Upgraded my 8800GT to a HD4850. More a downgrade !

[Sir_Real]

I recently swopped a gainward 8800GT 512mb for a gainward HD4850 512mb.
When i did the swop over (No changes to the rig & no overclocking on either gpu) I was surprised to see a slight drop in score on both 3dmark 2003 & 2006. Not that i'm really bothered like

But just av a look at the old school 3dmark2001se results.

You would of thought the HD4850 would kick ass over a 3 yr old 8800GT. Some surprisingly big differences in some tests.

 

[nafets]

Play games. You'll see a difference. 3DMark is just for maximizing e-peen...

 

[AltecV1]

my rig with a HD4850 gets 12 000 poinds in 3Dmark06 and P6638 in vantage(no overclock)!

 

[RX-7]

Also you may be limiting it with your processor a bit

 

[kyle2020]

Synthetic benchmarks mean f*ck all tbh. ATI cards tend to scale much better in AA drenched environments, so go play some games, crank the AA and see the GT struggle to keep up.

 

[qubit]

nvidia is better than ati

+1 to nafets and also I'm not surprised, as nvidia is better than ati (sorry if I've offended any ati fanboys there).

A while back, I had a 512MB HD4870. Sometime later I got an 8800GTX for about £60.

Now, if you've seen the benchmarks comparisons on TPU, you'll see that the 4870 spanks the 8800. However, in real-world gameplay the difference isn't nearly so marked and in some cases the 8800 beat it. And that's what, a card that's two generations newer than the 2900 that was supposed to compete with the 8800? The 8800 also runs with less software glitches, too.

I've now got a GTX 285 and haven't looked back. The 4870 ended up on eBay...

I recently got 3D Vision and that's really awesome. Where's ati with a competing product?

 

[r9]

Make clean OS install or driver sweep. And to see the difference you will need to bench on higher resolution than 1024x768.

 

[D3M0N-G4M3R]

I dunno i have to disagree with the 4870 beating the gtx260, i use a 4870 but theyre both pretty much equal i think. there aint no big gap in power betweeen them

 

[AltecV1]

I dunno i have to disagree with the 4870 beating the gtx260, i use a 4870 but theyre both pretty much equal i think. there aint no big gap in power betweeen them

i agree with you they both have the same performance

 

[kylzer]

Im talking about 192 not 216 BTW

 

[Darren]

Sir_Real,

Its not quiet as simple as switching the cards, 3Dmarks can be very temperamental, for example, when I went from my ATI x1600 Pro to Nvidia 9600 GT initially the performance increase was small and my 3Dmarks 06 score didn't change much, it was only when I did a fresh format and reinstall of the OS my scores almost tripled.

Also my 9600 GT was heavily overclocked, so its performance was about the same speed as a 8800 GT and it scored slightly lower than my current 4830 at stock in 3Dmarks 06, considering that you've got the 4870 the difference in score should be greater.

Also consider that Nvidia cards have an advantage in the "CPU" test due to its physics support and hence slower Nvidia cards tend to achieve better overall scores in 3D Marks. But in your situation I would ignore synthetically benchmarks generally, but typically the 4870 pisses on top of the 8800 GT, do not idiots like "Qubit" tell you differently you only need to look at the in realworld benchmarks online to disprove his nonsense.

1.) format your PC
2.) reinstall windows
3.) tweak the number of background processes
4.) disable junk like system restore
5.) install your latest ATI drivers
6.) you know the rest.

A while back, I had a 512MB HD4870. Sometime later I got an 8800GTX for about £60.

Now, if you've seen the benchmarks comparisons on TPU, you'll see that the 4870 spanks the 8800. However, in real-world game play the difference isn't nearly so marked and in some cases the 8800 beat it.

You are chatting rubbish. the 8800 GTX and 8800 GT are two different cards, indeed the 8800 GTX might be able to come closer to the 4870s performance but what has that got to do with the 8800 GT?

The 8800 has 112 shader units,
The 8800 GTX has 128 shader units

The 8800 GT has 16 ROPs
The 8800 GTX has 24 ROPs

The 8800 GT has a 256 bit memory bus
The 8800 GTX has a 384 bit memory bus

As you can see the GTX is alot more powerful than the GT so saying that "you are not surprised" illogical, although the 4870 is faster than both cards its only normal that the GTX would be nearer to the performance than the GT.

And that's what, a card that's two generations newer than the 2900 that was supposed to compete with the 8800? The 8800 also runs with less software glitches, too.

You are talking rubbish again, not all 8800s are the same, as I already pointed above the GTX is has a much higher shader and ROP count than the GT. Both the GTX and GT are slower than the 4870. A mere 4730 or 4830 would abolish the 8800 GT.

Edit:

I wish Qubit would unregister his account or for W1zzard to give him a ban for being a fan boy and persistent troll.

 

[Sir_Real]

1.) format your PC
2.) reinstall windows
3.) tweak the number of background processes
4.) disable junk like system restore
5.) install your latest ATI drivers
6.) you know the rest.

There really is no need 4 me to format & reinstall the OS. Its not in old install & my pc is kept clean of all junk & clutter. No crap running in the background !

I carnt agree the HD4850 is alot better than 8800GT. They seem to close in preformance for that ! However i should say a HD4870 would pull well ahead on a 8800GT.

I'm not putting the HD4850 down its is a cracking card for the price tag. Just i think it shows wot a good job NV did with the 8800GT & GTX.

 

[Cheeseball]

The HD 4850 is not a vast jump from a 8800GT, but the difference is still noticable in anti-aliasing and higher resolutions.

 

[newtekie1]

You shouldn't be seeing a performance drop in either benchmark going from the 8800GT to an HD4850. However, it isn't impossible. Do a completely clean of the old nVidia drivers, and re-install the ATi drivers. And play some games, and see how it does.

