the 3dfx FSAA. To support Nvidia's request, as soon as Nvidia officially releases a driver with FSAA, and signs on to support it, comparisons should begin again. Until that release, Nvidia will not have any FSAA to compare to 3dfx. I'm OK with that >:).

Is there a release of a driver with FSAA support pending so customers can get the support they deserve and the functionality Nvidia has promised? Is the shipping "BETA" driver shipping or BETA or both? Also, will that pending "released" driver include Microsoft WHQL certification? If not, does Nvidia encourage the use of Non WHQL drivers to OEM customers? Will FSAA only apply to retail versions of Geforce2 products and not OEM versions? I think a company that cares about quality and compatibility would want Microsoft WHQL certification to help insure trouble free use for its customers. I think OEMs require it, too. I know 3dfx has WHQLed our Voodoo5 drivers.

BB Senior PR Manager 3dfx, Inc."

(That's 15 uses of the name Nvidia in case you're wondering.) ®