I can't see a competitive x86 chip in the smartphone-world, and I can't see Intel being competitive outside of x86.
Their ASPs for Atom chips are only $25. Pricing shouldn't be a problem, as long as they stand on the premium ladder. Higher volumes of chips sold would help lower pricing(or increase margins at the same price). $25 for 40 million units/year. If they capture 20% of Smartphone marketshare, it would be 80 million units/year.
The problem is that the java machine for android contains some critical tuning for ARM neon architecture. On x86 and other architecture you end up with lower perfomance, poor battery and possibly with crashing (ANR Error).
We need someone to invest a huge amount of money on improving the JVM for x86 or MIPS. Oracle will not do it, and Google is worried with the explosion of cost of android development, so....
Steve Jobs initially hoped to create his own network with the unlicensed spectrum that Wi-Fi uses rather than work with the mobile operators, said wireless industry legend John Stanton.
Stanton, currently chairman at venture capital firm Trilogy Partnership, said he spent a fair amount of time with Jobs between 2005 and 2007. "He wanted to replace carriers," Stanton said of Jobs, the Apple founder and CEO who passed away recently after a battle with cancer. "He and I spent a lot of time talking about whether synthetically you could create a carrier using Wi-Fi spectrum. That was part of his vision."
One promising application for graphene is in making new parts of the radio-frequency spectrum available for consumer electronics applications, said Richard Doherty, president of Envisioneering Inc., an industry consulting firm.
“It allows you to tame a spectrum that before was the wild, wild West,” he said. For example, it might make possible a new class of WiFi-style communications gear for wireless applications, or allow set-top cable boxes to be redesigned to send and receive ever-larger amounts of high-resolution video and data.
So where does that leave Intel strategy?
One thing that concerns me is that Intel has a history of making some monumentally short-sighted strategic decisions.
Remember their first foray into the mobile phone market?
How about their intentions to dominate the HDTV set-top market?
And then of course we have Larrabee and the GPU market.
Netburst was a net bust.
Itanium did not turn out to be Itanic, it is still afloat, but it is infamous for this:
What history tells us is that Intel is not afraid of taking risks, and they aren't afraid of betting big and walking away from billion dollar investments that just don't pan out.
Who knows what big plan Intel has in the works, and what untold billions are allocated to supporting the effort. But will it stick?
I love that graph.
That graph is really sad...it's the part of the reason we are still stuck with x86 and all it's legacy crap and havn't moved on to a more future oriented ISA...without all the "baggage".
I know most people are hot in love with AMD64...but in the long term that ISA just postponed the most sane solution.
If anything is "shortsigthed"...it's AMD64...
Do you program in assembly? If not, why do you care?
BTW, try programming in Itanium assembly. It kind of sucks.
http://www.xbitlabs.com/images/cpu/core-i7-3960x-3930k/die.png
Notice how the majority of the chip, despite its 8 cores, is just gobs of fast cache, memory controller, I/O, uncore, etc.? The "legacy" parts of x86 are just so incredibly small compared to the rest of the chip.
I wouldn't worry too much about "cluttering".
that graph is hilariousOne thing that concerns me is that Intel has a history of making some monumentally short-sighted strategic decisions.
Remember their first foray into the mobile phone market?
How about their intentions to dominate the HDTV set-top market?
And then of course we have Larrabee and the GPU market.
Netburst was a net bust.
Itanium did not turn out to be Itanic, it is still afloat, but it is infamous for this:
What history tells us is that Intel is not afraid of taking risks, and they aren't afraid of betting big and walking away from billion dollar investments that just don't pan out.
Who knows what big plan Intel has in the works, and what untold billions are allocated to supporting the effort. But will it stick?
yeah it's basically impossible to get any sort of superscalar performance out of it without the perfect dataset to work on, and writing a compiler that can dynamically optimize (IE in the way a human can by looking at the code) turned out to be basically impossible. It was a worthy effort.BTW, try programming in Itanium assembly. It kind of sucks.
Would you mind makring out the parts that you mention?
And again, even if we were taling about 0,001% of the core it would be 0,001% too much.
(The actual figure is much larger)
yeah it's basically impossible to get any sort of superscalar performance out of it without the perfect dataset to work on, and writing a compiler that can dynamically optimize (IE in the way a human can by looking at the code) turned out to be basically impossible. It was a worthy effort.
To be honest I think they could have gotten really far if they dumped money into PhD students trying to write research papers on optimization techniques. The world will never know what Itanium could have been...
From a high level, I think the Radeon GPUs use a similar architecture. They seem to be doing OK for compute, with way more sales than Itanium ever had. Although, it would be interesting to see if AMD's Firestream line is outselling Itanium.
Would you mind makring out the parts that you mention?
And again, even if we were taling about 0,001% of the core it would be 0,001% too much.
(The actual figure is much larger)
From a high level, I think the Radeon GPUs use a similar architecture. They seem to be doing OK for compute, with way more sales than Itanium ever had. Although, it would be interesting to see if AMD's Firestream line is outselling Itanium.
yeah it pretty much worked (ie was superscalar) on large datasets and that's it.
That graph is really sad...it's the part of the reason we are still stuck with x86 and all it's legacy crap and havn't moved on to a more future oriented ISA...without all the "baggage".
I know most people are hot in love with AMD64...but in the long term that ISA just postponed the most sane solution.
Do you mainly use an x86 computer? If so, why?
Most of my time is actually spent on Octeon CPU' s(Multi-Core MIPS64 processors)...most of my time is spent inside a Juniper routing engine, I only have PC's (CPU and GPU's) as an hobby and a side "thingy" when I get tired of watching bits flow across our corenetwork.
But nice try
From a high level, I think the Radeon GPUs use a similar architecture. They seem to be doing OK for compute, with way more sales than Itanium ever had. Although, it would be interesting to see if AMD's Firestream line is outselling Itanium.
Good for you. And how much would it cost for me to buy an Octeon system? What kind of specs would I get for that price? How many GFLOPs can it do for that price?
Also, why don't you have one of these at home? or do you?
GFLOPS is the wrong metric...you should google what a routing engine does...and look at Gbit/s