• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Intel Broadwell Thread

Page 107 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
A tiny demand.

No, a strong demand. Likely about as strong as we're seeing for the top end 4 core mainstream SKUs today. Heck, a lot of people are even buying 4C/8T CPUs, despite that you think anything more than 4 threads is pointless except for rare corner cases.
 
Last edited:
No, a strong demand. Likely about as strong as we're seeing for the top end 4 core mainstream SKUs today. Heck, a lot of people are even buying 4C/8T CPUs, despite that you think anything more than 4 threads is pointless except for rare corner cases.

What are your cases? Rendering? Encoding? Compression? And else some chicken and the egg play?
 
If Intel provided a 6 core mainstream desktop SKU where they had replaced some of the 4 core iGPU die area with 2 extra cores at the same price as their 4 core SKUs there would be demand.

They won't do it the way you are suggesting. The only reason you see biggest iGPU on the biggest chip is since the rest of the chips are *derivatives* of the biggest chip and they just cut out the blocks not required for the new derivative. Creating a 6 core part without iGPU means creating yet another die.

The seperate die you are suggesting with 6(or 8 cores) will not be done unless they create derivatives of that too. So 2, or 4 cores without iGPU would be created. All that would do is create distraction for them and a better chance for Nvidia/AMD to have a share in the low-end GPU market that goes into $400-600 systems sold at brick and mortar stores. Nearly every chip they create has a high-end part and rest are derivatives. Including HEDT(which is the lowest end E5 part).

The problem is people expect that these companies are at the whim of the customers, but the reality is they only do that when they have a massive backlash, or an unexpected positive reaction.
 
So why are people buying 8 thread mainstream Intel CPUs? And why is Intel even providing them, since they are pointless except for corner cases according to you?

SKUs like Broadwell C is way different from what you are suggesting, which is yet another die from either current mainstream chips or HEDT ones. Also, even Broadwell C is a derivative is what's already existing. If one forum member is to be believed, Broadwell C was a quickly made part that was derived from GT3e mobile chips and hence the poor overclocking.
 
So why are people buying 8 thread mainstream Intel CPUs? And why is Intel even providing them, since they are pointless except for corner cases according to you?

Intel provides them because they are faster in some of those 'corner cases' which make up a disproportionate amount of benchmarks. And because it would require an unfeasible amount of effort to remove hyperthreading from the design, so it's not costing them extra die space given that hyperthreading is necessary for mobile and server parts.

Much as enthusiasts hate to admit such, there simply is zero reason for there to be a 6 core or greater mainstream processor. And Intel isn't going to make a separate die to cater to the enthusiast market when said market is going to buy derivative products regardless. That's not to say that there won't ever be 6 core processors, I just wouldn't expect to see it any time soon... Personally though I'm looking forward to 4 core U-series more than 6 core desktop.
 
Intel provides them because they are faster in some of those 'corner cases' which make up a disproportionate amount of benchmarks. And because it would require an unfeasible amount of effort to remove hyperthreading from the design, so it's not costing them extra die space given that hyperthreading is necessary for mobile and server parts.

Doesn't explain it. If 8 threads is pointless, why even design 4C/8T dies?

And even more important, why do people buy them? It may not cost Intel much more to manufacture them, but 4C/8T cost $100 extra for the customers compared 4C/4T. So why do the customers still buy them if it's pointless? Doesn't make any sense.
 
Doesn't explain it. If 8 threads is pointless, why even design 4C/8T dies?

And even more important, why do people buy them? It may not cost Intel much more to manufacture them, but 4C/8T cost $100 extra for the customers compared 4C/4T. So why do the customers still buy them if it's pointless? Doesn't make any sense.

You forgot frequency and cache.
 
Yep, that's the beauty of virtual monopoly. Hope Zen comes back and bites them in the ass. Just like happened previously when Intel got lazy.

Only the paranoid survives...

Yes, evil company blabla..doesnt want to give what I am entitled to for the price I demand. How dare they, death to them.

Did I miss anything?
 
At this point, I personally think I'm just gonna buy a 4790k from an enthusiast moving to Skylake.

