The ARM inherent efficiency myth

[beginner99]

2) Create Windows Professional OS aimed at x86-based convertibles, laptops and desktops, with full desktop mode, and with MS Office Home & Student Edition pre-installed in order to add productivity for more premium and higher cost end user devices.

MS would probably be sued if the did that.

But I agree. Windows RT and 8 are stupid. They should have made metro ui for RT only and no desktop. In contrast no metro on normal Win 8.

 

[Homeles]

MS would probably be sued if the did that.

But I agree. Windows RT and 8 are stupid. They should have made metro ui for RT only and no desktop. In contrast no metro on normal Win 8.

Oh no, there's additional functionality on the desktop version! It's the end of the world!

 

[ams23]

Oh no, there's additional functionality on the desktop version! It's the end of the world!

The additional functionality is fine, but how does a casual observer tell the difference between Windows RT and Windows 8 when looking at the metro/modern UI interface? Microsoft probably hopes that most people in the future get all their software from the Windows App Store (where the software is compatible with ARM and x86 processors) so that these differences are trivialized.

 

[Homeles]

I'd imagine price is a pretty big giveaway. I did state that I felt that Windows 8 is unpolished. It'll be interesting to see where Microsoft goes from here.

 

[ElFenix]

The additional functionality is fine, but how does a casual observer tell the difference between Windows RT and Windows 8 when looking at the metro/modern UI interface? Microsoft probably hopes that most people in the future get all their software from the Windows App Store (where the software is compatible with ARM and x86 processors) so that these differences are trivialized.

MS doesn't want you to tell the difference because it is attempting to force its way into the tablet market (and phone, for that matter) by banking on its dominance of pc OSes. If you're used to metro on the desktop then metro on the tablet is seamless (and familiar). Combine that with cloud services so that your appz docz gamez and filez are there with you where ever, and you have a pretty easy sales pitch once people are used to metro. And we will be, because it isn't going anywhere.

 

[ams23]

I'd imagine price is a pretty big giveaway. I did state that I felt that Windows 8 is unpolished. It'll be interesting to see where Microsoft goes from here.

In some cases that is true, but IIRC, the Clovertrail x86 Windows 8 tablets were just as cheap if not cheaper than the ARM Windows RT tablets (although the Windows 8 tablets did not include MS Office Home & Student Edition while the Windows RT tablets did). And with rumors that Microsoft has substantially cut it's OEM pricing for Windows 8, then the expectation is that we will see very low priced Windows 8 tablets this year.

On a side note, one wonders if Microsoft would have been better off doing what Google did with their Nexus lineup by working with various hardware vendors (such as Lenovo, HP, Dell, etc.) to create different tablets and cell phones that are sold and marketed by Microsoft, rather than doing it all alone with Surface.

 

[Homeles]

In some cases that is true, but IIRC, the Clovertrail x86 Windows 8 tablets were just as cheap if not cheaper than the ARM Windows RT tablets (although the Windows 8 tablets did not include MS Office Home & Student Edition while the Windows RT tablets did). And with rumors that Microsoft has substantially cut it's OEM pricing for Windows 8, then the expectation is that we will see very low priced Windows 8 tablets this year.

On a side note, one wonders if Microsoft would have been better off doing what Google did with their Nexus lineup by working with various hardware vendors (such as Lenovo, HP, Dell, etc.) to create different tablets and cell phones that are sold and marketed by Microsoft, rather than doing it all alone with Surface.

Microsoft's foray into the tablet market certainly hasn't been a graceful one, but I don't believe that it is the flop that people are making it out to be. My perception is that the majority of people are making it into some sort of "it's either a grand slam or it's the Titantic" dichotomy.

 

[ams23]

Microsoft's foray into the tablet market certainly hasn't been a graceful one, but I don't believe that it is the flop that people are making it out to be. My perception is that the majority of people are making it into some sort of "it's either a grand slam or it's the Titantic" dichotomy.

The Surface is a nice product no doubt (and even the lower spec Surface RT has probably sold over 1 million units by now). I just think that Google's approach is bit more friendly to other hardware vendors that rely on the Android OS, and if Microsoft used a similar approach with hardware vendors that rely on Windows OS, there would be less negativity from the likes of Acer and other companies that have a long history working with Microsoft. I do realize that Microsoft has more control and higher margins on their product by doing the Surface on their own, but it is actually in Microsoft's best interests to see their long-time partners thrive too, and there is some give and take when directly competing against your partners. And if the Windows App Store and Windows ecosystem thrives, then Microsoft will make a lot of money on the application software too.

