VIM4 vs Radxa RockPi 5

General Discussion
1

Will be interesting to see the performance difference between the VIM4 and Radxa RockPi 5

2

RK3588 (rockpi5 chip):

t7_an400 (code for A311D2) (vim4 chip)

  • note the t7_an400 is overclocked to 2.7GHz in performance governor

definitely pushing limits, but good to see

(just for comparison, one of the highest VIM3 scores on android)

cheers

3

VIM4 launches with the A53 cores clocked at 2.0 GHz and the A73 clocked at 2.2 GHz (see the sbc-bench listing with precise clockspeed measurements). RK3588[S] devices will be launched with the A55 cores clocked at 1.8 GHz and the A76 cores clocked at 2.4 GHz. Both might be subject to change over time once SoC vendors provide less conservative boot BLOBs that do DRAM initialization and determine clocking behaviour of the CPU cores).

But letā€™s simply compare GB listings with the aforementioned clockspeeds for the sake of simplicity.

So Geekbench is telling us RK3588 would be almost twice as fast as A311D2.

Now letā€™s take a closer look at the t7_an400 GB v5 listing @Electr1 wanted to use but instead posted two times some random RK3588 result. This t7_an400 entry claims to run at higher clockspeeds (A53 at 2.4 GHz and the A73 even at ā€˜2.78 GHzā€™ ā€“ LMAO) so letā€™s compare it to VIM4 launch day clockspeeds:

https://browser.geekbench.com/v5/cpu/compare/14170348?baseline=14570096

The AES-XTS score of the A73 core improved by almost 70% while the clockspeed has been only moderately lifted from 2.2 GHz to 2.8 GHz (less than 30%). As everybody can clearly see this isnā€™t possible since the AES-XTS scores are not affected by anything external like DRAM setup. But who cares? Geekbench never measures CPU clockspeeds but simply relies on some sysfs values that are reported as ā€˜truthā€™.

TL;DR: Geekbench sucksā€¦ or at least blindly trusting into ā€˜Geekbench scoresā€™ :slight_smile:

2 Likes
4

I donā€™t believe any of them (all on social media). Pretty simple to write code to make it test good or should I say make ā€œclick baitā€.

Best test is with the specific application running on your own bench.

NONE of them have ever posted the source code or mentioned compilers used for the codeā€¦Its all VERY flaky how they ā€œtestā€.

Social media is a good tool too steer and mislead folks AND works very well when competitors want to disparage a strong competitor in the market.

5

definitely geekbench scores are not a mapping to long term real world performance,
we can only get to see that when we have both devices on hands to test :slight_smile:

1 Like
6

Some YouTubers have both Khadas Vim4 and RK3588 machines. ETA Prime who does mainly game emulation tests, budget builds and new hardware reviews has used both.

Till software matures better to use which best runs the software, need you have. RK3588 on specs should be faster. But in general Vim4 makes a quick desktop.

7

Hi folks,
We confirmed that RK3588/S is higher performance then A311D2 and also price.

Have a good day!