The ARM v.s. Intel Thing - Let's Discuss It

[Nemesis 1]

http://www.anandtech.com/show/6433/intel-ssd-dc-s3700-200gb-review

Not quite. Intel just came back with a very strong product, entirely of their making. This could be just like the next gen atom and with process advantage.

Don't feed the troll . As bad as I can be . We have a new king troll. Go read all his post . HE lies alot . Proven
He uses outdated material as you just showed . He is Anti -intel which is fine . Its how he delivers the message that is a shameful act . Many many people that frequint this forum have a real stake in intels success . If intel wins they loose . Amd has SSD That they market . Intel actually makes the best controller and the memory itself that others buy. I going to find out who he is.

 

[Nemesis 1]

I would be just as slow to write off Samsung and Qualcomm as I would Intel. Samsung has the funds. Market Cap 199B. That is bigger than Intel. Twice the size actually. And Qualcomm IS the same size as Intel at 103B compared to Intels 100B.

I'm not saying that the money a company has directly correlates to how much R&D money goes to SoC work, but at the same time, the money available to Intel's competitors can't be ignored either.
ARM is, and will be, a formidable competitor in the mobile and SoC market. And this is a HUGE market constantly growing. Soon it will dwarf desktops if it hasn't already. I don't have those numbers so I can't make any good guesses there.

Anyway, Intel is a force to reckon with, but that doesn't mean Qualcomm or Samsung isn't.
If AMD, with it's pathetic R&D budget (pathetic in comparison to Intel) can put up somewhat of a fight against Intel (pre-conroe), well, nuff said.
Also look forward to seeing what Nvidia does as well.

This was one of your best ever post . Thoughtful and truthful as far as we know of these things . I really do see qualcomm buying AMD . The risk for apple is to high.

 

[Nemesis 1]

Eye opener? LOL

Did you not see the 40nm part being compared to a 32nm chip? One approaching its end of life in favour of 28nm Tegra 4? How about the fact Atom has barely been included in a single product its THAT good. What about the companion core which is turned off because its not supported by windows which normally gives Tegra its edge in power saving? Oh thats right we are comparing windows RT. A product that basically is a flop and stands little to no chance of getting anywhere just like WinPhone 8.

Samsung is testing 14nm Process nodes and im betting that they will want to catch intel up on that front eventually.

Intel claimed they were going to be a top GPU company and they failed. They also claimed they were going to be a top SSD company and they also failed at that too.

In Fact the only thing they have ever managed to beat is AMD. Now they have to face ARM which is a company with no physical production who have partners such as Samsung, Apple, Qualcomm, Nvidia, & many more. All have significant resources and cash on hand. ARM offer the chance to make bespoke SOC's What can Intel do? They made a special ULV chip for Apple once. Thats it.

Oh and Intel wont like working on the lower margins either and that will limit them always

Man I don't know how long your going to get away with this . I suspect not much longer . So Samsung did a test chip over 2 years to late .

http://www.xbitlabs.com/news/cpu/di...l_Runs_14nm_Test_Chips_in_the_Lab_Report.html

 

[Nemesis 1]

That is a pricing issue really. ARM processors are dirt cheap. Atom, not so much. Couple that with the fact that intel was VERY late into the mobile game and you can see why there aren't many Atom processors on the market.

The orginal Atom chip cost intel $6 to produce that was at 45nm . Intels margine include the cost building and operating fabs . Your world view needs expansion .

 

[simboss]

Samsung S3 Mini for example is 45nm. Its essentially only the top end phones that are 28/32nm.

Plus non smartphones still outsell smartphones with a huge margin.
I think for Samsung the ratio was 360/60 million for 2012 with a total of 420 shipped..

Obviously no one outside Samsung has the sales figure for Galaxy S3 and S3 mini, but I would not bet on the mini getting any significant sales compared to its older brother.

And as Apple has very well proven, it might be much better to sale reasonably high volumes of high end than huge volumes of low end. This applies to the SoC themselves as well.

The trick today is to find the sweet spot between performance/power/cost and probably IMHO more importantly time to market.
That is exactly what Apple and QC have done in the past few years, whereas Intel or TI have struggled (Intel being too high on power, TI too low on performance, and both equally late).
Interestingly both Intel and TI have also been pretty good on many other markets outside mobile, probably because time to market is less critical, whereas performance or cost can be decisive.

