Pixel 7 Pro benchmarks reveal unchanged CPU specs and a much-needed upgrade


The Tensor 2 chip that will power the Google Pixel 7 and 7 Pro may use up generations of CPU cores and offer only marginally better performance than OG Tensor, but better GPUs and better artificial intelligence can claim abilities.

Tensor was Google’s first in-house chip, built in collaboration with Samsung. Instead of using one large core, three medium cores and four small cores for the SoC like Qualcomm’s Snapdragon chips and Samsung’s Exynos chips, Google for the two cores to delegate the foreground processing duties, two medium cores and four small cores. went away.

It is also worth noting that the company chose the older Arm Cortex-A76 for the medium core, even though the A78 core was available at the time. developer Kuba Wojciechowski has found alleged benchmark scores for the Pixel 7 Pro and Dug Up code that indicates the Tensor 2 will reuse the same cores, even though Arm has likely created new designs that will support Qualcomm’s rumored next high-end chip. , will be powered by Snapdragon 8. Gen 2. Tensor G2, known internally as “gs201”, will have two Arm Cortex-X1 cores, two A76 cores and four A55 cores. On Geekbench, the G2-powered Pixel 7 Pro achieved a 10 percent higher multi-core score than the Pixel 6 Pro, which suggests that there is some improvement in performance.

Alleged Pixel 7 Pro Benchmark Scores

This can be attributed to two things: the G2 is apparently based on a 4nm process, which should make it more powerful and efficient than the first-gen chip, and the X1 cluster and A76 cores now run at higher clock speeds. Huh.


Also, as was the case with the first proprietary chip, Google is prioritizing power-efficient performance and responsiveness, and isn’t worried about breaking benchmark records.

Tensor G2 Builds a New GPU and Modem

Wojciechowski has also found that the G2 will have a Mali-G710 GPU, which is said to be 20 percent faster and 20 percent more power efficient than OG Tensor’s Mali-G78, and a 35 percent machine learning uplift. also provides. These improvements will be great for the camera and help the Pixel 7 and 7 Pro become the best camera phones of 2022.

The Tensor G2 reportedly has an improved TPU codenamed “Jaineiro”, which is probably meant for new AI experiences. Also, it appears that Google has improved the video encode/decode blocks and ISPs.

The device may pack a new Samsung-made S5300 modem and hopefully, this will help with connectivity issues experienced by Pixel 6 users.

Overall, even though the Tensor G2 is highly unlikely to be the fastest mobile SoC around, Google’s approach will likely ensure that it doesn’t overheat and as stated before, most modern smartphones are already pretty fast, And there’s more to the smartphone experience than other factors like raw CPU performance and software optimization, which Google is good at, also play a big part.

It is also learned that Google has started working on a third generation Tensor chip codenamed “Zuma” and will be based on Samsung’s 3nm process. Google will fully reveal the Pixel 7 on October 6.

Source


Related News

La soluzione magnetica rimuove le “sostanze chimiche per sempre” tossiche dall’acqua in pochi secondi

Scienziati in Australia hanno sviluppato una nuova tecnica interessante per rimuovere dall'acqua "sostanze chimiche per sempre" tossiche. Quando disciolto in acqua contaminata, il

Fortnite Capitolo 4: come evitare che le lobby siano piene di bot

Stanco dei robot nei tuoi giochi? Fortnite Capitolo 4? Skill-Based Matchmaking (SBMM) è stato coinvolto in polemiche da quando è stato rilasciato all'inizio

Come estendere la batteria dell’Apple Watch Ultra fino a 60 ore

Il nuovissimo Apple Watch Ultra è lo smartwatch più potente di sempre. Non solo a livello di prestazioni, ma anche in altri aspetti legati alla sua tecnica

Apple ospita il panel di sviluppatori Steam, educatori per promuovere le discipline umanistiche nello sviluppo di app

In onore del National STEAM Day, che mette in evidenza scienza, tecnologia, ingegneria, arte e matematica, l'8 novembre Apple ha interagito con cinque relatori che lavorano per