Forget Intel Haswell, Broadwell on the Way

Intel Launches the Intel Core M Processor
Intel launches three Core M CPUs, promises more Broadwell early 2015 | Ars Technica
Core M Broadwell Speeds, Feeds, And Performance
Intel’s Core M Strategy: CPU Specifications for 9mm Fanless Tablets and 2-in-1 Devices
Intel’s 14nm Core M Series “Broadwell-Y” Processors With 4.5W TDP Detailed – Powers Fanless and 2 in 1 Mobility Devices

Intel announces the 14nm Core M processor targetted at fanless 2-in-1 devices

 

Question, which CPU are they comparing it with?

 

Dufus, from the links already provided (footnote 3):

See:
Intel Launches the Intel® Core™ M Processor

 

Thanks.

So they compared the M-5Y70 which has a TDP of 4.5W and max turbo of 2.6GHz to a i5-4302Y which has a TDP of 11.5W and max turbo of 2.3GHz AND they did this by reducing the clock speed of the i5-4302Y to use a 4.5W TDP which would likely mean the i5-4302Y was running at something like 1.4GHz max turbo meaning a 40% reduction of it's normal speed.

IMO seems Intel are playing with numbers to make the newer Gen 5 look much better than it really is.

I can just see the users of the i5-4302Y thinking how good it would be to have up to 50% more processing performance with the M-5Y70 only to find there is very little difference other than saving some battery power.

 

I don't see it like that. They made the TDP equal and the performance of the newer chips are superior.

Win for users on battery life and 'thin' designs, correct?

And now, with no fans required.

 

IMO they should just say it runs similar performance at a lower power, not possibly mislead people by leading them to think they will run 50% faster than their current Gen 4.

 

Oh, I agree with your statement... but that is why learning to be cynical and thereby reading/understanding for comprehension is better than blindly trusting what companies and their marketing departments try to shovel down our throats.

Notice that they aren't technically lying - though it may seem like that to the people that may be reading/skimming at just at a surface level.

 

But that would sound nowhere near as sexy. Broadwell is Ivy Bridge all over again, incremental core improvements, large jump for iGPU (though it's a smaller jump than HD3000->HD4000), the biggest winners are the thin designs due to the more powerful HD graphics + eDRAM.
I'm more interested in performance of the switching IVR, if it is an improvement accross the whole CPU range (as opposed to Broadwell M only), it should reduce core temperatures due to the higher efficiency at high loads in addition to reducing idle power draw.

 

not really. All laptops will benefit...desktops see nothing really besides maybe slightly better overclocks since they run cooler but that will have to be seen. I can only get 3.6GHz at 45 TDP in IB on max load but on light load like single or dual thread i can get 4-4.3Ghz within the 45w TDP range. That is the max my M17x R4 can cool. With broadwell i bet we will see 4-4.3 GHz in 45w range (my guess. We will have to see) That would be pretty awesome considering Haswell desktops can only OC to 4.7 GHz :/

This means 45-55W TDP laptops will be able to almost be as fast as a full fledge desktop...kinda for the first time. This is good and bad...bad for desktops but pretty crazy still.

 

Anyone know which H processors are unlocked? I know some are because i saw on a review IIRC on anandtech showing the large igpu being TDP limited (so some have to be unlocked if they fitled with TDP). I am curious because a broadwell ultrabook would be tempting but my computer use is extremely single thread limited. I use a jerry rigged 4.3 GHz on my R4 and i would like to see a jerry rigged 4.5GHz on an ultrabook. I am note interested in an ultrabook with like an 850m/950m but i might be dreaming to have that option. IIRC ASUS zenbook/macbook pro are the only ones that offer that type of set up. It would kill me to be stuck with 3.8GHz on all 4 cores :/

I am debating if i will just abandon my desktop hopes until skylake because if broadwell doesn't solve some of this overclocking issues a laptop with 4.3 GHz vs 4.7-5Ghz desktop doesn't really warrant the expense considering i would just junk the thing for skylake E...assuming skylake E is merely a year away.

Also anyone know with broadwell laptops comiong if M.2, DP 1.3, and HDMI 2.0 will be standard? those are really key game changers in tech. Buying anything without those is really stupid if you plan on using it for several years. I would hate to buy a computer and junk it in 12 months. Waste of money and time if you ask me.

