What happened to Ivy Bridge?

That is already the case with the 3920XM. If you are not using the IGP, it is obviously not using power or giving off heat. So in that case the IGP plays no role in how much you can overclock.

For all the rest of the processors, heat output IS NOT the limiting factor in clockspeed. It is the multipliers that your chip has. Intel could make all the mobile Ivy Bridge processors could be 3920XM's or 3820QM's, but they are not so that Intel can charge more money for giving you greater clockspeeds. And Turbo-Boost is already supposed to account for if the IGP is enabled or disabled. So if you can't reach a high turbo multiplier because of heat given off by the IGP, disabling the IGP will remove the problem. Having an IGP on the die causes ZERO problems if it is not being used.

 

The biggest benefit of removing the IGP would be a cheaper chip because die-size is smaller. But I still don't see how removing it would allow for faster overclock, that's more the capability of the process and the chip and the system cooling.

 

Without an IGP means it requires discreet graphics, so that makes the total package more expensive on low end processors. For high end processors, shaving a tiny bit off the marginal cost of each processor is not a big deal when Intel makes a killing off of each one sold. Also lots of people want an IGP on their higher end processors. So Intel could save a little on the marginal cost of a processor, but it would cost a lot of money to have two product lines and two manufacturing lines for a tiny minority of users who would not ever want to use the IGP.

 

Talking non XM chips,
Removing the IGP like I said earlier, would mean that Intel can raise the turbo clocks higher because there is no extra resources using some of the thermal headroom. Everything will be for the CPU alone, which means that we can have higher clocked CPUs without IGP. I think the max turbo boost is based on the probability that the IGP is in use while the CPU is turboing hence why you can still reach the top frequency that Intel got for that chip, wether you are using the IGP or not.

You are looking at the multiplier on the chips out now with its own set of limits, but Intel can introduce a new line of CPUs with higher multipliers for the non IGP models.

 

He is right, when the gfx is gated the cpu does have the whole tdp. But smaller chips = cheaper or better samples.

 

Yeah I can agree that more thermal headroom will allow for higher overclocks because temperatures affects stability, however I think HTWingNut has a point. Even though there'd be more thermal headroom with overclocking an Ivy Bridge without an IGP, I doubt you will see higher overclocks if you put this chip and a regular Ivy Bridge under some dry ice.

Ivy Bridge is just a bit premature, especially with the crappy TIM under the IHS. Having an IGP just makes the chips a bit more profitable hence Intel's recent emphasis on improving the IGP. They want to dominate the low-end and possibly midrange GPU market. Sure they could remove the IGP and allow for a cooler chip but this may not be a profitable thing to do; if it isn't then it won't be done. Ivy Bridge has a lower TDP due to the voltage reduction however this is counteracted by the increased heat dissipation caused by less surface area. If there is less surface area for the heat to dissipate then it will become a hotter chip. So the only way Ivy Bridge could get any cooler really without adjusting the architecture and the CPU overall is if the IGP is removed so there is more surface area for CPU heat dissipation or if the TIM is improved.

 

Mobile ivy bridge has no heatspreader so no poor quality thermal paste and it still runs comparitively hot.

You can't just have more surface area since the heat comes from the transistors that are actually operating. We might need to see some clever thermal tricks to help spread out the heat across more of the chip more easily.

 

Yeah I was referring to the desktop chips mainly.

And it runs comparatively hot because there are more transistors per mm^2 of surface area...

Ivy Bridge: ~105 mm^2
Sandy Bridge: ~160 mm^2

Thus, there is less surface area for the heat to dissipate given the increased number of transistors. It will get hot one of two ways basically...a smaller chip and the same number of transistors or more transistors and the same size chip.

 

I.... But......

I wish people would read my posts :'(

 

I don't see what you mean. More transistors = less surface area per transistor...

 

Manufacturing the chip in two different SKUs might drive up the prices on producing both of them if you think about it since they'd probably need to alter their production a bit and have deal with supply/demand of the two parts. Its just more work for Intel in the end, so it all becomes one.

 

Theoretically, there can only be a single processor in Intel's lineup that is bottlenecked as you describe, the highest clocked on in each lineup. That would be the 3820QM. But that can be overclocked by 400Mhz anyway, making it a moot point. And also that is assuming that this it's maximum turbo multipliers are reaching the TDP limits only if coupled with full use of the IGP, which is actually a big assumption to make.

There is zero incentive for Intel to do this. Even if everything some of you guys believed was true, which I strongly disagree with, how many people would buy the version without IGP? Not many, that is for sure. So why make two product lines with tons of overlap just to satisfy a very small minority of consumers? The only thing that could benefit Intel is a smaller die size so a bit lower marginal cost on each chip produced. But no way would that be enough to offset all the money spent on making another production line for just a few customers who would not see any benefit anyway.

 

Who wouldn`t want a higher clocked CPU at the same TDP?
There are a lot of notebooks out there without Optimus and Enduro, which makes the IGP a total waste.

Looking at Sandy Bridge desktop models, all the ones without IGP is clocked 200MHz higher than the ones with IGP but with the same TDP. Not much though

 

ULV not great either: AnandTech - Intel Dual-Core Mobile Ivy Bridge Launch and i5-3427U Ultrabook Preview

 

interesting you are comparing the only windows ultrabook that doesnt fail in terms of battery life with a tech sample that was sent to you. Jarred Walton the brightest man in town.

now do please tell me why the ULV aint great? Im baffled.

 

All the people who have bought the 3610QM and 3720QM instead of the 3820QM. For some reason you are under the impression that Intel would sell higher clocked models without IGP for the same price as slower models with an IGP.

Why would Intel want to give you a clock speed boost without you paying more for a higher end processor? This has nothing to do with TDP or heat generated by the IGP because as I have already explained that is irrelevant to what speed your CPU clocks at.

That is absolutely not true.
At 130W TDP, none of the processors have an IGP because they are socket 2011.
At 95W TDP, the highest clocked model has an IGP.
At 65W TDP, they all have an IGP.
At 45W TDP, they all have an IGP.
At 35W TDP, they all have an IGP.

 

Think about it like this...

Not all chips are created equally. Some run hotter at a certain speed, some run a bit cooler at the same speed. If you remove the IGP, the CPU chip that runs hotter can be clocked a bit higher.

Check again.

i5-2320 3/3.3GHz 95W(IGP)
i5-2380P 3.1/3.4GHz 95W(No IGP)

i5-2500 3.3/3.7GHz 95W(IGP)
i5-2550K 3.4/3.8GHz 95W(No IGP)

 

The 2380p was released half a year after the 2320 and the 2550k was released a year after the 2500. The clock speed improvement was probably due to a more mature 32nm process rather than the removal of the IGP (e.g i5-2410m->i5-2430m, i7-2630qm->i7-2670qm, etc).

Plus, max clock speeds aren't limited by the TDP, the amount of time that the chip can stay at the max clock speed is. Max clock speeds are determined by the manufacturing process and how well the chip bins.

 

Obviously, but the point is that not a single processor with an IGP is at all being held back by the heat generated by the IGP.

What about the 2600K? 3.4/3.8Ghz, 95W TDP with an IGP. Or the 2700K at 3.5/3.9Ghz, 95W TDP with an IGP? Having an IGP is clearly not detrimental to clockspeeds in the eyes of Intel.

The 2380P is the same price as the 2400, which is clocked the same AND has an IGP.

The 2550K costs a lot more than the 2500. The 2550K also costs more than the 2500K. But even then the 2500k and 2550k both have unlocked mutlipliers so it doesn't even matter.

Some people seem to think that Intel should offer chips without an IGP on the die or with the IGP disabled at higher clockspeeds for the same price as chips with an IGP. That is never going to happen. If you want faster speeds, you need to pay for the next higher processor, even if you do not intend to use the IGP it has. And guess what, the fastest and most costly Ivy Bridge processor, the 3920XM, has an IGP and unlocked multipliers. If the IGP is holding back your overclock, you stop using it.

 

Check again(again):

2550K - $238.99 - Amazon.com: Intel Core i5-2550K Processor 3.4 4 LGA 1155 BX80623I52550K: Computers & Accessories
2500K - $218.98 - Amazon.com: Intel Core i5-2500K Processor: Electronics

Virtually the same price.

2320 - $189.99 - Amazon.com: Intel Core i5-2320, 4x 3.00GHz 3.0 4 LGA 1155 Processor (BX80623I52320): Electronics
2380P - $189.99 - Amazon.com: Intel Core i5-2380P Processor 3.1 4 LGA 1155 BX80623I52380P: Electronics

Oddly enough, the same price, again.

 

Funny, Qing suddenly switched to price discussion instead.

I won`t be participating in this discussion anymore since we clearly have one thats above all other

Here is the list of Sandy Bridge CPUs.
Sandy Bridge - Wikipedia, the free encyclopedia
i5s have the highest clocked CPUs, the top i7s also is without IGP.

I am pretty shure that Intel could have ramped up the clocks and had higher multipliers on a CPU without IGP which is the reason they offer the top Sandy bridge CPUs without IGP so the IGP doesn`t add additional TDP.

Bye

 

Sincerely you guys should look at the SB E line.

 

What about them? They are also without IGP

 

without the igp they could cram more performance and cores in there. But this is pointless in the notebook market.

 

No, not "virtually the same price." The 2550K is priced between the 2500K and 2600K. You are still paying more for higher clockspeeds, even if they disable the IGP. On Newegg where I looked, the 2500k was $209.99.

On Newegg the 2380P and 2400 are both $189.99. Amazon sells the 2400 for $199.99.

If it is not about price then what is it? Because if it is not about the price, the i7 3920XM has an IGP and unlocked multipliers. You can choose to not use the IGP and overclock it as high as you want. So please don't say it is not about the price, because all you have said is "higher clockspeeds / multipliers," but guess what, we already have clock speeds as high as your heart desires, with or without an IGP.

Wrong on two points.

1. The clock speeds are not any higher on processors without an IGP. Did you even look at that chart at all?

2. The socket 2011 i7's do not have an IGP because they are on socket 2011, Intel's high performance socket 1366 replacement. No processors on socket 2011 support an IGP. However, as I was saying before, the highest clocked processor on socket 1155 is an i7 with an IGP. And guess what, it has the same highest turbo multiplier as the fastest socket 2011 i7!

That is socket 2011, Intel's server / high performance socket. They don't have an IGP on the die BECAUSE it is the server / high performance socket. If they can cram up to 8 cpu cores on a single die and have a max TDP to play with of 150W, there is plenty of headroom for the relatively tiny and low power IGP in there. They don't have it because nobody would use it on a high performance server, workstation, or gaming platform.

 

Yeah, I'm still struggling with how more die area can relate to faster clocks. More cores, absolutely, but not sure how you can run faster just because it doesn't have IGP. The core size doesn't care if it's on a 10mmx10mm die or 100mmx100mm die, it wont help it move faster. By eliminating IGP you could either shrink the die size or add more cores, that's it. If you turn the IGP off it's consuming no power, yet the CPU won't be able to clock higher. If IGP is on, then boost time may be reduced slightly because IGP is consuming some power and generating heat along with the CPU.

 

not sure where you are getting the idea ,

our testings from 100s of 3612QM sold, indicate quite some cooler than 3610QM

 

I'm really not sure what you are referring to, but the 3612QM runs at a lower clockspeed/ lower voltage than the 3610QM, as well as set to work within a 35w TDP envelope, so no doubt the 3612QM's operate at lower temperatures.