ATI HD 4850 v/s Nvidia GTX 260m real game/demo benchmark comparison

There has been a lot of discussion about the relative power of the ATI Mobility Radeon HD 4850 and the Nvidia GTX 2xx series. I have been trying to find some apples to apples comparisons of the cards for a while.

Anothergeek and I ran some tests in the GT725 owners lounge and came to the conclusion that the GTX 280m is on the whole better (not surprising). The question remained, however, about the comparison between the 4850 and the 260m... and depending on who you asked around here, you could get three different answers.

So, when laptopNut and htwingnut started their CPU comparison using the Sager 8662 I though it was the perfect chance to compare. I ran many of the tests that they used and will post here the comparison to htwingnut's dual-core tests.

All tests are with stock-clocked GPUs at 1680x1050. The GTX 260m is paired with a P9700 @2.8 GHz. The 4850 is paired with a P8600 @2.88 GHz. The extra 80 MHz on my system and the extra 3 MB of L2 cache on htwingnut's should somewhat cancel, so the CPU influence will hopefully be negligible. Now to the results...

World In Conflict Demo benchmark
GTX 260m Very High - 22 High - 30
HD 4850 Very High - 23 High - 35

Crysis
GTX 260m Very High - 15 High - 26
HD 4850 Very High - 15 High - 26

GTA IV
GTX 260m High - 38
HD 4850 High - 33

Far Cry 2 - DX10 - Ranch Small
GTX 260m Ultra High - 34 Very High - 40 High - 42
HD 4850 Ultra High - 33 Very High - 42 High - 43

Far Cry 2 - DX10 - Ranch Long
GTX 260m Ultra High - 37 Very High - 45 High - 48
HD 4850 Ultra High - 40 Very High - 49 High - 51

Far Cry 2 - DX9 - Ranch Small
GTX 260m Ultra High - 29 Very High - 34 High - 43
HD 4850 Ultra High - 25 Very High - 36 High - 47

Far Cry 2 - DX9 - Ranch Long
GTX 260m Ultra High - 32 Very High - 37 High - 50
HD 4850 Ultra High - 31 Very High - 42 High - 54

ARMA 2 Demo Benchmark
GTX 260m Very High - 21 High - 25
HD 4850 Very High - 21 High - 25

The Last Remnant Demo Benchmark
GTX 260m 57.6
HD 4850 52.34

Resident Evil 5 Demo Benchmark
GTX 260m Variable - 49 Fixed - 45
HD 4850 Variable - 53 Fixed - 43

Amazingly similar actually. The GTA IV difference is probably attributed to Graphic Memory, since the 512 MB 4850 requires the -norestrictions prompt to run GTA IV on high. Benchmark results showed full graphic memory usage on my card and only 76% on the GTX 260m.

Settings for benchmarks can be found in htwingnut's original thread here http://forum.notebookreview.com/showthread.php?t=415786

 

thanks, very informative, + rep

 

Wow, that's really good. Thank you very much.

+ Rep as well.

 

didnt expect the results to be THIS similar actually.

good post man..+rep for ya

 

It's still imperfect, though. Saying that 80 MHz cancels addition 3MB L2 is... I don't know. It could be right, but I'm sure one chip will have the advantage over the other, and thus skew the results. But it's good to see those two chips competing competitively.

Oh, and I'm wondering, where the driver's optimized for both those machines? If, for example, the ATI had those modded or whatever drivers, and the 260m and 179.xx drivers, the scores would be screwed up. (I'm assuming they're optimized, since Geek and Neil aren't dummies, but yeah.)

 

Well, youre right, and the 80mhz wont really cover the 3MB of cache, and you cant really compare clock speed to cache, but at the same time as long as neither CPU was a bottleneck, then the test is quite accurate.

 

True. But off the top of my head, I know GTA4 is CPU dependent, and thus it could be skewed one way or the other. We could always just ignore that particular benchmark, though.

 

Thats what I was thinking too, and actually, at first glance one may attribute the increase in cache to the fact that the GTX 260m did better than the HD 4850m. This is a case where youre right, since like you said, it isnt perfect. Ignoring it is an option, but so is looking at the broad overall picture. And if the GTX 260m has a higher average FPS, then they can benchmark H.A.W.X as well.

 

Bottom line though, those results are very simular, and in my opinion close enough to conclude that these cards are near to neck and neck in performance, great comparison, really informative.