 

Skylake-E is at least another 3 years out, assuming it'll even exist.

 

broadwell E should be 2015Q4 IIRC so 2016Q4 would be skylake E right? new one after each year IIRC

IB-E 2013 Q3
HW-E 2014 Q3

so broadwell should be 2015 Q3/Q4 and skylake will be 2016 Q3/Q4

 

Sorry, if this sounds ignorant, but are any of these recent processors they were just announced more powerful than the 4940MX?

 

No. They only announced three 4.5W TDP mobile parts.

 

Skylake-E will most likely be on a new platform though. X58 to X79 took 3 years, X79 to X99 took 3 years, so I'm assuming X99 to X119 (or whatever Intel decides to call it) is going to be 3 years. And X99 was just released last week.

The E chips usually come out 12-15 months after launch of the mainstream desktop chips. If desktop Broadwell doesn't launch in Q4 this year then the chances of Broadwell-E in Q4 2015 are slim to none.

 

IB i7k 2012 Q2
IB-E 2013 Q3
HW i7k 2013 Q2
HW-E 2014 Q3

you maybe right but intel is also rushing skylake out the door with similar time-lines as if broadwell wasn't delayed.

Intel Desktop CPU Roadmap Confirms Broadwell-E In Q3 2015 - Broadwell K-Series in Q1 2015 Followed by Skylake In Mid-2015

Also this new roadmap that was released shows BDW-E as 2015 FYI


If i build a desktop it won't be a several thousand dollar one thats for sure. It'll probably be like a 1500 dollar one or so just to hold me over til skylake E. Then i'll use it as a server or a computer that doesn't touch the internet so i know its clean for what i deem to be sensitive work.

 

Since you mentioned you need single thread speed, why not do a Z97 build with 4790K? Stock single core is already 4.4GHz, and if you don't push all 4 cores you could easily get another 300-400MHz on single core. Haswell IPC is about +200MHz compared to IB, so 4.8GHz on 4790K is like having 5GHz on an Ivy (single core of course).

If you don't need to game then you could easily build a decent Z97 desktop for under $1500.

 

a 17% boost isn't really worth the money for the desktop i want* hence why i said i might just do a temp computer for 1500 give or take like you said while i wait for skylake E. It is just sad 17% is all i can get :/ Granted. it may be as high as 25% but that is pushing it to optimistic levels.

*I plan on going E with water cooling, PCIe SSD/RAM disk* if they ever make a new one of those. 2-4 way sli and what not.

*some company release the allcloud(?) but its a joke :/ Performance is throttled due to crap gen 1 PCIe and they want 15k too roflmao! It has to be a 100 dollar board lol. If ram drops to 2012 prices ($3.2/GB) again i would totally go for a 256-512GB ($1-2k) dram disk but no reasonable product exists :/ I read about people having the old gigabyte one and said even back then it was awesome and they wish a new one existed.

 

It is really really difficult to increase IPC, good lord, millions and millions of dollars of RnD is required just to eke out the 5-7% improvements we see yearly. We, as enthusiasts, really need to tame our expectations, Sandy Bridge was one of those rare boosts to IPC that only comes around once every several generations, akin to the unveiling of Core 2. I personally, am waiting for DDR4 to mature before upgrading since it's really expensive and exceptionally slow at the moment.

 

i know that but the lower over clock on Haswell was what made it equal to IB -_- so a complete side upgrade. IB was pretty good upgrade from Sandy bridge but haswell was a complete bust in max performance when you count overclocks. At least IB was able to hit 5GHz. Hell fox was pushing 4.5-4.9 on his m18x IB.

I am hoping skylake will at least be decent and hold me off until we change material in 7nm...that might finally offer higher freqs if we are lucky.


