5870 vs 460m. Some data from notebookcheck.

You ignored the price issue, which is the main reason it's an "upgrade." Your points are invalid. Please stick to actual relevant gaming benchmarks.

 

Yeah same here. This is my first ATi card and I am very happy with it but I still prefer nvidia. My card still runs about as cool as the day I got it though which is a little over 80c.

 

Very nice temp and yeah its not about what card when it differs from what price, im glad im experiencing ATI because I believe this is one of the best cards that they will release at the moment apart from the special 6970m but like that isnt going to come with its problems to start with

GTX cards are very powerful mind and I would always go for SLI over XFIRE but I will definetely be keeping the G73 for a good while especially when they think they can put a shallow upgrade with a sandy bridge and call it a G74 for 3 times the price. not likely!

 

Let's keep this discussion civil. No name calling ("fanboy"), no ad hominem, no personal attacks.

Counter with facts. And if facts are ignored, move on. Readers will be able to tell what's fact and what's fiction.

 

people are arguing about which is better than the other subjectively, but the truth is that 90% of people would actually be better off with a 5870. the other 10% use software that can utilize cuda or play lots of games that use physx and that makes them better off. therefor, I proclaim that I still think after reading all of this that the 5870 is generally, a superior graphics card!

 

87.34% of statistics are made-up.

 

the 5870 is capable of 1.12 Tflops. How many tflops is the gtx 460m capable of? This would be a great way to guage which card is more powerful.

 

The 460m is capable of 518.4 Gflops.

I personally think comparing processing power isn't the greatest yardstick for gaming performance, it's only relevant if the cards are used for brute-force hacking purposes or the like.

Source.

 

can you link your source for that figure? I can't seem to find any numbers on the mobile 460 card.

EDIT: In your source link, I can't seem to find the mobile version numbers for the 460m can you link to the exact page? I'd like to read into it. Thanks.

 

It's a big page so lots of people are too lazy to scroll downwards to the point where the mobile nvidia cards show up. Use the Ctrl-F function and type in GeForce 400M (4xxm) series and that'll lead interested readers to the correct section.

If you don't believe wikipedia, it's got the references at the bottom of the page.

For those trying so hard to justify the superiority of Nvidia cards over their ATI counterparts, the computing power differential between the 460m and 5870m should be quite surprising given the cards are roughly equivalent in gaming performance.

 

nvidias cards generally have lower raw processing power, but that does not hinder gaming performance. It is a bad way to rank gpus.

 

actually that raw performance is significant in guaging how many polygons it can push, and calculating polygons are crucial in any modern video game especially 3d first person perspective games. If the card can only calculate a certain amount before it starts to struggle then so will the games framerate in any given scene that has to be drawn on screen. After all the original reason for the GPU was to help relieve stress from the CPU and increase a games framerates to playable levels.

 

I thought the first rule for comparing video cards was the actual hardware on it i.e. Shader parts, memory used (and bandwidth) for the actual fill rate crunching. ATI and Nvidia may have different counting systems for their respective parts, even then it's hard to explain why current generation cards with wildy differing processing power ratings have similar gaming performance.

Polygon pushing?! I can't remember too many recently released FPS and RTS games still relying on polygons nowadays. If it's just an effort to promote one company's product over the other, a better analogy/reference should be used. As noted before, the 460m and 5870m are more or less equivalent in performance and it really depends on the price point and whether or not said user really needs CUDA in everyday tasks.

 

Besides 2D side scrolling games that rely on sprites, almost all 3D video games are built using polygons to build the models and environments. Since it's infancy up to modern games, polygons are used to give the 3rd dimension to what we see on screen, from Virtua fighter to Crysis 2. Infact the 3 different presets for graphical settings in Crysis 2, increases performance by reducing polygon count as one means to relieve calculating stress on the processors.

Here's an article from geforce.com on how to optimize Crysis 2 and it mentions how many polygons are rendered for each graphic preset from gamer to hardcore.

GeForce.com - Get the Most Out of Your GPU

EDIT: The link doesn't go to the specific section, which is located under "methodology" at the bottom of the page. Here's a clip of that article.

"Polygon count is also increased significantly between the modes - in Gamer each scene is comprised of roughly three hundred to four hundred thousand polygons; in Advanced five hundred to six hundred thousand polygons; and in Hardcore up to and above one million polygons."

So you see, the higher graphics settings the more the polygons it has to calculate, and this is where the raw horsepower of a GPU comes into play, and why it's important in guaging a GPUs power.

 

Odd then that the 460m has less than 2/3rds the processing power of the 5870m yet ends up outperforming the latter on certain gaming benchmarks. I'm not partial to selective copy and paste jobs from various gaming-related websites for proof of performance, polygon count doesn't convince me as the be-all and end-all of gaming performance.

