Comment by mtgx | Hacker News Reader

Comment by mtgx | Hacker News Reader

Parent Full thread

Not quite. They've been working together for a few years to do this. Qualcomm even based its motherboard design off of Microsoft's Project Olympus.

> QDT has been working with Microsoft for several years on ARM-based server enablement and has onsite engineering at Microsoft to collaboratively optimize a version of Windows Server, for Microsoft’s internal use in its data centers, on Qualcomm Centriq 2400-based systems. QDT’s OCP submission is the result of a multi-faceted, multi-level engagement between the two companies to facilitate ARM adoption into the data center. It encompasses multiple areas of hardware and software including board development, firmware, operating system, compilers and tools, and CoreCLR.

https://www.qualcomm.com/news/releases/2017/03/08/qualcomm-c...

Also, I doubt Microsoft went through the trouble of supporting full x86 programs on Qualcomm's ARM chips just to improve the relationship with Intel. If anything, finally bringing ARM to not just servers, but Windows notebooks is what will make Intel even angrier with Microsoft.

So far, the main reason why ARM couldn't enter the PC market was the Windows lock-in - if x86 programs wouldn't work on ARM, then they had no chance (save for UWP apps gaining huge popularity, which hasn't happened yet, and I don't expect to happen for at least a few more years, if ever).

Why doesn't apple move macbooks to ARM processors?

I could see that being a possibility in the future as the discussion moves away from firepower and more toward power consumption, which seems to be a focus of Apple products.

They introduced ARM to manage their fingerprint sensor and the TouchBar. Rumors had it that Apple is working on connecting the ARM chip to more subsystems so it can do background work like background refresh (or what that feature is called when the laptop is in sleep – PowerNap?). I see this as Apple slowly creating a hybrid (low energy ARM for easy tasks and faster x86 for speed). Maybe the goal is a complete transition to ARM and Apple starts to design their own MacBook (Pro) CPUs to gain more control (be less dependent on Intel’s varying release cycles).

I've been waiting for something like ever since I first got a Raspberry Pi.

Keep the x86 CPU around for active, high powered use, and then use a low power ARM CPU to run downloads and software updates while the PC is in an off or idle state.

What about number crunching on the GPU?

They'll probably take a hybrid approach before going all-in. Use the ARM CPU for low-power, now-CPU requirements to avoid draining the battery. The ARM chips in the iPhone and iPad are very good, but they're not quad-core i7 good.

They're far too wimpy today.

Which chips are you refering to, specifically? Because otherwise it's a blanket statement that is simply untrue.

> Which chips are you refering to, specifically?

Literally any ARM chip. What ARM chip do you think makes a fitting replacement for a high performance Intel core in a Macbook?

We're talking about servers here, which has nothing to do with the OS. If ARM chips were appropriate, we would see them in Linux, but time and time again, the power to performance just doesn't make sense for ARM, when compared to an Intel chip, for reasonable workloads.

Time and time? We do see them in Linux.

The ARM server ecosystem has been in the works for a few years. TechCon'16 we saw the announcement that the software is fully there: and this is not hype, all major distributions support all the major players out of the box. Download ISO, install, and use it.

HW-wise we are now in the 3rd gen of chips (QC Centriq, XGene3, ThunderX2) all of which compare with E5, not Atoms. First/Second gen were mostly proving grounds for the specs and software work, excepting for Cavium machines, which have some serious network I/O where the relative wimpyness of any one of the 96 cores simply doesn't matter for certain workloads. In a matter of two years we've seen private datacenters soak up XGene2 and Cavium chips. We've seen ODMs and OEMs get their shit together and start selling real kit (b2b.gigabyte.com).

It's funny, people expect these things to suddenly appear overnight everywhere and competitive with E7s and Power9. It takes decades, but the server guys managed to fit in nicely in about 5 years. I am excited.

> HW-wise we are now in the 3rd gen of chips

I stand corrected. It's been a time since I've looked at the performance. I didn't realize they caught up so quickly.

So if my code is parallel enough to take advantage of all the cores, like a web server serving 96 requests simultaneously, one per core, would I get better performance than Intel? Better performance per watt?

You would run 96 processes one for each core. Performance more depends on IO that CPU in reality.

What ARM will bring is competitive market and no Intel Tax. Average 1U in ARM will be massively cheaper. For servers OpenPOWER is more interesting[1]

[1]https://www.theregister.co.uk/2016/04/07/open_power_summit_p...

Reply on news.ycombinator.com