[ben6]

Let me start by saying , I agree with Nvidia and Kyle to a certain point. I think too much emphasis is placed today on synthetic benchmarks in MOST reviews.

In fact, those who know me , know that I've held this opinion since I started reviewing videocards in 2000 with the release of the Voodoo 5 5500. I have made a point of including experiences of games I currently play in addition to including benchmarks in every review I write. I've noticed , HardOCP has started including these as well. Which I like . Problems as well as good points. That's what a review is supposed to be about .

Having said that, I have strong reservations about the editorial written today at HardOCP. I'll have more on the subject a little later

[jb]

Due make sure to tell us good sir as you have always had a good view of things IMHO.

[Reverend]

Synthetic benchmarks have their uses. It comes down to the websites/medias on how to interpret such uses.

If 3DPulpit was still around, I won't use 3DMark at all (never have anyway) because I have a very specific focus for 3DPulpit and its targetted audience, and 3DMark serves neither 3DPulpit's focus nor its audience.

At Beyond3D, the focus and audience are different and 3DMark has its uses.

I'm quite amused by all the fuss generated by a few individuals (most noticeably Joe DeFuria) about HardOCP's "stance" on 3DMark2003 but that is probably because I look at things from the viewpoint of being part of a website and not the easily-influenced buying public.

[TheMightyPuck]

What is ironic, as people have noted, is the way the tables have turned (or rotated or something). Remember Nvidia was rah rahing T&L and the v5 was considered somehow inferior because it couldn't do it. Then someone noted that CPUs were pretty much just as fast for the T&L stuff. Now 3dMark has created a benchmark to specifically test graphics card capabilities (like if someone wanted to offload some work from the CPU) and Nvidia is moaning that that's not how people actually make games. Too funny.

[Chris123234]

thats not the only reason their 'moaning'.

BTW if youve noticed how the 9500's and above whoop the Gf4's its because of the weighting.

3DMark03 score =
(Game Test 1 frame-rate x 7.3) +
(Game Test 2 frame-rate x 37) +
(Game Test 3 frame-rate x 47.1) +
(Game Test 4 frame-rate x 38.7)

If you dont have DX9 capabilities then your score is gonna be LOW.

[Pete]

Just havin' read the blurb, allow me to express my reservations on the quality of English many internet writers display. ;P It's "slim pickings," thus the correct colloquial spelling should be "slim pickin's," not "pickens" (for all intensive purposes ).

While I'm being picky, it's do tell.

Offa nit-pickin' an' on to readin' the actual editorial....

[Sharkfood]

If you dont have DX9 capabilities then your score is gonna be LOW.

Identical shades of "Nature" from 3dmark2001. At the time, if you didn't own a Geforce3 specifically, your scores were going to be LOW.

The truly ironic thing about all this hubbub is that this is the 5th generation of this benchmark (3dmark99, 2000, 2001, 2001SE and now 3Dmark03), from which the same valid concerns and stipulations have always existed... it isn't until NVIDIA speaks up to voice these concerns that sources have decided to pursue this line of reasoning. That is what I label as a massive double standard.

NVIDIA's (original) concerns were very well formed and very debatable. It's just unusual that their statements concerning "3dmark fitness" as being used for a particular purpose have been the same case throughout the entire history of the benchmark... and even more unusual that people are now suddenly willing to stop the presses, step back and actually look at such concerns.

[Doomtrooper]

Originally Posted by Reverend

Synthetic benchmarks have their uses. It comes down to the websites/medias on how to interpret such uses.

If 3DPulpit was still around, I won't use 3DMark at all (never have anyway) because I have a very specific focus for 3DPulpit and its targetted audience, and 3DMark serves neither 3DPulpit's focus nor its audience.

At Beyond3D, the focus and audience are different and 3DMark has its uses.

I'm quite amused by all the fuss generated by a few individuals (most noticeably Joe DeFuria) about HardOCP's "stance" on 3DMark2003 but that is probably because I look at things from the viewpoint of being part of a website and not the easily-influenced buying public.

What I find Ironic when the synthetic benchmarks actually starts using more graphic card power certain websites that 'treasured it as the defacto standard' now don't condone it anymore...

