Here's how Google Tensor differs from other Android CPUs

Heres how Google Tensor differs from other Android CPUs.pngsignaturea3c855f06a01000750dc690b9be60b4d

The Pixel 6 and Pixel 6 Pro we officially announced today. Along with them, we got a confirmation of the phone's exciting new Google Tensor SoC - and how it's different from anything else currently available.

There are very few key ways in which Google differentiates itself. One of the most notable, as expected leaks, is the CPU fix.

Some background: Major Android processors like the Exynos 2100 and Snapdragon 888 use three types of cores. The Snapdragon 888, for example, uses one Cortex X1 (high power), three Cortex A78 (medium power), and four Cortex A55 cores (low power / high efficiency).

Where Google Tensor stands out is its use two of the most powerful Cortex X1 cores clocked at 2.8 GHz, which should make the phone better with some challenging functions.

On the other hand, the company confirmed another strange rumor: rather than triple mid-range cores, the Pixel 6 uses just two. Even more surprisingly, these will not even be like the latest A78, but instead let the older, less powerful and less efficient A76 cores (running at 2.25 GHz) back in 2022 .

Meanwhile, the phone still uses four low-power A55 cores running at 1.8 GHz as well, and the GPU is the Mali G78MP20, which should offer game performance as good as any on an Android device.

The use of the older A76 cores remains the largest headset, and we have not yet heard a clear reason for it.

For its part, Google claims that the processor is 80% faster than the Pixel 5; the GPU is 370% faster. But since the Pixel 5 was specified with Snapdragon 765, that's not saying too much, except that the processor should be relatively competitive with the existing major Android chips.

An interview with ArsTechnica with Google engineers sheds some light on the key format decisions.

Phil Carmack, VP of Google Silicon, explains:

We focused much of our design effort on how the workload is distributed, how the energy is distributed over the chip, and how the processors come in at different times. When it comes to a heavy workload, Android tends to be hit hard, and that's how we get responsive. "

It says “wso it's a stable state problem where, say, the CPU has a lighter load but it's still moderate, your dual X1s run, and at that level of performance, those are the most effective. ”

So there seems to be a greater bias towards mediocre tasks being run on the X1 cores, than the A76s when possible - leading to a more hopefully smart phone. By dialing down the powerful rights more often “a the workload that you would normally have done with dual A76s is barely increased, now tapping on the gas with dual X1s. ”

READ  Google Pixel bug prevents 911 users from making calls by Microsoft Teams

As pointed out in an ArsTechnica article, one big core is a recent development for ARM - based chips. Traditionally, these sinks have used two or more high-performance cores. Apple, for its part, has continued to adhere to a simpler separation of high-efficiency and high-performance cores.

Carmack, at the same time saying “iIf you want responsiveness, two large collars may be the quickest way to achieve that, and the most effective way to get high performance, "and it suggests that you don't have one big heart but good for criteria single-thread.

Google Tensor Chip

At least Tensor is not Apple's level jump in performance over competitors - at least when it comes to CPU and GPU core functions. Where a Tensor is about to shine is in AI and ML operations with its new tensor processing unit “TPU” or tensor (sometimes called a neural processing unit in other devices). There is plenty of unexplored potential in this regard.

Unfortunately, Google was vague in the specs, except to show off some Google Tensor - enabled devices, such as HDR video at 4K 60 fps, the new Face Unblur feature, and real-time implementation of Live Translate (read my hands-on for more). ).

The company suggests that they are willing to share numbers because the existing ML criteria are "looking back," but they basically want us to know. that Tensor is going to run Google's own ML algorithms in the best possible way. The company says that some of these machine learning activities cannot run effectively on other Android devices; I know what Qualcomm has to say about that.

I would still like to know more numbers though, just as a frame of reference. Rumors have also suggested that Google Tensor was created in partnership with Samsung, but the company made no mention of the fact when it was announced or published in the media; we have to wait for teardowns to see if people can see any Samsung components.

But more than anything, Google Tensor is an opportunity for Google to integrate Apple's best - known hardware and software - great control over every aspect of the user experience. It looks like the best level of Google Tensor is yet to come, and I can't wait to see how other manufacturers respond.

Related Posts

Deja una respuesta

Tu dirección de correo electrónico no será publicada.

Subir

We use cookies to ensure that we give the best user experience on our website. If you continue to use this site we will assume that you agree. More information