The main difference: one card has 1GB memory on it, but 850Mhz effective clock speed. The other has only 512 memory but 3588 Mhz effective clock speed. The price difference about $4~5 if I am not mistaken.
The fact remains they use the exact same GPU; "SuperClocking" one isn't going to give the (hopefully a typo) gargantuan boost of 850Mhz to 3588Mhz. The SuperClocked version is more likely to be somewhere around 1Ghz, despite what the specification sheet says (also note that the first one notes 850Mhz, 3.4gbps effective, whereas the other says "3588Mhz Effective" they probably cannot be compared directly.
what's more important? GPU effective clock speed or amount of memory coming with the card? (all other specs equal, that is)
Depends entirely on what it will be used for. And also, if for games, entirely on what games are played.
More ram on GPU allows for a smoother graphics on higher resolutions. But then again, more memory? But with the card having the higher memory amount, couldn't you just OC the card if you wanted to?
More RAM on a GPU doesn't allow for smoother graphics in higher resolutions. the memory used by the 2-dimensional screen pixels, even when doublebuffered, won't exceed about 16MB of space, even for a huge 1600x1200 screen. Video Card memory is used almost exclusively for storing textures. Textures whose quality settings in-game directly affect the texture resolutions; this means that more VRAM translates directly to either More Texture or higher- Resolution textures; using a texture quality above what your card's memory can handle means that many textures will be stored in system RAM, which means that the card will have to wait for a free DMA access to get to the data; needless to say reaching across the Bus, even if it is a high-speed PCI-E bus, is still a lot slower then simply reading from the high-speed VRAM. Clock-speed of the GPU will not make reading data from the System any faster. In fact, it means that even more of the otherwise faster cycles will be wasted, since it also has less texture memory. And this is only for textures; there are of course varying qualities to the other components, such as shaders, vertexes, polygons, and so forth; the code for a shader still has to be stored and accessed from video memory, just as the vertex data has to be stored in VRAM to be accessed by the rendering pipeline. a higher speed GPU doesn't mean that the memory will be accessed faster. The latency is still the same. The GPU processes the Vertex, texture, and lighting data and performs the hardware Transform, Lighting, and texture mapping, and nowadays it also calculates the normals of each triangle. faster GPU speed translates directly into the ability to display more polygons; but remember that the vertexes, textures, and so forth that the GPU is now working with faster still need to be stored in the Video Memory- having a faster GPU in no way "compensates" for less Video memory, just as having lots of Video memory cannot compensate for a slow GPU.
