24 replies to this topic

[jammin]

We have just published a roundup of the video cards we have reviewed over the past months, which you can read here.

You may find some of the conclusions slightly surprising.

[Waco]

What system was this run on? I don't see anything listed.


EDIT: The systems weren't the same? Awwww...

[GlimmerMan]

You really need to add the system specs (or at least CPU/RAM) or the review is pretty much a void (no offense to your hard work testing that many video cards).

[jammin]

The performance figures are taken from each product review

- Review list

So most of the data uses very similar setups (Q9450 with a Gigabyte X48-DQ6 + the same RAM)
Some of the older reviews did use a Q6600 and X38 based board.

Edited by jammin, 23 September 2008 - 01:39 PM.

[Waco]

- Review list

So most of the data uses very similar setups (Q9450 with a Gigabyte X48-DQ6 + the same RAM)
Some of the older reviews did use a Q6600 and X38 based board.

I saw that. I don't want to draw any conclusions about the cards used because the Q6600 and the Q9450 are so different.

[Bosco]

I saw that. I don't want to draw any conclusions about the cards used because the Q6600 and the Q9450 are so different.

They are not that much different at all and in Gaming they are so close there is not much of benefit. 266Mhz Higher Clock, 1333Mhz front side bus, and 3MB of cache more in gaming means nothing.

[Waco]

266Mhz Higher Clock, 1333Mhz front side bus, and 3MB of cache more in gaming means nothing.

That's a 10% clock speed increase (not even counting other stuff) and that's more than enough to see a difference on the latest high-powered cards.

Certainly not a massive difference but when it comes down to it even 10% can be quite a bit.

[Bosco]

In most of the games we test the 6600 and 9450 score exactly the same

ccokeman can provide the updated numbers on the stuff that was tested on the older system.

Pretty easy to do since I think there was only 6 cards used on the Q6600.

[Jhester]

That GX2 is still crazy

[ccokeman]

Lets look at one example. Just for kicks I ran the XFX 9800GTX not the plus mind you with the 177.79 nvidia drivers on the X48 and Q9450 to test your theory.

Here is the crysis link in the original review. Same run same game same settings 2xAA medium settings. for comparison


My current results for those settings are as follows

1024===== 57
1280===== 45
1680===== 35
1920===== 27


As you can see the difference in FPS from one cpu and chipset to the next results in a O FPS gain and in fact a loss in at least one instance. For our newest reviews the cards have been rerun with new scores based on new drivers and our standard test setup.

[Bosco]

Samething happens when you pit a E8400 VS E8500. :thumbs-up:

[Fogel]

This is awesome, especially since I've pretty much come to the conclusion I want to buy a new video card. And I love how compared both the price AND performance. First time I seen that done. But the conclusion DOES surprise me, not because of the previous generation winning thing. But because of this quote...

So, things change once again when we look at performance and value at the same time. At stock speeds, the Palit 9800GX2 is the clear winner, with all the HD4850-based cards nipping at its heels.

Cranking up the clocks we see the Sapphire HD4850 taking the lead, with the Palit 9800GX2 falling to 2nd place, again followed by the HD4850 cards.

Since everything is presented via graphical representation all I have to go from is those graphs and the card mentioned in the conclusion were neither the lowest or the highest - they were middle of the road - so no matter if "lower is better" or if "higher is better" ...I still don't see how you arrived at that conclusion.

Really digging the study and i love the direction it took, would like to see that added more often. One thing I was hoping to see in a recent review is a review of a 260 with more processing cores to see if the processing core upgrade makes much of a difference. Was talking with Bowtie and we think even the starting clock speed is less than the EVGA FTW version, if you OCed this bad boy what kind of difference would we see in performance:

EVGA - 260 w/ 216 Processing Cores
EVGA - 260 w/ 216 Processing Cores - clocked lower