Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Why is intel CPU's stagnating in the past 3 years
- Thread starter nec207
- Start date
- Sort by reaction score
Umm clock speed is not mean everything. I thought the mhz myth was dead. If you think clock speed is everything go get a late generation p4. They were at much higher clock rates.
Remember the die size of the chips have gotten smaller that means the info moves faster threw it even at a lower clock rate. There are tons of factors for speed
Remember the die size of the chips have gotten smaller that means the info moves faster threw it even at a lower clock rate. There are tons of factors for speed
It's not just speed. It's size, cores, battery life, etc. And it's not just CPU's, it's chipsets too. The average macbook pro from 5 years ago was much slower, and got 3 hours of battery life. Now they're way faster and have over double the battery life.
1. 50% "increase" (whatever you mean with this) isn't "stagnating".
2. The performance of the MBP 13" base model has increased more than 100% since I bought my MBP, mid 2010.
3. Intel's problem is probably that education is getting worse, and they find it hard finding engineers who can write two sentences without any mistakes, never mind finding engineers who can actually improve an already complex processor.
It's because they're also going for cooler.
The CPUs now use less watts than before, theres that, and the fact that AMD hasn't been much of a competition lately which gives Intel less of a reason to go at it faster.
With as much profit as Intel brings in I doubt they struggle to find engineers
. They just may not be from North America or Europe.Where did I say clock speed? It was intel that promoted faster clock speed and more cores = faster CPU that is wrong . Better architecture and design = faster CPU . Look computer engineering is about architecture and design not clock speed or cores. In past AMD had lower clock speed than intel but did more work per clock cycle.The part about clock speed it hit brick wall around 2004 with out it overheating.
Well has of now AMD phenom CPU's have way more cores than intel but intel better architecture and design do more work with less cores and much faster than AMD CPU's.If AMD does not bring out some thing really good out very soon they may go under has they are not doing well to intel.
Again faster clock speed and more cores does not always mean a faster CPU .The computer engineering is all about architecture and design . But do to the public and young people learning how computers work they would not understand this so the media hype is talk about clock speed and cores than architecture and design leading to faster CPU.
Also a 8 core and 16 core will do nothing if the software is not coded for it.Yes there are some dual core CPU's fater than 4 or 6 core CPU's.
Last edited:
Probably because when Intel was promoting the OMG FAST CPU MY GIGAHERTZ ARE HIGH idea, they were getting their asses handed to them by PPC and AMD CPUs, which could do considerably more at considerably lower clocks.
And even if you don't have any specific programs that can take advantage of more than 4 cores (truthfully, most consumer apps use two at most), any modern OS will use multiple cores for multitasking and other rather bland by very necessary low level tasks. So up to a point, more cores does equal a smoother better faster computing experience.
True but in past it would be more like 100% or 150% increase.
You talking about MBP 13" of 2010 to a MBP 13" of 2012 if that true that is 100% may be the video card making faster too and other things than yes that would be alot. And very fast.
Intel does not care they have a monoply just like Microsoft.
Part of the problem too is games hardly use CPU it needs a good video card .Where video editing and graphics and number crunching use CPU than video card .
Those very high in Mac pro with 6 , 8 and 12 core CPU are going to do really well with video editing and graphics and number crunching but not do so well with very high in video games.
Where people that spend $2,000 on high in PC with state of art video card is going to do really well for high in video games but really bad for video editing and graphics and number crunching .
Yes OS X and all of apple software and pro software for video editing and graphics and number crunching so on are going do really well has they are coded to use 6 , 8 and 12 core CPU .
That is not case with windows and software for windows .Also it was not really long ago before you had 64 bit windows , 64 bit software where apple been into 64 bit long ago.
True but in past it would be more like 100% or 150% increase.
During the generation jumps 10-20 years ago maybe. But now? Intel CPUs are so fast, the vast majority of the market isn't nearly as hard pressed for extra processing power. They've already got more than what they'll ever conceivably need. This isn't like in the days of the 386 and 486, when an upgrade was the difference between a computer that only barely chugs along while using a word processor, and a computer that can fly through text at the speed of abstract thought. This is in contrast to today, where you can still use an old P4 for Word 2007 almost as well as the latest i7.
These days, it's more about efficiency than raw speed upgrades. Lower power requirements, cooler temperatures, better battery life. This is what Intel is focusing on. The jump from Sandy Bridge to Ivy Bridge may have only netted a small increase in performance, but it's a much more efficient chip overall.
So no. Intel isn't stagnating, they've just switched gears.
I had my buddy render one of his architecture designs; a thirteen story building, on my 2012 13" mbp w/8gb ram. We did it 14 times simultaneously. Proc never peaked over 64% and ram peaked at 35%.
Anybody having issues with their new computer keeping up with what they are doing should consider wolf packing servers and surrender to he fact that a personal workstation is not enough computer for what they are doing. Or, they are doing it wrong.
Anybody having issues with their new computer keeping up with what they are doing should consider wolf packing servers and surrender to he fact that a personal workstation is not enough computer for what they are doing. Or, they are doing it wrong.
Nope. From a pure processing standpoint, a 4 core processor would do just as well as a 12 core. It just wouldn't use all the cores available.
I wouldn't say a gaming GPU would be really bad at it, just not as well suited. A Geforce GTX 690 is doing about the same thing as a Quadro 6000, it's just that the Quadro is geared more towards queueing high end materials for 3D textures and high res video clips. Truthfully, a Quadro these days is less some super specific production card, and more a low end Geforce with a ton more ram and better CUDA support.
