Curiousity about marketing (intel)

Risant

Risant

Limp Gawd
Joined
Jun 19, 2004
Messages
160
I understand that all, or nearly all of the unlocked SB processors are easily reaching 4.0 - 4.5+ Ghz, often with stock cooling. Why is it then that Intel doesn't sell these parts as 4+ Ghz processors. From a marketing standpoint it is insane they wouldn't take the opportunity to really stick it to AMD.

That is unless of course that they are using the overclockablity as their marketing sales point. (I'm sure that the non-K processors are likely the ones that just don't meet the standard and aren't stable at overclocking)
 
From a marketing standpoint it is insane they wouldn't take the opportunity to really stick it to AMD.
Click to expand...

1. They do not want to totally put AMD out of business so that governments do not come down even harder on them.

2. Also they limit the stock + turbo TDP on all lga1155 processors to 95W and with ivy bridge the new limit will be 77W. The reason for these limits is to try harder to not compete with lga2011 like lga1156 competed with lga1366.

3. I bet it's nice holding some reserve performance that can be tapped at any time if AMD ever gets its act together.
 
Last edited:
In addition to what drescherjm says remember intel has to ensure their processors are stable at their stock speeds in a motherboard that is right on the limit of compliance with memory that is right on the limit of compliance and a PCIe device that is right on the edge of compliance all while running a worst case workload.

Overclockers OTOH only have to ensure their processors are stable with thier motherboard, their memory and their PCIe device.
 
As stated, Intel has to think about specs, thermal and power use, as well as what every chip they make can hit, its a safe number for them. You also have to think about the places and uses they will be put in, OEM set ups, that are never cleaned, and abused, etc, so Intel putting out a 4.5ghz+ chip is not practical, sure, many of them can hit that, but you have to think about many other things than just what the chip can push on highend air cooling in a nice A/Ced room.
 
I'm pretty sure it's primarily TDW, and having to state that "X" fits into this power envolope. You can say the reverse for some of their other products that "didn't" overclock well, that they were much closer to their thresholds (in comparison to SB), but were able to say they got this many Ghz out of this much TDW.
 
It is about TDP and numbers. With a stock cooling solution a person who lives in the Northeast or Northwest doesnt have much problem with dealing with the processor reaching 4+ GHz while a person in Phoenix, Arizona who has a swamp cooler wont have the same success with the processor in trying to overclock it. In the end since we are a global company we have to be mindful of all the different environments that are processors are going to and build our projection for TDP on a wide range of environments that the processor can find itself in.
 
Yeah, TDP is the main thing that holds it back.

LGA1155 Ivy Bridge will be more of a legitimate target for the OP's question. Pulling back to 77W, which (most regular) desktop boards and cooling are already designed for 95W, means a lot of performance is held back vs release at LGA1155 SNB TDP levels (there is value with touting Intel's power reductions with 22nm/Tri-Gate process, using IB as the vehicle). Lower top speed LGA1155 IB will of course make Haswell look better that way, but limiting LGA1155 IB performance will be pure marketing.
 
Last edited:
The marketing behind the CPUs is fairly simple thing.
They are manufacturing 2 or 3 types of CPUs. They make wafers with just the highest one and the rest are downclocked, or have disabled features, or lower cache or something. It's in fact all currently made SNB CPUs are weaker versions of i7-2700K. i5 is i7 with HT disabled and 2 MB less cache, i3 has 2 cores disabled and has just 4 threads, celeron and pentiums are only dual cores with HT and turbo disabled and have lower cache and clock rate. All the ones which were made with some defects are also sold as weaker versions with the damaged circuitry disabled(IGP-less i5s, Celerons, Pentiums).
 
Acer_Sheep said:
It's in fact all currently made SNB CPUs are weaker versions of i7-2700K.
Click to expand...
Not exactly. There are 3 different desktop models (3 separate processors) plus the dual/quad core mobile versions, which are tweaked during manufacturing for lower power consumption instead of higher clock speed.

The 3 die are: the 216mm^2 one with 4 cores and 12EU graphics, 149mm^2 one with 2 cores and 12EU graphics and 131mm^2 one with 2 cores and 6EU graphics. With various tweaking done during manufacturing, there are at least 6 distinct SNB processor models for desktop and mobile use.

Also, binning of the CPUs to higher TDP (mobile especially) and turning off non-functional portions (i.e. less cache, lower spec graphics, lower clock speed, etc) can be done to increase yield from marginal parts. Or sometimes not... fully functional parts may be downgraded to fill demand of lower models.
 
TDP. That is all.

Wonder if we'll ever see the day when die shrinks are so small and power usage is so miniscule, especially with the dedicated lower TDP parts, that passive heatsinks will start being bundled again just like in them olden days.
 
DejaWiz said:
TDP. That is all.

