Advanced Micro Devices Inc (AMD)

[wbmw]

Combjelly, Re: Repeat after me, engineering silicon and 6 months to product launch.

Back to the AMDroid mantra again? Read the post from Ed Stroglio on Overclockers.com. He pretty much hits the nail on the head.

Opteron may be AMD's saving grace, since it could very well offer superior DP performance (providing they can ramp megahertz), something that Intel can't achieve with their current line of Xeons (unless they add cache or ramp megahertz). Opteron will be the first to get 64-bit x86-64 ported software, they will be the first to get solid DP infrastructure and be AMD's forceful entry into this market. Even if it is just hype, the hype seems to be working in the press for now, which is a good indication that Opteron will be successful.

Now comes Athlon 64. If you read Ed's post, it becomes clear that there is little room for improvement in the actual core. These synthetic benchmarks are not going to lie. They prove that actual core improvement is about 5%, while the greatest micro-architectural enhancements go towards the memory subsystem. This will help Athlon 64 to compete very well in memory intensive applications, while CPU throughput intensive applications will be no better than Athlon XP at the same clock.

This means that AMD's PR rated comparison will be all over the place. Either they'll have to guide rather conservatively, or they'll have made their QuantiSpeed nonsense a real spectacle. You can't say, "Wait for production silicon," this time, since the core is not likely to improve - just the frequency. You will end up with a 2GHz processor that deserves 3400+ in some apps and 2500+ in the rest (even compared to the Athlon XP), but AMD will still call it 3400+.

I plan to sell my shares as soon as the initial Opteron news passes, because there is little chance that AMD will be able to ramp the product for revenue before more bad news floats this way.