Ok so Some of you know and have seen a few of my posts in other threads. So you that have will already have a greater understanding of what theories I have been pondering.
First, information sucks. Neither AMD or Intel give specifics about their chips and overclocking of them. Now partly for the OCing aspect, it is prob to free them from fault if you blow your **** up. But still having an understanding of what each part of the chip does and its reasonable limits would help. Now I know they cant say things like the max OC is 5.2 or whatever, but knowing the max safe voltage and temp with minimum cooling apparatus would help. And aspects like the PLL voltage. There are gobs of speculation into the safe range of voltage and what it does but no hardcore facts from the manufacturer. Anyway that is why we exist on this thread like so many others: learn from what each other does.
That being said, my theory is as thus: How do we truly know the which of AMD and Intel is truly superior at any given time ( at any given time because it does change)? Well there is the argument that current benches would answer that. And that assertion would indeed be correct. Current benches would reflect current software with a few exceptions notably HWBOT, but it is a long standing ritual and even I wouldn't say tear it down. But then that opens up the argument, and this is what I have been injecting into other threads with some good response, how can we be sure that the bench is indeed fair as in compiled to give each architecture an equal run. Cinebench for sure is not. But it seems to be the bench of choice lately. Of course that could be because it favors the other manufacturer ( but that is another point to come further down). So now we are starting to see that the reason for AMD's somewhat poor benches compared to Intel is in fact how some of the benches are compiled and as ALL of us here now know x87 instruction set was disabled on AMD's recent CPU/APU offerings. ( I am very curious as to why this is myself. Seems they set themselves up to fail intentionally. x87 like it or not was still in use SKYRIM most notably.) But even with that fact there is one really good reason we cant scream FOUL!
Fact is Intel has a very huge market share in the low 80%s. So it is not reasonable to write software to the advantage of AMD. Now maybe I worded that poorly, I am not insinuating anything underhanded. But we have to accept that, and this part was hard to get one particular software writer/designer to understand, when any company writes new software, chances are that they wrote it on an Intel system. And that leads to the fact that as they were writing the new software they were checking its efficiency on an Intel computer. See how this goes and the whole process going in Intel's favor? I even stated to the afore mentioned writer that this is indeed how it must work and that he didn't write and test the program on a napkin. Probably the only place AMD gets some love is in the new game market. Even then though it just gets them a bit more favor but not to the extent Intel has. Really they are designing games to sell and the large portion/ nearly all are using Intel. Make a poor running game on Intel and have AMD in the catbird seat and see how it's sells do. But all this is just the fact of the moment <-- keyword for next part.
Now how do we get the change we need to compete. It isn't in designing a new processor. I believe that if we could truly test each manufacturer AMD might still be behind Intel in say 60-70% of specific CPU benches but that margin would be incredibly thin. And the other 30-40% would put AMD ahead. Dare I say the final score out of 100 for each would be Intel 94 and AMD 91 (just showing how close not an actual score <-- not for you AMD guys but any Intel troll that may lurk. Not you Alatar.
). Each architecture has its own advantages and disadvantages. But since I am talking true benches it would be hard to say which specifically, remember this is all theory. Moving on... Now again what do we change to compete = SOFTWARE. If anything about x87 or compilers used in benches has taught me it is that we need more software to be AMD friendly. We are in luck because that part is already starting. It wasn't the FX line that did it, it is the APU's. They have been getting a good foothold in the prebuilt lines of tablets, economy desktops, laptops, and even a few servers. In order to sell your product with these increased profit margin APU's you had better make sure they run included software well. So then guess what, next thing you know we have software that is AMD friendly. Benches with better AMD compilers start to surface and AMD sarts to look like a real competitor. But wait did AMD release a new architecture? No the same old will start to look like some new patch took effect or something. Of course this path does take a while. I figure by 2nd quarter next year we are gonna see a new direction that includes us for once.
Now I wanted to share with you guys this and see your thoughts. Keep in mind I am no programmer just someone that sees an issue and ponders the why it is that way, hence my air-conditioned cabinet and soon a new radiator design. BE GENTLE.