Discussion Google Tensor G3

FlameTail

Diamond Member
Dec 15, 2021
3,184
1,820
106
CPU:
1x Arm Cortex-X3 (2.91GHz)
4x Arm Cortex-A715 (2.37GHz)
4x Arm Cortex-A510 (1.70GHz)

GPU:
Mali G715 MP7 (?)

Node:
Samsung 4LPP

Benchmarks:
 
Reactions: krumme

Tigerick

Senior member
Apr 1, 2022
686
576
106
Cool, maybe TS could compile all the past Tensor SoC into the table to show the difference between G1, G2 and G3....

Seems like Google is serious into SoC design with rumor of upcoming G5 going to be made by TSMC's N3E process. We all know how expensive the process would be...Maybe we could see PC/Chromebook+ variant of G5 like Apple did.

 

hemedans

Senior member
Jan 31, 2015
207
102
116
Pretty horrible in real life gaming:
No improvement what so ever. In fact, it's worse than G2.
For those who don't know Goldenreviewer better avoid him for anything serious.

Android phones run Genshini Impact at different resolution, just look at that graph s23 run better than s23 ultra, iphone Se is most efficient phone ever released? Better than M1, 8 gen 2 and other phones, that's because s23 or iPhone Se has low resolution and run that game at lower resolution.

Comparing one phone which run game at 540p vs other phone which run game at 1080p then make conclusion on average fps it's wrong.

Even Andrei did criticise goldenreviewer before.
 

FlameTail

Diamond Member
Dec 15, 2021
3,184
1,820
106
For those who don't know Goldenreviewer better avoid him for anything serious.

Android phones run Genshini Impact at different resolution, just look at that graph s23 run better than s23 ultra, iphone Se is most efficient phone ever released? Better than M1, 8 gen 2 and other phones, that's because s23 or iPhone Se has low resolution and run that game at lower resolution.

Comparing one phone which run game at 540p vs other phone which run game at 1080p then make conclusion on average fps it's wrong.

Even Andrei did criticise goldenreviewer before.
Yeah the Genshin Test is whack. But the SPEC run is credible, no?
 

FlameTail

Diamond Member
Dec 15, 2021
3,184
1,820
106
IMO what they should have done:

Put MASSIVE CACHES

16 MB L3 + 32 MB SLC

Such huge caches will definitely offset the efficiency loss due to using Samsung's inferior 4nm node and Exynos IP.

And it would be feasible. Google is already saving a lot of money by using Samsung Foundry (and not TSMC), as well as using older ARM Core designs (D9300 and SD8G3 with next gen cores are launching very soon).

Also Google put the latest cutting edge LPDDR5X RAM in the Pixel 8 and 8 Pro. In my opinion this was a quite a non-sensible decision. LPDDR5X is very expensive. They might as well have put in 32 MB SLC in the Tensor G3 instead which would bring efficiency gains as well as provide memory bandwidth increase.

Google had the opportunity to do something great with the Tensor G3 but they fumbled it.
 

hemedans

Senior member
Jan 31, 2015
207
102
116
Yeah the Genshin Test is whack. But the SPEC run is credible, no?
G3 has worse efficient than Exynos 2200, sd 8 gen 1 and sd 888? there is No way A715 with improved 4lpp node will perform worse than A710 or A78 at same Samsung node.

Other reviewers got better Genshini score than him, so probably he has fault device.
 

FlameTail

Diamond Member
Dec 15, 2021
3,184
1,820
106
G3 has worse efficient than Exynos 2200, sd 8 gen 1 and sd 888? there is No way A715 with improved 4lpp node will perform worse than A710 or A78 at same Samsung node.

Other reviewers got better Genshini score than him, so probably he has fault device.

According to Geekerwan:

A710 was a regression in efficiency from A78, and A715 is back to the level of A78.

The A715s in the Tensor G3 might possibly be gimped versions with only 256kb of L2 cache and a measly amount of L3. Not a favourable comparison against a TSMC made A78 with 512kb L2 and big L3.

And from Andrei's reviews at Anandtech, we know the E2100's cache had worse latencies than the Sd888's cache, which points to the inferiority of Exynos IP.

Smaller cache capacities + worse cache latencies is heavy blow and the result he is getting might not be far from the reality.

But yeah, I am not ruling out the possibility he has a faulty device. For one, he said he couldn't get his device's Tensor G3's Prime Core to reach the max frequency even inside a freezer.
 

Attachments

  • Screenshot_20231012_062255_YouTube.jpg
    307.2 KB · Views: 8

Tup3x

Golden Member
Dec 31, 2016
1,012
1,002
136
For those who don't know Goldenreviewer better avoid him for anything serious.

Android phones run Genshini Impact at different resolution, just look at that graph s23 run better than s23 ultra, iphone Se is most efficient phone ever released? Better than M1, 8 gen 2 and other phones, that's because s23 or iPhone Se has low resolution and run that game at lower resolution.

Comparing one phone which run game at 540p vs other phone which run game at 1080p then make conclusion on average fps it's wrong.

Even Andrei did criticise goldenreviewer before.
Pixel 7 Pro and 8 Pro should be apples to apples. S22 (Exynos) actually ran the game at much lower resolution than what it should be unless one manually override that using Samsung game plugins. That wasn't the case with S23 so I guess they cheated since GPU was so lacklustre. Other manufacturers might have similar stuff so one should make sure that the game runs at correct resolution.

S23 and S23 Ultra are so close there so they are definitely using the same internal resolution. Genshin doesn't run at native resolution internally even when maxed out in game (it's well below 1080p). Since S23 Ultra has higher resolution display, there might be some overhead.

EDIT:
Also I doubt it's possible to make any meaningful cross platform comparison since I'm pretty sure that iOS and Android versions have differences in graphics even when both are maxed out.
 
Last edited:
Reactions: Tlh97 and hemedans
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |