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

[Arachnotronic]

What's there to leak that everyone doesn't already know?

14nm is super healthy, BW binning is spitting out 5GHz+ monsters, Intel doesn't want to do a paper-launch because they know the world+dog is going to want to upgrade immediately so they have spent the past 9 months stockpiling desktop skus.

Oh, and while we are living in fantasy land here for the moment, the S810 has no overheating issues, Qualcomm is just doing 20nm like a boss, and TSMC will never make another Apple chip again.

Its all true, I read it in my post on the internet after I pressed the Submit Reply button once.

This post put a genuine smile on my face. Thank you.

 

[Dave2150]

What's there to leak that everyone doesn't already know?

14nm is super healthy, BW binning is spitting out 5GHz+ monsters, Intel doesn't want to do a paper-launch because they know the world+dog is going to want to upgrade immediately so they have spent the past 9 months stockpiling desktop skus.

Oh, and while we are living in fantasy land here for the moment, the S810 has no overheating issues, Qualcomm is just doing 20nm like a boss, and TSMC will never make another Apple chip again.

Its all true, I read it in my post on the internet after I pressed the Submit Reply button once.

While accomplishing sarcasm, your whole argument hinges on what we believe we know.

Fact of the matter is we don't know anything about Broadwell-K, apart from the fact that it will be a 65w LGA1150 CPU.

We want to know other information, including clock speeds.

Whilst it's true that what we've seen from 14nm so far has been only low clocked skews, that doesn't necessary mean they can't make a high clocking 14nm part.

Intel could simply be waiting to clear 22nm Haswell stock, as they know there is no alternative for anyone who needs to buy a new CPU right now. Perhaps Intel just want to keep producing/selling the higher yielding 22nm parts as long as they can, to increase profits....

Fact is we don't know anything beyond the 65w 4C8T LGA1150, with Iris Pro graphics. Anything else is purely speculation at this point.

 

[Phynaz]

While accomplishing sarcasm, your whole argument hinges on what we believe we know.

Fact of the matter is we don't know anything about Broadwell-K, apart from the fact that it will be a 65w LGA1150 CPU.

We want to know other information, including clock speeds.

Whilst it's true that what we've seen from 14nm so far has been only low clocked skews, that doesn't necessary mean they can't make a high clocking 14nm part.

Intel could simply be waiting to clear 22nm Haswell stock, as they know there is no alternative for anyone who needs to buy a new CPU right now. Perhaps Intel just want to keep producing/selling the higher yielding 22nm parts as long as they can, to increase profits....

Fact is we don't know anything beyond the 65w 4C8T LGA1150, with Iris Pro graphics. Anything else is purely speculation at this point.

Dave, two things:

1. Do a little research into who you are replying to before you criticize.

2. Intel may not know these things. Speed bins may not be set yet.

 

[mrmt]

Intel could simply be waiting to clear 22nm Haswell stock, as they know there is no alternative for anyone who needs to buy a new CPU right now. Perhaps Intel just want to keep producing/selling the higher yielding 22nm parts as long as they can, to increase profits....

Intel isn't that dumb. They would release their 14nm parts if anything because they should have an improved cost structure compared to their 22nm counterparts. It is the failure to achieve this better cost structure, because of the 14nm problems, that prompted Intel to keep their 22nm parts longer than they should on the market.

 

[Dave2150]

Dave, two things:

1. Do a little research into who you are replying to before you criticize.

2. Intel may not know these things. Speed bins may not be set yet.

I didn't criticize his post - I simply questioned his opinions. Everyone is entitled to one

As for suggesting I research him, do you expect every poster to thoroughly research who wrote the post they are replying to? Good luck with that buddy.

I see he has thousands of posts, which suggests he spends many hours of his life on these forums, I'm not prepared to read enough of these posts to find out anything else, though feel free to list his credentials if it's that important to you.

 

[Fjodor2001]

Just wondering, what market is this 65 W TDP Broadwell-K SKU supposed to target? HTPC enthusiast? Seems like quite a small market to me.

 

[bullzz]

same market as a10-7850k. yes a small market

 

[jpiniero]

Just wondering, what market is this 65 W TDP Broadwell-K SKU supposed to target? HTPC enthusiast? Seems like quite a small market to me.

The K models having the best graphics and compute plus the edram makes sense. The TDP isn't really that big of a deal when you are talking about being able to overclock it.

 

[ehume]

Well, the boutique computer business has taken off of late. You get pre-built bombers from a number of manufacturers who are too small to make their own name brand cases. You've seen their ads -- Intel cpu, so-and-so case, so-and-so's gpu, etc. The business may be a bigger market than enthusiasts building their own rigs. Maybe the 65W Broadwell is for them.

 

[Ajay]

Just wondering, what market is this 65 W TDP Broadwell-K SKU supposed to target? HTPC enthusiast? Seems like quite a small market to me.

It all depends on CPU performance - of which we have no clue due date on what a desktop BW quad may offer. At this point, this is becoming a circular debate because there is no official data - only rumors from click-bait sites.

 

[DrMrLordX]

same market as a10-7850k. yes a small market

You may have a point, but not for the reasons that you think. Primarily, the Broadwell Iris Pro part appears to be aimed at the same people that bought the i7-4770R, which is a Haswell 4c/8t chip with Iris Pro graphics soldered onto a board (Broadwell Iris Pro has the advantage of being unlocked and socketed; presumably, the chip will also be 4c/8t).

On the flipside, by making sure that Intel releases an Iris Pro part for desktops with Gen8 graphics, Intel has introduced what may be the first Intel APU worthy of GPGPU functions. Sure, you can run some OpenCL code on HD iGPUs that aren't Gen8 Iris Pro, but Gen8 Iris Pro should be able to do it quickly enough to garner interest, especially if Intel supports Gen8 and Gen9 graphics with a developer-friendly software stack (AMD's attempts at producing an HSA stack have mostly led to disappointment).

I don't know if Gen8 iGPUs have all the capabilities necessary to do the same things that Carrizo can do, but I think Intel is at least getting closer to that point.

So in that sense, yes, Broadwell Iris Pro does go head-to-head with Kaveri and Carrizo.

 

[AtenRa]

same market as a10-7850k. yes a small market

You expect Broadwell-K GT3 Iris-Pro to be at $150 ??

 

[crashtech]

You expect Broadwell-K GT3 Iris-Pro to be at $150 ??

Similar intended use, but not same market.

 

[nemesismk2]

my athlon ii x2 270 dual core cpu is 65w and works very well and only cost me £35 for a bargain

 

[Dave2150]

Just wondering, what market is this 65 W TDP Broadwell-K SKU supposed to target? HTPC enthusiast? Seems like quite a small market to me.

Given that these are 5th generation desktop CPU's, as promised by Intel to be compatible with existing Z97 boards, logic would dictate that the 5th generation would be superior to the 4th generation.

So, you'd expect Broadwell-K to be superior to Devil's Canyon, in all aspects.

The news of it being a 65w part and having Iris Pro, coupled with the Intel press release that described it as a small factor PC CPU, leaves us with a problem; it doesn't seem like Broadwell K will outperform Devil's Canyon (Haswell).

I suspect alot of Z97 owners will be angry if this is the case, though Intel's has such a monopoly of the mid/high end desktop that these disgruntled Z97 owners will have no choice but to buy Intel next round anyway.

Or, we could all be surprised and this 65w Broadwell-K part could comprehensively outperform the 4790k, though I imagine it will cost a pretty penny if it does

 

[crashtech]

I don't know if there is conclusive evidence that the 65W part is the only unlocked SKU that will be offered.

 

[jpiniero]

I don't know if there is conclusive evidence that the 65W part is the only unlocked SKU that will be offered.

