How many years till Intel releases a new non-x86 uarch?

[cbn]

Be advised the poll is not public.

 

[cbn]

Some great posts from the old thread (that hopefully gets deleted due to fact I forgot a poll option)

itanium? I thought it was supposed to ultimately replace x86 as intel developed it, let people become more familiar with it on the super high end, and then let it slowly trickle down and replace x86 in the maintstream... but then AMD came around and say "wham bam watch us extend x86 even longer boyyyzzzzz" and that was the end of that. Someone correct me

Would be interesting to see Intel develop or liscense ARM architecture.

Does the mobile space have room for another architecture? Or is ARM getting to the point where its architecture is so entrenched it is the new intel of the mobile space and companies will only build systems around ARM?

Intel has ARM license, and they used to make ARM instruction based chips for phones. The first one was called StrongARM, the one they bought from DEC, and the successor was called XScale. They sold the XScale division to Marvell, which still uses the XScale name. They still use the ARM instruction chips in network cards for servers and I/O controllers.

Even Moorestown, Intel's Atom based handheld and tablet platform, uses a very simple ARM core inside the companion chipset.

 

[Edrick]

Intel i860 and i960 were a RISC based CPU in the late '80s early '90s.

Itanium, as someone else already mentioned.

Xscale, as someone else already mentioned.

So I do not really understand this poll.

 

[jones377]

Intel i860 and i960 were a RISC based CPU in the late '80s early '90s.

Itanium, as someone else already mentioned.

Xscale, as someone else already mentioned.

So I do not really understand this poll.

Those are the old non-x86 archs from Intel. I assume the OP was already aware of them? In which case, the poll makes perfect sense. It just assumes everyone else also know this already, a big assumption

 

[Edrick]

Those are the old non-x86 archs from Intel. I assume the OP was already aware of them? In which case, the poll makes perfect sense. It just assumes everyone else also know this already, a big assumption

Itanium and Xscale are current. And the fact that Intel spend billions on Itanium and it never did as good as they had hoped, I highly doubt Intel will move away from x86 anytime soon.

 

[Lonyo]

Considering that Intel has been talking about "x86 everywhere", it doesn't make much sense to release a non-x86 architecture, since that kind of counters the whole idea. Also Intel has somewhat of a lock on the x86 market due to patents, etc, so while they could do something with another architecture, or maybe come up with a new one themselves, they have a comfortable x86 position, so is there a need?

http://www.itnews.com.au/News/115934,intel-sees-x86-everywhere-in-future.aspx
http://www.osnews.com/story/22213/Intel_Aims_for_World_Domination_x86_Everywhere
http://www.bloggingstocks.com/2009/07/16/intel-intc-x86-everywhere/
http://www.anandtech.com/show/2493/3

Ironically, we are already moving away from x86, including in Intel's own devices, through the use of GPUs.
The real question should be "how many years until x86 doesn't matter because programs are made to work in a platform agnostic way, such as through .net, OpenCL or other means"?

 

[Wreckage]

I'm sure Intel has plenty of backup plans. ARM/Itanium/MISC.

If ARM totally crushes x86 in the next few years, Intel has plenty of experience with ARM and will just shift gears no problem. Although I imagine they will be cashing x86 checks for a long time to come.

 

[nyker96]

x86 ... live long an prosper!

 

[nenforcer]

IA-64

It will probably just be reinvented under a different name to avoid all references to Itanic.

 

[Voo]

IA-64

It will probably just be reinvented under a different name to avoid all references to Itanic.

So you're saying after investing billions into itanium which more or less was completely in vain (we won't even see another MS Server version for itanium, which pretty much says it all), they'll just start anew? Considering that some people in intel were opposed to the whole idea from the get go and seeing how the actual fate of itanium has confirmed their position, I don't see Intel trying another costly experiment in the near future.
That sounds hardly reasonable - a low power risc architecture sounds possible, but superscalar architectures are hardly known for their great power efficiency..


