[citation][nom]gim159[/nom]today, it is kind of unfair to try and stack a
AMD and Intel side by side... Its true, Intel chips are faster, but for the higher end chips you'll have to spend a outrageous amount to get your hands on one.. Meanwhile AMD went the way of the consumer and built sturdy cores, but at VERY low cost. True AMD might still beat the pants off of Intel, though I doubt it. Intel are and have been for servers in my book, and AMD are for Econo/gaming rigs... Almost no one runs a server with an AMD. And no one in their right mind runs a high end server with an AMD, you just don't do it![/citation]
Have you been living under a rock for the past 3 years? Seriously?!? I have built a TON of AMD opteron based server/blade systems and have been building a LOT of Intel based gaming rigs ever since the Core2 architecture hit the market. Even the Pentium D 805s were good performers albeit very very hot. Time to catch up with the times sir. Both chip makers see a fairly good amount of market share from all over the board. Some of the fastest supercomputers in the world are also utilizing Opteron/CELL setups. AMD tends to hold down the performance per watt in the server sector pretty well.
[citation][nom]jtt283[/nom]Exactly, Jaded. I was not being sarcastic in my earlier comment; this article WAS useful. Along with the forthcoming article on value, I hope it will lay to rest certain questions about spending this kind of money. It also opens up some good questions for future articles. If you add $500 to an existing system, what will provide the most benefit? Additional GPU? Water cooling? RAM? For example, suppose THG started with the existing $625 machine. What's the best, most cost-effective way to improve it (with an additional $xxx)? The forum is full of upgrade threads, and some articles along those lines would be very useful, likely including benchmarks and other hard numbers.[/citation]
I think in those kinds of scenario's you will always see a GPU ran out of power or CPU ran out of power depending on what you start with. THG has actually covered that quite a few times with their TRI-SLI / QUAD-CrossFireX tests. Generally you start at X, and you either run out of general computing horsepower or visual horsepower and it can be predicted only if you have some practical knowledge and experience with the hardware. Or if you just sit and read THG all day like me