Wonder if we'll ever see the day when die shrinks are so small and power usage is so miniscule, especially with the dedicated lower TDP parts, that passive heatsinks will start being bundled again just like in them olden days.
Click to expand...

they exist today. Called Atom processors :D
 
pxc said:
Not exactly. There are 3 different desktop models (3 separate processors) plus the dual/quad core mobile versions, which are tweaked during manufacturing for lower power consumption instead of higher clock speed.

The 3 die are: the 216mm^2 one with 4 cores and 12EU graphics, 149mm^2 one with 2 cores and 12EU graphics and 131mm^2 one with 2 cores and 6EU graphics. With various tweaking done during manufacturing, there are at least 6 distinct SNB processor models for desktop and mobile use.

Also, binning of the CPUs to higher TDP (mobile especially) and turning off non-functional portions (i.e. less cache, lower spec graphics, lower clock speed, etc) can be done to increase yield from marginal parts. Or sometimes not... fully functional parts may be downgraded to fill demand of lower models.
Click to expand...
Well you are right but it still does complies with what I said as well. All i7 and i5 desktop models (from 2700k to 2550K and to 2300) are technically same, they have just lowered parameters.
DejaWiz said:
TDP. That is all.

Wonder if we'll ever see the day when die shrinks are so small and power usage is so miniscule, especially with the dedicated lower TDP parts, that passive heatsinks will start being bundled again just like in them olden days.
Click to expand...
With that tri gate technology it is quite possible to achieve that in not very distant future.
 
DejaWiz said:
True, but I suppose I should have elaborated...

I was moreso meaning a processor of the likes of a SB i3 or even i5. :p
Click to expand...

Well in general processors are marketed by their performance, cost and clockspeed not on their ability to run passively cooled.

You'd be surprised though. Look at the Processor in my sig. It is cooled passively with an aftermarket Scythe Mini Ninja heatsink with no cooling fan anywhere near it what so ever and it runs extremely cool never exceeding 46C even under load. I only have 2 fans in the entire system and I suspect I could get by with only 1 fan if I had a higher efficiency PSU. I bet if you could cool this passively you could also do it with an i3 with ease. (they make a 35W TDP i3)
 
I suspect it's because making a smaller heatsink and putting a fan on it is cheaper than making a larger heatsink. Plus, a larger heatsink will not fit in all cases.
 
Tsumi said:
I suspect it's because making a smaller heatsink and putting a fan on it is cheaper than making a larger heatsink. Plus, a larger heatsink will not fit in all cases.
Click to expand...

Absolutely.
 
ZodaEX said:
Well in general processors are marketed by their performance, cost and clockspeed not on their ability to run passively cooled.

You'd be surprised though. Look at the Processor in my sig. It is cooled passively with an aftermarket Scythe Mini Ninja heatsink with no cooling fan anywhere near it what so ever and it runs extremely cool never exceeding 46C even under load. I only have 2 fans in the entire system and I suspect I could get by with only 1 fan if I had a higher efficiency PSU. I bet if you could cool this passively you could also do it with an i3 with ease. (they make a 35W TDP i3)
Click to expand...

Yes, however your HS is a performance aftermarket unit, not OEM from the CPU box. :p
 
DejaWiz said:
Yes, however your HS is a performance aftermarket unit, not OEM from the CPU box. :p
Click to expand...

It could be, but that's the point, a small HS with fan is almost always going to be cheaper than the larger HS needed for passive, stock HS's are very small and cheap, that's why you are not going to see passive HS with normal cpu's. However you do see them now on server class cpu's, most starting at the 2U size, going up to larger tower style coolers, and yes, they are made by intel, but no, you will not see them bundled with desktop chips, why would they? When what they use now "will do" and is far cheaper? It's a special use case, if you want or need passive, you will need to spend more, but for the masses, a HS/fan combo will do just fine and cost less.
 
Ok, I'll elaborate some more. Remember the passive sinks included with 386 and the first 486 processors that had a footprint the size of the die they were covering? Yeah, that size.
 
DejaWiz said:
Ok, I'll elaborate some more. Remember the passive sinks included with 386 and the first 486 processors that had a footprint the size of the die they were covering? Yeah, that size.
Click to expand...

If you want the same kind of processing power, with ease, you would not even need a HS at all. The amount of power we cram into a chip is shocking, sure we might reach a point of super low power we can do away with HS altogether on chips like I3 and I5, but by that time, new things will be out, and these chips will be as antiquated as the 386. It's a strawman.
 
DejaWiz said:
True, it really is just that.
Click to expand...

Granted, we would all love to see it I am sure...And well, never know, with things shrinking, we might see it some day (far far from now). But with that in reach, we would probably pack more and more in or bigger chips which would then need better cooling yet again. I think the HSF combo is here to stay, for the foreseeable future at least. Here's to waiting for 10mw CPU's that can play the latest games.
 
You must log in or register to reply here.
Back
Top