 

[krumme]

Samsung S3 Mini for example is 45nm. Its essentially only the top end phones that are 28/32nm.

Plus non smartphones still outsell smartphones with a huge margin.
I think for Samsung the ratio was 360/60 million for 2012 with a total of 420 shipped..

TSMC 28nm capacity was 5% for Q2 2012 - and at Q3 they predicted 20% for the years end.
http://www.taipeitimes.com/News/biz/archives/2012/08/11/2003539958

Samsung is probably one or two quarters back, so a guess 10% 32nm hkmg capacity by years end?

I have no clue why Qualcomm have the same market cap as Intel, but there must be a market we dont look at, at all?

What if 80% of all sold phones (inckluding non-smartphones) is even running on 65 and 90nm? and the 40/45nm takes the 80% of the rest of the 20%? - probably next to 95% of the non smartphones is running on 90nm fx. som low clock variant of Arm v11, like used in the Iphone 2 or nokia e72.

I had a hard time understanding NV tegra 3 when it surfaced on its old 40nm, but now, one year later i have to say, it was an excellent move by NV. (edit: in regards to time to market by simboss, tegra 3 was volume on time, with the great label as quad core)

 

[Nemesis 1]

i love how you cram your own version of facts alone with your fortune telling.

Apple is just as likely to completely ditch intel as they are to go with them. Last time i checked though iOS didnt run on x86. Also Apples ARM SOCs are pretty fast currently.

Intel has about 12-18 months ahead on process node and intel uses its best nodes for its desktop and laptop parts. Thats why Atom is at 32nm.

Intel will offer a all in one SOC maybe but they will have to sell it at $20 or it wont sell because thats the market price. Thats a long way from $200 retail prices of its desktop chips.

Right now the only partner intel has in this space is Motorola in 1 phone. Thats a long way to go.

Also i dont hate intel, i just dont like some of the stuff they do. Such as cheap out on the TIM on their $300 CPU's

Just stop , I am thinking of a lawsuite against you is needed. You can only go so far even on the internet. YOUR the biggest liar this forum has ever seen . Go ahead press the triangle . You are a problem for this forum . YA i going to lose points for this . But someone has to say it publicly . The amount of lies you have spread on this forum is an outrage . You can come up with up todate info on everthing except intel where you only give old information outdated . That only happens if its done purposely. Because of this post I likely gone befor you . But your time here is about over.

 

[ShintaiDK]

Be careful to use sales contribution vs actual output.

A large part (around 25%) of TSMCs output today is still 150nm+ for example.

Not to mention 8" wafers:

 

[Exophase]

I think I once heard that interpreters are almost entirely bandwidth limited, so the high memory bandwidth, small instruction size, and fast caches of Atom may be what contribute to their advantage in javascript.

None of these JS engines are interpreters, they're JITs. If they generate a lot of code per instruction they might still end up with a much larger

Memory bandwidth is about the same in many cases, and we don't see Tegra 3 at a huge disadvantage vs other Cortex-A9 SoCs so I doubt it's bandwidth limited. x86 doesn't really have a code density advantage vs Thumb-2, but I don't know if these JITs are targeting this instead of just normal ARM.

There's a lot of weird stuff with JITing a language like Javascript that makes it different from a lot of other conventional code. Since it's dynamically typed variable accesses have to be type checked at runtime unless analysis can determine they'll always be some type. JS engines vary on what happens when the type checking shows that the currently compiled code is in conflict, but they all will tend to modify or generate new code. This is a weakness for ARM vs x86 because the icache isn't coherent with the dcache; this wouldn't be a huge problem if you didn't have to be in system mode to perform cache flushes and invalidations, which requires expensive calls to the kernel. But I have no idea how much this actually plays a role in current JS code.

 

[krumme]

Be careful to use sales contribution vs actual output.[/img]

Thanx noted.

 

[Nemesis 1]

Read this post (bottom) and let us know if you feel the same.

http://forums.anandtech.com/showpost.php?p=34053183&postcount=570

Exactly dave . I took a differant approach . I delidded than I had the section of the lid machined out were the die is than we removed more metal from the top of the lid so it was 1/2 as thick and removed metal from the base of the lid to get a good seat for the waterblock to set on the die with tim . The only reason I used ther lid at all was to keep the block from rocking on the die . Its nothing more than a foundation . Unlike IDC my results don't matter , I use a silver based block and have my own circulation system in place using bad koolance parts and modding them to perfection . Running threw a 2loop single rad AQUACOMPUTER 420. So telling my results is useless to anyone else but me and a few rich ball players. I did give my grandson a complete Nemesis case and watercooling setup for xmas He won't have it done for awhile . waiting for haswell . This 5 year old has to do all the assembly under suppervision . His mother is a video making fool along with grandma . Oh the path these 2 have lead me down . But I sure they will do a video on it . Even tho I asked them not to . The women are out of my control lol .

