Anyone else underwhelmed with their 4870 crossfire?

sgrinavi

Diamond Member
Jul 31, 2007
4,537
0
76
I have not played a bunch of games, just run some benchies, but I was expecting the 4870 CF to be more than 1% better than the 4850 CF

19554 3DMarks 4870 xfire
Aug 21, 2008 20:41 CDT
CPU:Intel Core 2 Duo E8400
GPU:ATI Radeon HD 4870
OS:Microsoft Windows Vista

19327 3DMarks 4850 xfire - e8400
Jul 14, 2008 21:41 CDT
CPU:Intel Core 2 Duo E8400
GPU:ATI Radeon HD 4850
OS:Microsoft Windows Server 2003

certainly not worth an extra $200...

What am I missing here?
 

IcePickFreak

Platinum Member
Jul 12, 2007
2,428
9
81
Originally posted by: sgrinavi
What am I missing here?

The resolution and settings you tested at? At the default settings that looks about right honestly. I also noticed two different OS's; 3DMark scores change with wind direction. I wouldn't be giving it that much weight to begin with, play some actual games and crank the res/AA/AF up.

 

OCGuy

Lifer
Jul 12, 2000
27,227
36
91
Originally posted by: sgrinavi
I have not played a bunch of games, just run some benchies, but I was expecting the 4870 CF to be more than 1% better than the 4850 CF

19554 3DMarks 4870 xfire
Aug 21, 2008 20:41 CDT
CPU:Intel Core 2 Duo E8400
GPU:ATI Radeon HD 4870
OS:Microsoft Windows Vista

19327 3DMarks 4850 xfire - e8400
Jul 14, 2008 21:41 CDT
CPU:Intel Core 2 Duo E8400
GPU:ATI Radeon HD 4850
OS:Microsoft Windows Server 2003

certainly not worth an extra $200...

What am I missing here?


GT200

(hides behind wall)
 

chrismr

Member
Feb 8, 2007
176
0
0
I was rather underwhelmed with 2 x 4850 in crossfire over a single 4870, and looking at reviews and benchmarks the 4870 crossfire not very far ahead of the 4850's, so I imagine I would be disappointed.

It's not that they were slow/unstable/bad.

The highs were a fair bit higher than the single 4870, but the lows were not much better (if any). I would want crossfire for a smoother game experience, which it did not offer me.

Though I could see the benefits to a higher res, whereas I am on 1680x1050
 

BFG10K

Lifer
Aug 14, 2000
22,672
2,817
126
Even before I clicked this topic I just knew the premise was going to be founded on 3DMark.

I feel sorry for people that base their upgrade decisions on meaningless synthetic benchmarks.
 

sgrinavi

Diamond Member
Jul 31, 2007
4,537
0
76
Originally posted by: BFG10K
Even before I clicked this topic I just knew the premise was going to be founded on 3DMark.

I feel sorry for people that base their upgrade decisions on meaningless synthetic benchmarks.

You feel sorry for people? lol.

Well, save it for someone else. I do not base my upgrades on benchmarks, I just use them as a tool to compare -- that's what they are for, right? In this case I would have expected more than a 1% increase in that particular app, wouldn't you?

Or, perhaps, I'm CPU bound at this point?
 

AzN

Banned
Nov 26, 2001
4,112
2
0
Nearly 20000 3dmark and you are not satisfied? You might need a GTX6000.
 

chizow

Diamond Member
Jun 26, 2001
9,537
2
0
At 1280 with current parts 3DMark06 is certainly showing its age and is severely CPU bottlenecked. Depending on your resolution you still may not see much difference in FPS, you'll just get more "free AA".
 

Hugh H

Senior member
Jul 11, 2008
315
0
0
Originally posted by: lavaheadache
Ever hear of cpu limitation? Cause that is most certainly wants going on in 3dmark

+1

I've found out the hard way that I'm CPU limited in certain situations with an E8500 @ 4.00Ghz. 3dMark seems to be one of those apps that benefits from 4 cores.

 

BFG10K

Lifer
Aug 14, 2000
22,672
2,817
126
Originally posted by: sgrinavi

You feel sorry for people? lol.
Yep. I mean look at what you?re doing - instead of enjoying your rig in real games with what is a clear upgrade over your previous kit, you're here, scratching your head over worthless synthetic benchmarks.

I just use them as a tool to compare -- that's what they are for, right?
Compare what? What do 400 marks mean to gaming? And how many marks are required before the upgrade is worth the $200 you put down?

Tell me, do you buy hardware to play games or to get as many marks as possible?
 

lavaheadache

Diamond Member
Jan 28, 2005
6,893
14
81
I "upgraded" from 2 512 8800GT's.... Sure my 3dmark score went down but Now I can play with 8xAA with playable fps on most games, where as the GT's were only really good at 4xAA with playable fps... so to me this is an upgrade. Also I don't have to deal with sli, which sometimes is a pita and sometimes not
 

sgrinavi

Diamond Member
Jul 31, 2007
4,537
0
76
Yep. I mean look at what you?re doing - instead of enjoying your rig in real games with what is a clear upgrade over your previous kit, you're here, scratching your head over worthless synthetic benchmarks.

Again, it is not my primary focus, just a tool to compare parts.

Compare what? What do 400 marks mean to gaming? And how many marks are required before the upgrade is worth the $200 you put down?

