Re: [linux-sunxi] performance difference between two A20 boards how to debug?

2014-03-25 Thread Olliver Schinagl 

On 03/25/2014 07:27 PM, Rajesh Mallah wrote:


Hi ,

i am using tinymembench to bench mark 2 A20 based 1GB boards
MELE finished in 6mins and the
noname one finishes in 12mins.

pls tell how to systematically debug the issue.

at least dram settings are same. where else should i see?
The mbus can be different, this is a compiletime option for u-boot. Then 
again, mbus changing might not work on a10 iirc.


But try using the same bootloader for a fair comparison ...

Olliver



MELE:

root@ltspmele:~# ./a10-meminfo-static
dram_clk  = 408
dram_type = 3
dram_rank_num = 1
dram_chip_density = 4096
dram_io_width = 16
dram_bus_width= 32
dram_cas  = 9
dram_zq   = 0x7f
dram_odt_en   = 0
dram_tpr0 = 0x42d899b7
dram_tpr1 = 0xa090
dram_tpr2 = 0x22a00
dram_tpr3 = 0x0
dram_emr1 = 0x4
dram_emr2 = 0x10



NONAME:

root@ltspenjoy:~# ./a10-meminfo-static
dram_clk  = 408
dram_type = 3
dram_rank_num = 1
dram_chip_density = 4096
dram_io_width = 16
dram_bus_width= 32
dram_cas  = 9
dram_zq   = 0x79
dram_odt_en   = 0
dram_tpr0 = 0x42d899b7
dram_tpr1 = 0xa090
dram_tpr2 = 0x22a00
dram_tpr3 = 0x0
dram_emr1 = 0x4
dram_emr2 = 0x10
dram_emr3 = 0x0

--
You received this message because you are subscribed to the Google
Groups linux-sunxi group.
To unsubscribe from this group and stop receiving emails from it, send
an email to linux-sunxi+unsubscr...@googlegroups.com
mailto:linux-sunxi+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.


--
You received this message because you are subscribed to the Google Groups 
linux-sunxi group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to linux-sunxi+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [linux-sunxi] performance difference between two A20 boards how to debug?

2014-03-25 Thread Siarhei Siamashka 
On Tue, 25 Mar 2014 23:57:22 +0530
Rajesh Mallah mallah.raj...@gmail.com wrote:

 Hi ,
 
 i am using tinymembench to bench mark 2 A20 based 1GB boards
 MELE finished in 6mins and the
 noname one finishes in 12mins.

You can't compare the times of running tinymembench. That's because
the amount of work they do may differ! The number of repeats for each
individual test increases if they are poorly reproducible. The test
results may be poorly reproducible if you have some other background
activity in the system.

To sum it up: just look at the numbers in the log and don't try to
time the whole program.

And some random advices for getting more stable test environment
with better reproducible results:

1. Ensure that both systems are configured to use exactly the
   same screen resolution. Or blank the display to disable
   framebuffer scanout by running:
  echo 1  /sys/devices/platform/disp/graphics/fb0/blank
  echo 1  /sys/devices/platform/disp/graphics/fb1/blank

2. Check for any possible IRQ spam in /proc/interrupts

3. Run the benchmark as the init process to make sure that
   nothing else is interfering

-- 
Best regards,
Siarhei Siamashka

-- 
You received this message because you are subscribed to the Google Groups 
linux-sunxi group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to linux-sunxi+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.