 

[podspi]

On a side note, one wonders if Microsoft would have been better off doing what Google did with their Nexus lineup by working with various hardware vendors (such as Lenovo, HP, Dell, etc.) to create different tablets and cell phones that are sold and marketed by Microsoft, rather than doing it all alone with Surface.

It absolutely blows my mind they did not do this. This is one idea from Google that they should shamelessly steal ASAP. Best of both worlds: You get (some) design control, and you don't piss off all your partners.

Funny, it seems like Google itself is moving in the opposite direction with Motorola. You know what they say, the dollars are always greener...

 

[Homeles]

The Surface is a nice product no doubt (and even the lower spec Surface RT has probably sold over 1 million units by now). I just think that Google's approach is bit more friendly to other hardware vendors that rely on the Android OS, and if Microsoft used a similar approach with hardware vendors that rely on Windows OS, there would be less negativity from the likes of Acer and other companies that have a long history working with Microsoft. I do realize that Microsoft has more control and higher margins on their product by doing the Surface on their own, but it is actually in Microsoft's best interests to see their long-time partners thrive too, and there is some give and take when directly competing against your partners. And if the Windows App Store and Windows ecosystem thrives, then Microsoft will make a lot of money on the application software too.

Yeah, Microsoft could really use some work when it comes to OEM relations. Apple doesn't suffer from this problem, and the disadvantage is only going to grow as Apple becomes increasingly more vertically integrated.

I do see Android fragmentation as being a big problem though. I definitely have not been pleased with being stuck on Gingerbread. Custom ROMs are great and all, but only when they exist for your device. I'll still be choosing an Android smart phone when I go to upgrade next month to a Galaxy S IV, and hopefully the phone fares better in terms of software support than my Droid 2 did.

They all have their quirks though. Microsoft is definitely the worst, but I hope they see the light.

 

[Olikan]

Switching to a more frequent release cadence puts them in trouble? I do hope that you're on drugs.

Windows development has always been so slow and predictable... becoming more nimble is exactly what they need.

Well, i thouhgt that the abismal W8 sales, made M$ rush widonws blue...
didn't thought that it was an strategy....

 

[Homeles]

Well, i thouhgt that the abismal W8 sales, made M$ rush widonws blue...
didn't thought that it was an strategy....

They're shifting towards a rapid development cycle, more akin to OSX, iOS and Android. OSX, for instance, gets new releases every year. Windows on the other hand probably averages around a new OS version every 3 years.

The gap from XP to Vista was nearly 6 years long, and look how wonderful the release of Vista came out to be. Look at AMD Phenom and Bulldozer. Long development cycles that bring a large number of changes leave room for more disaster. Smaller, evolutionary changes leave less room for disaster. There's more flexibility.

Intel's done well here, and so has AMD ever since they've switched. Apple's done well with it. Android's done well, if we ignore the unrelated problem of fragmentation.

Smaller, more frequent changes are the way to go.

 

[poofyhairguy]

Well, i thouhgt that the abismal W8 sales, made M$ rush widonws blue...
didn't thought that it was an strategy....

They learned from Vista.

First you have to put out the release with the new stuff that people will hate. Then you re-release it with the same stuff but some cosmetic differences and everyone will be so happy its not the hated OS so they don't notice by then that you didn't really undo the thing that pissed everyone off.

Since Windows 98 every other consumer release had been a flop.

 

[inf64]

I don't see it posted yet(maybe deserves its own thread?), but AT's Johan tested the Calxeda's ARM server:
http://www.anandtech.com/show/6757/calxedas-arm-server-tested

ARM based servers hold the promise of extremely low power and excellent performance per Watt ratios. It's theoretically possible to place an incredible number of servers into a single rack; there are already implementations with as many as 1000 ARM servers in one rack (48 server nodes in a 2U chassis). What's more, all of those nodes consume less than 5KW combined (or around 5W per quad-core ARM node). But whenever a new technology is hyped, it's important to remain objective. The media loves to rave about new trends and people like reading about "some new thing"; however, at the end of the day the system administrator has to keep his IT services working and convince his boss to invest in new technologies.