Even with a 5-15% decrease compared to Broadwell or Skylake, Devil's Canyon will last me quite a long time.


If 5820Ks are still $300 at microcenter in one month when I head to WI I'm gonna swing by the twin cities and pick one up. I have this gut feeling that Intel may have lost a bit of revenue from selling a low priced unlocked 6 core chip and they might not be so willing to try that again, meaning the next 6 core chip on the enthusiast "E" platform, assuming they even offer one, will be more expensive... starting at say $500.

They seem perfectly content sticking with dual and quad core chips given the rumors going on with Kabylake. Going into 2016-2017 I would have expected dual core chips to not exist by now, looking back from say 2010 or so into the future, with the smallest number of cores at 4, with 6 and 8 being an option, but clearly this hurts their server revenue too much to offer these at mainstream prices.

Besides, I expect them to debut the 6700K part around $400, more than I'm willing to pay for a quad core these days. With no competition there's no reason to sell it for less.
 
Last edited:
Yes, evil company blabla..doesnt want to give what I am entitled to for the price I demand. How dare they, death to them.

Did I miss anything?

Nope not evil, just what companies do when given the opportunity of monopoly like standing. But it's still bad for the consumer.
 
Doesn't explain it. If 8 threads is pointless, why even design 4C/8T dies?

And even more important, why do people buy them? It may not cost Intel much more to manufacture them, but 4C/8T cost $100 extra for the customers compared 4C/4T. So why do the customers still buy them if it's pointless? Doesn't make any sense.

Uhmmmmm, as said? Because the die is going to be 4C/8T by default if it has 4 cores. Then, instead of disabling hyperthreading, keep it on to increase performance in those few actual workloads where it's useful and, of course, the larger proportion of benchmarks where it helps. If Intel didn't design hyperthreading into its cores for the purposes of servers and, somewhat, to allow dual core mobile to be tolerable, then I'd expect both i5 and i7 would be 4C/4T with other features to distinguish between. But since hyperthreading is there regardless, why not enable it and charge more?

As for why people buy them... It's in part the marketing, in part ignorance, and in some cases because they'll actually make good use of the additional features (cache and hyperthreading primarily.) There's a good reason why Intel spends so much on marketing - it does pay for itself.
 
A tiny demand.

Because of the price.

You forgot frequency and cache.

Cache differences on i7 vs. i5 are basically undetectable.

Right, because the market for 6-8 core consumer CPUs is so huge? Oh wait...

Their may be people wanting a 6C cpu but they don't want/need it enough to pay the HEDT price.

Considering the tiny size of cores on 14nm (10mm^2 for one BW core + 2 MB cache) adding a 6C GT2 cpu to the lineup is relatively minimal. I expect that a broadwell 4C + GT2 model would be ~105 mm^2 and a skylake 4C + GT2 die around 110-115 mm^2.

10nm, as far out as it is, will bring even smaller dies with 4C + GT2 around ~70-75 mm^2. A core + L3 would likely be around 5-5.5 mm^2.

Simply put its not a matter of the market needing or demanding it, its a matter of competition. Nobody 'needed' a quad when the first quads were launched either but they were launched because of competition.
 
As for why people buy them... It's in part the marketing, in part ignorance, and in some cases because they'll actually make good use of the additional features (cache and hyperthreading primarily.) There's a good reason why Intel spends so much on marketing - it does pay for itself.

So 8 threads CPUs sell well after all. Nuff said.
 
Considering the tiny size of cores on 14nm (10mm^2 for one BW core + 2 MB cache) adding a 6C GT2 cpu to the lineup is relatively minimal. I expect that a broadwell 4C + GT2 model would be ~105 mm^2 and a skylake 4C + GT2 die around 110-115 mm^2.

10nm, as far out as it is, will bring even smaller dies with 4C + GT2 around ~70-75 mm^2. A core + L3 would likely be around 5-5.5 mm^2.

Simply put its not a matter of the market needing or demanding it, its a matter of competition. Nobody 'needed' a quad when the first quads were launched either but they were launched because of competition.

+1
 
Back
Top