Also we should not forget that the $revenue /invested R&D $ has shrunken and shrunken the last few years..

No I think we'll see more CPU/GPU integration and more power efficient x86 architectures (new atom) in the next few years. There's also x86 for phones on the way, let's see how that turns out.

PS: Though someone could check all the Intel research papers to see if they done (well released) some serious research about ARM - not aware of any papers in that direction, but then there's lots of papers.

 

[cbn]

Intel i860 and i960 were a RISC based CPU in the late '80s early '90s.

Itanium, as someone else already mentioned.

Xscale, as someone else already mentioned.

So I do not really understand this poll.

I knew about Itanium, but wasn't sure about ARM. (I didn't know about i860 and i960. Thank you for mentioning them.)

Basically I am wondering if Intel has incentives to build something completely new (from the ground up) for mobile? Maybe something even more power efficient than ARM that could also be used to replace their x86 servers later on?

 

[P4man]

Everything intel ever tried to get away from x86 has been a failure, ranging from minor multi billion dollar failures to colossal multi billion dollar cockups.

Most have already been mentioned, they tried so often, I dont think they will try again. On the contrary, they seem to have realised that x86 is all they do well, and as a result are now trying to shoehorn x86 in to gpus (remember larabee?) and phones, so far with about as much success as their attempts to get away from x86.

intel will slowly sink along with x86. That might take 30 years though, and perhaps even some stuff might keep afloat when the x86 ship goes under. Perhaps it even has a bright future, fabing ARM chips designed by Apple, microsoft and nvidia

 

[cbn]

Everything intel ever tried to get away from x86 has been a failure, ranging from minor multi billion dollar failures to colossal multi billion dollar cockups.

Most have already been mentioned, they tried so often, I dont think they will try again. On the contrary, they seem to have realised that x86 is all they do well, and as a result are now trying to shoehorn x86 in to gpus (remember larabee?) and phones, so far with about as much success as their attempts to get away from x86.

intel will slowly sink along with x86. That might take 30 years though, and perhaps even some stuff might keep afloat when the x86 ship goes under. Perhaps it even has a bright future, fabing ARM chips designed by Apple, microsoft and nvidia

I was quite surprised when I found out MS didn't even bother to port Office to Windows Itanium. Can anyone explain why that didn't happen?

 

[P4man]

I was quite surprised when I found out MS didn't even bother to port Office to Windows Itanium. Can anyone explain why that didn't happen?

Itanium never managed to scale down outside the highend server market. Office has absolutely zero relevance there. The only itanium windows machines that ever sold where big iron running ms sql server.

 

[cbn]

Lets say the Battle between Google/ARM and MS/x86+ARM stabilizes at some point in the future.

Who could Intel use as a software partner for a new and more efficient Intel developed uarch? Maybe Amazon?

I don't know very much about Amazon other than the fact they make a nice high battery life device (with Free 3G for life) in the Kindle 3. Would it make sense for Intel to strategize with that company for a move against both Google and MS at some later point in time? (focusing on high performance rather than basic e-reader/web browser functions)

 

[BD231]

As soon as they stop paying multiple billions to companies they don't want running their code. x86 is a cash cow in ways RISC will never be and lets face it, by the time Intel gets rid of x86 they'll be moving to something far better than RISC processing anyway.

 

[cbn]

A couple of other reasons Intel may be developing a new uarch:

1. IMHO 16nm and 11nm may be less profitable for x86 in this age of competition from ARM and Google (particularly if Google has plans to invade desktop). I'm not saying Intel won't use 16nm. I just wonder if the ROI will be a good deal less than with past fabs.

2. Advancements in Low power screens will put a stronger spotlight on power efficient Operating systems and Chip Design. Here is a good article discussing the future of low energy consumer displays. For example the Qualcomm Mirasol Display uses only 1 mw compared to 2000 mw for a standard LCD display.

P.S. The display article I linked talks about a "yellow tint" on the Mirasol display. Looking at other YouTube videos on Mirasol it appears that problem has been fixed.