Re: x86 has relatively few general purpose registers unless it is in 64 bit mode.
Which is an unimportant consideration to performance, thanks to register renaming. x86 processors continue to be the best performing processors in 32-bit consumer code in the world.
Re: Code size is a big factor in embedded. It is often one of the metrics used when choosing a processor.
Except, this is an unimportant consideration for Netbooks.
Re: x86 code tends to be pretty branchy with a branch typically every 6 to 8 instructions.
What makes one architecutre more branchy than another? x86 compiled code shouldn't generate any more branches than ARM. I don't know about this "instructions per branch" metric that you're talking about, but I don't see how it's relevant to performance.
Re: With the growing popularity of web apps, all the netbooks and MIDs really need to do is run a browser. If they evolve into laptop replacements, well, that is different. It is also a tougher slog for Atom, because processing power goes up under those kinds of loads.
I don't know what you're thinking, CJ, but when someone sees one Netbook marketed as "Internet web browser" and another marketed as "Internet web browser as well as support for all the applications on your PC", I can't imagine the end user not seeing more value in the latter.
Re: What is pretty clear is Atom almost certainly doesn't have a significant edge on IPC. It definitely draws more power at a given clock speed and occupies significantly more silicon, even on a smaller process.
I don't buy this. I've looked at the OMAP datasheets, and I haven't found any significant power or die size advantages relative to Atom. If anything, Atom seems more power and area efficient, but that's due to the 45nm process. All else being equal, ARM is a more efficient architecture, but Atom has made a revolutionary leap for x86 architecture, and while not much of this is obvious at the platform level due to the lack of chipset integration, I already see a lot of promise.
Re: I will point out that AMD seems to be doing ok at 45nm without Hi-K dielectrics.
Based on overclocking experiments, or on anything else that would be relevant to this discussion?
Re: Besides, it doesn't need to exactly match speeds with Atom, closing the gap some is enough.
Actually, ARM needs to bring a lot more value to the table if it wants to compete in a kind of device that promises to bring all the capabilities of a PC to the end user. (Well, almost - we all recognize that Netbooks will be slower than mainstream notebooks, but otherwise they run the same software). For ARM to compete in a market where its power benefits aren't exactly obvious to the end user, it needs to bring substantial benefits, not just "close the gap". I see some vendors pushing battery life, which is good for starters. All day battery for web browsing is a nice feature for some, but losing the PC compatibility is a big tradeoff for the larger volume of end users.
Re: On most netbooks tasks, an OMAP34xx running at a high clock rate is going to give better performance than most Atom implementations. And draw less power and occupy less board real estate.
Before going any further in this part of the discussion, I'd like to see a quantitative comparison. As I said, in my own private investigations, I did not see any disadvantages from Intel at the core level. Of course, on the platform level, Intel has power and board area disadvantages, but these are not as obvious in devices with 7+ inch screens, since even Atom with an i945GME + ICH8 chipset can fit in small form factors with 7" or larger screens, while offering excellent battery life.
Re: Now true, if WinXP or Vista becomes the OS of choice for netbooks and MIDs, then Atom wins. If it doesn't, then it has a fight on its hands.
Why wouldn't the average end user prefer Windows? It's the staple of the PC, and Netbooks are like small PCs. Now, MIDs are different, and Intel is trying to compete with Linux here, where it's a more level playing field. It will definitely be a much harder battle on this front. |
