Analyst: Intel Should Buy An ARM Chip Company

Status
Not open for further replies.

Haravikk

Distinguished
Sep 14, 2013
317
0
18,790
This seems a bit premature; Intel's mobile strategy is only really just taking off, as Broadwell is the first generation that's going to really deliver on the efficiency and size improvements needed to reach the tablet market. If that's a flop then sure, sound the death bell for Intel in the mobile market, but while some of their past chips have been okay, their future offerings look very promising on both a performance and efficiency standpoint.
 

edlivian

Distinguished
Dec 10, 2009
96
2
18,635
This is freaking blasphemy!

The ATOM is better than ARM garbage. When will people learn.

on straight out performance its currently equal to high end smartphone socs, but battery life is miserable compared to arm socs.
 

ivyanev

Distinguished
Jan 29, 2011
101
0
18,680
If intel decide they want ARM processors - they can manufacture them if they want. But when the trend is - more power, more battery(capacity not life) what is the point when customers want powerful CPU and x86 can deliver.
 

icemunk

Distinguished
Aug 1, 2009
628
0
18,990
Why would they do that? This analyst seems to be out-of-touch. Intel is rapidly pushing into mobile, and within the next 6-12 months will have a huge lead on ARM chips. Note the $100 intel Z3740s Windows 8 tablets popping up - these are faster than the most ARM android tablets, and Intel is just warming up.
 

InvalidError

Titan
Moderator

The IGPs in Intel's current SoCs are still horrible jokes though and Intel will not have anything worth writing about in that department until late-2015.

Once Intel stops slacking off on the IGP, things could definitely get scary.
 

ddpruitt

Honorable
Jun 4, 2012
1,109
0
11,360
x86/64 will never be able to compete in the mobile space. The architecture has become so large and complicated that it can be made to run at the efficiency of ARMs architecture. Most of the power in big x86/64 systems comes from the fact that compilers and code have been optimized to take as much advantage as possible. The problem is the rules aren't the same for the mobile space and the same techniques used to give killer performance at the high end (massively superscalar, complex cache systems, etc) hurt performance for smaller devices. Atom is to ARM as AMD is to Intel, the performance is subpar but people use it because they can get it cheap. Intel can get there but their attempts seem half-hearted at best.
 
Ignoring any comment about ARM vs x86 etc., the big problem here is really that Intel already owns part of the company called Rockchip, who is the 3rd largest ARM SoC company in China behind Qualcomm and Allwinner.

Not only does it own part of Rockchip, it has signed agreements that is making Rockchip design and create Intel based x86 SoCs for future release.

So Intel is already stepping up its mobile plans, but besides that if it had to buy an ARM company it would be Rockhip not Mediatek.
 

InvalidError

Titan
Moderator

Tell that to Intel's 2-3W Atom chips.

Intel has managed to bring x86 down to power budgets nobody though x86 would never fit into and Intel's ultra-low-power x86 chips perform about on par with same-power ARM chips in CPU-centric benchmarks.

The biggest problem for Intel is the large number of non-portable popular Android apps compiled using the Android/ARM NDK instead of the portable Android DK... if you buy an Intel-based Android phone or tablet, about two thirds of the most popular games are unavailable because they are ARM-NDK builds.
 

icemunk

Distinguished
Aug 1, 2009
628
0
18,990


Oh yeah, I thought I heard something about that. Rockchip cpu's are actually pretty awesome. The RK3188 is a dirt cheap little tv box/tablet chip that does a great job. You can get little quad-core 1.6ghz RK3188's with 1GB of ram right now for $65. The RK3288 is even better, but it's pretty new to market, so prices are still a bit higher. $130 will get you a beast of a RK3288 tv box though. Rockchip is great.

 


Yes indeed it is, shop around you can do better on Amazon. $50 will get you an RK3188, a quad core A9 1.8Ghz little beast with a Mali-400MP4 clocked at 600Mhz, backed by 2GB RAM and 8GB internal ROM.
There new stuff is cool, but it will be a while before this seems too slow for video streaming and web browsing. But Intel I think only cares about getting Rockchip to put Intel into tablets, don't think they care too much about the mini-PC market. Its too low end for most companies to directly target.
 

ldo

Reputable
Aug 21, 2014
21
0
4,510
Intel has been trying to respond to ARM since it brought out its first Atom chips in 2008, back in the Netbook era. Remember Intel trumpeting its “Full Internet Experience” slogan? They said they were two years ahead of ARM at the time. Now they’re no longer quite as complacent.

