Broadwell Vs Haswell (Devil's Canyon)

[XtremeAero426]

So I'm researching for my upgrade that I'm currently investing for that will take place in late December or early January. I was wondering, at that time, will Broadwell be worth it over Haswell(Devil's Canyon) in terms of Overclocking and performance?

Importance in Choice:
- Main Factor: Overclocking

Things I already know:
- Broadwell has little to no performance increase.
- Broadwell offers the new 14nm lithography and lower power consumption.

Key Questions:
- How much higher can I overclock a Broadwell processor than a Haswell (Devil's Canyon), specifically the 4790K? (I have a Cooler Master V8 GTS CPU Cooler with Tuniq MX-2 Thermal Paste)
- [strike]Will Broadwell be supported by Z97?(Because if not I'm definitely not getting it.)[/strike]
A: Broadwell is compatible with Z97. (Thanks guys!)

NOTE: I do not have an Intel Processor. This is a switch from AMD to Intel.

 

[DarkSable]

If you already have a Devil's Canyon CPU, you would have to have WAY more money than brains to upgrade up just one generation. People are still finding Sandy Bridge chips to be more than powerful enough for gaming when overclocked. Just saying.

 

[Skylyne]

If you have a Haswell chip right now, the only benefit you'd get from going Broadwell is that you'd be running the newest line of Intel CPU. There's no point in upgrading from Haswell to Broadwell, from what has been released.

As far as OC'ing is concerned, until the Broadwell chips come out, I don't think anyone can accurately tell you how well it will OC. If you can wait until the OC'ed reviews of the Broadwell chips start to pour in, that will tell you more than speculation will. Even if they had stated their intended goal to be better OC'ing, you never really know what the final product will give you in the real world. OC'ing capabilities are mostly going to be little more than pure speculation at this point, and it's a bad idea to base buying decisions on speculations in this instance.

On the upside, the Z97 boards will be Broadwell compatible.

 

[EasyLover]

Can't answer your query as of now because broadwell is not yet out so no benchmark or performance info is available.

H97/Z97 supports broadwell.

There is no point of updating from haswell to broadwell. If you have not yet purchased CPU then you can wait for broadwell but if you have haswell based CPU, stick with it in my humble opinion.

Users having sandy bridge or ivy bridge may consider upgrading to braodwell though sandy/ivy bridge chips are still powerful enough to game on.

MHO.

 

[XtremeAero426]

If you already have a Devil's Canyon CPU, you would have to have WAY more money than brains to upgrade up just one generation. People are still finding Sandy Bridge chips to be more than powerful enough for gaming when overclocked. Just saying.

Either way, I will be getting at least a 22nm Haswell Processor with a Z97 motherboard. Also, I am not upgrading from Haswell to Broadwell. I am upgrading from Richland(AMD) to Haswell (It's a huge performance jump). So, to get a general idea of the OC'ing difference, how much higher were people able to overclock with a 22nm processor instead of a 32nm processor? Does anyone know?

 

[XtremeAero426]

If you have a Haswell chip right now, the only benefit you'd get from going Broadwell is that you'd be running the newest line of Intel CPU. There's no point in upgrading from Haswell to Broadwell, from what has been released.

As far as OC'ing is concerned, until the Broadwell chips come out, I don't think anyone can accurately tell you how well it will OC. If you can wait until the OC'ed reviews of the Broadwell chips start to pour in, that will tell you more than speculation will. Even if they had stated their intended goal to be better OC'ing, you never really know what the final product will give you in the real world. OC'ing capabilities are mostly going to be little more than pure speculation at this point, and it's a bad idea to base buying decisions on speculations in this instance.

On the upside, the Z97 boards will be Broadwell compatible.

Thanks for the answer to the compatibility issue. As for the answer to OC'ing, I think looking at the OC capability jump from 32nm to 22nm would tell me. And I'm not upgrading from Haswell to Broadwell. I'm upgrading from Richland(AMD) to Haswell/Broadwell.

 

[Skylyne]

So, to get a general idea of the OC'ing difference, how much higher were people able to overclock with a 22nm processor instead of a 32nm processor? Does anyone know?

Once again, going on this kind of speculation will not give you an accurate idea of what to buy. If you're okay with waiting for the Broadwell chips to come out, you might as well wait for the OC'ing reviews to be published; it shouldn't take too long for that to happen. If you are anxious, the Haswell chips will do fine.

 

[EasyLover]

If you have a Haswell chip right now, the only benefit you'd get from going Broadwell is that you'd be running the newest line of Intel CPU. There's no point in upgrading from Haswell to Broadwell, from what has been released.

As far as OC'ing is concerned, until the Broadwell chips come out, I don't think anyone can accurately tell you how well it will OC. If you can wait until the OC'ed reviews of the Broadwell chips start to pour in, that will tell you more than speculation will. Even if they had stated their intended goal to be better OC'ing, you never really know what the final product will give you in the real world. OC'ing capabilities are mostly going to be little more than pure speculation at this point, and it's a bad idea to base buying decisions on speculations in this instance.

On the upside, the Z97 boards will be Broadwell compatible.

Thanks for the answer to the compatibility issue. As for the answer to OC'ing, I think looking at the OC capability jump from 32nm to 22nm would tell me. And I'm not upgrading from Haswell to Broadwell. I'm upgrading from Richland(AMD) to Haswell/Broadwell.

In that case stick with your current setup and wait for Broadwell and its reviews. Make your decision accordingly.

 

[Cazalan]

Broadwell for desktop won't be out for like 6 months. That should simplify your options.

 

[EasyLover]

Broadwell for desktop won't be out for like 6 months. That should simplify your options.

Rumors were circulating that Intel may release Broadwell startup CPUs near Xmas. Any latest news/update on that.

 

[Skylyne]

It wouldn't surprise me if some Broadwell chips made it to market around the xmas holiday. Either way, if the OP can get by without an upgrade until the Broadwell chips are released, I don't see what the real rush to get a Haswell chip is. If there isn't a real need, then why buy a Haswell now, and risk missing out because you were impatient? It might not be the biggest improvement in OC capabilities, but it might be something the OP wants.

 

[Cazalan]

Broadwell for desktop won't be out for like 6 months. That should simplify your options.

Rumors were circulating that Intel may release Broadwell startup CPUs near Xmas. Any latest news/update on that.

The Broadwell parts shipping now are the Broadwell-Y aka Core-M processors for the 5W range. These are low end mobile parts. They will be available as early as this month from companies like Lenovo/Dell.

You're looking for desktop gaming Broadwell which will be much later. 2015Q2 or 2015Q3.

 

[EasyLover]

Broadwell for desktop won't be out for like 6 months. That should simplify your options.

Rumors were circulating that Intel may release Broadwell startup CPUs near Xmas. Any latest news/update on that.

The Broadwell parts shipping now are the Broadwell-Y aka Core-M processors for the 5W range. These are low end mobile parts. They will be available as early as this month from companies like Lenovo/Dell.

You're looking for desktop gaming Broadwell which will be much later. 2015Q2 or 2015Q3.

Great. Thanks for update.

 

[EasyLover]

So, OP, here are your options: -

Continue with current specs and wait for broadwell

or

Simply buy Haswell Refresh with good mobo and settle with it.

Your call.

 

[ebalong]

So, any updated information on when more powerful Broadwell Mobile (Core M) processors might debut? I am tempted to pull the trigger on a laptop with an i7-4810/4910MQ chip, but would wait if something (besides the ultra-low Core M) was coming down the pipeline in the near future - as in, the next few months.

 

[Fr33Th1nk3r]

So I'm researching for my upgrade that I'm currently investing for that will take place in late December or early January. I was wondering, at that time, will Broadwell be worth it over Haswell(Devil's Canyon) in terms of Overclocking and performance?

Importance in Choice:
- Main Factor: Overclocking

Things I already know:
- Broadwell has little to no performance increase.
- Broadwell offers the new 14nm lithography and lower power consumption.

Key Questions:
- How much higher can I overclock a Broadwell processor than a Haswell (Devil's Canyon), specifically the 4790K? (I have a Cooler Master V8 GTS CPU Cooler with Tuniq MX-2 Thermal Paste)
- [strike]Will Broadwell be supported by Z97?(Because if not I'm definitely not getting it.)[/strike]
A: Broadwell is compatible with Z97. (Thanks guys!)

NOTE: I do not have an Intel Processor. This is a switch from AMD to Intel.

If you already have a Devil's Canyon CPU, you would have to have WAY more money than brains to upgrade up just one generation. People are still finding Sandy Bridge chips to be more than powerful enough for gaming when overclocked. Just saying.

AMD chips are better for overclocking than haswell+ but if you get lucky on the dice roll you can get a good binned DC and go 4.8-5.0 with the proper cooling solution. A thick 240mm 2x rad or liquid nitrogen.

4790k and 4770k are the same price right now. So just go with the newer chip and try to squeeze out a 4.7 OC with a good liquid cooler.

 

[synphul]

It may only be one factor and is still speculation without real world attempts on a cpu that doesn't exist yet, but going down in lithography can present problems with overclocking. Part of the reason current chips are running out of the oc'ing headroom they had on larger dies. Each transistor or circuit generates x amount of heat. Overclock the chip and it pushes its frequency beyond the engineering design. More current means more heat. Now drop that package almost in half from 22nm to 14nm. Heat build up potential (which is what kills overclocks) increased by quite a bit. (look how much faster your car warms up than your home with the heat on). While it may not be the only factor, it's kind of a big one. Overclocking potential with reduced die sizes keeps diminishing. In addition, the gains are becoming less. 1ghz overclock on a 3.5ghz chip are less impressive than 1ghz overclock on a 1ghz chip. In order to achieve the same sort of impressive overclocks on newer chips we saw on much older ones, we'd need to have 'easy' overclocks of 2ghz or more, which would make a current i5 4690k able to easily hit 5.5-6ghz. It's not happening. Even with a good quality chip, 4690k's are topping out around 4.5, maybe 4.6 at the most. Some even less depending on the 'chip lottery'.

 

[GlitchySpiritX]

Well from the looks of it I have a Core i7 4770k and the broadwell successor to the 4770k wont be out until Q4 2015
So intel announced this way too early
not only that
but Probobly by 2016 Q1 Broadwell Will be obsolete so you wont have the newest CPU anymore
Also Skylite 6th gen intel which will be out in the end of 2016 Q1
wont have LGA1150 anymore but will have a new socket LGA1151
Im going to wait for skylite because in 2016 intel must improved the Graphics
So yeah waiting for the I7 6770k is the best choice right now because broadwell wont be different
also chromebooks will be Switching over to the intel core M CPUs

 

[CompuTronix]

... going down in lithography can present problems with overclocking. Part of the reason current chips are running out of the oc'ing headroom they had on larger dies. ... Now drop that package almost in half from 22nm to 14nm. Heat build up potential (which is what kills overclocks) increased by quite a bit. ... Overclocking potential with reduced die sizes keeps diminishing. ...

Guys,

Here's another major consideration which shouldn't be overlooked:

Intel discontinued their manufacturing process of a soldered IHS on Ivy Bridge, Haswell and Devil's Canyon 22 nanometer processors in favor of TIM, because of the cost of using "Indium" solder. Indium is an exotic material which is quite expensive, but has excellent thermal conductivity. It's also a carcinogenic hazardous material.

Intel's bean counters saw an opportunity to save a dollar by switching to TIM, so they worked it out with engineering and management. With the exception of the 140 watt "High End" 5xxx processors, which Intel reverted to Indium solder, we have 77, 84 and 88 watt 22 nanometer processors that are very thermally ill behaved, as we're all disappointingly aware.