And the HD4850 isn't really an upgrade from the 8800GT, overall you are only looking at about a 10-12% increase overall.

And while we are on the subject, the HD4870 definitely doesn't beat the GTX260, not even the 192SP version. See Here.

 

[aCid888*]

The order of cards goes:

8800GTX < 8800GT/4830 < 9800GTX/4850 < 4870/GTX260.


Anyone who says the 8800GTX is faster in anything should be beaten with their own keyboard...the 9800GTX/4850 spank the older GTX in games and 3dMark06..end of.



Benchmarks mean nothing either as the ATI cards dont bench as good as the nVidia ones...its called marketing so when a n00b sees "A" card scoring 18000 and "B" card scoring 16000 they will buy the one with the best score even if in real-world gaming its actually slower. :shadedshu


Edit:

And while we are on the subject, the HD4870 definitely doesn't beat the GTX260, not even the 192SP version. See Here.

Apart from thats a 4730 and not a 4870.

This is a 4870:

 

[newtekie1]

1.) The 8800GTX and 9800GTX perform almost identical overall. That was one of the major complaints when the 9800GTX came out, that it was actually slightly slower than the 8800GTX.

2.) Benchmarks definitely give a very good indication of real world performance. Anyone that says otherwise is a fool. Though I will put a lot more weight in an actual game benchmark than something like 3Dmark.

Edit:


Apart from thats a 4730 and not a 4870.

This is a 4870:
http://img.techpowerup.org/090805/Capture018.jpg

Look a little bit lower on the graph, get past the highlighted portion, I know it is easy to get drawn in by the pretty colors...

And the problem with the graph you posted was the the HD4870 was using newest drivers while the GTX260 was not, read the review.

 

[aCid888*]

So my 4870 scoring 16500 in 06 and my 9800GTX+ scoring 17500 is an indication of performance is it? :shadedshu


Yeah, OK.

 

[Cheeseball]

You got it slightly wrong.

8800GT/HD4830 < 8800GTX < 9800GTX/4850 < 4870/GTX260

The 8800GTX has more shaders and a 768bit GDDR3 bus. If it was upgraded to G92 architecture, it would most like surpass the 9800GTX+/GTS 250.

 

[aCid888*]

Look a little bit lower on the graph, get past the highlighted portion, I know it is easy to get drawn in by the pretty colors...

And the problem with the graph you posted was the the HD4870 was using newest drivers while the GTX260 was not, read the review.

And with the new magic drivers nVidia pulled from their ass, the GTX260 gained at least 47.3% over the 4870. :shadedshu


Both have newer drivers, want to compare them now then I'm sure you will see the same results.

 

[jamesrt2004]

So my 4870 scoring 16500 in 06 and my 9800GTX+ scoring 17500 is an indication of performance is it? :shadedshu


Yeah, OK.

+1

ignore the troll... let him be ignorant to the real world -sigh-

 

[newtekie1]

And with the new magic drivers nVidia pulled from their ass, the GTX260 gained at least 47.3% over the 4870. :shadedshu


Both have newer drivers, want to compare them now then I'm sure you will see the same results.

I did compare them now, which is exactly why I used the latest review for a basis of my comments. If you want to dwell on 5 months ago, thats fine, but the situation today doesn't back up your statement.

So my 4870 scoring 16500 in 06 and my 9800GTX+ scoring 17500 is an indication of performance is it? :shadedshu


Yeah, OK.

Your benchmark scores mean nothing because they were not done in a controlled environment, or probably not even in the same machine.:shadedshu

Look at a real reviewers 3dMark scores, and they will be a lot closer. Completely accurate? No, definitely not, which is why I said I would put more weight in a real game benchmarks.

 

[Mussels]

3dmark is useless. and i mean that 100% seriously - there is NO use comparing seperate cards with these programs.


Test in games - you'll see a difference. (Not a huge one, since there isn't too much of a performance difference from an 8800GT to a 4870, but 30% or so at higher res, maybe higher with AA)

 

[aCid888*]

I'm sorry if I offended your bias view towards the GTX260, I didn't mean to hurt your feelings.

That being said, its obvious to anyone and everyone that the 4870 and 260 are neck and neck....one wins by a few FPS in one game, the other wins in another game.

I need not say anymore about this as its pretty well documented about this already and it isnt helping the OP.

 

[Mussels]

I'm sorry if I offended your bias view towards the GTX260, I didn't mean to hurt your feelings.

That being said, its obvious to anyone and everyone that the 4870 and 260 are neck and neck....one wins by a few FPS in one game, the other wins in another game.

I need not say anymore about this as its pretty well documented about this already and it isnt helping the OP.

the cards are completely equal in gaming. and i do mean equal - at stock speeds they're a dead heat*, and while the 260 tends to OC better it also costs more.


The OP just made a mistake in trusting 3dmark. that is all. you cant compare nvidia to ATI in those programs, because nvidia cheat a lot more in them than ATI do.


*excluding the games that are biased towards Nvidia or ATI

 

[newtekie1]

I'm sorry if I offended your bias view towards the GTX260, I didn't mean to hurt your feelings.

That being said, its obvious to anyone and everyone that the 4870 and 260 are neck and neck....one wins by a few FPS in one game, the other wins in another game.

I need not say anymore about this as its pretty well documented about this already and it isnt helping the OP.

I have a BIAS towards the truth, and that is all. What you are saying simply isn't true today. The GTX260, even 192SP versions, outperform the HD4870. Wiz's latest review proves that. They are very close, and given the option I would take the cheaper HD4870, because the difference wouldn't be noticeable, but it is there. When you break it down statisitcally, and not just "one wins sometimes, the other wins sometimes", the GTX260 comes out ahead.