It makes sense, given the 4770R lineage. Probably saves Intel validation time.

I suspect alot of Z97 owners will be angry if this is the case, though Intel's has such a monopoly of the mid/high end desktop that these disgruntled Z97 owners will have no choice but to buy Intel next round anyway.

How many people out there with DC would even be considering buying anything newer at this point? Even then, I imagine Intel would rather have them buy Broadwell-E.

 

[crashtech]

So far every Tick and Tock has brought with it one top rated mainstream CPU, most recently the 2700K, 3770K, and 4790K. So if there is no 84W+ part offered, this would obviously be a departure. Perhaps the true successor to 4790K will be Skylake-K in time for Christmas.

 

[Abwx]

This should allow to extrapolate the caracteristics of a 4C/8T variant :

http://www.notebookcheck.com/Test-Intel-NUC-5i5RYH-Mini-PC.138750.0.html

 

[mikk]

This should allow to extrapolate the caracteristics of a 4C/8T variant :

http://www.notebookcheck.com/Test-Intel-NUC-5i5RYH-Mini-PC.138750.0.html

You can't extrapolate.

 

[crashtech]

I'ts clockspeed is too slow, and it might throttle due to the enclosure. We'll just have to wait and see.

 

[Abwx]

You can't extrapolate.

The chip run the benches at full frequency, that s not difficult.

Obviously it has a 25W thermal headroom wich allow it to sustain 2.5GHz,
assuming 45W at this frequency with twice the core count you ll end at roughly 90W for a 3.5Ghz/4C/8T chip.

BDW voltage/frequency variation between 2.893GHz and 3.093GHz is 4%/6.9% if you want to do the maths, i dont think that it can improve when getting closer to 3.5, so my estimation is quite cautious.

I'ts clockspeed is too slow, and it might throttle due to the enclosure.

The article say otherwise, as well as the CB 11.5 score, unless FP IPC was improved by roughly 50% compared to Haswell...

 

[Enigmoid]

This should allow to extrapolate the caracteristics of a 4C/8T variant :

http://www.notebookcheck.com/Test-Intel-NUC-5i5RYH-Mini-PC.138750.0.html

Throttles.

I have no idea how you are extrapolating anything. Intel already has 45W chips...like the mobile quads...which easily hit >3.3 Ghz on CPU only load in a <45W envelope.

 

[mikk]

The chip run the benches at full frequency, that s not difficult.

There is no evidence. It's out of thin air from your side. There is not a single frequency log from one of the games for example. Also the 25W headroom is nothing more than speculation from your side. Furthermore the turbo clock is quite low for both CPU and GPU. And speaking for GT3e edram is another unknown factor. You can't extrapolate.

 

[Abwx]

Throttles.

I have no idea how you are extrapolating anything. Intel already has 45W chips...like the mobile quads...which easily hit >3.3 Ghz on CPU only load in a <45W envelope.

What is the use to deny the obvious.?.

The article clearly state that it doesnt throttle during benches, only when Prime 95 + Furmark are running both.

There is no evidence. It's out of thin air from your side. There is not a single frequency log from one of the games for example. Also the 25W headroom is nothing more than speculation from your side. Furthermore the turbo clock is quite low for both CPU and GPU. And speaking for GT3e edram is another unknown factor. You can't extrapolate.

The power delta at the main imply roughly 25W drained by the CPU assuming it consume 0W at idle, so it s obvious that the chip has a 25W thermal headroom, unless it has something like 55% better perf/Watt than Haswell..

Now try to find a single HW or Core M laptop at Notebookcheck that score more than 2 on CB 11.5 MT and that consume about 25W at most in the stress test, you wont find a single one, at 15W a HW wont score more than 1.65-1.8 while a Core M is in the same range, that s all extended thermal headrooms as a smoke screen for efficencies that are well below the hyped values, hence the shocks once i post numbers that burst the usuals irrational beliefs.