Tensor chip which is called the biggest mobile hardware innovation in the history of Google.
So let's explore on what does that mean?
Google stated a goal for building tensor is to bring ai breakthroughs directly to pixel and drive forward its vision of always available technology things like ambient computing.
Google considers phones as the central control device of an ambient system. In a interview to The verge, Rick Ostello said that work started back in 2017 after the company came to the realization that it couldn't take a piecemeal approach things like building a single co-processor for example the pixel visual and the neural core to boost its ai models rather they needed an optimized chip that is for the desired tasks or the tasks needed on pixel smartphones.
Google went into tensor details and explicitly touted the inclusion of two high performance arm cortex x1 calls at 2.8 gigahertz and they are also joined by two mid 2.25 gigahertz a 76 cpu, these are based upon a five nanometer process rather than the seven nanometer process. This allows Google to throw more power at the workloads that are of medium intensity in a normal cpu and the mid cores could handle such tasks like google lens visual analysis, but be maxed out google says using two x1 cores in that scenario would be more efficient and that's what sensor is optimized for in real terms it's 80 faster than the pixel 5 snapdragon 765g from last year which is quite a big jump when you think about it.
There's also a 20 core gpu that google says delivers a premium gaming experience for the most popular android games out there. it's 370 percent faster than the pixel 5 which uses the adreno 620 gpu, so gaming shouldn't be an issue for the pixel 6 series.
The Tensor security core is a cpu based system that's isolated from the application processor and dedicated to running sensitive tasks and controls. It works with the dedicated titan m2 security chip which isn't actually part of tensor itself but google touts it as being resilient to things like advanced attacks like electromagnetic analysis, voltage glitching and laser fault injection.
Original titan m chip just works in conjunction with software to stop your phone from being rolled back to an older version of android that might have security vulnerabilities, it also does things like prevent bootloader unlocking and verifies your lock screen passcode for that extra layer of security.
The tensor processing unit and the machine learning engine is said to be custom made by google research and built based on where the machine learning models are heading not where they are currently today. This chips image signal processor or isp features is an accelerator that runs the hdmi algorithm and a key reason the pixel 6 and 6 pro can do things like live hdr plus video at 4k 60fps more conveniently.
Other phones on the market will also brings machine learning to the ultra low power domain and it allows always on display, now playing and other ambient experiences to run all the time without necessarily draining your battery.
Google with Tensor prioritizing, what they consider total performance and efficiency. This specifically involves excelling at heterogeneous computing tasks that require various parts of the soc to work together for example lens makes use of the cpu, the gpu, the isp and the tpu to run efficiently and this is where google hopes that tensor will excel compared to other chipsets out there.
In regard to camera features tensor also allows other things to be done besides the live hdr plus and does make colors more accurate and vivid at 4k 60 fps. Tensor also allows computational photography and video features like motion mode and google camera action pan which blurs background while long exposure works on the subject to ensure it stays in focus.
Face detection is more accurate on the pixel 6 and works faster due to the integrated subsystems while apparently consuming half the power compared to the pixel 5. Also in these modes assistant on tensor is also using what google considers the most advanced speech recognition model ever released by the firm.
It also uses half the power and the high quality asr or automatic speech recognition model is also used to transcribe voice commands as well it's in long-running applications things like recorder and live caption without quickly or significantly draining the battery.
The assistant voice typing for editing and transcribed in an entirely hands-free manner. Live translate with the pixel's translation quality improving up to 18 percent, this level of improvement typically takes multiple years of research according to google's.
Geekbench listings suggest that tensor sits at top of the single core benchmark leaderboard with a score of over 1030 and a multi-core score of around 2750 and that places it just around the same levels as devices with the qualcomm snapdragon 865 at least in multi-core score. Overall that seems pretty good for the first internal chip from google but it's in other areas that raw power and the google kind of touts the tensor chip itself.
Google will be making more chips of phones and potentially other form factors have been rumored and we would expect to see more tensor powered smartphones and then maybe hopefully some other hardware in the not too distant future.
Google's tensor chip is very exciting for the pixel 6 and 6 pro because of the new cpu. Hope this article helped you to gain some information about Tensor chip.
0 Comments