HardOCP and Doom 3 benchmarks

Just FYI, that info came from this thread:

http://www.beyond3d.com/forum/viewtopic.php?t=5776

 

Thomas Pabst obviously hasn't been successful inspecting Nvidia. There are still a lot of reports of Doom III comming out through reviewers about Nvidia. Unless ATI gets free and unfettered access soon I think they no choice but to invade. Hopefully, they can liberate gaming populace in the process and keep IHV casualties to a minimum. King of Spades, Jen-Hsun Huang, must be brought before the independant FutureMark Beta coalition to pay for his marketing crimes!

 

While HL sold an order of magnitude more copies then Q3 IMO Q3 sold many, many more video cards then HL.

 

Hmmm, has this issue ever been fixed:

http://www.hardocp.com/article.html?art=NDQ0LDM=

I'd say that SC looks much worse on NV hardware in those shots. This may have been fixed though, I don't have an Nvidia board so I don't know.

The sites running the Splinter Cell benchmark are also running the same codepaths for both hardware when benchmarking. Will they do the same with Doom3?

 

I am guessing their next drivers will kill performance to the 10 fps range. Why, because we saw what the next available drivers (Cat 3.4) did. These are the next drivers from ATI released with the R9800-256. From what I can tell, these will infact be the next release of the drivers. But they were "broken" for all the review sites. I still want to know why none of the sites could get a meaningful number using the D3 using these "public" drivers.

If the Cat 3.4s really are "fully standard compliant," why did they tank when the last set worked fine? Or are there things broken in the drivers that maybe none of the "standard" benchmark programs have yet revealed? Or is it that the version of D3 released didn't properly recognize the drivers? That seems like a possiblity, but that raises a lot more questions.

 

Does it really matter? D3 isn't a "public" game yet.

 

Please refer to my reply/comment in the same thread quoted above. I don't know what exactly is going on and until I do, I hope you guys won't speculate too wildly.

 

id corresponded with me, not Dave, regarding this and Dave's "Talk to NVIDIA" comment (link just above this post) should not be miscontrued, as is evident by Joe's thoughts/comments. I don't know what exactly is going on and if I don't, neither Joe, Dave nor anyone else can claim to know (based on the info in the link provided above).

Don't speculate about this.

 

You need to get nvidia's permission to run / acquire a Doom3 benchmark. Full stop. Correct? No speculation there.

Asking us to not speculate WHY nVidia has a say in this is out of line, IMO. There would be no speculation involved, if id was "allowed" to release their own benchmark without nVidia's "blessing".

 

Perhaps nVidia will be the ones held accountable if there is a leak.

Edit-

Meant to post this and forgot, from Carmack-

 

B3D: "Is there any way we could work together on a similar basis as per what was accorded to HardOCP and AnandTech?"
"

ID: "Anthony,

NVIDIA chose those outlets - we did not. It would be best to place this
request with them."


I see nothing about the Demo, I see asking for a meeting.

 

What would id do to hold them accountable...give another build to ATI?

 

Except Z bias! :? (Until DX9 that is.)

 

Incorrect. From the way I read id's reply to me, it meant id prefered I liase with NVIDIA to arrange for a benchmarking session. Big difference.

Again, you're either speculating or jumping the gun.

PS. 2AM, going to bed, so don't expect any further quick responses from me.

 

Holy crap! I'm in agreement with Doom! I think iD is thinking of it in terms of Beyond3d wanting to benchmark with equipment provided by NVIDIA. iD has no control over who gets the equipment, which is why he said: go ask NVIDIA if you want to meet here and use their equipment and run my 'demo'.

(I think you chicken little types are a bit off base on this one)

 

One more quick note b4 bed :

Note that id chose to use the word "outlets" when they replied to me.

That, to me, means the websites NVIDIA chooses to arrange what [H] and Anand experienced. Again, refer to what I said at http://www.beyond3d.com/forum/viewtopic.php?p=114825#114825

 

Translation: "nVidia picked a build from some point in the development process, and as far as we're concerned, that's random enough."

Translation: "We realize that for the past 10 months ATI has had the clearly superior hardware, so providing a .plan update that indicated that, or worse, providing a benchmark "during that time" to show it, wasn't important for comparison purposes. On the day that nVidia introduces a competitive part, the defintion of "important at this time" has been fulfilled."

Translation: "We're not COMPLETELY off our rocker and realize that we could not spin enough to justify actually using an nVidia created demo, and have the community accept that."

Translation: "nVidia was not nervy enough to give ATI the heads up and some time to compete with tuned drivers."

Translation: At some point, there will be a benchmark version that is actually useful. But for now, accept this as the marketing gimmick that it is.

 

Not so big.

Why should anyone have to liase with any IHV for a benchmarking session using id's software?

It would not be an issue, if you can go to ATI and ask THEM for a "benchmarking session" as well. Go ahead and try it. (Serious request...if id allows ATI to also orchestrate a benchmarking session, then my concerns are in fact unfounded.)

Indeed...provided that ATI is allowed to arrange a similar benchmarking session with Id's blessing.

I'm more interested in a response from ATI and nVidia at this point.

 

There is no such thing as "ARB2 extensions" either. It's either ARB or non-ARB.

ARB2 is simply the name Carmack gave the second path using ARB extensions. The first ARB path he used is using other ARB extensions, such as GL_ARB_texture_env_combine, that is, fixed function (DX7 level) capabilities. The second path uses extensions such as GL_ARB_fragment_program and GL_ARB_vertex_program.

 

To avoid leaks. nV has the bench with them when they arrive at the site. They take it with them when they leave.

Perhaps if you substituted S3 or Matrox that would be a good point. Throwing the demo up on a website likely wouldn't be the best way to avoid leaks if that is the reason they are doing it this way