The larger die size and excellent thermal conductivity of the Indium soldered 95 watt 32 nanometer Sandy Bridge processors prevents them from approaching throttle temperatures with AVX code, such as Prime95 versions later than 26.6, or apps like Handbrake.

As I see it, Intel's experiment with TIM on 22 nanometer processors was a HUGE blunder that has resulted in a big black eye, and more RMA's than they ever imagined, not to mention a considerable loss of profits. In my opinion, the only way Intel can control the thermals on 14 nanometer processors, is to revert back to Indium solder.

CT

 

[XtremeAero426]

... going down in lithography can present problems with overclocking. Part of the reason current chips are running out of the oc'ing headroom they had on larger dies. ... Now drop that package almost in half from 22nm to 14nm. Heat build up potential (which is what kills overclocks) increased by quite a bit. ... Overclocking potential with reduced die sizes keeps diminishing. ...

Guys,

Here's another major consideration which shouldn't be overlooked:

Intel discontinued their manufacturing process of a soldered IHS on Ivy Bridge, Haswell and Devil's Canyon 22 nanometer processors in favor of TIM, because of the cost of using "Indium" solder. Indium is an exotic material which is quite expensive, but has excellent thermal conductivity. It's also a carcinogenic hazardous material.

Intel's bean counters saw an opportunity to save a dollar by switching to TIM, so they worked it out with engineering and management. With the exception of the 140 watt "High End" 5xxx processors, which Intel reverted to Indium solder, we have 77, 84 and 88 watt 22 nanometer processors that are very thermally ill behaved, as we're all disappointingly aware.

The larger die size and excellent thermal conductivity of the Indium soldered 95 watt 32 nanometer Sandy Bridge processors prevents them from approaching throttle temperatures with AVX code, such as Prime95 versions later than 26.6, or apps like Handbrake.

As I see it, Intel's experiment with TIM on 22 nanometer processors was a HUGE blunder that has resulted in a big black eye, and more RMA's than they ever imagined, not to mention a considerable loss of profits. In my opinion, the only way Intel can control the thermals on 14 nanometer processors, is to revert back to Indium solder.

CT

At this point it's far too late as I've already made an upgrade and Skylake processors are out already but I was wondering about the TIM vs Indium solder change you were mentioning. If you delidded a 22nm Haswell processor and used metal TIM instead of the TIM that Intel was originally using would that make up for the Indium solder change and/or increase OC potential at all?

 

[synphul]

Not sure which you ended up upgrading to, broadwell or skylake. Both use the same type tim approach as haswell. Thermals may not be as good as they could have been though I've been lucky enough neither of my devil's canyon i5's seem to suffer from horrible thermals even using a tim material instead of solder. They approach a voltage limitation before thermal issues.

If heat was your issue, you could delid and try running either without the lid or applying your own tim and it will probably help though it depends on what thermal compound you use. Skylake was delidded and intel's silver based tim material removed and replaced with prolimatech pk3 which got a 4c drop. If you're looking for 'extreme' best temps and 4c matters to you enough to risk delidding (and possibly damaging) your brand new cpu then it's worth a try. Otherwise in my opinion 4c isn't enough to warrant the hassle and potential to bork a cpu and void the warranty with no recourse.

http://www.guru3d.com/news-story/intel-skylake-de-lidded.html

They also did the same test with coolab's liquid pro which provided a 20c drop but that stuff is highly conductive (now introduced to the more sensitive bits under the ihs), if it's in contact with dissimilar metals will cause severe corrosion and of course there's the risk that comes with delidding in general. As you pry off the ihs you have to be careful it doesn't get warped at all or you may end up with worse temps than it was to begin with.

I still think you'll run into a voltage limitation. Even if you can keep temps below 70c (well within safe during a stress test), to what end of overclocking? If the chip hits its design limits and begins requiring 1.38 and 1.42 or 1.43v on the vcore to hit the next multiplier it won't matter if temps are at a nice cool 68c. The issue will be from having so much voltage run to the cpu for regular day to day use. You can't push 1.6v or 1.7v to the vcore and keep upping the multiplier just because temps are within limit.