Intel Broadwell Chip Release Pushed Back to 1Q 2014

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
too add @ingtar33, even your article states that Intel has not officially confirmed that there WON'T be a Haswell for desktop.
 


how about you show me something from an article newer then last year Christmas time.
 


As opposed to your article that states that Intel has NOT confirmed that there WON'T be a Broadwell Desktop? Ontop of the articles where a leaked roadmap is showing a Broadwell-K leading to a desktop version. Even Fudzilla's article stating that their "sources" (unconfirmed) are saying there won't be a Broadwell-D

 
It all goes back to Moore's Law, folks. You can only shrink the die so much before **POOF!**

22nm is good enough for desktops. To be honest, going lower for the sake of it is more than likely a hassle at this point. I'd leave the 14nm for mobile devices, if they're able to sort it all out. Refine what you have and don't press your luck so quickly.
 

The funny thing about "process nodes" is that in the 350nm days, the process name used to match the size of the smallest discrete feature on the chip... but today, at 22nm, there is very little relation between the process name and the smallest feature: on Ivy Bridge/Haswell with Intel's FinFET flavor, the smallest feature is the 8nm gate width. Gate lengths on the other hand have not shrunk much in recent years due leakage increasing too much.

Other things like copper traces have also reached their practical limits due to the passivation layer required between the copper and silicon to prevent copper from leaching out.
 
Here's a more correct headline:

"INTEL KNOWS THAT IT HAS NO COMPETITION SO IT'S WAITING AT LEAST THREE MONTHS TO RELEASE THE NEXT GENERATION SO IT CAN MILK THIS ONE FOR JUST THAT MUCH LONGER."

Related Headlines:

"CONSUMERS BUYING INTEL BECAUSE THEY DON'T KNOW ANY BETTER IN MOST CASES."
 

It does not really matter anyway since Broadwell is little more than a Haswell die shrink and based on what happened with IB, it seems unlikely Broadwell will perform substantially better than the 7-10% improvement we have been getting from annual Intel product refreshes since the first-gen Core iX... we got the huge 40-60% kick over Core2Duo/Quad largely thanks to the Integrated Memory Controller reducing RAM latency from 120-140 cycles to ~60 cycles and that was pretty much it for major performance boosts if we ignore AVX1/2/3 extensions.

As I have been saying for a long time, if improving CPU cores was so easy that one could reasonably expect Intel to continue improving CPU performance by 50-70%/year, why is AMD who is so far behind on per-core performance only managing ~10%/year as well? Simple: it just is nowhere near that easy to push per-thread performance much beyond where it is at today.


If by "thermal issue" you mean the thermal paste and gap, my guess is they won't go back to soldered IHS on mainstream chips - it gives enthusiasts one more shove to pay up for LGA2011 or whatever the next Extreme socket might be. The entry-level Extreme chip and motherboards are about on par with the i7 and mid/high-end motherboards for both price and performance with the Extreme having the benefit of extra PCIe lanes and memory channels so it seems like a relatively reasonable push to me.
 
actually intel itself has said there will be zero IPC improvement with broadwell... they're putting all the extra stuff into the igpu for this release. so don't expect even a modest improvement in the ipc for broadwell.
 
Interesting times ahead, for sure. My understanding of Haswell-E is that there will be no Quad chips, only 6 & 8 cores available.

But that doesn't make any difference for me at this point. I want one of my last builds to be the best one possible, after having dealt with AMD for so long. It's time to go back to Intel once again...
 
I equate this more to "AMD has nothing that can beat our CPUs so we're going to milk the current one for another three months. We'll say the next one will not be delayed until we decide that AMD still can't compete and we'll push that one back 6 months instead of three. If things still don't change, we'll milk that one for an extra year. Have a nice day!"
 

With only 5-10% performance gain per chip generation, chips from a few years ago already being overkill for most people's everyday computing needs and PC/laptop sales dropping by 8-13%/year, the market simply is not in a state that can sustain unnecessary major desktop chip launches every ~12 months.

BTW, Mobile Broadwell was originally supposed to start production in late 2013 so even with a ~3 months delay, that still puts it ahead of the 1-year cycle from Haswell's launch. The late-2014 launch for desktop Broadwell might be a few months "late" but then again, Intel's original plan seemed to be forgoing desktop Broadwell altogether so, not counting Haswell Refresh, that would make it ~24 months between Haswell and Skylake on desktop.

The current state of things is not just due to Intel being dominant. Getting extra performance out of individual cores has pretty much reached the end of the line, finely threaded mainstream applications to take advantage of extra cores and hardware threads are nowhere in sight, most recent developments points to parallel processing on PCs and even SoCs moving to IGPs, computers from 3-5 years ago are still viable for an increasingly large chunk of the market so new chips face increasingly stiff competition from previous generations of hardware, mobile devices are taking over most media consumption roles, etc. So the market growth opportunities in the desktop space to justify further major investments in the diminishing returns pure-CPU desktop processing power simply aren't there anymore. This is true for AMD too and seems to be confirmed by a lack of roadmap updates for the AM3+ platform.
 
Silicone? Is Intel going to give up on CPU's when they need a sub nanometer process and switch to making breast implants?
Click to expand...
...for ANTS!
 


Very sexy ants.

I thought that Graphene was on deck as the next base material for the CPU. Pilot tests have already demonstrated the capability of a 1 terahertz clock, uses photons instead of electrons so less power consumption, and an atom of silicone has almost 2.5 times the mass of carbon, so just think of the transistors you could put on a graphene chip.
 

Discovering that something can be made to work under laboratory conditions as a one-off proof-of-concept experiment where mass-production scalability and unit costs are unimportant is a completely different story from designing something for (relatively) inexpensive mass manufacturing. Countless technologies get shelved or get restricted to premium specialty products due to high production costs.

One of the other breakthroughs in semiconductors is replacing wiring with nanotubes but this requires a fast, cheap, efficient and reliable way of putting or creating exactly the right nanotube at exactly the right place. Many self-assembly mechanisms have been demonstrated in labs with various degrees of success but none of them are anywhere near cheap, fast and reliable enough for consumer-level mass manufacturing - at least not yet nor in the foreseeable future.

Theoretical breakthroughs failing to reach mass manufacturing due to requiring further breakthroughs to make them actually viable is extremely common. Ex.: supraconduction is an awesome breakthrough but room-temperature semiconductors (required to make supraconduction practical for consumer use such as hybrid car electric motors and regenerative braking energy storage) are still elusive.
 


Completely agree. It should be interesting to see what pans out though, whether is graphene (AKA carbon-nanotubes) or DNA microprocessors. Nobody can deny though that the current fabrication process will be hitting a wall soon, both Intel's former chief architect and several theoretical physicists., i.e. Michio Kaku, think that good old Moore will be irrelevant by the inside of 2022. To me that signifies that in order to move forward a paradigm shift needs to occur-think the discovery of the double helix or the germ-theory of disease, either that or we just keep puttering along.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom