my 9800 article

[ben6]

http://www.motherboards.org/articlesd/hardware-reviews/1240_1.html


It was a lot of fun. I just wish I had more time with the card. Ah well... Note I reran the tests several times and could not account for the results. As I only had a few hours with the card, and I had done a fresh install of the drivers , I ran with it.


Hopefully, ATI will get me a card out for a more detailed look. The screenshots rocked though


Ben

 

[Hellbinder]

I am sorry but i simply disagree that you should be posting benchmark results like this that you HAVE to know are innacurate.

They do not reflect any of the rest of the reviewers results. Anyone reading your review is going to get a severely skewed picture of the 9800pro.

I simply cannot accept that a veteran like yourself could not get the obvious performance/installation issues resolved. I am not saying anything other than its pretty damn peculiar.

 

[gkar1]

Agreed

Also i note the lack of consistency in his work. The unfinished image quality comparison between the Geforce4 and 9700, and now this. Pretty good track record eh?

 

[ben6]

Hi guys, appreciate the feedback.

Believe it or not, I agree with the points. But that is all I will say on the matter.

As to the 9700/Geforce 4 article, it was done a LONG time ago. Again that's all I will say on the matter.

 

[ben6]

I'm not sure what the issue is with performance. The scores were repeatable, even after resetting the settings and rebooting. It was strange.

 

[Althornin]

I happen to DETEST statistic stretching graphs like this one:

Anyone glancing at that sees what looks to be a 30% deficit!
Uh-oh, its really only ~7.5%!
Here is another shining example (and there are plenty more in the review). I think this type of misleading graph is wrong.

This one looks like a nice 50% difference, yet is only 10%

 

[Ichneumon]

ya, I think my major dislike about the review is the graphs not being normalized, so even minor differences in framerate/score look huge.

 

[ben6]

Hrm that's a good criticism

 

[Althornin]

Whats next:

I was compelled to whip this up, sorry

 

[ben6]

heh, it's been a issue before .

 

[Typedef Enum]

Ben,

I don't know what to say...but it seems like something is not quite right. I can't honestly say that I've seen a 9800 Pro underwhelm as much as it did in your article.

Heck, even when they were fairly close, the 9700 Pro would come out on top! I mean, that seems pretty inconceivable?

Oh yeah...you really should fix those graphs

 

[Fuz]

Hey Althornin, nice graph.
Actually made me laugh quite a bit.

 

[gkar1]

Althorin that was comedy gold. LOL

 

[demalion]

Nitpick: you said the GF FX was bigger than the 9800, but you said "chip" not "card".

Issue: it is up to ATI to specify the justification for listing "displacement mapping" as a feature the 9700/9800 have and the GF FX don't...you shouldn't repeat it without having that justification established to your satisfaction.
Given nvidia's refusal to, for example, enable n-patches or their native hardware HOS support in their drivers, it might be true for some basic DX 9 "fixed function" displacement mapping specification, but I haven't seen that established anywhere.

It annoyed me greatly when I saw it mentioned at driverheaven, and it still annoys me to see it repeated without an explanation. You shouldn't take any company purely at their word in their marketing material, even if it isn't as bad or slanted as another's, or repeat an item by itself unchallenged if the validation for it is not common knowledge.

EDIT: I'd be saying something about the graphs except it has already been said and you've accepted it (unsurprising behavior for you) and I'll just mention I share the sentiment expressed rather strongly in case hearing another voice would help cement it in your mind to nevereverever do it again.

 

[Nebuchadnezzar]

The Card
Hardware Features

Memory Speed (ns)
3.5

Shouldn't that be 2.5 ns?

 

[demalion]

I'm not sure what the issue is with performance. The scores were repeatable, even after resetting the settings and rebooting. It was strange.

Are you sure that the 6x settings were "taking" the same for each card?

Also, it could have made the questionable results useful if you'd had a discussion about verifying AGP settings, or maybe even an AGP benchmark (I know someone asked for one...did we find one?) in the context of sanity checking your results.

 

[ben6]

There is some question in my mind as to whether the settings were correctly set . Dronezmark, for example was all over the place with results and I had a hard time telling the difference ingame visually with AA on/off. The settings were correctly identified in the game, but there was no preset.

A combination of factors lead to this situation, not all of it ATI's or my fault.

1. I was scheduled to be briefed on 9800 Pro along with someone else by ATI PR . Suffice it to say, signals got a little crossed, which was why the article wasn't up on the review lift date. Worse, the person who was briefed, followed his agreements so therefore I got 9800 Pro information at a far later time than would have been prudent for a review scheduled to go up on the 6th (most of the information I had was from the previews posted on the 6th sigh). And, the requests I sent for more information from ATI weren't answered in a timely manner.

2. As I stated before, I had a very limited time with the card. The question being, should I have written an article without more time with the card, and time to troubleshoot? Well there was no hope for it. Believe it or not , the above situation contributed to this as well as to matter of lifting of the review date, and contributed to the lateness of the review (had everything worked out, the review should have been up on the same date as everyone else's and with more time with the card)

3. Unfortunately other things also got into the picture, which aren't germaine to the discussion here

Therefore, should I or shouldn't I have written the article? Given this situation, we did the best we could with what we had to work with.

Anyway, if the same situation presents itself in the future, I won't do it period.

 

[jb]

Therefore, should I or shouldn't I have written the article? Given this situation, we did the best we could with what we had to work with.

I see and understand your point. I guess what you can do is more of a tech review of the card. Some screen shots, maybe some pixel/vertex shaders scores or other scores that ephise the tech vrs the Common benchmark score and stuff like that??

 

[Reverend]

ben6, you probably should've stated all you stated above in your review (maybe at the end, as a small "disclaimer"). Believe me, it helps. Unless you didn't know all of these were "problems" before you read this thread

demalion, as much as we, as 3D enthusiasts/geeks, may be interested in HOS, it will not take off with the current generation of hardware, rendering it a worthless feature ATM. I would not use ATi's implementation nor NVIDIA's hidden implementation at all right now. With the current generation, it's pretty much DOA. I think this should be evident when viewing its support in current/upcoming games. It's just not worth the time/effort.

 

[Dave Baumann]

Is there an N-Patch test anywhere? I keep meaning to see if that was 'fixed' in R350 or not.