Google Tensor G2 benchmarks feature image

Google says it’s “very comfortable” that Tensor doesn’t beat the standards

Robert Treggs / Android Authority

tl; DR

  • A company executive said Google is “very comfortable” with Tensor chips not winning the criteria.
  • The manager said Google is measuring software workloads that will actually run on Pixel phones.

The Google Pixel 6 And the Pixel 7 Phones operate on a semi-custom system tensor processors, but these SoCs don’t really take the fight to Apple, Qualcomm, and other chip makers when it comes to most benchmarks.

Now, Google Silicon’s Senior Director of Product Management Monika Gupta has spoken about the benchmark performance of the Tensor line in the official Made by Google Podcast (free: 9to5Google).

Gupta suggested that standards are only part of the story when it comes to smartphone processors:

I think Classic Standards served a purpose at one point, but I think the industry has evolved since then. And if you look at what Google is trying to do by pushing AI innovations into the smartphone – because we feel this is an approach that will provide useful experiences like some of the ones I just mentioned – the classic benchmarks were composed at a time when AI and phones didn’t even exist . They may tell some stories, but we don’t feel like they’re telling the whole story.

She added that Google measures “actual software workloads” that will run on its chips rather than relying on standard applications with artificial workloads.

“We know exactly what we’re building for, and if that means we won’t beat the standards or we don’t look good on the standards, we’re totally comfortable with that, because the end result speaks for itself,” Gupta explained.

“Like the Pixel 6 and Pixel 7, you can see all the amazing innovations that we’ve come up with, and many of them were firsts on the Pixel. So we’re very comfortable with this approach.”

Correct approach to Tensor?

To be sure, benchmarks do not fully represent real-world results and workloads. First, many benchmark tests focus on peak performance you’ll rarely see rather than continuous performance. In fact, our own Tensor G2 test It shows that the Pixel 7 chipset has fallen behind competitors when it comes to extreme results but actually beat some Snapdragon 8 first generation The phones are under constant testing. And sustainable performance is a must for tasks like gaming and photography.

What do you think of standardized tests?

3478 votes

Google is also pushing machine learning in a big way for tasks like computational photography, and benchmark apps like Geekbench and 3DMark don’t take these devices into account. There are some AI standards, but the changing nature of AI devices and workloads means that even these tests don’t tell the whole story. For what it’s worth, the AI-Benchmark rating shows that the Pixel 7’s Tensor G2 SoC slots are behind the Snapdragon 8 Gen 1 and Dimensions 9000 phones.

It must also be said that this is not a new approach for Google. The original Tensor chipset seen in the Pixel 6 uses the company’s Cortex-X1 CPU core observed at that time This approach was intended to maximize efficiency in “medium” workloads. The company confirmed that one Cortex-X CPU core (as seen on competing SoC designs) was only useful for benchmarking.

#Google #comfortable #Tensor #doesnt #beat #standards

Leave a Comment

Your email address will not be published. Required fields are marked *