KITA

anantksundaram said:

But, of course...

Can someone name one original idea from Microsoft?

Of course what? Of course Appleinsider misquoted Bloomberg to make it sound as if Microsoft's objective was to follow Apple? Or, of course Microsoft wants to remain competitive in the cloud computing industry in which they are a major player?

There are plenty, but a good and recent example is HoloLens 2 with Azure Remote Rendering and Dynamics 365. Of course, Apple doesn't compete in the cloud computing or commercial augmented reality market, so I wouldn't be surprised if most of the audience here hasn't the slightest idea these groundbreaking technologies even exist.

AppleInsider said:

According to sources of Bloomberg, Microsoft is working on an in-house processor design, one that could be used in its Surface lineup, but may also end up in servers. Thought to be ARM-based, the chip design unit is said to report to Jason Zander, the head of Microsoft's Azure cloud business.

That's a weird spin in the title to make it sound like Microsoft is following Apple to ARM.

First paragraph in the bloomberg article:

Microsoft Corp. is working on in-house processor designs for use in server computers that run the company’s cloud services, adding to an industrywide effort to reduce reliance on Intel Corp.’s chip technology.

Microsoft is going the ARM route for servers (Azure), similar to their largest competitor Amazon. AWS has already been using Amazon's own second generation ARM Graviton2 processors which, as Amazon claims, offers customers 40% better price performance than x86 based instances. A third generation Graviton3 (likely based on Neoverse V1 or Neoverse N2) is expected in 2021.

Seriously, this isn't all about Apple all the time.

sflocal said:

Apple's chip beats out Qualcomm chips?  Wake me up in a few years when Qualcomm actually does something.

This is embarrassing.

This is a bit deceiving given the claim seems to hang mostly on peak CPU and peak GPU performance.

While I understand it's going on what's been provided, the Snapdragon 888 likely has a more powerful GPU in sustained performance (claimed and historical implementations demonstrate this).

Note - the higher end Snapdragon 865/+ based devices have typically offered the same peak and sustained performance. Meanwhile, the A13 and especially the A14 can be a bit of a mess despite that impressive peak performance.



The 888 also has a better integrated 5G modem and an extremely impressive ISP. In terms of AI? The 888's DSP (+CPU/GPU) offers 26 TOPS, the A14 only offers 11 TOPS.

It looks like Qualcomm will have to continue to rely on ARM's Cortex X program if they want to catch up (or at least try) to Apple in raw CPU performance. The underclocked Cortex X1 is an impressive jump, but not exactly ARM's ideal implementation (larger L2 and higher clock). Even in the ideal case, it would still fall short of Apple.

techconc said:

Do you think the version of something like Geekbench has different workloads from the PC version of Geekbench?  Do you think the various workloads in these benchmarks don't represent common or even professional workloads of desktop apps?  Perhaps you should take a look the different workloads in Geekbench 5, then tell me how they don't apply to desktop apps, etc.   https://www.geekbench.com/doc/geekbench5-cpu-workloads.pdf

Just a note from Anandtech in their recent Zen 3 deep dive: 

Most of the people in our industry have a love/hate relationship when it comes to synthetic tests. On the one hand, they’re often good for quick summaries of performance and are easy to use, but most of the time the tests aren’t related to any real software. Synthetic tests are often very good at burrowing down to a specific set of instructions and maximizing the performance out of those. Due to requests from a number of our readers, we have the following synthetic tests.

...

GeekBench 5:

As a common tool for cross-platform testing between mobile, PC, and Mac, GeekBench is an ultimate exercise in synthetic testing across a range of algorithms looking for peak throughput. Tests include encryption, compression, fast Fourier transform, memory operations, n-body physics, matrix operations, histogram manipulation, and HTML parsing.

I’m including this test due to popular demand, although the results do come across as overly synthetic, and a lot of users often put a lot of weight behind the test due to the fact that it is compiled across different platforms (although with different compilers).

The Surface Laptop 3 was their competitor to the MacBook Air. The Surface Laptop Go has a 12.4" 3:2 display and is 2.4 lbs.

Surface Laptop 3 (2019) - $999

13.5" 2256x1504 display
Intel Core i5-1035G7 (4 cores / 8 threads with G7 graphics)
8 GB LPDDR4X-3733
128 GB SSD (user replaceable) 
2.79 lbs

MacBook Air (2020) - $999

13.3" 2560x1600 display
Intel Core i3-1000NG4 (2 cores / 4 threads with G4 graphics)
8 GB LPDDR4X-3733
256 GB SSD
2.8 lbs