VisionFive 2 performance and power consumption benchmarks

I spent some time to benchmark my super-early-bird VisionFive 2, looking both at CPU performance and power consumption.

The results are really good, it’s much faster than the VisionFive 1 while also consuming much less power!

Here is the full article: Bits of networks · Benchmarking RISC-V: VisionFive 2 vs the world

And here is a copy of the conclusion (the performance results are about a xz decompression benchmark):

When looking at single-core CPU performance, the VisionFive 2 is roughly 75% faster than the original VisionFive. Since it has twice the core count, that means a +150% total performance increase. And taking into account power consumption, it is 4 to 6 times more energy-efficient! So that’s definitely a very big improvement.

Compared to the HiFive Unmatched (which is not even technically a SBC), the VisionFive 2 still outperforms it by 50%, and is 15 to 18 times more energy-efficient. So, as far as I can tell, the claim about it being a “high-performance RISC-V SBC” is true.

When comparing with Raspberry Pis, the VisionFive 2 is much better than a Raspberry Pi 3B+, but still 46% slower than a Raspberry Pi 4, and 40% to 70% less energy-efficient.

Compared to low-power x86_64 systems, the VisionFive 2 is of course slower when looking at raw performance, but at the same time it is 2.75 to 3.4 times more energy-efficient. This is a general advantage that SBCs have over more complete systems: they have much less peripherals, are less extensible and have generally lower performance, but they are much more energy-efficient.

Overall, the VisionFive 2 is a big step in the right direction, and this kind of RISC-V hardware can definitely compete with modern ARM boards since they have very similar performance-energy tradeoffs.

9 Likes

Nice testing, and it’s looking fairly good despite we’ve only got early drivers and there’s likely not much power management going on yet.

Raspberry Pi 4 did much, much worse at launch.

Sorry, I found and fixed several issues in the measurements:

  • the power measurements of the VisionFive 2 were too low. I have now re-made all SBC measurements with the exact same setup to allow comparisons
  • all RaspberryPi 3B+ results were completely wrong because of a faulty USB cable (I didn’t notice the “undervoltage” LED being on)

I have updated the article. The new conclusion goes in the same direction as before, except that the VisionFive 2 is now basically equivalent to a Raspberry Pi 3B+ (roughly same performance and same power consumption).

When comparing with Raspberry Pis, the VisionFive 2 is about as fast as a Raspberry Pi 3B+, although much slower on memory-heavy benchmarks, and also as energy-efficient. However, it is still 46% slower than a Raspberry Pi 4, and two times less energy-efficient. As far as I can tell, both SoC are 28 nm, so we would ideally expect the same energy-efficiency.

2 Likes

What about a head to head on the GPU’s for power usage ?

1 Like

I think if we use newer gcc and binutils and also enable Zba Zbb extensions, we should got higher performance.

3 Likes

yeah, we do have plan to upgrade gcc and binutils. stay tuned.

2 Likes

The use of USB as a power source for Pis is a curse, there’s a table of results for USB cables where, as with the TIMs that function worse than ketchup or toothpaste, some cables are described as “long thin resistors” rather than cables. So always used known-good cables to power Pis.

1 Like

thanks @zorun , very detailed report.
BTW. we also test the power comsupmtion for idle state of visionfive 2, it is almost 12v@210mA measured by DC power

2 Likes