Why doesn't Intel move to 3-5 year desktop CPU product cycles?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

nyker96

Diamond Member
Apr 19, 2005
5,630
2
81
Since Intel currently is improving CPU performance by ~8% per year, why not move to a 3-5 year product cycle for desktop PCs? I mean, for most people it's pointless upgrading for anything less than atleast a ~50% performance improvement. Also, at current TDP levels further TDP decreases are not that important for desktop PCs, unless you're building a fanless AIO system or similar.

Moving to a 3-5 year product cycle would save a lot of development costs. Also it is not so strange actually, since that's the product cycle time Intel used to have earlier when going from 286->386->486->Pentium. And still, then the CPU performance improvements per year were actually higher than currently.

For reference on which Intel CPUs were released what year, see:

http://www.intel.com/pressroom/kits/quickrefyr.htm

it's kinda sad, it used to be like +30-40% or more per generation. now it's like 5-15% is already pretty good. i mean for me if it's not 50%+ it's like wasting money, but even more, most of stuff I do now are not every cpu limited. maybe we just need a killer app that will use up all the cpu cycles to push desktop sales.
 

WhoBeDaPlaya

Diamond Member
Sep 15, 2000
7,415
404
126
The current refresh system works because we don't all buy at the same time. I just put together an Ivy system because I don't like testing motherboard BIOS revisions.
Honestly, I'm pleasantly surprised with the Z87 (well at least the MSI Z87-GD65) thus far.
Other than 1st gen SandForce drives going nuts on it (50-50 blame), it's been running very smoothly.

I've usually done what you did in the past - jump onto a new platform >= 6 months (or thereabouts) after launch.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
it's kinda sad, it used to be like +30-40% or more per generation. now it's like 5-15% is already pretty good. i mean for me if it's not 50%+ it's like wasting money, but even more, most of stuff I do now are not every cpu limited. maybe we just need a killer app that will use up all the cpu cycles to push desktop sales.

I also remember Doom as a beautiful game. But when I load it up its huge squares.

The last time you had 30-40% was from P4 to Core 2.
 

Fjodor2001

Diamond Member
Feb 6, 2010
4,224
589
126
With a 30-40% bump in power consumption every time?

We would have 1000W CPUs now if that had to continue.

True. But since the TDP started at such a low level, TDP increases was not such a big problem back then. So we could see 50-100% CPU performance increases on a year to year basis to a large degree driven by frequency increases (but also uarch changes).

Also, the TDP increase per frequency increase is much higher now than back then. If we were to double the frequency now (from 3.5 GHz => 7 GHz), the TDP would be insane. In fact it is not even possible, since the chips would not be stable. :eek:
 

Wall Street

Senior member
Mar 28, 2012
691
44
91
Nehalem's (e.g. i7-870) quadcore die size is 296 mm2 while Haswell's (e.g. i7-4770k) quadcore die size is 177mm2. I see a big reason for Intel to keep iterating their process and technology. Although there is little market for a higher priced/higher performance chip, Intel has been able to deliver a slightly faster chip that is much cheaper to manufacture for Intel each year, which means they profit margins remain healthy even given the recent slowdown in PC sales.

Also, yes, Intel could still compete with a original Core i series processors in 2013, so in a way a lot of their development cost recently has been less useful. However, outside of performance improvements, Intel developed Finfet transistors (a.k.a. "3D") which are now progressing towards 14 nm next year with 10 nm in the pipeline, they have figured out how to improve their integrated graphics with an on-package L4 cache (a.k.a. IRIS), and they have integrated voltage regulation that promises to allow them more granularity in power gating in the future (FIVR). It should take the rest of the industry years to catch up to these new technologies.
 

blackened23

Diamond Member
Jul 26, 2011
8,548
2
0
Yes, but prior to that it was quite common. Just the frequency increases alone did a lot to improve performance back then (although with increased TDP as a penalty).

Check out the history here:

http://www.intel.com/pressroom/kits/quickrefyr.htm

What am I looking at here? You linked a list of processors as if it's some type of evidence to tie in to your argument but it is ONLY a list of processors. This does NOT help your argument - you linked a list of processors with no corresponding data. 30-40% IPC increases on a yearly basis has certainly never, ever been common. In fact, in decades past IPC increases were minimal with all increases coming from raw architecture MHz clockspeed. Beyond that, once multi core became common - 30-40% increases on a yearly basis certainly were not the norm.
 
Last edited:

Fjodor2001

Diamond Member
Feb 6, 2010
4,224
589
126
What am I looking at here? You linked a list of processors as if it's some type of evidence to tie in to your argument but it is ONLY a list of processors. This does NOT help your argument - you linked a list of processors with no corresponding data. 30-40% IPC increases on a yearly basis has certainly never, ever been common. In fact, in decades past IPC increases were minimal with all increases coming from raw architecture MHz clockspeed. Beyond that, once multi core became common - 30-40% increases on a yearly basis certainly were not the norm.

I never said we had 30-40% IPC increase per year back then. I said we had 30-40% performance increase per year (or actually more than that some years), which is something completely different. The performance increase back then to a large degree came from frequency increases and not IPC increases, which I also said before.

In the end, if all you case about is CPU performance increase, it does not matter whether you get it through IPC or frequency increase (or both).
 
Last edited:

blackened23

Diamond Member
Jul 26, 2011
8,548
2
0
I never said we had 30-40% IPC increase per year back then. I said we had 30-40% performance increase per year (or actually more than that some years), which is something completely different. The performance increase back then to a large degree came from frequency increases and not IPC increases, which I also said before.

In the end, if all you case about is CPU performance increase, it does not matter whether you get it through IPC or frequency increase (or both).

