Intel Demonstrates 65W Broadwell-K Socketed Processors at GDC 2015

[Yuriman]

Isn't heat density just a symptom, the underlying problem being that electrical parameters are not keeping up with density increases?

I've seen it commonly speculated that 22 nm did not overclock as well as 32 nm because of heat density issues, but heat density has been increasing for quite some period of time now. I'm skeptical of the idea that it's only just now become significant roadblock. Certainly it is the nature of exponentially growing "bad transistor behaviors" to be a non-issue one node, and a severe headache the next, like gate oxide leakage at ~130 nm. However, you'd think that the credible authorities would have highlighted that heat density is, in fact, what's holding us back, rather than whimsically pointing out that we've got heat density on par with nuclear fission reactors, and doing nothing more than that.

My experience has been that I run into heat problems when I start to enter "dangerous" voltages (i.e. 1.3v+). Even with significantly lower temperatures, I doubt I'd want to run at that kind of voltage.

 

[tenks]

I think they're ill-informed. There was an i5 and i7 GT3e Haswell SKU, with a TDP of 65W.

Pretty sure it's a Broadwell chip. Seems like you are ill-informed.



Anandtech > Wccftech

 

[DrMrLordX]

So... basically an unlocked desktop i5 with Iris Pro graphics and a giant-as-hell L4 cache?

Killer HTPC chip. I'd definitely consider it.

Actually, if this is an i7-4770R replacement, then wouldn't it have HT? That makes it more than an i5.

 

[NTMBK]

Pretty sure it's a Broadwell chip. Seems like you are ill-informed.



Anandtech > Wccftech

He was talking about Haswell R, which came in i5 and i7 variants.

 

[Fjodor2001]

who says this is a single SKU?

Have you heard of any other 1150 desktop Broadwell SKU, or seen any on the roadmaps? We all know there will not be a usual full desktop i3/i5/i7 range of Broadwell CPUs, and that's the point. Broadwell will in essence not be a mainstream desktop CPU.

The interesting question is why that is the case? A theory is that heat 14 nm density issues are preventing them from reaching sufficiently high frequencies. Do you have any other probable explanation?

 

[Idontcare]

My experience has been that I run into heat problems when I start to enter "dangerous" voltages (i.e. 1.3v+). Even with significantly lower temperatures, I doubt I'd want to run at that kind of voltage.

I run the opposite problem. Can't seem to overvolt my chips enough to make them die quickly enough so as to need to replace them frequently enough.

Don't know what Intel is so worried about in their QRA department but the voltage ratings they give their chips seem (from experience at least) to be woefully overly conservative.

If the TJmax will allow it, the first thing I'd do with a 14nm unlocked chip is set the voltage to 1.5V and see what kind of clockspeed it gives me. If it dies in two years then that will be a first.

 

[ClockHound]

I run the opposite problem. Can't seem to overvolt my chips enough to make them die quickly enough so as to need to replace them frequently enough.

Don't know what Intel is so worried about in their QRA department but the voltage ratings they give their chips seem (from experience at least) to be woefully overly conservative.

If the TJmax will allow it, the first thing I'd do with a 14nm unlocked chip is set the voltage to 1.5V and see what kind of clockspeed it gives me. If it dies in two years then that will be a first.

IDC, you just keep handing out the brilliant ideas - Broadwell is a stop-gap anyhow - I might get one, run it at 3V for 12 months* then get a Skylake-K. Win-Win! ;-)

* or milliseconds.

 

[VirtualLarry]

If the TJmax will allow it, the first thing I'd do with a 14nm unlocked chip is set the voltage to 1.5V and see what kind of clockspeed it gives me. If it dies in two years then that will be a first.

O_O

I've seen E5200 (2.5Ghz stock, 45nm Core2Duo) CPUs degrade (in about a year) @ 1.4v load, 1.425v BIOS, overclocked to 3.75Ghz.

I guess you're joking, but I certainly wouldn't want to run a 14nm CPU above 1.3v, nevermind 1.5v.