Wuh? And they're not for Windows? PS, Pro Tools, Maya, etc, all utilize the same amount of cores and leverage the processor the exact same way on a Windows machine as it does a Mac. There isn't anything about OSX that makes it more multiple core friendly.
And I might be wrong about this, but I believe Windows has been "true 64-bit" for a little longer than OSX has.
I see a lot of people take the wrong point, he's not saying the speeds are getting stagnant, just that the performance increases are lower than they were before.
But as i said it's due to intel focusing on using lower power, just go and look at how much lower the TDP has gone down, from 95W down to 77W and the new parts are still a bit faster.
And also i think ivy bridge and the move to 22nm was slower due to an ineffective AMD as true competition
But as i said it's due to intel focusing on using lower power, just go and look at how much lower the TDP has gone down, from 95W down to 77W and the new parts are still a bit faster.
And also i think ivy bridge and the move to 22nm was slower due to an ineffective AMD as true competition
I was saying before it was like 100% or 150% increase and now is like 50% increase CPU in 2009 to CPU 2012 .
Windows XP is not 64 bit OS and nether is windows vista that came out in 2006 only some windows vista is 64 bit.
Windows XP does not support multiple core only SP 3 for Windows XP and still will not be is good has windows 7 .
Most people do not use more than 4 cores for PC
It is not about clock and gigahertz... Companies now make their chips much more efficient in terms of instructions and executions per second regardless of current clock cycle.
Re -read the post above .
computer engineers do not talk in terms of clock or cores that is marketing term for the average public.
A 2.2 GHz quad core can be 2 times faster than 8 core 3.5 GHz if the architecture is better and do more work per clock cycle
----------
No this information is wrong both AMD and intel are still making fast CPU every year it just not fast like they where in the past .
And part that is video cards now are super fast than what they where 2 or 3 yaers ago and this what most people use for games.
People not into games use CPU racks and cloud computing for massive CPU power now.
And as i've said a lot, Intel is making cooler CPUs not focusing solely on performance. those cooler cpus use less watts, so actually performance per watt could still be increasing at the same level as before.
Maybe because there's a point where you reach the smallest/fastest transistor that can be built:
One and done: Single-atom transistor is end of Moore's Law; may be beginning of quantum computing
Intel's Secret Sauce
From 2009 to 2012, Intel CPU based products have improved significantly, but not just through CPU architectural changes. Their secret sauce is their process technology.
Intel has gone through two "Tick-Tock" or "Tock-Tick" cycles, Nehalem/Westmere and Sandy Bridge/Ivy Bridge. Each full cycle usually gives a 1.2X cycle for cycle performance increase. Or about a conservative 1.4X improvement.
Second, Intel has integrated more of the system. In Nehalem, Intel incorporated the North Bridge, or memory controller. Doing this allows them higher IO speeds, with less latency. In Sandy Bridge, they brought in the South Bridge, which includes 2D/3D graphics. More functionality, reduced power, improved reliability, and longer battery life are the primary benefits.
Third, from 2009 to 2012, Intel has stayed on Moore's Law, and steadily drove their processing technology, being the first to introduce High K Metal Gate CMOS in their 45nm process, and first to productize 3D FinFet transistor technology in their current 22nm process. From a simplistic V-Squared x Scaling Factor, it looks something like:
Process Voltage Scaling Factor
45nm 1.2V 1X
32nm 1.1V 0.8X
22nm 1.0V 0.56X -> 1.44/0.56 = 2.57X
So, CPU Performance*Power = 1.44 x 2.57 = 3.7X
But, with all this, I do believe that Intel could improve their cycle for cycle performance dramatically, but they would have to lose some of their backward compatibility. If Intel could start a new, it would be frightening how advanced they could be. Let's see how Haswell/Broadwell turns out.
From 2009 to 2012, Intel CPU based products have improved significantly, but not just through CPU architectural changes. Their secret sauce is their process technology.
Intel has gone through two "Tick-Tock" or "Tock-Tick" cycles, Nehalem/Westmere and Sandy Bridge/Ivy Bridge. Each full cycle usually gives a 1.2X cycle for cycle performance increase. Or about a conservative 1.4X improvement.
Second, Intel has integrated more of the system. In Nehalem, Intel incorporated the North Bridge, or memory controller. Doing this allows them higher IO speeds, with less latency. In Sandy Bridge, they brought in the South Bridge, which includes 2D/3D graphics. More functionality, reduced power, improved reliability, and longer battery life are the primary benefits.
Third, from 2009 to 2012, Intel has stayed on Moore's Law, and steadily drove their processing technology, being the first to introduce High K Metal Gate CMOS in their 45nm process, and first to productize 3D FinFet transistor technology in their current 22nm process. From a simplistic V-Squared x Scaling Factor, it looks something like:
Process Voltage Scaling Factor
45nm 1.2V 1X
32nm 1.1V 0.8X
22nm 1.0V 0.56X -> 1.44/0.56 = 2.57X
So, CPU Performance*Power = 1.44 x 2.57 = 3.7X
But, with all this, I do believe that Intel could improve their cycle for cycle performance dramatically, but they would have to lose some of their backward compatibility. If Intel could start a new, it would be frightening how advanced they could be. Let's see how Haswell/Broadwell turns out.
Unfortunately, the market would never allow for that. x86 has built up too much momentum. Remember Itanium?