Slow down Rpi 2 to work at same speed as Rpi B+

[cmarty]

Hello.
Sounds strange. But is there a way how to slow down the Rpi 2 (cpu frequency, disable quad cores, get rid of 512mb of ram) to work at speed similar to Rpi B+.
I have project playing several video files and switching between them and Rpi 2 is visibly faster in that than old Rpi B+.
In some cases I need all Rpi (Pi 2 and B+) to work at the same speed.
Thanks for any help.

[DougieLawson]

Try

Code: Select all

isolcpus=1,2,3

in /boot/cmdline.txt

Top shows:

top - 00:00:49 up 0 min, 1 user, load average: 0.60, 0.18, 0.06
Tasks: 93 total, 1 running, 92 sleeping, 0 stopped, 0 zombie
%Cpu0 : 0.7 us, 0.3 sy, 0.0 ni, 99.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu1 : 0.0 us, 0.0 sy, 0.0 ni,100.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu2 : 0.0 us, 0.0 sy, 0.0 ni,100.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu3 : 0.0 us, 0.0 sy, 0.0 ni,100.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem: 884388 total, 108212 used, 776176 free, 13244 buffers
KiB Swap: 511996 total, 0 used, 511996 free, 56356 cached

It's still 900/700 times quicker than a A/B/B+/A+ (because of the increased CPU frequency) but closer to what you want.

[W. H. Heydt]

Why not just get another B+? They're not only available, but several companies have started discounting them since the Pi2B came out.

[mattmiller]

That'll save a bit of hardware unplugging and pluggin during testing :0

Is there a similar setting to restrict RAM that can easily be altered as well?

[cmarty]

I have try isolcpus=1,2,3 in config.txt do the reboot and no change
top says:
top - 08:45:08 up 10 min, 1 user, load average: 0.71, 0.38, 0.19
Tasks: 83 total, 1 running, 82 sleeping, 0 stopped, 0 zombie
%Cpu0 : 1.0 us, 4.5 sy, 0.0 ni, 81.7 id, 12.8 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu1 : 1.0 us, 17.6 sy, 0.0 ni, 81.4 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu2 : 8.1 us, 4.4 sy, 0.0 ni, 78.7 id, 8.8 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu3 : 0.3 us, 2.0 sy, 0.0 ni, 97.6 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem: 234660 total, 135652 used, 99008 free, 6148 buffers
KiB Swap: 0 total, 0 used, 0 free, 78916 cached

I also reserved 768M for GPU to have the same amount for the system as in Raspi B+

[ktb]

in /boot/cmdline.txt

[cmarty]

It is my mistake. Sorry for misreading. When I put it into the cmdline.txt it works now.
It is better now. But processor speed is still quicker than Rpi B and this is still visible.
Is there a way how to underclock the 900Mhz cpu to 700Mhz.

[cmarty]

I even tried to setup on Rpi 2 in /boot/config.txt
arm_freq=600
sdram_freq=400
but Rpi2 is still faster than Rpi.
I notice first difference in cache reading from SD card (same SDcards):
On Rpi2 and Rpi is emmc clock same :
vcgencmd measure_clock emmc
frequency(47)=250000000

On Rpi2:

hdparm -t /dev/mmcblk0
/dev/mmcblk0:
Timing buffered disk reads: 54 MB in 3.03 seconds = 17.81 MB/sec

hdparm -T /dev/mmcblk0
/dev/mmcblk0:
Timing cached reads: 554 MB in 2.00 seconds = 276.80 MB/sec

On RpiB+:
hdparm -t /dev/mmcblk0
/dev/mmcblk0:
Timing buffered disk reads: 52 MB in 3.00 seconds = 17.32 MB/sec

hdparm -T /dev/mmcblk0
/dev/mmcblk0:
Timing cached reads: 376 MB in 2.01 seconds = 187.11 MB/sec

Why there is so such difference in cache reads?

[joan]

The bcm2836 has cache which wasn't present in the bcm2835.

In particular the new 512kB L2 cache dedicated to the core.

See http://makezine.com/2015/02/02/eben-upt ... erry-pi-2/ (search for cache).

[DougieLawson]

The bcm2836 has cache which wasn't present in the bcm2835.

In particular the new 512kB L2 cache dedicated to the core.

See http://makezine.com/2015/02/02/eben-upt ... erry-pi-2/ (search for cache).

Could that be crippled by massively over specifying GPU memory?

Code: Select all

gpu_mem=528 # or bigger

in /boot/config.txt

[dom]

"total_mem=512" will hide half the memory.
"arm_freq=700", "sdram_freq=400" will remove the higher frequencies.

However Pi2 will still be a lot faster than Pi1. The large/fast L2 cache is a lot quicker, and clock-for-clock, then newer arm architecture gets more work done each cycle.

[cmarty]

That is true. After slowdown frequency, restricted process to only one core, use the same amount of memory, the new raspi 2 is still more powerfull than old Raspi.(especially in video/image decoding). It is really with help of 512kb L2 cache and use of NEON instructions. I compared it when switching between several video files. The new video file is decoded and showed to user more faster than in previous Raspi.