• 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.

[reuters 5-18-14] Intel CEO promises Broadwell on shelves for holidays 2014

Page 9 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Sure, it may improve a bit. But it won't ever come close to desktop PC productivity levels. For starters, it's impossible to make a 5/10" screen a 24" screen. 😉 So why settle for something inferior if you don't have to?

Tablets/phones have they use cases, but they will not replace desktop PCs for serious work (coding, video editing, etc).

Just look around in a typical engineering office. Do you see lots of people developing on phones/tablets? Close to none, right. Ever wondered why, despite that they have been around for years now, so their pros/cons should be known by now?

There is no "one size fits all" in compute. Tablets won't eliminate desktops, mobile phones won't eliminate tablets, etc. etc.

But there is certainly a sizable number of consumers whose compute habits are such that they have the option of migrating from platform to platform on the basis of personal choices that transcend straight-up numbers.

Speed, weight, battery life, screen size, etc. These are all becoming more and more negotiable on more and more platforms (phone, tablet, laptop, desktop) such that ultimately it is up to the personal preferences of the end-user to decide what matters most to them.

That's the benefit of having choices as a consumer. There is no "one product to rule them all", which is great.
 
Broadwell with 20-40% better performance,

The figures should be over current Y models set to 6-7W using CPU bound applications. My research suggests that all the models using Y are using figures between 6-7W, that varies depending on the manufacturer. The Sony Vaio Tap and the XPS 11 for example, are set at 6-7W(not sure) for CPU + GPU load. The system does not permit CPU temperatures to rise above 65C, which means for the Y systems, SDP is really their TDP. No system exists yet that uses full 11.5W TDP. My guess is that manufacturers thought the trade-off between performance/cooling wasn't worth it to justify going little lower to 11.5W, when they could either use 15W or go much lower at 6-7W.

My assumption came from following few facts:
-"Uncore", or whatever that is uses fixed(well, not much changing on load) ~4W on U and Y series chips
-On a 7W use, 4W is uncore, and 1-2W is CPU and 1-2W is GPU

The Broadwell Y SKU(which is likely what Core M is) has a 4.5W TDP. I assume they can do the following:
-Cut the Uncore TDP to 1/2, or 2W
-Keep the CPU TDP same
-In Broadwell GPU generation, its rumored that the GPU improves perf/watt by 4x! They can cut power to 1/4 and combining all will allow ~4.5W, with roughly equal performance to Haswell Y

It's likely that because Haswell was targeted at Ultrabooks at 15W TDP, engineers deemed that reducing Uncore power wasn't necessary. Broadwell targets Tablets first, so that would necessitate a change. Because CPU TDP stays the same, they can improve the performance thanks to 14nm.
 
Last edited:
If apple put broadwell (which they prob won't) in a "tablet" it wouldn't be called an iPad it would be something else.


The iPad is a tablet not a desktop or a notebook and apple has made it pretty clear that they haven't been interested in putting OSX in that form factor.



The real question about putting broadwell into a tablet for apple is why bother when they already have a processor with better low power characteristics and equal or better IPC in the A7, and it's likely the A8 (which will be quad core, unlike these Broadwell-U and Y skus at that low TDP) will be even faster than anything intel can put in that power envelope.
 
and it's likely the A8 (which will be quad core, unlike these Broadwell-U and Y skus at that low TDP) will be even faster than anything intel can put in that power envelope.

Haswell has 60-100% better perf/core/clock. Probably it'll be similar at heavy multi-threading, but Broadwell should end up much faster everywhere else.
 
Those would be interesting figures if they weren't totally fabricated http://forums.anandtech.com/showthread.php?t=2343525

Funny, but Haswell's faster in more real world applications: http://forums.anandtech.com/showpost.php?p=36426761&postcount=82

(I posted that in response to you, did you forget?)

