5850 vs GTX260M

Forget about DirectX for a moment.

Which Gpu is better? 5850 or GTX260M?

 

Fairly certain 5850.

 

GDDR3 or GDDR5 version?

My understanding is GDDR3 5850 < GDDR3 GTX 260m < GDDR5 5850.

All are pretty close together in power, though.

 

GDDR5 5850 > GDDR3 5850 > GDDR3 GTX 260m

EDIT: fixd

 

How can a GDDR3 GPU be better than the same GPU on GDDR5?

 

Good lord Forge I thought you'd know by now how much better a GDDR5 card is to a GDDR3.

 

I think he mixed up the < with the >
Anyway, the GTX260M is better than the GDDR3 version of 5850.

In any case, the 5850 with the GDDR3 in the 18.4" Acer gets about 8500 3DMark06 paired with an i7 quad core, whereas the GDDR5 version in the MSI GX640 paired with an i5-430M gets 11k to 12k. I'd say the GDDR5 version is at least 50% better than the GDDR3 version. In comparison, the GTX260M needs to be paired with at least an i7 quad core to get anywhere near 11k. Given the CPU dependency of 3DMark06, this means that the HD 5850 GDDR5 destroys the GTX260M.

 

What exactly are you basing this on? Because Notebookcheck.net shows the GDDR3 5850 getting 200 more points in 3DMark06, and Vantage GPU scores. The 5850 is HIGHLY memory bound, and GDDR5 fixes that, showing almost 7500 in the GPU score vs. the 260GTX's 4200.

 

?

That's what I wrote.

 

I'm not sure how but I quoted the wrong person Forge...I was going to quote the other guy. I think this means bedtime for meeeeee

 

Forge, when you're talking math, the arrows go the other way. Big end goes to the bigger/better number/card.

10 > 5 is correct. 5 < 10 is also correct. 5 > 10 is not correct.

 

But isn't that what I wrote?

is "A GDDR3 5850 is less than a GDDR3 GTX 260m, which is itself less than a GDDR5 5850".

Ah, whatever. Stupid syntaxes.

Gonna find you just like Freddy Krueger. Sleep tight!

 

Well... GDDR5 is better than GDDR3 that's a fact. And if both cards have the same amount of memory, GDDR5 should always give better performance. I wouldn't trust those benchmarks.

EDIT: Damn I screwed up the direction of the > I meant to say the GDDR5 was better.

 

No worries. Happens to the best of us.

 

lol not to take this thread off topic but there has been a complete < and > mass confusion here. Anyway back to on topic I am curious how the DDR3 5850 stacks to the 260GTX

 

From my [retarded] understanding, the "power" of the card is pretty good, but the 128-bit buss is hurting it. So they need GDDR5 to counter that. I suppose a 256-bit GDDR3 would work, too, but who wants to do that.

And I think Pita's screwing up Random with Forge.

SomeRandomPitaForgeDudeBred. Rock on.

 

I always use the Pacman reference(as in, > is a Pacman mouth). Whatever Pacman is eating is the biggest thing because Pacman is hungry

Anyhow, yeah I agree with the order then lol ^_^

 

Doh. My turn to plead confusion

 

And I thought I was the only one who learned it like that. Well, instead we used ">" as a fishie's mouth.

S'Okay. Still love you.

Besides, the last couple of posts were a whole glop of confusion.



Which laptops carry a GDDR3/5 5850 and an Core ix CPU, anyway? An Asus G51j could be a good comparison machine against a 5850.

 

http://www.notebookjournal.de/tests/laptop-review-acer-aspire-8942g-728g128twn-nkmo-1084/2

 

Ya, that. I keep quoting that benchmark, but kept forgetting the source.

Anyway, the benchmarks are all given at 1280x800:

Acer Aspire 8942G (i7 quad) with GDDR3 version of HD5850: 9200 marks
Alienware M17x (i7 quad) with GDDR5 version of GTX260M: 11500 marks.

Now, I'd say those would translate to something like 8000-8500 and 9500 - 10000 at 1280x1024.

Compare that with the HD 5850 GDDR5 with an i5-430M in the GX640 getting 11200 at 1280x1024:

http://forum.notebookreview.com/showpost.php?p=6115332&postcount=356

Or the Thai benchmark of the same notebook getting 12000:
http://www.notebook4game.com/web/2010/01/review-msi-gx640-highpower-corei5-ati-hd5850/3/

I'd expect an i7 quad core paired with the HD 5850 GDDR5 to score something like 13000-14000.

 

The only GDDR5 5850 I know of is in the MSI GX640

 

The GTX 260M doesn't have GDDR5...

 

Not even the GTX285 has GDDR5, it's GDDR3.

Only Nvidia I've seen with GDDR5 is the GTS360

 

My bad, that's a typo it was late at night..