 

Great thread, I am amazed at how well the HD4850 keeps up with the GTX260. They are basically the same card.

I am more of an ATi fanboy myself, though I use both competitors alike. So yeah, I wish my G51 had an HD4800 series GPU

 

For the sake of better drivers and easier overclocking I'd stick with the Nvidia.

 

MMmm Yummy ATI 4850 goodness.. om nom nom +rep nice to know how this card fares with the competition

 

Hmm so could the 4850's memory be bottlenecking on GTA4? Maybe when someone gets hold of the 1gb model, that would get a bit more out of it.
And the CPU cache theory is definately accurate, the more the merrier for that game.

 

Great to see more threads like this. The results are surprisingly close.

 

+Rep. Good comparison and interesting to see.

 

Thanks for the positive feedback guys. I know the different CPUs aren't ideal, and I would have preferred identical CPU's, but frankly I got tired of waiting for someone with a P8600 to run a bunch of game/demo benchmarks.

One thing I noticed both before and during this set of benchmarks is that WIC seems to favor ATI (or favor Nvidia less?). It has been consistently one of the stronger benchmarks, even against systems with more powerful CPUs.

 

GTX 260m was running 186.81 WHQL

HD 4850 was on either 9.7 or 9.9 (RE5 needed 9.9)

 

+1 Rep for you here. Another good test I wish somone would try would be to compare idling and running temperatures, so we can see which one runs cooler. God Bless

 

If you're comparing the Sager 8662 with the MSI GT725 the Sager runs a lot cooler with more OCing headroom. Asus 260's are a different story.

 

Fantastic post! I expected the 4850 to edge out the 260M a little more.

 

Overclocked the 4850 beats the 260



the 4850 has scored OVER 7K in Vantage

 

You really would need a 1GB version of the HD4850 vs GTX 260m. A doubling of the memory is going to show an improvement in most games at 1680 x 1050. That is the point where it starts to make a difference. It should also be running the same processor, memory configuration, and FSB. Differences in on-die cache are most readily seen in games. The 3MB vs 6MB could be 10% or more FPS difference. As it stands, the test seems to be fairly biased against the HD4850.

 

Awesome thanks for comparison. This is some great info for anyone considering a Core 2 Quad, Core 2 Duo, GTX 260m, or HD 4850. At least now we have some hard real numbers to compare. +rep!

 

this is a pretty good comparison on the cards. Though the cpu's used are probably not all the same. But it still pretty good.
http://www.notebookcheck.net/Which-games-run-fluidly-on-notebook-graphics-cards.18488.0.html

 

hmmm..so Nvidia is the winner then..

 

At equal price points, yes, I would choose Nvidia. However, having spent $200 less than I would have for a similarly equipped 8662, I have no regrets.

 

No, it isn't. This test tells us that they are in the same ballpark but the memory (both graphic and system) and processor differences make the test inconclusive. The notebookcheck link that naticus posted shows the 4850 getting significantly better scores than the 260m almost all across the board, but again we don't know how much of that is due to other hardware. I would buy a 1GB 260m over a 512MB 4850 if I had a resolution higher than 1366 x 768, but with a 1GB 4850 it would be a different story.

 

Which can play these games?
-Pac-man
-MS.pac man
-tetris
-nas car
-megaman
-space invaders
well thats all for now hope the nvidia can handle these with a steady 10 fps atleast maybe even less since fancy pac man needs gddr5 since its so hard to emulate.What ever happend to the nvidia series for 10,20,30,40...They just skipped every number..

 

nice thread for all the info. i'll keep an eye on this and see how it progresses.

 

basically 260 gtx or HD4850 will run games good. go with what you are comfortable with if its gaming both these cards do good. For the budget conscious types you can't go wrong on an MSI GT725. If you want linux Nvidia is way way way better.

 

Hmm i would say you should have used the DOX drivers for the GTX260m might have shown quite a good increase also was this with fraps on? if so ill be playing away on very high when i get my Sager 8662 (Custom Config) also anyone know how smoothly this will play GOW

 

Thanks
Very helpful

+29 rep

 

It wouldn't have.

They used the built-in benchmarks for all of these.

Flawlessly.

 

I would never cite notebookcheck.net for anything specific. It's good for general ballpark figures, but we've seen huge differences between their "results" and real world.