VoluMark

[Voxilla]

Edit : major new Volumark version:

• Now runs in 64 bit if supported.
• Possibility to configure number of cores used.
• Analysis of CPU throttling by running x10 or x100 times.

VoluMark is based on pure CPU volume rendering.
It is highly optimized with SIMD and multi-threading.

When hitting the BenchMark button benchmarking starts.
The rendered volume is a micro - CT scanned head of a mouse of 500x500x400 voxels.

During benchmarking the volume is rotated for 10 seconds. At the end the score is displayed.
Rendering is at 500x500 pixels and the image is upscaled to fit the display.

The benchmark has been optimized both for ARM and x86 processors.

Measured VoluMark scores are:

97 Samsung ARM A57 Exynos 5433 Quad 1.9 Ghz
80 Samsung ARM A15 Exynos 5420 Quad 1.9 Ghz
72 Nvidia Denver K1 Dual 2.5 Ghz
58 Qualcomn 805 Krait Quad 2.7 Ghz
58 Intel Bay Trail Z3745 Quad 1.86 Ghz
55 Qualcomn 801 Krait Quad 2.5 Ghz
45 Qualcomn 600 Krait Quad 1.9 Ghz
43 QualComn 410 ARM A53 Quad 1.2 Ghz
29 Nvidia Tegra 3 ARM A9 Quad 1.2 Ghz
26 QualComn 400 ARM A7 Quad 1.2 Ghz
22 RockChip RK3066 ARM A9 Dual 1.6 Ghz
19 QualComn 400 Krait Dual 1.4 Ghz​

 

[Infinisearch]

Congratulations on the release, hope its a success for you.

 

[Lightman]

Great!
Galaxy Note 4 (Qualcomm) scored 56.

Not a great score compared to Galaxy Note 4 Exynoss edition but will have to do.

EDIT: Correction! On cooled down phone I've got 58! Same as Intel Atom Z3745

 

[Osamar]

Just benched!!
16 for Onda V973v2 Allwinner A31 quad core ARM A7 / 1,2 GHz.

 

[OlegSH]

I got scores from 77 to 96 on Galaxy Note 4 with Exynos 5433 - 1.9 GHz quad core A57 cluster + 1.3 GHz quad core A53 clusters (Heterogeneous Multi-Processing mode)

 

[lanek]

Samsung Galaxy S3, I9505 stock version ( international version with Qualcomm quad core ): 45

This said i will do some nightmare for this rat,

 

[itsmydamnation]

36 on my sony XTZ (Quad-core 1.5 GHz Krait)

seems kinda low............

 

[Voxilla]

Samsung Galaxy S3, I9505 stock version ( international version with Qualcomm quad core ): 45,

That's probably a Galaxy S4 instead of a S3.

 

[Voxilla]

36 on my sony XTZ (Quad-core 1.5 GHz Krait)

seems kinda low............

The Krait of lanek does 45 at 1.9Ghz, and 45*1.5/1.9 = 35.5, close enough.

Anyway I've the impression the Qualcomm processors run at way lower clock speed as their advertised clock because the benchmark is extremely taxing on the CPU.
For highest score, you should start with a fresh 'cold' device.

 

[OlegSH]

60-72 for dual core Denver@2.5 GHz in Tegra K1. I wonder whether the benchmarks is capable to utilize all 4 Denver's NEONs

 

[Voxilla]

60-72 for dual core Denver@2.5 GHz in Tegra K1. I wonder whether the benchmarks is capable to utilize all 4 Denver's NEONs

That's a pretty good score for a dual core processor, certainly the highest IPC so far. To stay in the race K2 definitely needs 3 or 4 cores though.
There is plenty of NEON code in the benchmark, using all available hardware resources.

 

[Plano]

Here are some scores from my devices:
OnePlus One: 55
Nexus 7 (Tegra 3): 29
HTC One Mini (2x1.4GHz): 19

 

[lanek]

That's probably a Galaxy S4 instead of a S3.

YOu know what, i should really stop to drink, we approach christmass, and so i go from a fest drink to another... you are 100% right.. S4 not S3

 

[Lightman]

I got scores from 77 to 96 on Galaxy Note 4 with Exynos 5433 - 1.9 GHz quad core A57 cluster + 1.3 GHz quad core A53 clusters (Heterogeneous Multi-Processing mode)

My brothers Note 4 also had score of 97 when plugged in to charger Same phone as mine but nearly twice as fast thanks to 20nm Exynoss 5433 :/ Now I'm not happy with my Qualcomm version, anyone wants to swap?

 

[OlegSH]

Don't hurry to swap, Exynos 5433 version has slow Mali T760 with slow driver - https://twitter.com/sonicadvance1/status/541297486174760960 and https://twitter.com/sonicadvance1/status/542383750328250369

 

[itsmydamnation]

The Krait of lanek does 45 at 1.9Ghz, and 45*1.5/1.9 = 35.5, close enough.

I didn't mean that it was wrong, just compared to the numbers in your first post i expected a higher score for krait, in this benchmark (assuming linear scaling) krait is no better then A9. Is that RockChip A9 a regular A9 (2 wide, 64 bit NEON stock ARM cache etc)?

 

[Voxilla]

I didn't mean that it was wrong, just compared to the numbers in your first post i expected a higher score for krait, in this benchmark (assuming linear scaling) krait is no better then A9. Is that RockChip A9 a regular A9 (2 wide, 64 bit NEON stock ARM cache etc)?

The reason for the lower as expected results for Krait, is it runs not at maximum clock speed. At least not all cores, and not all the time.
Here a graph made with CPU Spy (measured during 4 runs of Volumark). This is for a 1.9 Ghz Krait. You can see the clock is all over the place.
Average only is 1.4 Ghz.


On the other hand for example the Z3745 runs at full speed of 1.86 Ghz (actualy I thought it was only running at 1.33 Ghz, so I'll change that in my first post)

 

[Ethatron]

Samsung Note Pro 12.2: 79
That's with two profiles running, and about 10 apps in the background. 55% battery. First run.

 

[liolio]

I got 43 on a Snapdragon 410 @1.2 (Huawei Ascend y550).

 

[Voxilla]

I've updated the list of scores in the first post, with all the scores posted here.

Later I also plan to release a 64-bit version, but currently I've no device to test with...