LukaP

CPu wise it was very competitive in the ZF2, if they keep that up, while improving the GPU to Apples level (which they are very capable of) then i dont see what else. their modems are also very good, and getting integrated, and they are selling at cost to help penetrate Its mostly the same as the Intel Inside branding on PCs. its branding, that the public recognises, without knowing what it means

oh no, the c stands for JOOOOHN CEEEEENAAAAAAA

IDK people say im smart...

that is what we need. What the public wants is Snapdragon! now with 30% more dragon!

not really a given. in any case Kyro seems like nothing groundbreaking, and hence not something that will help QM in any tangible way

i was talking single core performance if i recall corectly.

Exactly. There is no Light, there is no dark, there is only force Though i firmly believe with good moral standards, the "dark side" powers are much more effective than most "light side", but both come to their best when combined

i love you

got mine maybe? i think i lost it during reinstalls

ITS NOVEMBER FFS. Atelast wait a week cmon

id wager otherwise. Not because samsung is some great scientific company (they are), or because theit RND budget is bigger then some countries entire budgets (it is), but simply because their biggest product is memory, and developing new tech for the company wont exactly be a very public process. when they are a year or so from release, we will hear of it and no sooner

http://www.realnebraskagirls.com/Portals/0/0a3332_f7eb.gif

Im not going to go into the coding part today, because sick, but i can comment on the high level perspective. The reason oMP isnt used (or even just simple C++ threading, is because most programmers in games studios have been there since the beginnings and they didnt need to thread anything then, because they only had one CPU. And then when we started getting more and more possible threads, their companies became big and started relying on a release schedule. What that means is you have very little time to learn new things between development/support cycles. And so it comes to be that while most of them are great programmers in general, fresh blood is needed to bring in ideas for new language features that are useful, but the old garde just hasnt had the time to figure out yet. Then there is the team aspect. they have to sync between eachother, and it takes time to setup properly for every new API/language they decide to use

meh with v8 both are basically CISC

number of transistors/area stacking helps, but its really not feasible for logic ciurcits. those heat up too much

Thanks, awesome stuff

well not if you can have finer voltage control, which you can have, if you improve the voltage controller when was that? got a link? naaah. its just the lenght or width of the drain/source. the whole transistor (what from source to drain means) is in the hundreds

voltage takes time to build up. you need a set voltage to switch the state of the transistor. the smaller the transistor, the lower the voltage needed to switch. from that we extrapolate its faster to switch a smaller transistor. same concept applies to exotic transistors made from other materials, like graphene (max achieved switchrate was 40GHz iirc, with silicon doing 10)

Mainly because they used solder (actual metal) to seal the contact between the die and the IHS. now with smaller transistors, solder kills chips (as its applied hot) and they use a special TIM instead. that is the biggest reason for the "higher" temps. Otherwise its placement of functional units that determines hot spots that need to be cooled. usually chips have some dead transistors around those units to help dissipate heat more efficiently these days. actually the number of transistors per core doubled till haswell (give or take a few thousand, and not count the die shrinks as generations).

just to clarify, its not the diameter, its the smallest feature. usually either the lenght of the gate, or the source/drain channels. the whole 22nm transistor from intel is some 120nm across, so there is still lots of room to shrink things (also why samsung and TSMC are lying about their process sizes slightly, as the transistors are still 22nm class) Also, with a good uarch, you can manage the heat by placing high heat units away from eachother (things like ALUs, FPUs, VPUs, anything that does work).

you of all people should know Amdahl doesnt agree with the second statement We will probably see arch go to average instruction time come down to 1.x cycles, and L3 latency come down to 1.x cycles. Then its up to the programmers to utilise the speeds, properly start multithreading everything, and utilise SIMD (and wiht the combination of all of that, MIMD as well). After that happens, Von Neumann is dead, and quantum realms need to be explored

I thought somone was giving away a fighter jet...

just a note agreed not really, as long as they have some travel distance, they are millions of times better than touchscreens. and you get used to a keyboard pretty fast. i have had flash completly disabled on all my computers since April, and i havent encountered one site that didnt work. Nope, i prefer them, because the screen is big enough, and atleast they have all the features there... As said before, most users dont feel the need to organise everything the way they want, they are quite happy to have the "move to ipad", "move from ipad" options, and then have the OS file it where it has to be... byebye

Not saying it doesnt have to be reported. But this is something that needs to be reported to apple, by users. And if then nothing is done, reported by a reputable tech media source that goes deep, researching the problem. No need to be all "iPad is shit and will fail because of this one small thing hapening to 100 out of hundreds of thousands of users"

umm the benchmark i linked is offscreen. The "lowres" doesnt matter. And wow, apple has great software/API/driver stack for its GPU, guess that makes it unfair on android based phones right? Wrong.