Originally Posted by malmental
funny.... using your link one of the first things I read is this..
so I think your math not sure if correct or not but your off on the reason I said what I said..
...Did you even read it?
192bit cards native RAM (ie. when you have an equal amount of equally sized chips on each memory controller) is either 384MB, 768MB, 1.5GB or 3GB, nVidia doubled up on some of the RAM chips to get 2GB on the 660Ti, meaning that without any interleaving 1.5GB will be at full speed, but that last 512MB will be at the max speed of one memory controller, 3GB would have an equal number of equal sized chips on each memory controller meaning each runs at full speed...a 3GB 660Ti is probably going to be a little faster than a 2GB one assuming all else is equal, actually. It wouldn't make any sense in nVidia's lineup (It'd have more vRAM than all but the most expensive GTX 680s/670s, or lower their performance by having an uneven amount of vRAM on those two cards to match it)
Originally Posted by malmental Testing For Memory Interface Limitations: 1920x1080Testing For Memory Interface Limitations: 2560x1440bottom line is the GTX 660 Ti is not capable of pushing out 3GB of VRAM....
the 192-bit interface is a detriment at higher resolutions and high levels of AA is death.
I feel like deja vu around here today..
and the 3GB version is slower than the 2GB version...
Brutuz - you need to stick to Radeon dude....
it's a good thing for everyone that Radeon is better @ Folding but for you specifically it's a good thing.
for you brother have no clue about what's going on here...
Firstly, the 660Ti is memory interface limited...But the fact remains it's still around as fast as a HD7950 which can
use its extra vRAM, it could use the extra vRAM but would still lose performance.
Secondly, running out of vRAM while the cores hungry means that it's going to be loading from the much slower HDD, the 660Ti would still be able to pull playable FPS, especially with a good vRAM OC while benefiting from the extra vRAM. (See: Skyrim with lots of texture mods. It doesn't load them all at once but a lot of HDD/SSD reads are saved by preloading textures into the vRAM even if they're not being used, then just grabbing them when necessary)
Thirdly, nVidia will have optimisations in the drivers to make up for 512MB of the 660Tis vRAM being at 64bit bus speeds, the 3GB card will likely be doing all of that without any need on the Galaxy 3GB card, which has 1024MB on each MC vs the 2GB cards with have 512MB on 2 MCs and 1024MB on the 3rd MC.
And finally, even if nVidia does have allow for 3GB 660Tis without those optimizations (I doubt it) then what are the actual RAM chips onboard? What's the betting Galaxy has used cheap chips that run looser timings in order to keep their margins on that card decent?
You're arguing against common sense here...You don't see many 3GB 256bit cards because it's not easily done with maximum performance, without optimisation (That would hurt performance for a 4GB version of the same card) then you're going to get 2GB at full speed and 1GB at half speed.
"Because NVIDIA has disabled a ROP partition on GK104 in order to make the GTX 660 Ti, they’re dropping from a power-of-two 256bit bus to an off-size 192bit bus. Under normal circumstances this means that they’d need to either reduce the amount of memory on the card from 2GB to 1.5GB, or double it to 3GB. The former is undesirable for competitive reasons (AMD has 2GB cards below the 660 Ti and 3GB cards above) not to mention the fact that 1.5GB is too small for a $300 card in 2012. The latter on the other hand incurs the BoM hit as NVIDIA moves from 8 memory chips to 12 memory chips, a scenario that the lower margin GTX 660 Ti can’t as easily absorb, not to mention how silly it would be for a GTX 680 to have less memory than a GTX 660 Ti."
"Of course at a low-level it’s more complex than that. In a symmetrical design with an equal amount of RAM on each controller it’s rather easy to interleave memory operations across all of the controllers, which maximizes performance of the memory subsystem as a whole. However complete interleaving requires that kind of a symmetrical design, which means it’s not quite suitable for use on NVIDIA’s asymmetrical memory designs. Instead NVIDIA must start playing tricks. And when tricks are involved, there’s always a downside.
The best case scenario is always going to be that the entire 192bit bus is in use by interleaving a memory operation across all 3 controllers, giving the card 144GB/sec of memory bandwidth (192bit * 6GHz / 8). But that can only be done at up to 1.5GB of memory; the final 512MB of memory is attached to a single memory controller. This invokes the worst case scenario, where only 1 64-bit memory controller is in use and thereby reducing memory bandwidth to a much more modest 48GB/sec."
Read those quotes, please, before you comment again...nVidia will have done optimisations/tricks to make up for the asymmetrical memory which is why the 3GB card there won't perform as well, the drivers see it's a 660Ti and run that through nVidia's codepath for the 660Ti.