Making sense out of MadOnion's results database

  • Thread starter Thread starter Rick S.
  • Start date Start date
R

Rick S.

I'm in the market for a high-end AGP card, and I'm confused by
results posted in the various 3DMark databases.

First I searched 3DMark2001SE, sorted by performance rating,
to get a feel for DX7 and 8 performance, and ATI's X850XL PE
is close to the top of the list, among other cards such as Nvidia
7800 GT's etc.

But then I searched through the 3DMark2006 database for DX9
results, and *all* of the first 15 pages of results are 7800's. The
highest performance ratings are just over 10,000, while the
X850 XL PE doesn't show up at all until results are under 5000.

So what's the deal? Is a X850 XL PE really half as fast in DX9
as the 7800GT? Or is this just a result of some benchmark
optimizations in Nvidia's driver?
 
it has long been rumoured that Mad Onioin has catered to Nvidia cards in
that benchmark..but I would go to sites like Toms Hardware and look at the
performance in games and such.
 
So what's the deal?
Click to expand...

3dMark gives you a big bonus score for PS3 support, which puts "old" ATI
cards at a substantial disadvantage. Whether PS3 is important or not yet
is a different matter, 3DMark is always about pushing the boundaries.

3DMark2005 is a better test for current games, though I find AquaMark3
more useful as a real world benchmark for current generation games.

Andrew McP
 
Check for "co-operative" adapters. That indicates an SLI (nVidia) or
Crossfire (ATI) system with two graphics cards.

That aside, there are some crazy numbers in the tables. The top performer in
3DMark06 claims to have overclocked a dual core Athlon 64 family CPU to over
3.4 GHz. (I believe that the FX60 is normally at 2.6 GHz.)

Also note that some of the 7800GTX cards are of the 512 MB variety, which
comes with a higher GPU clock than the lesser cards.

There are checkboxes in the list that permit you to search for only single
graphics card systems, but they're available only to paid subscribers.

My 7800GTX (256 MB version) gives about 4500 marks in 3DMark06 without
overclocking.

The fastest single 850XT is at approximately 2800. The fastest single nVidia
6800 ultra comes in it about 3500.

Perhaps the benchmark favors nVidia cards, but it's not all *that* lopsided.


Address scrambled. Replace nkbob with bobkn.
 
What operating system are you using?

I thought 3DMark2001SE doesn't work on anything higher then WindowsME.
 
I'm in the market for a high-end AGP card, and I'm confused by
results posted in the various 3DMark databases.

First I searched 3DMark2001SE, sorted by performance rating,
to get a feel for DX7 and 8 performance, and ATI's X850XL PE
is close to the top of the list, among other cards such as Nvidia
7800 GT's etc.

But then I searched through the 3DMark2006 database for DX9
results, and *all* of the first 15 pages of results are 7800's. The
highest performance ratings are just over 10,000, while the
X850 XL PE doesn't show up at all until results are under 5000.

So what's the deal? Is a X850 XL PE really half as fast in DX9
as the 7800GT? Or is this just a result of some benchmark
optimizations in Nvidia's driver?
Click to expand...

I think that the moral of the story is that such tests are to be taken
with an exceedingly large pinch of salt.

I do use these tests to compare changes to my own system. My scores
are exceedingly low, it would seem and this may be in part due to my
motherboard (L7S7A2) rather than my graphics card.

In terms of buying a new graphics card, you will pretty much get what
you pay for and that is as good a guide as many tests.
--

Julian Richards
computer "at" richardsuk.f9.co.uk

www.richardsuk.f9.co.uk
Website of "Robot Wars" middleweight "Broadsword IV"
 
Dazzer wrote on Mon, 23 Jan 2006 06:35:17 GMT:
What operating system are you using?

I thought 3DMark2001SE doesn't work on anything higher then WindowsME.
Click to expand...

Works fine on Win 2000 and XP.

Dan
 
Back
Top