 

in 3dmark06 and general gaming

260m gtx>all single cards except hd5870 and 280m and 285m gtx.

in 3dmark vantage only.

360m gts>260m gtx
hd5850>260m gtx
hd4870>260m gtx
285m gtx>260m gtx
280m gtx>260m gtx
hd4850>260m gtx

the 260m gtx is still a great card with or without an overclock. its not like having an extra 2 fps in games will add to a games substance or make it more enjoyable.

I own the 260m gtx, and run every game maxed at around 25fps at least, even crysis gets 20-28 fps.

I also owned the mobility hd2600 gddr3 card, and after seeing how well that card performed in comparison, and how the games look much better. I'd pick ATI anyday.

ati cards have better drivers, better image quality(lighting), and perform much better when you overclock in comparison to nvidia.

pick the hd5850 and overclock it.

my hd2600 was set to 500/600, and I safely overclocked it for 2 years to 635/930. 3dmark05 went from 6300 to 9558, games gained about 10fps.

my 260m gtx is stock at 500/800/1250, and I safely overclock it to 580/960/1510. 3dmark05 went from 16200 to 17024, games gained about 2 fps(gtaIV gained 7fps).

hd2600 crashed about 10 times in 2 years
260m gtx has crashed 10 times per day on average

hd2600 maxed temps at 80c with cooling fan and thermal paste
260m gtx maxed temps at 91c with cooling fan and thermal paste

hd2600 played crysis at 30fps with low level 1 config
gtx 260m plays crysis at 50 fps with low level config

crysis has not gotten any funner at 20fps faster.
masseffect is not funner at 62fps vs 30fps
gta IV has not gotten anyfunner or smoother even though I average 10fps more

metro 2033 and vantage run like dirt with the 260m gtx.
metro 2033 and vantage wont run with the hd2600 unless you call 1000 points running it.

long story short, now is not exactly a good time upgrade your laptop if you have anything close to the hd2600 gddr3.

also.

BF:BC2 looks like crap all settings low
looks like crap all settings high

take your pick

if you can find a laptop for under 1000 bucks that has any high end card, then you are in good shape for the current set of games available.

but don't expect future games to run faster than 20fps at high rez and details just because you decided to spend 1800-3000 on a gmaing laptop thinking 5870 is twice as good as the 260m, therfore i'll get twice the framerates.

the gtx260m is at least 3 times(300 percent) better than the hd2600 gddr3, however 0 games show this, I get 40 percent max gains.

so if people are saying this card or that card is 40-80 percent faster, I can promise you, they are not getting double the framerates, or even close.

percentage gains when it comes to synthetic benchmarks, are indicative of nothing when it comes to actual gaming compatability or performance.

also, if your thinking of buying a new laptop, think about native rez and features what matters and what doesnt for your needs.

for example, laptops shouldnt really be larger than 15 inch. its suppossed to be a portable computer, not a breifcase sized monster.

the g51vx-rx05 is on sale for 750 dollars, its native rez is lower. and contains the 260m gtx. it will be future proof at 1378x768 much longer than any laptop out even the 5870 at 1600x900. forget about dx11 for this generation of gpus.

 

eazy, please stop spreading your tripe about the 260M GTX. The 5850 will pound the heck out of the 260GTX, and the 5870 even more so. The 260GTX is not a bad card, it's just not in the same class as the 5850/5870. It's roughly comparable to the 5830 with GDDR3.

Your framerate numbers are all completely irrelevant to this discussion because the 2600 is a 5 year old card, not to mention your assertion that the 5870 is not twice as fast as the 260M GTX. It's very close to twice as fast. Notebookcheck has FEAR2 on a 5870 at high at 110fps. On the 260M GTX it's 69fps. Crysis GPU benchmark on the 5870 is 56fps. On the GTX, it's 38. Risen at medium? 60 vs 40fps. Those are all at the same settings. You have to measure things at the same resolution. A 260M GTX showing the same framerate at 1024x760 as a 5850 at 1680x1050 does not mean that the 260M GTX is "just as fast". The 5850 (especially with GDDR5) will play things faster at 1600x900 than the 260M GTX will do so at 1366x768.

Please stop spreading misinformation about the 260M GTX. It is not a perfect card, and nowhere near as fast as you lead people to believe.

 

So both memory versions of the 5850 beat the GTX 260m? Hmm.

And I haven't read too much on the 5830, but I do remember seeing on <strike>that unreliable site</strike> notebookcheck that it wasn't ranked all that high, at least as enthusiast cards are concerned.

 

In the rankings it's not high, but the 3DMark Vantage GPU scores show the 5830 about even with the 260M GTX. That's why you shouldn't take the numerical rankings as anything too meaningful on Notebookcheck.net (just the relative class rankings), but the benchmarks are more or less valid.