Pretty amazing perf./watt in targeted workloads. Now we see why AMD wants 64bit ARM based opterons that use "freedom fabric" in 2014. The perf./watt proposition is great.

 

[R0H1T]

I don't see it posted yet(maybe deserves its own thread?), but AT's Johan tested the Calxeda's ARM server:
http://www.anandtech.com/show/6757/calxedas-arm-server-tested


Pretty amazing perf./watt in targeted workloads. Now we see why AMD wants 64bit ARM based opterons that use "freedom fabric" in 2014. The perf./watt proposition is great.

If only people would take their blinders off & not take ARM/AMD as also ran especially considering the monolithic giant that Intel is alongwith the enormous resources at their disposal !

 

[Cerb]

They learned from Vista.

First you have to put out the release with the new stuff that people will hate. Then you re-release it with the same stuff but some cosmetic differences and everyone will be so happy its not the hated OS so they don't notice by then that you didn't really undo the thing that pissed everyone off.

Except that so far, they have fixed what was really wrong. How many UAC prompts do you need to install updates in Windows 7? How about Vista? Exactly--it's way less annoying to use, and will actually stay up to date. OTOH, they still haven't learned to get SPs to install (I have yet to see a tech-illiterate's computer with a newer service pack than it came with).

XP could go years w/o instability, unlike 95, 95b, 98, 98SE, or ME. 7, FI, doesn't require 2-3 UAC prompts just to get Windows updates (despite all the apologists, I've yet to see a Vista install with that crap gone).

MS hasn't really learned anything, despite reacting, and are acting like a large corporation with enough inertia that they don't feel they should have to care about the user experience, because people need Windows.

They don't have flops for any strategic purpose. The managers that get to make them just happen to be the guys with big ideas, big egos, and plenty of blame-passing skills. But, that only works well for products when they are also visionaries, which most corporate cultures try to weed out, rather than nurture. They are big enough that they don't see short-term risk, and thus can't make quality efforts towards long-term goals that actually make for truly better products.

 

[krumme]

I don't see it posted yet(maybe deserves its own thread?), but AT's Johan tested the Calxeda's ARM server:
http://www.anandtech.com/show/6757/calxedas-arm-server-tested


Pretty amazing perf./watt in targeted workloads. Now we see why AMD wants 64bit ARM based opterons that use "freedom fabric" in 2014. The perf./watt proposition is great.

Why not use jaguar? - its only twice size per core sans l2 compared to A15

 

[zebrax2]

Why not use jaguar? - its only twice size per core sans l2 compared to A15

ARM is probably AMDs backup plan in case ARM gains traction on markets other than mobile. If ARM turns out to be the next big thing at least they would be there and maybe have a few IPs under their belt

 

[Vesku]

Atom actually stacks up reasonably to Calxeda's quad core, if you ignore the chipset. If anything we can expect x86 and ARM SoCs to be very competitive with each other in the 5-10W range.

 

[beginner99]

I don't see it posted yet(maybe deserves its own thread?), but AT's Johan tested the Calxeda's ARM server:
http://www.anandtech.com/show/6757/calxedas-arm-server-tested


Pretty amazing perf./watt in targeted workloads. Now we see why AMD wants 64bit ARM based opterons that use "freedom fabric" in 2014. The perf./watt proposition is great.

As posted in the comments, the choosen scenario is ideal-case for such a server. In most cases you won't find 24 physicals servers with almost identical CPU needs. However since they are physical you will end up with some going almost unused and others not having enough CPU power. This can be much better dealt with in a virtualized environment. You can use it for 1 "high load application" and several mostly idle ones. This is not possible using physical servers.

 

[Haserath]

I don't see it posted yet(maybe deserves its own thread?), but AT's Johan tested the Calxeda's ARM server:
http://www.anandtech.com/show/6757/calxedas-arm-server-tested


Pretty amazing perf./watt in targeted workloads. Now we see why AMD wants 64bit ARM based opterons that use "freedom fabric" in 2014. The perf./watt proposition is great.

I was going to say that it wasn't so amazing, then I saw the ARM cores are 40nm.

But the way they allocated the cores would vary performance much more than the Xeons.