This is a classic case of disruptive technology: Intel can see the writing on the wall, but it is simply incapable of responding to it rationally.
 

icemunk

Distinguished
Aug 1, 2009
628
0
18,990



Yeah, I got my RK802II for $54 back in November 2013; its Mali-400MP4 pumps out about 20 gigaflops with is faster than a Tegra 3 - so it can handle pretty much most 720p encoded videos I throw at it. I wouldn't expect it to handle compressed 1080 or 4K videos. For web browsing, its zippy; I use it with a little air-mote that works quite well. I think I'll wait a bit for the RK3288's to drop to $60 or so and buy one of them. They have a Mali-T760 MP4 and produces 80 gigaflops, so will be able to handle higher quality 1080 videos, and higher if you wanted.

My two little 7" cheap Chuwi V17HD tablets use the RK3188s, and are great for having around the house for occasional surfing/streaming/movies. I can't complain for $65.. 1GB ram, 8GB storage, 1024X600 screen, 3000mah, RK3188
 

ZolaIII

Distinguished
Sep 26, 2013
178
0
18,690
Intel should invest in Imagination technologies so that they have royalty money back.
License GPU and CPU cores and fight back ARM.
They always can deliver more truth manufacturing technology. Why not make money?
Intel mised smartphone & tablet market. After five years of loses they will only become relatively competitive in tablet market with new gen of atoms bat not thanks to architecture bat huge engineering effort & manufacturing process. This won't help them in smartphone market and they will again totally miss wearables. To put it simple as it gets a x286 (16 bit) its bigger than cortex M4 (With DSP, full 32 bit). The x86 cores will never be competitive to ARM or ever RISC. If some smart ass wants to remained me that M series are for microcontrollers I would just remind them that x286 is not good for anything this days and its still bigger.

I don't care for Intel, I want to see fight.
Fight will bring progress & will be beneficial to users.
 

Jaroslav Jandek

Honorable
Jan 13, 2014
103
0
10,680
on straight out performance its currently equal to high end smartphone socs, but battery life is miserable compared to arm socs.
They have Atom Z3770/3740 now - way more perf./watt than any ARM SoC. The GPU is weaker than the ones in high-end ARM SoCs, though.
Still, you can even play games like Starcraft 2 or Xenonauts on them at 30+ fps, office work is (of course) no problem and even working in Visual Studio is fine (if you plug a display in) - that's what I use my ASUS T100 for when I'm away from my PC.
 

amk-aka-Phantom

Distinguished
Mar 10, 2011
3,004
0
20,860
Some comments here are a LOT of BS. That is forgivable since many of Tom's readers don't seem to know much about mobile SoCs. Let's clarify some things.


1) Intel's new SoCs kick a lot of ass (just read some Notebookcheck reviews of devices with them) and are used in many new tablets and phones, especially Lenovo and Asus. Intel doesn't need ARM, it just needs to improve x86 more like it's already doing and ARM will squeal. A few years ago people were saying that Intel won't be able to release chips that can compete with ARM on power efficiency and performance - right in your face with Bay Trail. What people don't get is that besides having tons of money, Intel has some of the brightest minds in the world working for them. And beating ARM is just a challenge for them, one that I'm confident they'll win. Just wait for Broadwell-Y and Cherry Trail. At some point current ARM makers will go bankrupt unless they're smart like nVIDIA with its K1 (especially Denver version).

BTW:

The IGPs in Intel's current SoCs are still horrible jokes though and Intel will not have anything worth writing about in that department until late-2015.

Nonsense. Their current Atoms use scaled-down Ivy Bridge GPU that performs a bit behind Adreno 320 and their new Moorefield Atoms will use PowerVR G6400 (dual core models) and G6430 (quad core models). That's the same one as in Apple A7 chip in iPhone 5S. Moorefield is coming out in Q4 this year.

They (Intel's management) better not hear of this. They'll probably engineer buying out ARM itself.