3Dmark 2001 was horribly unbalanced and was a platform benchmark, yet I didn't see a editorial about that, in fact the site in question made statements like:

You guessed it, 3DMark 2001 is the next generation DX8 benchmark from MadOnion. It is more heavily weighted on the VidCard performance, as it fully utilizes some of the special DX8 instructions. Now after seeing the 3DMark 2000 scores, we were surprised to see the Radeon 8500 do so well. While it still falls up to almost 500 points behind the Ti500, it certainly produces a respectable score at 1024x768x32, though much less so at 1600x1200x32.

We are also dealing with a individual that freely accepted a application from a competing IHV to cause as much havoc as possible on a companies launch of their video card, a application looking for a Quake reference that has been in there since original Radeon vivo....used 3Dmark 2001 in every review

Trash Interenet Journalism or T.I.J

[Joe DeFuria]

I'm quite amused by all the fuss generated by a few individuals (most noticeably Joe DeFuria) about HardOCP's "stance" on 3DMark2003...

That's funny indeed! I'm amused at the fuss that nVidia is making over 3DMark2003, and even moreso at the web-sites (um, web-site?)that "coincidentally" share that identical stance with nVidia...

[Chris123234]

Originally Posted by Sharkfood

Identical shades of "Nature" from 3dmark2001. At the time, if you didn't own a Geforce3 specifically, your scores were going to be LOW.

The truly ironic thing about all this hubbub is that this is the 5th generation of this benchmark (3dmark99, 2000, 2001, 2001SE and now 3Dmark03), from which the same valid concerns and stipulations have always existed... it isn't until NVIDIA speaks up to voice these concerns that sources have decided to pursue this line of reasoning. That is what I label as a massive double standard.

NVIDIA's (original) concerns were very well formed and very debatable. It's just unusual that their statements concerning "3dmark fitness" as being used for a particular purpose have been the same case throughout the entire history of the benchmark... and even more unusual that people are now suddenly willing to stop the presses, step back and actually look at such concerns.

I SAID BTW. BTW <-- see BTW. I DID NOT SAY NVIDIA IS COMPLAINING BECAUSE OF THE WEIGHTING. I SAID BTW YOUR SCORES WILL BE LOW. READ THE FREAKING POST BEFORE YOU COMMENT ON IT. I DID NOT SAY THIS WAS ONE OF "NVIDIA's (original) concerns"

BTW: If I read yours wrong, my bad. But I dont think I did cause you wouldnt have put "NVIDIA's (original) concerns".

And to Joe, the points nVidia and ]H[ pointed out are legit. So its not like their mad cause the Gf4's cant compete with the 9x00's.

[Pete]

I agree, given that nVidia uses (do they still?) a Futuremark bench (3DM02SE, no?) on their site to determine whether you should upgrade your card or not, their stance on 03 is suspiciously timed.

[Joe DeFuria]

And to Joe, the points nVidia and ]H[ pointed out are legit. So its not like their mad cause the Gf4's cant compete with the 9x00's.

Legit in who's opinion?

Even if they are legit, the point is nVidia is highly hypocritical with this stance. If they are legit now, they would be legit 2 and 4 years ago. I'd say it's very likely they are "mad" that the GeForce4 Ti can't compete with the 9500, and the GeforceMX line can't compete with the 8500 and 9000 in this test.

Thing is, 3DMark03 more or less highlights the technology gap between the current line-up of nVidia compared to ATI. And one would be foolish to just assume there is no realtionship between that fact, and nVidia's stance.

[OpenGL guy]

Originally Posted by Chris123234

thats not the only reason their 'moaning'.

BTW if youve noticed how the 9500's and above whoop the Gf4's its because of the weighting.

3DMark03 score =
(Game Test 1 frame-rate x 7.3) +
(Game Test 2 frame-rate x 37) +
(Game Test 3 frame-rate x 47.1) +
(Game Test 4 frame-rate x 38.7)

If you dont have DX9 capabilities then your score is gonna be LOW.

Note that the weightings are such that each test gives about an equal contribution. Test 2-4 give much lower numbers so they are scaled higher to give a more equal contribution. I mean, look at the difference between GT1 and GT4... If you're trying to measure future game performance, then you can't have GT1 outweighting all the others.

[Doomtrooper]

Nothing Nvidia stated holds any water at all because the same arguement can be used for the 3Dmark 2001...simple as that.

Doesn't reflect game peformance..Neither did 3Dmark 2001

Complaining about the use of flight sims..you have GOT to be kidding.

I won't even continue...it is marketing damage control at its best,

this sums it up nice, if kyle can do it, so can other websites:


http://www.gamersdepot.com/ed/geforcefx/001.htm

[Pete]

Ugh, I'd be careful quoting rants that include gems like this:

You’d think that building it on a .13-micron process would’ve allowed NVIDIA to run a much cooler chip, thus not requiring a card so big. If ATI can make a DX9 chip – which keeps pace with the NV30 – on a .15-micron, have it run cooler, than why can’t NVIDIA do better?

[Sharkfood]

I DID NOT SAY NVIDIA IS COMPLAINING BECAUSE OF THE WEIGHTING. I SAID BTW YOUR SCORES WILL BE LOW. READ THE FREAKING POST BEFORE YOU COMMENT ON IT.

Absolutely no relation was implied- sorry if I was misunderstood on that point.

By NVIDIA (original) commentary, I am referring to their original points concerning the applicability of single-textured tests (well formed), an alleged stenciling that causes unusual bottlenecks in the vertex engine (quite possibly valid, need sourcecode or more data to prove) and other legitimate concerns voiced. They all simply tie to a synthetic benchmark's similarities with actual games and any possible deviation these benchmarks may have in relation to games. This has been the case for all 5 generations of 3DMark, with nothing new or unusual with 3DMark03.

This differs from NVIDIA's (revised/more detailed) commentary concerning the applicability of flight sims as they are somehow not valid, or that developers will never choose PS1.4, but instead only opt for PS1.1 or PS2.0. (which is nothing but unfounded rubbish on all counts).

[Dave H]

GT4 is actually weighted less than the other three tests, at least on Futuremark's baseline platform, which presumably centers around a 9700 Pro. GT4 is weighted to provide 20% of the overall score on that baseline platform, with the other 3 tests good for 26.7% apiece.

OTOH, GT4 appears to be very bandwidth-invariant, so it makes up more like 25% of the 9500 Pro score. But still. The 9500 Pro gets about 900 points on a GF4 Ti4600 due to the ability to run GT4. But it gets about 1000 points on a GF4 from beating it on the other 3 tests.

[Reverend]

Originally Posted by Joe DeFuria

That's funny indeed! I'm amused at the fuss that nVidia is making over 3DMark2003, and even moreso at the web-sites (um, web-site?)that "coincidentally" share that identical stance with nVidia...

You have a different interest in the Internet participation of discussing about the entire video card technology than mine, Joe.

[JD]

These type of benchmarks come handy when you're comparing relative scores from previous runs like when changing computer components or when tuning up your computer w.r.t. other person with same config, etc.

[martrox]

This is all about nVidia's hypocricy. I doubt you will find ANYONE that has posted here for any lenth of time that will tell you they believe that 3DMark is/was THE reason to buy one videocard over another..... unlike nVidia which pused it BECAUSE it showed their cards in a very good light. Didi ATI complain about how unfair it was? Now that 3DMark shows ATI in a good light, nVidia has no use for it.

[Bjorn]

Didi ATI complain about how unfair it was?

Maybe they didn't think it was unfair.

[Hyp-X]

Originally Posted by martrox

Didi ATI complain about how unfair it was?

When the R8500 came out 3dMark 2001 was the only test they could beat the competition's card.
So why should they complain?

[martrox]

Maybe I should have said....did PVR complain?.......

[BoddoZerg]

Originally Posted by martrox

Maybe I should have said....did PVR complain?.......

PVR fans sure did...

[16Valve]

Must be "deja vu all over again" for poor Brian Burke. Wasn't too long ago that he was backpeddling for 3DFX.