 

[Nemesis 1]

And discrete gpus are not going faster with the next process shrink?
nVidia's slowest 28nm mobile card is 30-50% faster than HD4000. Haswell will only catch up with the GT640m...

SO what . Intel Igpu needs only be as good as the next consols , Nobody is going to make games for a shrinking discrete GPU market . The smaller that market becomes the higher priced the GPU will become as they lose market share to IGPUs .

 

[Nemesis 1]

Khato:

I'm in agreement with you there. It's probably Z2580 based.

We can't derive much from the Nenamark 2 scores, because it seems its refresh rate capped at 60Hz. It may well be the SGX544MP2, because that's what the Z2580 is.

What about 4g AT article says 2g

 

[Nemesis 1]

Z2580 is 32nm. 22nm comes with Merrifield. We won't be seeing "Silvermont" in a phone/tablet until Q1 2014.

"Avoton" SoC for micro-servers, however, is a 2013 part. That's where we first see that new core debut.

GO back to jan 2012 read AT article on medfield see what he says when to expect merrifield. I may be wrong but isn't the A2580 in the tablets. I to lazy to look . Again I seen no roadmaps that intel normally releases none.

 

[Nemesis 1]

Thing is, Medfield and Clover Trail are both 32-bit only, and it looks like Bay Trail-T will be too, meaning the smartphone Silvermont will probably follow suit. Since they're OoO they might be better suited to it, but programs can still be penalized if they're compiled for the current Atoms instead..

Read silvermont article referring to Atom server chip its 64bit . Thats the server chip . Whats on the other silvermont chips have to wait and see.

 

[Homeles]

Very true, an excellent demonstration of microarchitecture triumphing over process technology and thriving despite the inherent weaknesses.

Dothan accomplished the same on Intel's fabled 90nm node, making do with what Prescott could not.

It is too bad that AMD didn't find a magic combination for GloFo's 32nm. Maybe they will with its 28nm?

GloFo's 32nm was a total stinker. Their track record hasn't been exactly inspiring, so I'm not getting my hopes up for 28nm. I'd imagine that bulk will be easier to work with, although I'm not well informed in this area.

I'm not sure what to think of GlobalFoundries. They say they're doing better with 28nm, and given the circumstances and what I know about that node, I do believe them. But they really need to ensure that AMD does not run into any or minimal headaches with them to ensure that they'll have AMD around as a partner. And I'm worried about how 28nm stacks up against 32nm from a performance and power perspective -- area savings be damned.

You know, after reading about how Intel has always used gate last... I'm beginning to wonder if that's the reason behind why they're so good at avoiding fab disasers. TSMC certainly did much better this time around, and we'll likely see the Common Platform Alliance pull off a decent 20/22nm node as well, when they make the switch.

 

[Exophase]

Read silvermont article referring to Atom server chip its 64bit . Thats the server chip . Whats on the other silvermont chips have to wait and see.

I said BayTrail-T. Tablet chips. Not server chips. Intel has had 64-bit Atoms for years, of course they aren't going to suddenly not have them in the segment that needs them most. Of course the core design of Silvermont is going to be 64-bit. That isn't the issue, it's about whether or not they fuse off 64-bit support for tablet and phone SoCs. If that slide is correct that is what they will be doing (barring the bizarre scenario of having 32-bit only tablet chips alongside 64-bit phone chips).

Even in early 2014 this shouldn't be a huge issue for RAM use, but lacking x86-64 also puts it at a performance disadvantage. I can only imagine that Intel is doing this to save on TDP.

 

[Nemesis 1]

IF you can show me a bay trail roadmap with intel written on it . That would be a good start . Intel not going to change formats for graph they put out . I have never seen a oad map from intel without intel written on it . On other than roadmaps yes I have seen nameless graphs but never a roadmap graph . Well if intel can turn on 64 bit . I fail to see the point . Its is alot of words about nothing is it not .