And if ARM poses a real threat, they will. People tend to forget this too.

x86/64 will never be able to compete in the mobile space. The architecture has become so large and complicated that it can be made to run at the efficiency of ARMs architecture. Most of the power in big x86/64 systems comes from the fact that compilers and code have been optimized to take as much advantage as possible. The problem is the rules aren't the same for the mobile space and the same techniques used to give killer performance at the high end (massively superscalar, complex cache systems, etc) hurt performance for smaller devices. Atom is to ARM as AMD is to Intel, the performance is subpar but people use it because they can get it cheap. Intel can get there but their attempts seem half-hearted at best.

... right. So you are saying that code and compiler optimization is bad. Are you an Android developer by chance? :D But seriously, just look at what you wrote here. How is ARM different from x86 in this regard? The most powerful ARM chips are those with most cores, highest clock and biggest caches. And THAT is how x86 gets most of its power, by having much more headroom to operate in, although of course optimization counts. In fact, the ONLY way to make any architecture powerful is optimizing applications for it, and that's what ARM has to count on the most - and so does x86. Intel's contributions to Linux kernel aren't just for fun, you know, they want their Atoms to work under Android properly. And in case you haven't noticed, unlike AMD, Intel doesn't simply try to stuff "more of everything" into their chips anymore. They've realized that modern x86 processors are more than enough for most applications and are now heavily focused on power efficiency, integrated graphics and heat management. To say that Intel's attempts are "half-hearted at best" is blind. They simply chose to first secure (read: crush the competition) the market sector they make most profit from (laptop and desktop) and only then dive into something they haven't touched before (mobile). I'd say that's a sound strategy. You don't leave the house without making sure you lock the doors first.

Intel has been trying to respond to ARM since it brought out its first Atom chips in 2008, back in the Netbook era. Remember Intel trumpeting its “Full Internet Experience” slogan? They said they were two years ahead of ARM at the time. Now they’re no longer quite as complacent.

First Atoms failed because of poor quality of the boards they came on (had to throw away about twenty of them at work in the last two years) in case of desktops and bad form-factor in case of laptops, combined with rapid increase in hardware requirements that Windows Vista and 7 brought (primarily RAM, those damn chips always came in devices only with 1-2 GB of RAM). Windows XP worked fine on my Asus Eee PCs, but I got rid of them both because they were too small and eventually, too slow. Rome wasn't built in a day, first Atoms were a good start and Intel has learned well. They are STILL ahead of ARM, in fact even more than ever before, because there are now actual mobile devices that use Intel chips, a lot of devices. MS got tired of Windows RT and asked Intel to make chips to run x86 Windows 8 on tablets, Intel delivered :D Deal with it!

2) Mediatek is a steaming pile of garbage. Their chips might cost less than what customers perceive as Qualcomm equivalents, but perform worse, heat up more, consume more energy and have worse graphics. Moreover, they're buggy, search for "Mediatek SMS attack". Just messaging an = sign to someone with Mediatek based phone will cause it to reboot. MOREOVER, that <expletive> company refuses to release source code necessary for 3rd party ROMs like Cyanogenmod to be made for devices with their chips. It's despicable, because most stock ROMs - even pure Android, as I found out after buying a Moto G (can't even change Quick Tiles order and contents, what the heck) - are garbage and most companies abandon their devices after 1-2 years. Samsung refused to release KitKat for S3 and S3 Mini (except LTE version) because TouchWiz UI needs more than 1 GB RAM to run on KitKat. FYI: pure KitKat needs 512 MB. I'm running it on my Galaxy S1 (thanks, Cyanogenmod team) and it's smooth and stable. The only hangups I get are because of the CPU, but that's to be expected. On my GT-P6200 tablet with a stronger dual-core CPU that already doesn't happen.

TL;DR - Intel laughs at your underestimation of its potential and misinformed opinions; compared to Qualcomm, Mediatek chips run hotter, consume more power, are buggy and they won't release source code.
 

Marlin Schwanke

Honorable
Apr 23, 2013
48
0
10,540
Last I hear, Intel has an architecture license from ARM (just like Apple) and they can design build ARM chips of their own, whenever they want. Also, IIRC, Intel already has an investment in Imagination Technologies one of the better mobile graphics firms so I think they could get their hands on that tech too. I think it is just VERY important to Intel to move forward with the x86/x64 architecture rather than ARM in order to bolster/protect their desktop/laptop/server processor business.
 
Status
Not open for further replies.