Oshyan, They're the value proposition because they are cheaper overall ;) (Yes, I know the P4 is generally not as fast as the K8 -- I was just making fun of Intel because during the Northwood's heyday (i.e. last year ;)), the Athlon was the value option). Hellborn, to answer your questions: 1) More memory doesn't add power, it's basically a bigget cache. If you're using a lot of large textures, you might see a difference, otherwise not. In other words, it's a definite maybe :) The card's ram has to hold a frame buffer (what you see), a back buffer (what the card is drawing to), a z-buffer (depth map, used during the drawing process), a bitmap cache so that it can redraw windows and the cursor quickly, and scene data (geometry). Whatever is left over it can use for caching textures. That's why the high-end cards have more memory. 2) You won't find any Athlon64 dual boards, only Opteron... but they're the same processor. So a dual Opteron would crush a dual AthlonMP like a bug, because the AthlonMP is severely memory starved. Comparing it to a single Athlon64 is hard, but since the Athlon64 will give you a huge improvement in graphics performance, it would probably be a huge winner if you plan to use it as a workstation. 3) Xeon. The current Xeon processors are based on the same Prescott core (current = Nocona) as the P4, using the same memory bus. So everything everyone has told you about P4 vs Athlon64 applies directly to Xeon vs Opteron, but with one exception: the Opteron's memory subsystem scales better from one all the way up to four processors. After that they both pretty much hit the brick wall of needing external logic, after which the better choice depends as much on the system as on the processors themselves, making the comparison a lot harder. But then you're also looking at the bargain basement price of $50k for the barebones system, so that's not likely to be a problem anytime soon, I imagine :)