 

[Smoblikat]

My prediction:
Another overpriced and underpowered CPU from Intel with a graphics chip on it that most people will never use. I would be surprised if it was worth it over a sandy bridge processor.

O_O

I've seen E5200 (2.5Ghz stock, 45nm Core2Duo) CPUs degrade (in about a year) @ 1.4v load, 1.425v BIOS, overclocked to 3.75Ghz.

I guess you're joking, but I certainly wouldn't want to run a 14nm CPU above 1.3v, nevermind 1.5v.

When I was done with my E5200, it had about 4 BLACK spots on the contacts I was running a similar OC to what you described. It still worked fine though, just couldnt handle the clocks like it used to

 

[IntelUser2000]

It will be interesting to see how the Iris Pro 6200 integrated GPU performs. Its predecesor, the 5200, wasn't quite as good as the Nvidia GT 650M on most games (though it did surprisingly well in GPGPU).

Its likely not due to memory bandwidth, but lack of performance and good drivers. Maxwell improved despite having less bandwidth for some parts.

AMD and Nvidia does constant game optimizations. Intel basically only does bug fixes, and does game optimizations for 3-4 of them every quarter or so.

Iris Pro 5200 lagged GT 650M some what. Iris Pro 6200 probably just catches up to the GT 650M. I'd be surprised if they achieve even the 40% range(which is still mediocre) they promised. It should have been twice as fast.

 

[AtenRa]

If Broadwell-K will only have those GT3e 65W TDP SKUs, then people that bought a 1150 motherboard to upgrade to Broadwell will be very disappointed.

 

[III-V]

Have you heard of any other 1150 desktop Broadwell SKU, or seen any on the roadmaps? We all know there will not be a usual full desktop i3/i5/i7 range of Broadwell CPUs, and that's the point. Broadwell will in essence not be a mainstream desktop CPU.

The interesting question is why that is the case? A theory is that heat 14 nm density issues are preventing them from reaching sufficiently high frequencies. Do you have any other probable explanation?

Why do you think there will be only one BDW-K SKU? Since Sandy Bridge, when BCLK overclocking was ended, there have always been at least two "K" series SKUs, an i5, and an i7.

Pretty sure it's a Broadwell chip. Seems like you are ill-informed.

Anandtech > Wccftech

Ugh, no. I don't pay attention to that pile of garbage. As NTMBK pointed out, I was talking about Haswell-R.

If Broadwell-K will only have those GT3e 65W TDP SKUs, then people that bought a 1150 motherboard to upgrade to Broadwell will be very disappointed.

That's what Broadwell-K was always slated to be. The time to have been disappointed would have been something like 6+ months ago, for those paying attention.

 

[Ajay]

If the TJmax will allow it, the first thing I'd do with a 14nm unlocked chip is set the voltage to 1.5V and see what kind of clockspeed it gives me. If it dies in two years then that will be a first.

Lol! Good luck!

Anywho, great for AIOs. As far as desktop goes - gak! Bring on Skylake-K!

 

[Fjodor2001]

Why do you think there will be only one BDW-K SKU? Since Sandy Bridge, when BCLK overclocking was ended, there have always been at least two "K" series SKUs, an i5, and an i7.

Does it make any difference if there's one or two? The point is that Intel is skipping desktop with Broadwell, apart from that (or those few) rare low TDP SKU(s).

What's interesting is why they are doing that. Heat issues, problems reaching high frequencies, ...? If there were no such problems, don't you think we'd have seen the regular Intel i3/i5/i7 desktop armada by now on 14 nm?

 

[Arachnotronic]

I run the opposite problem. Can't seem to overvolt my chips enough to make them die quickly enough so as to need to replace them frequently enough.

:biggrin:

 

[Phynaz]

If Broadwell-K will only have those GT3e 65W TDP SKUs, then people that bought a 1150 motherboard to upgrade to Broadwell will be very disappointed.

Why?

Now the people that bought boards to handle 125W AMD CPUs to find out they couldn't actually handle the power draw because they were actually 140w were disappointed.

