Can I get Ivy Bridge motherboard for Haswell too?

[IntelUser2000]

The thing is rumors are saying Haswell will have very flexible overclocking options, and Ivy Bridge is already looking better in that area than Sandy Bridge. It's hard to see then that it will be limited by VRM. I can't see why they would purposely limit it, and if they do, revise hearing from motherboard manufacturers.

Which is why I was suggesting certain high end motherboards might still have the regular VRM addition to handle the higher power. And power usage isn't a problem with those folks of course.

 

[Edrick]

This sort of reminds me of 2 years ago before SB was released. Everyone was saying how Intel was killing OCing by removing the BCLK from SB1155. Here we are talking about Haswell in the same manner due to iVRM. I have faith that Intel is not going to kill OCing as some may fear.

I would put my money on the iVRM only being on laptop/ultrabook CPUs (SOC).

 

[taltamir]

The sad part is, I find it completely believable that they simply won't.

Extreme overclockers are not that big of a market, and this move will be hugely positive to pretty much everyone else. If that means dumping us, it would be pretty daft of them not to.

The higher-end socket ones will likely have much more VRM capacity, and since it's all semiconductors, it will probably be able to handle a lot more power when it's colder, but that's about it.

Well... would it be hugely positive enough to make stock speeds better then what an OC would have been without it?

 

[Tuna-Fish]

The thing is rumors are saying Haswell will have very flexible overclocking options, and Ivy Bridge is already looking better in that area than Sandy Bridge. It's hard to see then that it will be limited by VRM. I can't see why they would purposely limit it, and if they do, revise hearing from motherboard manufacturers.

For most people, it wouldn't be the limiting factor. The proportion of people who run their cpus at 300W is rather low.

Which is why I was suggesting certain high end motherboards might still have the regular VRM addition to handle the higher power. And power usage isn't a problem with those folks of course.

The problem is, this would require the cpu chips to be radically different. If they are not putting any low voltage pathways into the cpus, then the MB cannot supply low voltages, no matter how you design it.

Well... would it be hugely positive enough to make stock speeds better then what an OC would have been without it?

Probably not. While it would increase the headroom of the chip some (because less power spent while distributing power = more power spent by the chips themselves), the biggest gains would be in the lower end.

This sort of reminds me of 2 years ago before SB was released. Everyone was saying how Intel was killing OCing by removing the BCLK from SB1155. Here we are talking about Haswell in the same manner due to iVRM. I have faith that Intel is not going to kill OCing as some may fear.

They did shift OC into the most expensive cpus of the line. A lot of people used to get the cheapest cpu with all cores/cache enabled, and oc that past the stock best cpu. You cannot do that anymore -- now you have to buy the more expensive OC-enabled gear.

This would only be an extension of that policy. I fully expect that the S1150 platform will have awesome OC features for "reasonable" overclocks, but no way of raising the voltage above ~30% of stock. While the actual overclocking will be limited to the high-end sockets. (which have beefier VRM for better stability in server loads.)

I would put my money on the iVRM only being on laptop/ultrabook CPUs (SOC).

Servers would love the power savings, which means it will be found from the high-end desktop socket too. At that point, why would you design the desktop cpu differently?

 

[IntelUser2000]

They did shift OC into the most expensive cpus of the line. A lot of people used to get the cheapest cpu with all cores/cache enabled, and oc that past the stock best cpu.

And if Ivy Bridge brings fixed bclk settings back, that would allow limited overclocking on the lower end parts, no? 133MHz bclk would allow 33% overclock potential. Does current Sandy Bridge systems allow lower multipliers with locked parts or are they completely locked?

The problem is, this would require the cpu chips to be radically different. If they are not putting any low voltage pathways into the cpus, then the MB cannot supply low voltages, no matter how you design it.

Can't you just allow manufacturers to bypass the integrated VR?

 

[amenx]

Overclocking is dead with Haswell

It was 'dead' with Sandybridge too.

 

[Tuna-Fish]

Can't you just allow manufacturers to bypass the integrated VR?

No. The whole point is that the integrated VR is evenly distributed in the chip, so most of the power distribution inside it happens at 12V. The external interfaces of the silicon chip will take in 12V. To allow the manufacturers to bypass this, they'd have to add in a power distribution network designed for ~100A, instead of the one designed for 10A, and that would defeat the point.

 

[Tsavo]

They should bring back COAST modules.

 

[shellx]

I'm trying to imagine what will be the layout of the motherboard with no chipset. The producers will just make sure to have something to differentiate the base models and top models. And above all names to be attached to motherboards. Up to date are based on the chipset name.

 

[IntelUser2000]

The only Haswell parts without chipset are the Ultrabook parts. Desktops where we get to actually look at a board and care about it, still have the chipset on the motherboard.