When Does Intel Usually Do the First CPU Step Up

[youghmama]

Seeing some varying results regarding overclocking and heat dissipation of different IB units, it makes one wonder if the next step release will offer a significant improvement. So how long do we normally have to wait for the first step after a CPU release?

 

[NathanP2007]

+1 I am curious too

 

[pxc]

.

So how long do we normally have to wait for the first step after a CPU release?

A stepping may not fix the "problem." The higher power portions of the chip occupy less area than 32nm SNB, and that seems to be a problem. (The average watt-density of a 95W/221mm^2 32nm chip would be nearly the same as a 77W/170mm^2 22nm chip, for example. But the "hot spots" on the 22nm chip would occupy less area, which could be limiting the 22nm chips overclocking and region cooling.)

I do wonder how Intel will approach the same issue when it hits 14nm, because it will likely be even worse. I wouldn't be surprised if flip-chip packages come back on the desktop, or if some type of 2-sided, stacked cooling arrives. Or worse, if chips start shipping even more TDP restricted.

 

[AthlonXP]

I would say they would go with the TDP restrictions first.

 

[youghmama]

.A stepping may not fix the "problem." The higher power portions of the chip occupy less area than 32nm SNB, and that seems to be a problem.

True, but some of the reviews have been saying that the earlier chips could overclock well but had heat issues. While the more recent ones dont have the heat issues but can't OC like the ones with the heat issues. From this, I was thinking that Intel is still playing around with things, and maybe they will play around with things more after the initial release.

In the "old days", going back to when the flip chips were around, I recall everybody discussing which "step" you had and how one step would OC better than the other. I dont think many peeps got involved in exactly what the differences were between the steps. Obviously, OCing has gotten a bit more sophisticated. LOL

I am thinking that this release is something of a dud for everybody except Intel which can now make more money because they get more chips per wafer. Improvements are marginal and really price is going to control the Sandy v. Ivy question.

 

[pxc]

Yeah, the chance, or from Intel's perspective, the necessity of a fix might not be coming. IMO, the heat issues, at 14nm if not at 22nm, are going to require some new fix that may not come from the silicon itself.

 

[aphexcoil]

I am thinking that this release is something of a dud for everybody except Intel which can now make more money because they get more chips per wafer. Improvements are marginal and really price is going to control the Sandy v. Ivy question.

I don't think the chip is a dud. Intel knows that the future is in mobile devices. How many users really mess around with overclocking? 5% perhaps? Possibly 10%? The rest of consumers who are purchasing ultrabooks just want a fast chip with decent integrated graphics and more battery life. IB gives them all of that on a die shrink plus a modest 10% speed increase clock-per-clock over SB. That's pretty impressive for something that should just be a die shrink.

The enthusiast market has always been a niche market. However, since this is a completely new process that is supposed to take Intel to the 7nm node eventually, I would assume that they're aware of the heat dissipation issue and are looking at new ways to get that heat out of the transistors and to the chip's surface.