 

I know. Meant to say I looked at benches.

I swore I saw the actual benches being noticeably lower. Well, guess I'm wrong. I do notice the GTX 260m readings are a little skewed, since at least half of the benches are from Asus machines, which are downclocked to GTX 9800m levels. In fact, only one out of the nine 3DMark06 benches are explicitly coming from normally clocked 260ms. Similar problem with the Vantage readings.

Stupid Asus, haha.

 

My biggest gripe with the GTX260M is the absurd amount of power it consumes. The 5850 with GDDR5 consumes 30W-39W according to ATI, whereas the GTX260M goes up to something like 60W+. That's why ASUS had to underclock it even when putting it in 7lb+ 16"s.

 

Hmm... good to know. It's not like the 260 isn't a GTX 9800M anyway, though A properly clocked 260M GTX is quite likely a bit faster than the 5830 then (may swap, depending on clocking and all), but the point stands that a 5850 is still a good bit faster than a 260M GTX, and there's just no comparison to a 5870.

Aye. Especially considering that the 5830, which it performs closely to, is 24W with DDR3. That's more than twice the power to do just about the same thing.

 

These values are all TDP, are there any real power consumption values? In desktops, ATI has low idle but high load power, while nVidia typically has slightly higher idle but similar load power.

 

We can safely say that the HD 5850 will be <= 39W in power, correct?

Now, why would they make a GTX260M if it draws significantly, say, 20W less than their stated maximum of 60W? I'm on the same page as you if we were comparing say 39W with 45W, but 39W and 60W is pretty big of a difference.

 

I'm still skeptical on whether a GDDR3 5850 will outperform a GTX 260m, though. I'm looking at specs, and while it's a newer architecture, the (I think) fewer shaders and half-sized buss don't look all that hot.

Then again, I need to do more research.

By the way, what's the rule of thumb for converting between Radeon shader count and nVidia shader count?

 

This is what it is.. signs are correct and u ppl should really do more maths

Yup , the acer which has 5850 is GDDR3..

Also GTS250M in qosimo is GDDR5 i think...

 

Last I heard it was divide by 5. So 800 ATI shaders ~ 160 nVidia ones.

 

The 5850 kills the 260m, especially gddr5 (MSI book.)

Gddr3 fails on a bus that big.

 

The GDDR3 5850 won't come close to beating the 260M. Neither would the 5830.

 

Whereas someone who has had actually had both calls bull, says the 5830 and the 260M are similar: http://forum.notebookreview.com/showpost.php?p=5959817&postcount=41

Perhaps he's talking an underclocked 260M or a GTS, not sure, nor am I sure if it's a 260M GTX with GDDR5 or GDDR3.

 

GTX 260M vs 5830:

 

How fast does it go DX11? Oh, right... linked post says an average of 4fps faster than what RyanHurtt finds for DX10. And your card won't even run DX11. "Those exact settings" seem to be tuned very much against an ATI card and for the Nvidia. I can pour diesel in an F1 cars tank and get it to not run, too.

I'll run that benchmark later tonight. Think you could run the Unigine Heaven 2.0 too, 1024x768, and whatever you find appropriate, and I'll do the same?

 

One thing I've noticed with the 5xxx series even against the GTX480, ATi does a much better job at sustaining FPS.

If you look at some of the benchmarks, whether it's a G92, G200 or the GF100, the minimum in games are atrocious. Even with G200 can lead to stuttering. Few times, the maximum is excellent, which is what Nvidia marketing is all about.

If you look at the 4xxx and 5xxx series in comparison, the maximum is not that awesome. The average is sometimes higher than the Nvidia and the minimum is very close to the average. Meaning the gameplay is smooth without slow down and stuttering through out.

That's the biggest difference I've noticed looking at game benchmarks.

The new drivers for ATI are doing this even better. For example one person who installed the supposed 10.5 Catalyst noted that while his maximum FPS did not increase in Vantage, but his minimum was reduced even more! So in the scene where the girl jumps onto the boat, where your minimum is heartbreaking to watch, he used to get around 30, now the minimum is 60!!!

 

I much prefer in-game benchmarks, but what the hey:

 

Hey guys,

I've been a longtime lurker and the information on this forum has been invaluable with my decision making. Thanks everyone.

Now for my question... so I've looked through benchmarks of the 5850 and 260m GTX and many members say that the GDDR5 5850 blows the 260m GTX out of the water.

Here's my dilemma, I JUST bought a used Sager NP8662 with a P9600 and 260m GTX (not even here yet) for $1000CAN and noticed that MSI recently released the GX640 with the 5850 and i5-430m for $1200CAN.

