Snapdragon 765G vs Snapdragon 845: The Google Pixel 5 could offer worse performance than the Pixel 3

[Redaktion]

The Pixel 5 is expected to be powered by a sub-premium chipset, which a host of rumors claim could be the Snapdragon 765G. How does a sub-premium SoC compare to the two-year-old Snapdragon 845 on the Pixel 3? Well, it's a mixed bag.

https://www.notebookcheck.net/Snapdragon-765G-vs-Snapdragon-845-The-Google-Pixel-5-could-offer-worse-performance-than-the-Pixel-3.467706.0.html

[joseph]

The 700 and 600 series always receives significantly worse GPUs. The choice of the 821 in some tablets last year might be odd, they would still have a certain advantage over the 600 series, such as when compared to the 660

[Ricci Rox]

The 700 and 600 series always receives significantly worse GPUs. The choice of the 821 in some tablets last year might be odd, they would still have a certain advantage over the 600 series, such as when compared to the 660

Yes. The SD765G only matches up to the SD835 GPU-wise.

[Roy]

First of all, the 765G does support 4k 60fps. Secondly, If Google decides to go with the overclocked version the 768G then we're looking at a comparable gpu performance to the 845 but with a newer architecture, 10nm vs 7nm which is way more efficient in battery.

[Ricci Rox]

First of all, the 765G does support 4k 60fps. Secondly, If Google decides to go with the overclocked version the 768G then we're looking at a comparable gpu performance to the 845 but with a newer architecture, 10nm vs 7nm which is way more efficient in battery.

Author here!

There's no mention of 4K60fps recording on Google's product sheet.
https://www.qualcomm.com/products/snapdragon-765g-5g-mobile-platform

The 786G only has 20-25% improved GPU vs the 765G. That's still way less than the 845.

"It's important to note, though, that the Snapdragon 765G is likely a far more efficient chipset."

[Bomp]

Decent article
"It's important to note, though, that the Snapdragon 765G is likely a far more efficient chipset."
This should not be understated; benchmarks are usually conducted in air-conditioned rooms but in the real world the 845 is likely to throttle slightly at room a temperature, badly if the phone chassis is not well designed for heavy workloads. Meanwhile the 7nm lithography and general architectural evolutions would likely see 765G practically never throttling unless used in hot climates.
Aside from this, newer SoC may carry higher bandwidth connections, and crucially newer accelerators for "AI" and for images, which are increasingly relevant and better suited to their tasks. (Hexagon "685" vs "696")
Regardless, personally I would always take the new chipset for connectivity reasons... Due to server-side processing potentially taking significant portions of compute work - 5G can essentially be considered a contributing factor to a phone's performance now and more in the future.

[S.Yu]

Usually don't expect a midrange SD to match a flagship under 2 years old, the segmentation they maintain is fairly consistent.

First of all, the 765G does support 4k 60fps. Secondly, If Google decides to go with the overclocked version the 768G then we're looking at a comparable gpu performance to the 845 but with a newer architecture, 10nm vs 7nm which is way more efficient in battery.

Don't flaunt a yet to be announced new model matching the performance of a 2 year old predecessor like it's something to brag about. Unless it's priced notably lower, we'll see who's stupid enough to buy the new one.

[Mckillio]

How does memory speed compare?  Assuming it has UFS 3, it will be a big gain in performance over the 3.  I also wonder if Google will have more custom silicone this time, possibly making up for some of these inherent shortcomings. 

[Gianni]

The efficiency of the 765G should not be understated. It is 3x more efficient than the 845, and can sustain peak performance much more easily & for much longer. Battery life is also greatly improved

[Mckillio]

Usually don't expect a midrange SD to match a flagship under 2 years old, the segmentation they maintain is fairly consistent.

First of all, the 765G does support 4k 60fps. Secondly, If Google decides to go with the overclocked version the 768G then we're looking at a comparable gpu performance to the 845 but with a newer architecture, 10nm vs 7nm which is way more efficient in battery.

Don't flaunt a yet to be announced new model matching the performance of a 2 year old predecessor like it's something to brag about. Unless it's priced notably lower, we'll see who's stupid enough to buy the new one.

It's supposed to be $100 less than the 3/4.  As long as it's better overall it won't be a bad decision to get it, especially on sale.

[Arief]

I don't think common user would feel the differences in real life, unless for gaming. But who buy pixel for gaming tho. I think it's a good news. Qualcomm monopolize the market for years now. Lack of competitor makes their chipset so pricey these days. I'm afraid Qualcomm become the next Intel. Glad to hear more news from Exynos with their rdna, ryzen mobile, and Mediatek. For Mediatek, please share the source just like Qualcomm. Your affordable chipset it's great just lack of customizable. Ok that's my opinion thank you

[Dan6]

Most of the people expect an android flagship from Google. But now instead of just droping Soli and adding bigger battery, wide camera and 4k60 video they decided to drop top snapdragon? Is it really so hard, Sundar? Google is big and smart company, but looks like they have no idea what to do with the hardware. Surface Duo is probably a better buy this year.

[Chris Welton]

SoC?  My pixel works fine without this medical nomenclature about the hypothalamus. :). Get a copy editor. 

[Ricci Rox]

SoC?  My pixel works fine without this medical nomenclature about the hypothalamus. :). Get a copy editor.

You lost me there, mate.  ;D

[S.Yu]

SoC?  My pixel works fine without this medical nomenclature about the hypothalamus. :). Get a copy editor.

You lost me there, mate.  ;D

Lost me too. I froze for a second and realized that there's probably no such structure, turns out I'm right, there's SC, SO, OC, but not SoC.
https://msu.edu/~brains/brains/human/hypothalamus/index.html