UPDATE:
Additionally, one of the main purposes of getting a desktop and the timing of getting one was trying to hope on the 20nm or whatever is the next GPU die shrink. Though, that is turning out to be a big unknown because everything i have read says that we won't see a die shrink even in 2015. Further, if i have read articles correctly TSMC and Global Foundries both have said that their won't be any high performance 20nm at ALL. So it looks like 20 nm very well maybe even skipped for GPUs, which means 14/15/16 nm or whatever will be the next die shrink. This further cements the likely hood that i might not as well bother blowing the big bucks til 2016 (for a desktop that is). So modern 4K gaming won't happen for me for a long time :/ Might as well build my range and buy my gear and weapons over a highend desktop. :/ I will be looking forward to the news over the next 6-12 months to see whats going to happen. That even means i think AMD won't even have a 20nm CPU. They might be stuck at 28 nm for a super long time.

 

I seriously doubt Intel will stick to schedule... With the transistors being shrunk down so much, any further shrinkings are becoming very costly so its in Intel's interest to stall as much as possbile.. With AMD being so far behind, even more so, Intel has no incentive to release Skylake anytime next year..

 

It could be argued, though, that as an architecture change and not a process shrink, Skylake would have no real technical reason for being delayed. My best guess is that IF Skylake is delayed, it will be for business reasons rather than technical ones (companies want time to sell Broadwell systems before they're obsolete).

 

and support for that? Everything so far on their roadmap checks out...unless you have read anything i haven't

On a side note, on the SB-IB overclocking if IB didn't offer a decent boost in performance over SB then I am thinking of laptops with overclocking. So i could be confused but don't have time to compare SB and IB to fact check. I did a quick google but didn't find a good review.

they will have a year for the most part. It is just the k chips and some laptops chips that are getting shorted. Laptops with U get 3 quarters instead of 4 like normal and desktops get locked skylake a quarter after broadwell k get released.

 

Broadwell was originally scheduled to launch at the end of 2013, according to a Techreport article published back in Oct 2013.

Then it got delayed to "around 2014".

Then it got delayed to "very end of 2014 for Core-Ms (small ultrabooks and tablets), spring 2015 for regular ultrabooks and laptops, and fall 2015 for desktops".


TSMC and GF haven't been doing that well either. I'm fairly sure Nividia and AMD are annoyed that they have to port Maxwell/Tonga to 28nm instead of producing them on 20nm.

 

you realize the beginning of your post makes no sense right?

So Haswell 4770k releases Q2 2013 and months after that release people are saying broadwell will come out? I dont have a good memory but i am pretty sure no one with a brain thought that or would believe that.

 

Here's the article from Oct 2013: Intel pushes Broadwell production into early 2014 - The Tech Report

"Intel demonstrated a 14-nm Broadwell processor in action, and it said the processor was on track to ship by the end of 2013.

That's no longer the case. Speaking to investors during the company's quarterly earnings conference call, Intel revealed that its Broadwell product has been pushed back one quarter."

 

ok its an earnings call...intel is always last minute on saying something we already know. They don't give us a realistic up to date timeline....ever? The point is no one in October expected broadwell til 2014 because the actual release date on haswell was evident that it was never going to happen and everyone had enough ideas due to rumors and leaks. Its like saying we expect 20nm GPUs this winter because the last official word said so 1-2 years ago...obviously that's not the case.

 

First Actual Intel Broadwell Core M Benchmarks From IDF 2014 - HotHardware

 

My point is that Intel previous set the schedule for 2014, and moved it to early 2015 (for Core M).

My bigger point is that the silicon die shrinks' delays are going to get more frequent as we move below 20nm/14nm. And there's a certain point where economics won't permit a silicon die shrink of the cost per transistor goes up (which is very possible below 10nm).

 

that may happen and that may not happen it all depends. Silicon it self only has one more shrink so the delay might not exist or be very short. Assuming those applied mechanics or whatever slides are accurate. skylake though shouldn't be delayed. Cannonlack is a different issue. I do see buying a laptop on broadwell being a good investment like IB was. The battery life and higher performance over SB was pretty good. Also we know broadwell E edition is still on track for a 12 month cycle so skylake should still be the same given that its just a new arch

 

I'm still waiting for my 10Ghz Pentium IV. Any timeline that Intel says is completely bogus and they know it. Idk how they can get away with it when every single release for the past 6-7 years has been "delayed" by months or years.

 

Seems like you can't see the forest for the trees.