Last time I checked benchmarking was a valid method of comparing two sets of hardware. Pehaps, as pointed out above, I am CPU limited and this is not a valid way to test, but it has served me well up untill now.

Tell me, do you buy hardware to play games or to get as many marks as possible?

I don't understand why this is giving you heartburn, it has nothing to do with anything but a simple way to compare hardware. Did you notice that in the original post I asked if I was missing something here.....

 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Originally posted by: sgrinavi
I have not played a bunch of games, just run some benchies, but I was expecting the 4870 CF to be more than 1% better than the 4850 CF

19554 3DMarks 4870 xfire
Aug 21, 2008 20:41 CDT
CPU:Intel Core 2 Duo E8400
GPU:ATI Radeon HD 4870
OS:Microsoft Windows Vista

19327 3DMarks 4850 xfire - e8400
Jul 14, 2008 21:41 CDT
CPU:Intel Core 2 Duo E8400
GPU:ATI Radeon HD 4850
OS:Microsoft Windows Server 2003

certainly not worth an extra $200...

What am I missing here?

if you have the full/pro version of 3DMark06, up the resolution from the default 12x10 :p

try at least 16x12

and add 4xAA/16xAF

then report back
rose.gif


 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
You can just force 8AA in the control panel and see if this makes a difference. In general though 4870CF isn't going to give you much faster performance than already acceptable 4850 in CF until you scale beyond 1920x1200 with 4AA where the 4870s bandwidth advantage comes into play. Even still this depends on the benchmark - 4870 CF vs. 4850 CF

Using 3dMark 06 is POINTLESS simply because 4870CF is meant for 30inch monitors, not 12x10 resolutions with no AA.

You should overclock your CPU as well to get the maximum minimum frames out of this setup.
 

sgrinavi

Diamond Member
Jul 31, 2007
4,537
0
76
Originally posted by: Azn
Nearly 20000 3dmark and you are not satisfied? You might need a GTX6000.

My lack of satifaction is based on the difference between the performance of the 4850CF and the 4870CF.. Nothing more.

Besides, I was thinking about going with 6870x4
 

LOUISSSSS

Diamond Member
Dec 5, 2005
8,771
54
91
its because you're running the 4850 CF with a quad core.

CPU power is a strong determinant in 3dmark06 score...

those are both very good scores and u MUST MUST play some real games. i haven't installed 3dmark in the past year and gave up benchmarking.
 

MarcVenice

Moderator Emeritus <br>
Apr 2, 2007
5,664
0
0
This is like playing commander keen in 320*240, with tri-sli GTX280's, and then asking why am I only getting 10 million fps and not 100 million -_-

Ok, so it's not exactly the same, but it's a pretty weird question nonetheless :p
 

sgrinavi

Diamond Member
Jul 31, 2007
4,537
0
76
Originally posted by: apoppin

What am I missing here?

if you have the full/pro version of 3DMark06, up the resolution from the default 12x10 :p

try at least 16x12

and add 4xAA/16xAF

then report back
rose.gif

Thanks, excellent idea.
 

sgrinavi

Diamond Member
Jul 31, 2007
4,537
0
76
Originally posted by: LOUISSSSS
its because you're running the 4850 CF with a quad core.

CPU power is a strong determinant in 3dmark06 score...

those are both very good scores and u MUST MUST play some real games. i haven't installed 3dmark in the past year and gave up benchmarking.

No, the test was done with the same PSU at the same 4.05 clock.
 

HOOfan 1

Platinum Member
Sep 2, 2007
2,337
15
81
Originally posted by: sgrinavi
Last time I checked benchmarking was a valid method of comparing two sets of hardware. Pehaps, as pointed out above, I am CPU limited and this is not a valid way to test, but it has served me well up untill now.

No, not really, if that benchmark is 3dMark. 3Dmark scores don't always translate into gaming performance.

If I remember right the 2900XT scored higher in 3Dmark than the 8800GTX, yet the 8800GTX would wipe the floor with the 2900XT in games. So no 3dMark benchmarking only is not really a valid method for comparing hardware.

No more than comparing the dynomometer results of a 600HP Mercedes AMG to a 480HP Porsche Turbo and thinking that determines who has better on track performance.
 

sgrinavi

Diamond Member
Jul 31, 2007
4,537
0
76
Originally posted by: HOOfan 1
Originally posted by: sgrinavi
Last time I checked benchmarking was a valid method of comparing two sets of hardware. Pehaps, as pointed out above, I am CPU limited and this is not a valid way to test, but it has served me well up untill now.

No, not really, if that benchmark is 3dMark. 3Dmark scores don't always translate into gaming performance.

If I remember right the 2900XT scored higher in 3Dmark than the 8800GTX, yet the 8800GTX would wipe the floor with the 2900XT in games. So no 3dMark benchmarking only is not really a valid method for comparing hardware.

No more than comparing the dynomometer results of a 600HP Mercedes AMG to a 480HP Porsche Turbo and thinking that determines who has better on track performance.


I was comparing two sets of video cards from the same generation and same manufacturer using the same processor at that same speed. Are you telling me that it is not a valid comparison?

 

HOOfan 1

Platinum Member
Sep 2, 2007
2,337
15
81
Originally posted by: sgrinavi
Are you telling me that it is not a valid comparison?

I suppose it is, if getting the best 3dMark score is what you care about.

If you want someone to agree with you that you wasted another $200 then try posting several game benchmarks.