On Sunspider Haswell is... 13.8% faster
On Kraken Haswell is... 88.7% faster
On Google Octane V2: Haswell is... 104.1% faster
On WebXPRT: Haswell is... 76.5% faster

Core i3 4330(2 cores, 4 threads, 3.5GHz, no Turbo), vs iPad Air A7(2 cores, 1.4GHz)

Sunspider Javascript: 137 vs 389.9(less is better, 184.6%)
Mozilla Kraken: 1224 vs 5773.2(less is better, 371.7%)
Google Octane: 27089 vs 5308(higher is better, 410.3%)
WebXPRT: 2369 vs 537(higher is better, 341.2%)
 
Last edited:
Funny, but Haswell's faster in more real world applications: http://forums.anandtech.com/showpost.php?p=36426761&postcount=82

On Sunspider Haswell is... 13.8% faster
On Kraken Haswell is... 88.7% faster
On Google Octane V2: Haswell is... 104.1% faster
On WebXPRT: Haswell is... 76.5% faster

Core i3 4330(2 cores, 4 threads, 3.5GHz, no Turbo), vs iPad Air A7(2 cores, 1.4GHz)

Sunspider Javascript: 137 vs 389.9(less is better, 184.6%)
Mozilla Kraken: 1224 vs 5773.2(less is better, 371.7%)
Google Octane: 27089 vs 5308(higher is better, 410.3%)
WebXPRT: 2369 vs 537(higher is better, 341.2%)


So 8 years of software optimization for core architecture netted them 13% sunspider .



Sounds like Apple should def throw in the towel. 13% from a desktop socket processor against a mobile SoC is impressive.


You know these are all browser benchmarks. All it says is Haswell had better software optimization.
 
Last edited:
Sounds like Apple should def throw in the towel. 13% from a desktop socket processor against a mobile SoC is impressive.

Sunspider, not Kraken. Only thing probably simple enough to fit in the A7's dedicated cache memory, which no ARM chip has.

And its 13% per clock.

Let's see another benchmark: http://www.anandtech.com/show/8035/qualcomm-snapdragon-805-performance-preview/2
This time against the A15 core in the Tegra 4 for Asus Transformer

Per clock advantage of A7

Sunspider: 112.7%
Kraken: 26.6%
Google Octane: 36.6%

Apple's version of A57 is A7.

That doesn't mean Apple will use Intel chips of any sort, or they have to. They have financial reasons for making their own SoC.
 
Last edited:
Sunspider, not Kraken. Only thing probably simple enough to fit in the A7's dedicated cache memory, which no ARM chip has.

And its 13% per clock.

Let's see another benchmark. This time against the A15 core in the Tegra 4 for Asus Transformer

Per clock advantage of A7

Sunspider: 112.7%
Kraken: 26.6%
Google Octane: 36.6%

Apple's version of A57 is A7.

That doesn't mean Apple will use Intel chips of any sort, or they have to. They have financial reasons for making their own SoC.


Haswell IPC has been known to be roughly equal to A7, this is something that has been discussed many times before.


You keep quoting irrelevant browser benchmarks when the hard data on A7 IPC is right in front of you in the thread I quoted.


Broadwell is useless to apple in a tablet form. You can quote me on that.
 
Haswell IPC is far better TreVader and you know it. Stop cherry pick a benchmark optimized for the A7 config to try prove otherwise. Its about equal to cherrypick the handfuld of benchmark that BD/PD/SR doesnt suck at vs Core series.
 
Haswell IPC is far better TreVader and you know it. Stop cherry pick a benchmark optimized for the A7 config to try prove otherwise. Its about equal to cherrypick the handfuld of benchmark that BD/PD/SR doesnt suck at vs Core series.


Haswell IPC is actually slightly inferior to A7 and the vast majority of people agree including Anand. I would link you to the thread but you posted the same denials in that very thread and I'm sure are aware of the misinformation.

Broadwell should be interesting for higher TDP but Apple already is working towards replacing Intel in its MacBook line.

It's even possible we will see an A8 or A8X powered MacBook Air in the fall.

Enough. This isn't the Haswell vs. Cyclone thread
-ViRGE
 
Last edited by a moderator:
Haswell IPC is actually slightly inferior to A7 and the vast majority of people agree including Anand. I would link you to the thread but you posted the same denials in that very thread and I'm sure are aware of the misinformation.

Broadwell should be interesting for higher TDP but Apple already is working towards replacing Intel in its MacBook line.