Do you think the people that bought 1150 boards will have this problem?

 

[Enigmoid]

Does it make any difference if there's one or two? The point is that Intel is skipping desktop with Broadwell, apart from that (or those few) rare low TDP SKU(s).

Because they plan on releasing Skylake?

:whiste:

 

[greatnoob]

Not pin compatible with Z87 motherboards? Really?!

 

[Fjodor2001]

Because they plan on releasing Skylake?

:whiste:

Yeah, but that's only because Broadwell got so delayed we are now getting into Skylake time frame. Effectively, desktop Broadwell therefore was canceled.

We'd seen desktop Broadwell disappearing from leaked roadmaps, and the reappearing again, only to disappear for good except for this/these rare 65 W TDP SKU(s). Likely Intel never got desktop Broadwell to perform as desired within reasonable time.

 

[AtenRa]

Why?

People were made to believe they could upgrade to Broadwell-K with those 1150 boards. And although they may be able to install a 65W TDP Broadwell-K, those SKUs are not to be consider upgrades over 88W TDP Haswell SKUs.

 

[Sweepr]

People were made to believe they could upgrade to Broadwell-K with those 1150 boards. And although they may be able to install a 65W TDP Broadwell-K, those SKUs are not to be consider upgrades over 88W TDP Haswell SKUs.

People were made to believe they could upgrade to Carrizo using FM2+ too. We're at page 2 and you already said that they will be overpriced and not so much of an upgrade compared to Haswell (care to share the specs please?). Careful there, others might think you are angry because Intel will finally have top notch CPU + graphics performance in a socketet form.

 

[Fjodor2001]

Careful there, others might think you are angry because Intel will finally have top notch CPU + graphics performance in a socketet form.

All at 65W TDP? CPU performance on par with 4C 84W TDP Haswell too? I highly doubt that.

 

[AtenRa]

People were made to believe they could upgrade to Carrizo using FM2+ too.

Yeap and it sucks big time if they will not.

We're at page 2 and you already said that they will be overpriced and not so much of an upgrade compared to Haswell (care to share the specs please?). Careful there, others might think you are angry because Intel will finally have top notch CPU + graphics performance in a socketet form.

Well with a huge iGPU + huge L4 extra chip that die is not going to be cheap.
Even if they will have the fastest iGPU in the desktop, the high price will make it irrelevant.
People were bashing AMD for selling kaveri at $180, i will really like to see all the bashing Intel will get for $200-$300+ Broadwell-K GT3e SKUs.

 

[jpiniero]

People were made to believe they could upgrade to Broadwell-K with those 1150 boards. And although they may be able to install a 65W TDP Broadwell-K, those SKUs are not to be consider upgrades over 88W TDP Haswell SKUs.

I don't think Intel really expects people to upgrade to Broadwell-K from Haswell-K. The socket compatibility is mainly for OEMs to make their lives easier.

 

[Sweepr]

Yeap and it sucks big time if they will not.



Well with a huge iGPU + huge L4 extra chip that die is not going to be cheap.
Even if they will have the fastest iGPU in the desktop, the high price will make it irrelevant

People were bashing AMD for selling kaveri at $180, i will really like to see all the bashing Intel will get for $200-$300+ Broadwell-K GT3e SKUs.

If they keep the current Core i5 and i7 prices consumers will be able to choose between the tested and proven >4GHz Devil's Canyon (GT2) and Broadwell-K (GT3e) - top notch CPU performance or slightly slower CPU performance (just my guess) with greatly improved graphics - unlike A10 7850 vs Core i5 you won't have to give up quite a lot of CPU performance in order to have the best iGPU @ similar prices. I like choices. Also if you choose to plug in a discrete GPU you will still have the huge eDRAM acting as cache to improve certain applications (and the >5% better IPC will also help offset the clock gap vs DC). In that scenario, they will be praised to offer GT3e for much lower prices than they did in the Haswell generation and for releasing it as a LGA chip, not a niche BGA-only chip.