Talk:Triage Reevaluation Project
[Aradan - 6/29/2012] One comment about Low Voltage processors - traditionally they were actually sold for a premium (sometimes a lot) over equivalent (or even faster) standard voltage parts because only the best-behaved chips can actually sustain decent clock speeds at low voltage. So they were generally treated as premium parts with a particular use case in mind (namely, performance per Watt), rather than budget parts. I guess that fits with the idea of shifting from a "low-end" category to a "budget" category pretty well, but the chips themselves might have more value than higher-clocked, standard voltage parts, even if they have lower performance.
[Aradan - 6/29] Reading it back now, I'm not actually sure what my point was above.
[Aradan - 7/27/2012] I think what I meant above is that we should think of processors that have low absolute performance in two subcategories: low-spec/budget models and efficiency models. Furthermore, as improvements in features, instruction sets, and power/heat envelopes are brought to to processors, even low-spec models include a lot of them, so while the general performance of a recent budget model may not be much different from low-end or mid-range procs of yesteryear, they may run cooler, use less power, and be significantly better at certain tasks or support modern features (e.g. include vastly superior integrated graphics, or dedicated fixed-function hardware for video transcode, or an on-die memory controller that supports DDR3 rather than DDR2, etc.). This is where the impetus for breaking the paradigm of performance determining spec levels comes from - absolute processing performance matters only to the extent that it is "good enough" for the intended task, whereas the feature set and platform shape what use cases that device is well-suited to. So maybe we should drop the Low-Spec/Budget as a broad category, and High End as a broad category, and basically have Freekbox Spec and then a bunch of function-driven categories like Efficiency, Budget (as in, the use case is we simply want to be able to provide cheap computers as a way of increasing access), Gamer Box, Workstation, etc. Basically, just have Freekbox and Too-Fast-To-Give-Away, and Everything Else as the broad categories and then create the use-case driven archetypes from there. Buh.