Performance-wise, a 10GHz Pentium IV was surpassed years ago. And although specific releases have been proven to vary vs. the official timeline Intel originally offered, overall, they're on schedule.


See:
http://www.anandtech.com/show/8510/idf-2014-intel-demonstrates-skylake-due-h22015


Look for the "Intel's Tick-Tock Cadence" table in the link above.

 

looking back to SB

the E edition has this timeline so far
SB-E 2011 Q4
IB i7k 2012 Q2
IB-E 2013 Q3
HW i7k 2013 Q2
HW-E 2014 Q3
BW-E 2015 Q3?
SL-E 2016 Q3?

IB-E was the one that got delayed a lot but it seems that they might be back to a 12 month cycle if the last leak holds to be true, which might be the case

There tick tock maybe on a year basis on schedule but if you look at it by quarters i believe* they are 3Q behind. they used to do hanuary releases IIRC and now its Q3/Q4 IIRC

 

Would have been good to have had the PL1,PL2 and PL3 settings, core temps and actual power consumption with the benches. Maybe in the next few weeks.

If Skylake isn't that far behind Broadwell it might be better waiting, at least for the performance CPU's. 512bit vectors and some posts saying the integrated voltage regulator is getting dumped! Wonder why that might be?

 

IVR is being dumped in skylake. I dont know why but it has been know for awhile. Maybe IVR was too soon? I like the concept though. I think it has merit or at least in theory.

BW will definitely be good for notebooks but for desktops i am 99% sure it won't offer anything good in performance wise (max performance), which is sad :/

 

TSMC Said to Manufacture Apple’s A9 Chip on 16nm FinFET Process in Q1 2015 | Mac OS X | Buy iMac

TSMC is starting 16 nm trial production and will have full production starting Q4 2014/Q1 2015. So maybe we will see 16nm GPUs and 20/22nm will be skipped. Maybe we will see 16nm AMD CPUs too.

 

I like FIVR, gave laptop users much more control over how to run their chips. Still quite not desktop level tweaks but getting there. Too bad it just added more heat to the already overheating Haswell.

 

Well the Broadwell FIVR (V2) is supposed to bring some enhancements such as dynamically adjusting the CPU input voltage. Seems like a big step backwards with Skylake.

 

Yeah, Broadwell supposedly brings a switching design for the FIVR which should be much more efficient (i.e. less heat output). No idea why Skylake would forgo the FIVR, it is pretty much responsible for Haswell's awesome energy efficiency.

 

It would be a shame if Skylake included locked voltages, preventing undervolting like what Sandybridge did.

 

So a Broadwell super low voltage (6W) CPU stronger than a 15W Haswell or what?

 

it might also have a TDP cup or something. I doubt its almost 3 times more efficient

 

Stupid question... anybody heard anything about if the Spring '15 chips are going to be Socket G3... or slightly off-topic if there might be a G3 Skylake later? I bought a Sager for its ease of upgrading, and I'm really hoping Haswell Refresh i7-4940MX won't be the end of the road for it.

 

With notebooks, I'd worry more about BIOS support/CPU microcode updates than socket compatibility.

 

I would be very surprised if you could put a Broadwell or a Skylake chip in your Haswell based platform, no matter how upgradeable it once was.

How would Sager stay in business if you were able to do that on your own?

If you want what's coming next and especially if you want Skylake... start saving for a new system.

 

thats irrelevant.


Has any notebook even been socket compatible or upgrade-able? If so it hasn't been in recent times.

 

Maybe irrelevant to you, but not to a business.

Your second point is exactly my point/first sentence.

 

sure i would love everything to be compatible but that tends to not happen for a number of reasons but to simply blame and say its sager being *insert word here* is naive. IIRC desktops partly change socket because they have to for them to function correctly....granted i am sure some or a lot of ulterior(SP?) motives exist.

 

Dude, Sager (more specifically, Clevo) have some of the best track records for upgradeability provided the CPU uses the same socket due to their excellent policy of updating the BIOS microcodes. Even if Broadwell is directly pin compatible, most other OEMs won't provide the microcode updates.

I'm dubious of Skylake being compatible due to the rumoured FIVR changes but I'm expecting Broadwell to be akin to Ivy Bridge.

 

So, we're all saying the same thing here?