AMD Radeon HD 6850 CrossFire

[Wing_Zero]

Originally Posted by bear jesus

When i was reading through the review and looking at the frame rate increases from one to two cards i was expecting the kind of numbers that are shown on the 460 chart you just linked to.

the real number are much better than the 460 SLI
if you cut out bench and 2 or 3 games that are not well supported: it will read an AVERAGE 90% of increase

[Bhairava]

I'm sure W1zzard, who is respectable and reliable person, and one of the best reviewers we have, will soon explain the reason why he used those numbers.

[bear jesus]

Originally Posted by Bhairava

I'm sure W1zzard, who is respectable and reliable person, and one of the best reviewers we have, will soon explain the reason why he used those numbers.

I'm sure w1zzard will be able to explain, that's one of the reasons i didn't put the effort into trying to work it out for myself as i'm sure i would have just missed something... plus i hate math but only as i suck at it

But right now all i can do is look at and think about the numbers untill my 6870's arrive during the week and then i won't care about the numbers... i will care about my eyefinity and crossfire problems

[Wing_Zero]

Originally Posted by Bhairava

I'm sure W1zzard, who is respectable and reliable person, and one of the best reviewers we have, will soon explain the reason why he used those numbers.

I'm sure too
My purposose was only to show a possibile error in the result summary, to let Wizzard correct it

[pantherx12]

Wing zero are the 6870 scaling graphs correct?

[Wing_Zero]

Originally Posted by pantherx12

Wing zero are the 6870 scaling graphs correct?

If I had to be honest, I don't like to waste my time to re-do the math also for 6870 crossfire.
Sorry, that will take time.

Otherwise, they are very stupid calcs, I'm sure that you can calculate the average scaling increase by yourself

[bear jesus]

Originally Posted by pantherx12

Wing zero are the 6870 scaling graphs correct?

I had mainly been looking at the 6870 numbers and i really think they are off as well, they seam to have as many games that scale between 90% and 100% yet have lower scaling percentage in the end of the rerview than the 6850 does.

I wish w1zzard would come set us straight

[forces]

where is power consumption and temperatures? a great review has always power consumption and temperatures, if it doesn't it's just a "review" without "great"...

[pantherx12]

Originally Posted by Wing_Zero

If I had to be honest, I don't like to waste my time to re-do the math also for 6870 crossfire.
Sorry, that will take time.

Otherwise, they are very stupid calcs, I'm sure that you can calculate the average scaling increase by yourself

I could, hell I wouldn't even need to do it myself Computers do things like this for me now XD

Will try.

[Wing_Zero]

I've sent a pm to w1zzard

Waiting for a public answer

[W1zzard]

yes, something seems wrong with the calculation but i'm still looking into where the problem is with my script that generates these numbers

[Wing_Zero]

Originally Posted by W1zzard

yes, something seems wrong with the calculation but i'm still looking into where the problem is with my script that generates these numbers

Tnx for the fast answer!

I'l try to show you where is the error:

If whe have these values:

A: is the fps number on single card

B: is the fps number in Crossfire

The right calc is this:

Pecentage Scaling performance from A to B is: [ (B-A)/A ] * 100

In example:

A= 60 fps (single card)

B= 115 fps (Crossfire cards)

Scaling performance is: [ (115-60)/60 ] *100 ---> this is equal to : +91,66%

I think instead that you've done [ (B-A)/B] * 100 (instead of A)

[HTC]

Originally Posted by Wing_Zero

Tnx for the fast answer!

I'l try to show you where is the error:

If whe have these values:

A: is the fps number on single card

B: is the fps number in Crossfire

The right calc is this:

Pecentage Scaling performance from A to B is: [ (B-A)/A ] * 100

In example:

A= 60 fps (single card)

B= 115 fps (Crossfire cards)

Scaling performance is: [ (115-60)/60 ] *100 ---> this is equal to : +91,66%

I think instead that you've done [ (B-A)/B] * 100 (instead of A)

Maybe my math skills suck because i divided the biggest number by the smallest one * 100 - 100.

Take a look @ post # 86 of the 6870 crossfire numbers: AMD Radeon HD 6870 CrossFire

@ the time, i didn't include the 3DMark benches as well as Unigine bench (editing and including them now). Did i forget anything else? Maybe we're missing something in our calculations.

[Wing_Zero]

Originally Posted by HTC

Maybe my math skills suck because i divided the biggest number by the smallest one * 100 - 100.

Take a look @ post # 86 of the 6870 crossfire numbers: AMD Radeon HD 6870 CrossFire

@ the time, i didn't include the 3DMark benches as well as Unigine bench (editing and including them now). Did i forget anything else? Maybe we're missing something in our calculations.

oh yes, that is the same.
You can also do B/A (where B is the crossfire fps, and A is the single card fps) and then you'll have a number like like "1,XX" where XX is the percentage of increase

look:

B= 115 fps
A= 60 fps

B/A= 1,91 --> 91% of performance increase (if you had as result 2.00 it means a 100% of performance increase). the you moltiply for 100 = 191 and you cut off 100 --> 91.

It is the same thing i had described in my post ( B/A *100 ) -100 IS MATHEMATICALLY THE SAME OF [(B-A)/A] *100 .

Guys, after all that is only a simple operation. I can't believe at all the problems that this simple op created ^^

[Wing_Zero]

SO HTC, you've done for the 6870 the same thing i've done for the 6850 Good!

So we can say that the scaling summary for both reviews are wrong :/

[HTC]

Originally Posted by Wing_Zero

SO HTC, you've done for the 6870 the same thing i've done for the 6850 Good!

So we can say that the scaling summary for both reviews are wrong :/

Seems so, hehe

EDIT

Question is: is it wrong for only these reviews?

Assuming the same script was used for other scaling reviews (ATI and nVidia), this means all those reviews are wrong as well, no?

[Wing_Zero]

Originally Posted by HTC

Seems so, hehe

EDIT

Question is: is it wrong for only these reviews?

Assuming the same script was used for other scaling reviews (ATI and nVidia), this means all those reviews are wrong as well, no?

...it is strange but for the NVIDIA-side, they are correct.

EDIT: W1zzard, I hope that you'll update it very soon...there many people I know that are having a bad idea of 6850 & 6870 crossfire, simply reading the result summary.

[pantherx12]

Several people including myself have posted correct scaling for 6870 on the crossfire review comments.

HTC
AMD Radeon HD 6870 CrossFire

Me : no unigine BM or 3dmarks AMD Radeon HD 6870 CrossFire

Including everything
AMD Radeon HD 6870 CrossFire




I see that HTC has already said anyway, cool!