 

[Arachnotronic]

IF you can show me a bay trail roadmap with intel written on it . That would be a good start . Intel not going to change formats for graph they put out . I have never seen a oad map from intel without intel written on it . On other than roadmaps yes I have seen nameless graphs but never a roadmap graph . Well if intel can turn on 64 bit . I fail to see the point . Its is alot of words about nothing is it not .

See the investor day materials at intc.com

 

[Exophase]

IF you can show me a bay trail roadmap with intel written on it . That would be a good start . Intel not going to change formats for graph they put out . I have never seen a oad map from intel without intel written on it . On other than roadmaps yes I have seen nameless graphs but never a roadmap graph . Well if intel can turn on 64 bit . I fail to see the point . Its is alot of words about nothing is it not .

http://images.dailytech.com/nimage/Intel_Atom_ValleyView_Wattage.png Intel is written in the corner. Is that good enough for you?

Although it doesn't outright say that there won't be x86-64 support on the tablet variant this is strongly implied by the indication that it only supports the 32-bit version of Windows.

Other than that I don't really know what you're saying.

 

[Nemesis 1]

http://images.dailytech.com/nimage/Intel_Atom_ValleyView_Wattage.png Intel is written in the corner. Is that good enough for you?

Although it doesn't outright say that there won't be x86-64 support on the tablet variant this is strongly implied by the indication that it only supports the 32-bit version of Windows.

Other than that I don't really know what you're saying.

Yes it is. Thanks. I would assume the same as you did .So I take it that AMD included 64 bit in temesh. Will arm have 64bit at this time . I noted that only The Soix had 32. So in phones it won't matter . I don't believe temesh will make it into phones on the 1st chip . So I really don't see why it matters at all. On the tablet thing it clearly states it supports 32-64 just as IB does.
2 core phones the next one is mobile the next is server

 

[sontin]

Anandtech updated the power consumption test with a Krait Win8 RT tablet and the Nexus 10:
http://www.anandtech.com/show/6536/arm-vs-x86-the-real-showdown

No suprise, Krait has a much better perf/watt ratio than Atom.

Overall Atom has a 40% better perf/watt ratio than Tegra 3. A shrink to 28nm would be enough to beat Clover Trail in that aspect.

 

[Arachnotronic]

Anandtech updated the power consumption test with a Krait Win8 RT tablet and the Nexus 10:
http://www.anandtech.com/show/6536/arm-vs-x86-the-real-showdown

No suprise, Krait has a much better perf/watt ratio than Atom.

Overall Atom has a 40% better perf/watt ratio than Tegra 3. A shrink to 28nm would be enough to beat Clover Trail in that aspect.

How do you get that "Krait has a much better perf/watt ratio than Atom"?

Krait consumes less power, but Atom performs better. To do "performance per watt", you need to factor in the "performance" part.

 

[wlee15]

I said BayTrail-T. Tablet chips. Not server chips. Intel has had 64-bit Atoms for years, of course they aren't going to suddenly not have them in the segment that needs them most. Of course the core design of Silvermont is going to be 64-bit. That isn't the issue, it's about whether or not they fuse off 64-bit support for tablet and phone SoCs. If that slide is correct that is what they will be doing (barring the bizarre scenario of having 32-bit only tablet chips alongside 64-bit phone chips).

Even in early 2014 this shouldn't be a huge issue for RAM use, but lacking x86-64 also puts it at a performance disadvantage. I can only imagine that Intel is doing this to save on TDP.

It's because only the 32-bit version of Windows 8 supports Connected Standby. Since you're not allowed to simultaneously support ACPI S3 and Connected Standby Bay Trail-T would be in major disadvantage in 64-bit mode.

 

[Nemesis 1]

Anandtech updated the power consumption test with a Krait Win8 RT tablet and the Nexus 10:
http://www.anandtech.com/show/6536/arm-vs-x86-the-real-showdown

No suprise, Krait has a much better perf/watt ratio than Atom.

Overall Atom has a 40% better perf/watt ratio than Tegra 3. A shrink to 28nm would be enough to beat Clover Trail in that aspect.

I am out of this one . not because It wasn't great . But Anand conclusion section about floored me! I been aced we all been aced . When he talks about haswell at 4 watts . I mean jesus I not the only one gona read it . I mean shit man . I just don't no what to say we been trolled