This is also NOT true.
 

CHADBOGA

Platinum Member
Mar 31, 2009
2,135
833
136
I never said we had 30-40% IPC increase per year back then. I said we had 30-40% performance increase per year (or actually more than that some years), which is something completely different. The performance increase back then to a large degree came from frequency increases and not IPC increases, which I also said before.

In the end, if all you case about is CPU performance increase, it does not matter whether you get it through IPC or frequency increase (or both).

This is also NOT true.

Pretty sure it was true from around 1994 to 2002ish

We had a Pentium 1 at 66Mhz and ended up with a Pentium 3 at 3Ghz
 

Ratman6161

Senior member
Mar 21, 2008
616
75
91
Ever since the core 2 Duo the vast majority of people have had a computer that was "fast enough". For home users, the increased availability, higher speed, and lower prices of broad band connections have probably had more of an impact than improvements in computers themselves. But the thing is, just because they make it, doesn't mean you have to buy it.

Personally, though I rank myself as an "enthusiast" for my home equipment I usually skip generations - sometimes several generations. The system before the one I have now was a core 2 quad 6600 @3.2 Ghz. I skipped some and my next (and current) system is a 2600K @ 4.4 Ghz. I skipped IB and will skip Haswell too. But....I always have my eye on whatever the latest and greatest is and keep watching for just the right time to upgrade. Others like me may have had something newer than my C2Q and so skipped SB but did get IB. Everyone ends up in different places in the cycle and the constant incremental improvements give us the ability to choose our "just right" time in terms of price and performance.

The PC's I manage at work however, are a completely different story. We buy on a set 4 year cycle. So the 3.0 GHz Core 2 Duo desktops we bought in 2009 are now being phased out and replaced with i5's. Thing is, no one was complaining about the performance of their 2009 systems. In fact a number of people are actually complaining about "having" to get a new computer because its all the same software (sticking with Windows 7) and they have to get sorted out on a new machine that doesn't actually do anything more for them.

As enthusiasts and IT folks, we would get excited if someone told us we could get an IB i5 quad core in place of the old C2d - but for my users its just a pain with no gain. They can't type their Word documents or send their mail any faster than they could before. We really only replace the old machines because their support/warranty has expired so if they start breaking now I have to fix them myself rather than picking up the phone and making Dell send someone out. So we buy whatever is the current generation at the time of purchase. In that regard, steady incremental improvements are good. I don't have to feel bad because I bought 55 new machines with Ivy Bridge CPU's just before Haswell came out.
 
Last edited:

tential

Diamond Member
May 13, 2008
7,348
642
121
Ever since the core 2 Duo the vast majority of people have had a computer that was "fast enough". For home users, the increased availability, higher speed, and lower prices of broad band connections have probably had more of an impact than improvements in computers themselves. But the thing is, just because they make it, doesn't mean you have to buy it.

Personally, though I rank myself as an "enthusiast" for my home equipment I usually skip generations - sometimes several generations. The system before the one I have now was a core 2 quad 6600 @3.2 Ghz. I skipped some and my next (and current) system is a 2600K @ 4.4 Ghz. I skipped IB and will skip Haswell too. But....I always have my eye on whatever the latest and greatest is and keep watching for just the right time to upgrade. Others like me may have had something newer than my C2Q and so skipped SB but did get IB. Everyone ends up in different places in the cycle and the constant incremental improvements give us the ability to choose our "just right" time in terms of price and performance.

The PC's I manage at work however, are a completely different story. We buy on a set 4 year cycle. So the 3.0 GHz Core 2 Duo desktops we bought in 2009 are now being phased out and replaced with i5's. Thing is, no one was complaining about the performance of their 2009 systems. In fact a number of people are actually complaining about "having" to get a new computer because its all the same software (sticking with Windows 7) and they have to get sorted out on a new machine that doesn't actually do anything more for them.

As enthusiasts and IT folks, we would get excited if someone told us we could get an IB i5 quad core in place of the old C2d - but for my users its just a pain with no gain. They can't type their Word documents or send their mail any faster than they could before. We really only replace the old machines because their support/warranty has expired so if they start breaking now I have to fix them myself rather than picking up the phone and making Dell send someone out. So we buy whatever is the current generation at the time of purchase. In that regard, steady incremental improvements are good. I don't have to feel bad because I bought 55 new machines with Ivy Bridge CPU's just before Haswell came out.

It almost gets annoying to hear even "enthusiasts" say they need faster CPUs. A lot of the enthusiasts here only game. Last thread this came up and I asked "What applications would benefit from this increase in speed?" The answer was literally benchmarks.... Maybe 1 person had a legit reason. Last time I checked, intel sold processors so that people could use them, not so they could be ran through some arbitrary benchmark.

I'm happy with the direction intel is going. I don't think desktops have too long of a future at this rate. Because most applications aren't CPU bound at all, we just can make CPUs more efficient. I'm guessing 2025-2035, somewhere inbetween that, Desktops will be a relic, and enthusiasts will have moved towards using fully customizable notebooks, while the rest of the population is on tablets(converters/addon keyboards for those who need keyboards)/smartphone computing.
 

escrow4

Diamond Member
Feb 4, 2013
3,339
122
106
I've given up on CPU upgrades. My gaming box runs a 3930K @ a moderate 4.1GHz, and my work system runs a non K 3570 @ 3.4GHz. Neither will be upgraded for years at this rate. Even for gaming, I'm skipping Ivy-E/Haswell-E. Sure it will be faster, but is it really worth upgrading to? That 3930K has no hesitation keeping minimum and average frames up . . . .