A Comprehensive Raspberry Pi 3 Benchmark

RE: A Comprehensive Raspberry Pi 3 Benchmark

gpolder — Thu, 10 May 2018 19:47:04 GMT

Hi [mention:e533a280d27446f88d5eca64e725a4fb:e9ed411860ed4f2ba0265705b8793d05] in your table with the sysbench results for 4 threats, I expect the second column is on the Pi 2 in stead of the Pi 3?

Which version of Raspbian did you use for these tests, I got comparable results with jessie, but with wheezy the performance was worse (RoadTest Review a Raspberry Pi 3 Model B ! - Review ).

RE: A Comprehensive Raspberry Pi 3 Benchmark

andrewgoh0 — Fri, 16 Dec 2016 10:00:53 GMT

it has come to time & fashion that r-pi would need 1 of those monster cpu coolers lol

Optimized CPU Cooling Performance Top-Down Heatsinks

RE: A Comprehensive Raspberry Pi 3 Benchmark

bonestaak — Mon, 17 Oct 2016 12:14:50 GMT

What should be the Passmark score for the Pi3 (passmark is used to compare speed for Plex server)

RE: A Comprehensive Raspberry Pi 3 Benchmark

rive — Sun, 22 May 2016 17:05:41 GMT

There was also another NEON related test, which was the Linpack Single Precision Benchmark. This related the speed to MFLOPS, the Pi 2 came in at 299.93 MFLOPS and the Pi 3 at 462.07 MFLOPS.

462 Mflops is equal to .4 Gflops.

This is to inform, that the Linpack Bench mentioned above is not even close to being accurate. On a stock Pi3 that has a proper power supply, and is properly cooled, the Linpack bench is 6 Gflops (or 6 billion operations per second). A moderately overclocked Pi3 can achieve around 6.8 Gflops, as shown here:

Ultimate Guide for Raspberry Pi 3 Performance and Cooling

Run it yourself, and see (may require 'over_voltage=1' in /boot/config.txt):

Install Linpack Bench:

sudo apt-get install libmpich-dev
wget http://web.eece.maine.edu/~vweaver/junk/pi3_hpl.tar.gz
tar -xvzf pi3_hpl.tar.gz
chmod +x xhpl

Reboot, then Run Linpack Bench:

./xhpl

A result of less than 1 Gflop benched on the Pi3, means it throttled, overheated, and/or was undervolted.

Stock Pi3 Linpack Bench:

pi@raspberrypi:~ $ ./xhpl
================================================================================
HPLinpack 2.1 -- High-Performance Linpack benchmark --   October 26, 2012
Written by A. Petitet and R. Clint Whaley, Innovative Computing Laboratory, UTK
Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
Modified by Julien Langou, University of Colorado Denver
================================================================================

An explanation of the input/output parameters follows:
T/V    : Wall time / encoded variant.
N      : The order of the coefficient matrix A.
NB     : The partitioning blocking factor.
P      : The number of process rows.
Q      : The number of process columns.
Time   : Time in seconds to solve the linear system.
Gflops : Rate of execution for solving the linear system.

The following parameter values will be used:

N      :    8000
NB     :     256
PMAP   : Row-major process mapping
P      :       1
Q      :       1
PFACT :    Left
NBMIN :       2
NDIV   :       2
RFACT :   Right
BCAST :   2ring
DEPTH :       0
SWAP   : Mix (threshold = 64)
L1     : transposed form
U      : transposed form
EQUIL : yes
ALIGN : 8 double precision words

--------------------------------------------------------------------------------

- The matrix A is randomly generated for each test.
- The following scaled residual check will be computed:
      ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
- The relative machine precision (eps) is taken to be               1.110223e-16
- Computational tests pass if scaled residuals are less than                16.0

================================================================================
T/V                N    NB     P     Q               Time                 Gflops
--------------------------------------------------------------------------------
WR02R2L2        8000   256     1     1              55.37              6.166e+00
HPL_pdgesv() start time Sat Apr 23 15:14:17 2016

HPL_pdgesv() end time   Sat Apr 23 15:15:12 2016

--------------------------------------------------------------------------------
||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)=        0.0025941 ...... PASSED
================================================================================

Finished      1 tests with the following results:
              1 tests completed and passed residual checks,
              0 tests completed and failed residual checks,
              0 tests skipped because of illegal input values.
--------------------------------------------------------------------------------

End of Tests.

RE: A Comprehensive Raspberry Pi 3 Benchmark

Former Member — Tue, 15 Mar 2016 14:37:22 GMT

Great Summary!! Thanks for the write up...