Odd that the post above this one talks about Crysis 2 and not its predecessor, perhaps it's my imagination but many of the forum posters here have widely derided the sequel for its use of "console lighting tricks" on the PC platform whereas its predecessor was built purely from scratch on the PC platform.

 

yet I'll bet that even though people say that Crysis 1 is "poorly coded" they don't understand what that means. The "poorly coded" means that the developers didn't limit polygon counts, but went through the roof with it to make the best looking, most interactive environments at the time. I'll try and find an article of how many polygons the original crysis uses, and I'm sure if I do it would be like in the 10-100 million plus lol.

 

As far as Crysis polygon count, I found some discussions on other forum sites where users documented 4 million polygons on a given scene and that made the framerate drop to 7-10 fps. That's alot of triangles that need to be calculated and moved on the fly and then on top of that calculate the textures, lighting, shading, AI etc.

 

Well my theory on this is that, the games that the 460m outperforms the 5870 on is because those games probably had polygon counts that both GPUs could calculate without struggling, the deciding factor must've been the DX 10/11 special effects that needed that calculating, and the Nvidia card was stronger in that area.

To me, this was the purpose for DX 9, 10, 11 etc. to negate the use of high polygon counts and instead substitute polygons with better looking models and environments using tesselation, lighting, etc., that wouldn't tax the GPU as much as calculating the polygons, but instead of capping the polygon count to say 500,000 max, which would allow older cards to calculate and sustain at 30 fps, game developers use even more polygons and then implement the DX 11 special effects which kills even more performance.

So going back to why the 460m would outperform the 5870 even though the Amd card has more raw horsepower to calculate polygons, is that the Nvidia card has better hardware to calculate the DX special effects.

 

And yet, at the end of the day, polygon counting, fill rates, bandwidth etc, etc, etc means little if not put together in the big picture: how many FPS can this card push with respect to others in real games? (well, this picture at least is accurate for people comparing cards for gaming purposes...).

 

But, what are GFlops and TFlops? I know they are Gigaflops and Teraflops, but what do they measure? What do you mean with "raw processing power?

 

I did i lot of research between 5870 and gtx460m before i got the gtx460m, and i have never regret my decession....nuff said

 

I like how people start to comparing some "graphic processor numbers" and completely ignoring REAL data from todays games posted on first page (or notebookcheck for that matter). Performance boost seems at the very least noticeable. Sometimes like +20-25% which is hellalot.

Did your research involved going to notebookcheck and seeing that 5870 give way better fps?
Sure it might not make a huge difference NOW because 30 vs 40 fps isnt noticeable. But in tomorrows games we are looking at 20 vs 30 fps. Does 460m "better image quality" and "better build quality" (which people kind of failed to proove yet) make up for not being able to play some metro 2033 v2 on medium if it will come out tomorrow? Granted that 5870 will be able to dish some 30fps there. (not to mention that 460m cant run metro 2033 on high now and 5870 can — how is it not better?)
Thats something to consider at least.

 

Did your research involved going to notebookcheck and seeing that 5870 give way better fps?


I studyed ever review online with the 2 GPUs, and also noteboockcheck. I never use 1 review to make a decession

 

Care to show tests / proofs that show that 460m DO give better fps or better image quality in various games?
(note: not SHOULD give but actually DO give)

Also you are aware that benchmark data is NOT a "review", right?

 

Listn. I know the 5870 is a few fps faster in some games, BUT it isnt like playing very high on the 5870 and high on the gtx460m. Both GPUs is good and i am sure you would be happy no matter what you choose.

Google is your friend as they say. I dont have the links now, but i am sure if you google you would find much info about the gpus

 

I see all talk but no real hard facts, why has no one posted any gaming bench marks, I'm looking at you jacob808 I wanna see how much better the 460m is compaired to the 5870

 

This thread needs to die.

Let me sum this whole thing up:

Both are about equal........

 

I vote that Nvidia and ATI get together and have a Graphic card orgy and create the ultimate Hybrid Atividia.

This would be the result:

Madd cores: 3600 Cuda cores: 492
Fermi DragonAge2 sux technology
Core: 1250mhz
Memory: 2000mhz
VRam: 5GB GDDR5,
Bus: 512bit
Wattage: 240watts, 8nm
Information: 4D, 42 monitor capability, Physex, More than HD HD.
2.512 Nero flops


Or we could stop dreaming that the GTX is better than the ATI and agree that for the burn you pick the ATI and it owns or if you dont want to burn a hole in your laptop and live with less power but are able to sit at your laptop and have an epileptic fit wearing a pair of goggles you choose the Nvidia. Both own in their own way.

Simples.

 

Alrighty then...

Thread closed.