It's even possible we will see an A8 or A8X powered MacBook Air in the fall.

You already got proven facts that A7 is far below Haswell IPC:
http://forums.anandtech.com/showpost.php?p=36426761&postcount=82

The next MacBook Air is Broadwell based. So your hope of death to Intel have to wait a while longer.

But again, x86 sales are in growth again. Not dying anytime soon.
http://newsroom.intel.com/community...ll-year-revenue-and-gross-margin-expectations


Enough. This isn't the Haswell vs. Cyclone thread
-ViRGE
 
Last edited by a moderator:

"Intel's next-gen processor, Broadwell, is (mostly) a 2015 thing"

I think it would be easier if they just skipped Broadwell completely and went straight to Skylake, since Broadwell is so seriously delayed. What's the point of having Broadwell around for a few months if Skylake will be released just after that? Skipping Broadwell would mean less SKUs to maintain.
 
Skipping BDW would mean another 6-12 months delay. Would be stupid from Intel.

More like 3 months for the desktop CPUs, and 3-6 months for there rest of the SKUs (except for Broadwell-M).

It will be stupid for OEMs to base their products on such a short lived CPU. Those products will soon end up on the discount shelves when Skylake is released.
 
Last edited:
"Intel's next-gen processor, Broadwell, is (mostly) a 2015 thing"

I think it would be easier if they just skipped Broadwell completely and went straight to Skylake, since Broadwell is so seriously delayed. What's the point of having Broadwell around for a few months if Skylake will be released just after that? Skipping Broadwell would mean less SKUs to maintain.

Given lack of competition and the inherent nature of ROI I think we can infer that skylake will be receiving a delay of its own.. or at least a delay in the segments where broadwell is still catching up (the dollar bills).
 
Given lack of competition and the inherent nature of ROI I think we can infer that skylake will be receiving a delay of its own.. or at least a delay in the segments where broadwell is still catching up (the dollar bills).

"Lack of competition"? In tablets/2 in 1s Intel is competing against the most dangerous competitor in years, Apple. How long do you think it will be before Apple brings out an iPad with a 20nm Cyclone? Given that new iPads tend to launch in October, it's likely to be here before Broadwell...
 
Last edited:
"or at least a delay in the segments where broadwell is still catching up" - i said trying to cover my butt ... Yea, cyclone, interesting, sucking up more power, needs software rewrites etc.. adoptation may take time or Apple may have it locked down allready (anyone can do it, spouse its apple).
As I read Anand, cyclone is more of a desktop-move than anything else (ultimately replacing x86 for OSX? -Which again means that for a given power envelope(mac books) Apple does not see Intel/x86 competing anytime soon).
 
Like how Haswell and Bay Trail also were more a 2014 thing?

He said "including Core M". Maybe that means "only Core M" or maybe he was just stressing Core M since that's a new name they introduced a few weeks ago.

"Intel's next-gen processor, Broadwell, is (mostly) a 2015 thing"

I think it would be easier if they just skipped Broadwell completely and went straight to Skylake, since Broadwell is so seriously delayed. What's the point of having Broadwell around for a few months if Skylake will be released just after that? Skipping Broadwell would mean less SKUs to maintain.

Great idea, now tell Intel to skip and waste those billions of dollars they spent on Broadwell. If they start releasing Skylake in Q2-Q4, that would be a very long time between 2 generations, too. So I guess their current plan is quite decent.
 
Great idea, now tell Intel to skip and waste those billions of dollars they spent on Broadwell. If they start releasing Skylake in Q2-Q4, that would be a very long time between 2 generations, too. So I guess their current plan is quite decent.

Don't get trapped in the "sunk costs" fallacy. 🙂 It's a common mistake in business- confusing past spending on a project with its current value. Sometimes a project should just be canned. (Not that I am saying that this is the case for Broadwell! Just discussing business principles in general.)
 
@AtenRa - there were many haswell laptops end of 2013. surface pro 2, yoga and few more from lenovo too

That maybe true in the US, the rest of the world didnt have any haswell Laptops until early 2014. You could only find one or two extremely high-priced ultra-books at best. I remember even Newegg only had Apple Haswell laptops until October-November 2013.
 
Back
Top