RE: A Comprehensive Raspberry Pi 3 Benchmark

chappy1978 — Sun, 13 Mar 2016 16:30:02 GMT

Great detail to testing. It really has sold me on the PI3. What I am wonder is where the PI Zero would fall into this testing.

RE: A Comprehensive Raspberry Pi 3 Benchmark

Former Member — Fri, 04 Mar 2016 22:46:34 GMT

Thank you very much for the benchmarks and comparisons.

I am still disappointed in the RAM available on RPi3 (same mistake as for RPi2). 1GB is insufficient for a "PC" or a "desktop". Try running LibreOffice and Chrome at the same time; this will chew up memory fast and you'll be swapping to a very slow disk.

Also, there will be no benefit in configuring /etc/fstab tmpfs (RAM disk) entries for frequently updated file system folders like /tmp and /run.

Is the jump to RPi3 worth it?

For science/engineering applications, Yes.

For creating a Linux or Unix desktop, No.

RE: A Comprehensive Raspberry Pi 3 Benchmark

amgalbu — Tue, 01 Mar 2016 16:45:01 GMT

Hi Christopher

Great review!

Just a question: I can't understand the mAh values you reported for the three boards (00006, 00103 and 00002)

Cheers

Ambrogio

RE: A Comprehensive Raspberry Pi 3 Benchmark

Former Member — Mon, 29 Feb 2016 21:38:01 GMT

Interesting the difference in default idle power between a P2 and Pi3 is 200mA v 220mA and that you think the increase may be due to the Wifi/BT chip.

I clicked on your link to the USB-current meters. Most seem to be 3 digit, with only 10mA resolution. To accurately measure an approx? 20mA 'difference' between Pi2 and Pi3 idle current I would have thought a current meter with 1mA resolution/accuracy ?

I am not familiar with the BCM4348 WiFi/BT chip, could not find a datasheet for it (nearest on broadcom site is BCM4342), but if it indeed takes about 20mA idle, then unlike some of the higher power chips on the Pi, I suppose there will be no means of depowering it.

It would be interesting to know that the 'difference' between Pi2 and Pi3 is when both run at the same, or even lower, than the default 900MHz SoC Frequency of the Pi2, in order to get a better energy comparison between the BCMBCM2836 and BCM2837.

It would also be interesting if you were able to make some WiFi speed vs range tests, perhaps using similar 5, 10, 20 meter ranges used by a number of Router comparison sites.

After writing then login to paste,, noticed Liliputing with link to your benchmark gave wrong part no for the WiFi chip, should be bcm43438 so may search later for datasheet on current consumption

RE: A Comprehensive Raspberry Pi 3 Benchmark

jack.chaney56 — Mon, 29 Feb 2016 19:52:00 GMT

Would really like to see if the processor speed boost improves any of the video throughput, or if the limiting factor is still the I/O channel.

Jack

RE: A Comprehensive Raspberry Pi 3 Benchmark

DAB — Mon, 29 Feb 2016 19:29:11 GMT

Very nice summary.

Your numbers are about what I expected. I did notice that the memory was the same speed, so that was no surprise.

I think your biggest benefit would come if you are doing a lot of extended math calculations. The 64Bits give you a much better accuracy and the math hardware should help speed things up.

If you are going to do a lot of IO, then I suspect that the RPi 2 B would do just fine.

DAB

RE: A Comprehensive Raspberry Pi 3 Benchmark

m.ratcliffe — Mon, 29 Feb 2016 13:53:35 GMT

Great Summary!! Thanks for the write up.

RE: A Comprehensive Raspberry Pi 3 Benchmark

bwelsby — Mon, 29 Feb 2016 10:55:13 GMT

Thanks for the testing, how warm does the cpu get especially when it's being thrashed [View:/resized-image/__size/16x16/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-a4b9a967-b441-4ff2-989b-9f2e693ef015/contentimage_5F00_1.png:16:16]

RE: A Comprehensive Raspberry Pi 3 Benchmark

sarunaszx — Mon, 29 Feb 2016 08:34:19 GMT

LoL, and I ordered Pi 2 just a week ago.. What a loss [View:/resized-image/__size/16x16/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-a4b9a967-b441-4ff2-989b-9f2e693ef015/contentimage_5F00_2516.png:16:16] built in WiFi is a great addition IMO.

RE: A Comprehensive Raspberry Pi 3 Benchmark

ipv1 — Mon, 29 Feb 2016 07:58:55 GMT

The extra horsepower and BLE support makes it completely worth it...