Projected R3 versus R2 bechmarks

Nvidia changed their shaders quite a lot between the two.

The DX10 shaders were stronger per shader than the DX11 ones.

 

The GTX 460M is around 25% faster than the GTX 280M. That's more than "slightly ahead".

Terrible analogy. The differences found between those desktop GPUs are far greater than the ones you see with the 335M and 540M.

 

Eh? Read it again. I was comparing desktop GPUs. With proof.

 

Let me break it down for you.

First off, your analogy makes no sense, because no one compares the GTX 280 and GTX 460 in the first place. You either compare the GTX 280 to the GTX 480, or the GTX 460 to the GTX 260.

Second, all of those GPUs have far greater differences than shader counts and clocks. The GTX 280 is 512-bit GDDR3, to the 460's 256-bit GDDR5. With the 335M/540M, this is not so; both of these GPUs are on the 128-bit bus. This makes it extremely easy to project their performance ceilings.

Their architectural differences do not supersede the fact that they are both crippled by 128-bit DDR3. Therefore, it indeed comes down to 72 vs 96, and clock speeds.

The stated 20% performance gap will stand up to testing.

 

I would defend my analogy but that goes way off topic.

This will be seen in the user reviews.

 

Really? No one? I compared them. I has the chance to buy a notebook with a 280m or a 460m and had to compare to decide. I'm sure there's someone who has compared a freaking 485M to a 130M. Anyways, there is a point that they both have the same memory and busses, but you forget ROPs and what not.

 

Really? The same memory AND bus. Unless ofc you count size, width and memory type then sure!

GTX280 - 1GB 256bit GDDR3
GTX460 - 1.5 GB 192bit GDDR5

GTX460M supports DX11, is made on a new 40nm process so uses less power and has enough shaders to overcome them being slightly less efficient.

Also all of the backend stuff is FAR newer tech than the GTX280M which is based off the 8000 series from nvidia released in 2006!

Now true it was a pretty epic arch, but it's time has come and gone.

 

I meant the 335M and 540M.

 

GTX 280 is 1GB 512bit GDDR3.
GTX 460 can also be 1GB 256bit GDDR5.
Same bandwidth clock for clock.

You guys are over analysing my point. My point is that despite the 460 significantly more shaders and clocks, it comes out slightly ahead of the GTX 280. It is not a direct comparison between them both- that is not what this thread is about. It is a point about how the different architectures of the cards are not comparable and how comparing both cards is not as easy as 96 shaders > 72 shaders, 540M IS WINRAR.

I drew the parallels between the desktop and mobile core because they are related, cut down and/or shrunken. I chosen the GTX 460 because it has more shaders, ROPs similar memory bandwidth and looks faster on paper as well as because the 540M is based on this. As such, it has the best fit to compare to the GTX 280, which the 335M is based off of.

 

It's just a mess of an angle. The 540M and 335M are not based from either of those cards, unless you use the metaphor of them being second or third cousins. And this is where it all falls apart.

Forget about it and let's move on to more logical discussions.

 

GPU Engine Specs:
CUDA Cores 96
Processor Clock (MHz) 1344 MHz
Texture Fill Rate (billion/sec) 10.8
Memory Specs:
Memory Clock (MHz) 900
Standard Memory Config DDR3
Memory Interface Width 128-bit
Memory Bandwidth (GB/sec) 28.8

The challenge is to find a desktop card to compare it and I can't find one. The old 430 is much slower and the 440 oem compares with the nvidia 555mGT.

I think like other newer cards, the drivers will focus on dx10 and 11 performance and dx9 performance will not be the strength. I could be wrong...

The newer SB cpu's seem to get higher numbers in passmark. The lowest 2630qm outscores the I7-920xm by a margin but I don't trust those numbers. My "guess" is that if you compare game benchmarks between the two you will see about a 14-16 percent increase in frames per second.

Example if your current R2 scores 30 fps in "whatever" game the R3 will score 34-35 fps.

Just a guess... StevenX

 

Actually the GT430 desktop is the same. Same core just clocked slightly higher.

Therefore the GT430 should be around 3% ahead of the 540M.

 

You're still over analysing my point. 540M is based off a tweaked GF100, much like the 460. 335M is based of the GT215, a cut down and shrunken GT200b.

Nonetheless, my point still stands. No one can directly compare both cores on paper.

How is the GT430 much slower? It looks very similar if not faster.

GT430: GF108
CUDA cores: 96
Processor clock: 1400Mhz
Texture fill rate (billion/sec): 11.2
Memory Specs:
Memory Clock: 900
Standard Memory Config: DDR3
Memory Interface Width: 128-bit
Memory Bandwidth (GB/sec): 28.8

The difference between those Sandy Bridge CPUs and these ULV type is that these are already throttling so it may be slower.

 

Let's just drop the desktop comparisons. There's plenty of 540M benchmarks out there already. The chip is relatively "old", at this point.

 

@Kevin - Just trying to give some folks some more figures and search opportunities to look at. I searched and only found 3 real reviews for the m11x -r3. I may have missed one but that's all I found. Searching for the 430 nets a few more 'figures' opportunities.

@Davy - the 430 would be close. I was just giving the 540 the benefit of the doubt as the fill rate is higher and it's the 'next gen' product and hopefully more efficient.

Best Wishes, StevenX

It's all speculation until it's in your hands and you can see what your OWN MACHINE can do!

 

The 540M and 430M both share the same GF108 core, difference being in the clocks.

You may have missed this.

 

A renaming does not make a next gen product lol.

Plus the fill rate on the GT430 desktop is higher too.

 

In some regions like Canada, the GT 430M with 1.0GB or 2.0GB of memory are the only options in the M11x-R3.

The Dell Online Store: Build Your System

 

Thats interestingly because here in Austria i had no choice except for the 2GB 540M