If I only plan on playing Starcraft 2 and Diablo 3 (in the future), would I see a vast increase in frames going to the MSI? From my understanding, the Sager can run Starcraft 2 Beta at medium very smoothly... would the GX640 be able to run it at high smoothly? Would it also be worth the $200 premium (I would think the Sager has superior build quality; build is important to me)?

 

I'd say the GTX260M is a bit ahead of my 5830 at higher resolutions due to memory bandwidth issues but the higher clocked GDDR3 5850 would probably pull pretty even with the GTX260M. If you get the GDDR5 5850 it'd be a fair amount better than the GTX260M.

Also to the poster above - I believe that is the benchmark in which you upgraded your CPU and it raised your scores. Really the benchmark that matters is the "Sun Shafts" as that one is the most GPU bound...

With that being said my graph is glitchy and I believe my average on "Sun Shafts" is 11.6 and not 16?

Just ran the test as showed above with DX11 - 25.3, 23.5, 25.2, 13.7

I believe the engine that STALKER is built on is one of the only that is optimized this way though.

 

It could be all down to testing circumstance if he called that, for example, resolution lets say...

 

When we talk about one GPU being better than another, it's usually a slight difference. The GTX 260m can knock around games for quite a while. And Clevo has excellently-built laptops.

 

Alright, fun times here. All your settings where possible, and stock clocks except in the Heaven benchmark. Dunno what was wrong with RyanHurtt's setup, but it was obviously something. Might have just been drivers, since these are run on the Catalyst 10.5's:

Heaven benchmarks, noted in image (ran into image upload limits )

Stalker, DX10:

Stalker, DX10.1:
Looked slightly better than DX10. Nothing remarkable, though.

Stalker, DX11, tesselation enabled:
Much prettier, couldn't take screenshots, but you are seriously missing out. The clouds look 100x better, all the lighting is nicer, ground textures and all.

I couldn't get screenshots in SCOP, but I could get them in Heaven, so here's the difference between DX10 and DX11. Pay attention to the dragon, as well as all the cobblestones around him and such:
DX10:
DX11:

The most damning thing I can say about the 260 is that it's minimum framerates are barely faster than what the 5830 does. And if you look at the DX11 results, where the 5830 is much better looking, it's not even funny how close the framerates are, and how badly the 260 GTX loses the sun shafts run. Note also how in the DX10 runs how close the minimum framerate is to the average framerate as compared to the DX11 run on the 5830. Means that the DX11 version has a higher framerate during more of the demo, and doesn't drop as low. I'll stand by the 5830 being very close to the 260M GTX in what people will perceive.

And to get back on topic, the 5850 with GDDR5 will eat the 5830 for lunch, evidenced by how much the Heaven benchmark changed with just a 20% increase in memory bandwidth. The 5850/GDDR5 has over DOUBLE the bandwidth, not to mention likely being clocked another 20% or so higher on the core.

5850:
* Engine clock speed: 500-625 MHz
* Processing power (single precision): 0.8-1.0 TeraFLOPS
* Polygon throughput: 500-625M polygons/sec
* Data fetch rate (32-bit): 80-100 billion fetches/sec
* Texel fill rate (bilinear filtered): 20-25 Gigatexels/sec
* Pixel fill rate: 8-10 Gigapixels/sec
* Anti-aliased pixel fill rate: 32-40 Gigasamples/sec
* Memory clock speed: 1 GHz GDDR5 or 900 MHz GDDR3/DDR3
* Memory data rate: 4.0 Gbps GDDR5 or 1.8 Gbps GDDR3/DDR3
* Memory bandwidth: 64 GB/sec (GDDR5) or 28.8 GB/sec (GDDR3/DDR3)
* TDP: 30-39 Watts (GDDR5) or 31 Watts (GDDR3/DDR3)

5830:
* Engine clock speed: 500 MHz
* Processing power (single precision): 0.8 TeraFLOPS
* Polygon throughput: 500M polygons/sec
* Data fetch rate (32-bit): 80 billion fetches/sec
* Texel fill rate (bilinear filtered): 20 Gigatexels/sec
* Pixel fill rate: 8 Gigapixels/sec
* Anti-aliased pixel fill rate: 32 Gigasamples/sec
* Memory clock speed: 800 MHz DDR3/GDDR3
* Memory data rate: 1.6 Gbps DDR3/GDDR3
* Memory bandwidth: 25.6 GB/sec
* TDP: 24 Watts

5850M/GDDR5 >> GTX 260M ~> 5830M

 

This is quite interesting. The 5830's high shader count and architecture are really well utilized, despite its 128-bit bus.

If one of you would run a few of these, we will really get down to the bottom line. I don't see this as a competition, as much as I'm obsessed with tech knowledge. If the 5830 really is as fast as the 260M, the Envy 15 jumps up the list of recommended gaming machines.

I also have a fresh Crysis run: