How to run two different task parallely on two standalone processors on Zedboard

I am new to Zed board.I want to run two different task on two different standalone processors without load any OS on those processors.Like one processor will do sum of two integers and other processor will do multiply two nos. at a same time in Simple C program.suggest me some solution.Thank you.

0 ejubenville over 12 years ago

Take a look at appnote 1079. It address bare-metal applications running independently on both Cortex A9 processors on the ZC702 development board. I seem to recall seeing instructions to adapt that example for a ZedBoard.

http://www.xilinx.com/support/documentation/application_notes/xapp1079-amp-bare-metal-cortex-a9.pdf
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Former Member over 12 years ago

Also check out the latest information wiki.
http://www.wiki.xilinx.com/XAPP1079+Latest+Information

What you want to do is actually very simple. Alot more simple than involving Linux.

XAPP 1079 involves writing 2 programs though, and putting them on separate cores. Not sure if that's how you want it...
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Former Member over 12 years ago

Thank you.Please help me in the another scenario like sum of two nos. done in first processor and then result should go to another processor and multiplication it with another integer on second processor.
Example:
First Processor:result=a + b;
Second Processor: result1=result*c;
Can you please how to manage inter processor communication?
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Former Member over 12 years ago in reply to Former Member

Default OCM (On Chip Memory) locations are 0x00000000 to 0x0002FFFF (low 64kB) as well as 0xFFFF0000 to 0xFFFFFFFF (high 64kB). You can see this in the default linker script (lscript.ld) under ram_0 and ram_1. OCM is common to both processors. So you could store your result in OCM and then both processors could access it. You'll need to explicitly identify what address you want to store your variable at. So do something like this (in both CPU0's and CPU1's source code):

volatile u32 *result = (volatile u32 *)(0xFFFF0000);
CPU0: *result = a + b;
// set flag telling cpu1 that first operation has completed
CPU1: result1 = *result * c;

or do something simpler with a #define
#define RESULT *(volatile u32 *)(0xFFFF0000)
CPU0: RESULT = a + b;
// set flag telling cpu1 that first operation has completed
CPU1: result1 = RESULT * c;

In both cases, you'll have to have a flag to communicate between the two processors (which will also need to be common to both cores, in OCM), telling each processor when the other has completed its task. If you try and operate on the same memory location with both CPUs at the same time, results will be unpredictable and probably undesirable.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Former Member over 12 years ago in reply to Former Member

Thank you.This will be helpful.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Former Member over 11 years ago

I implemented and run xapp1079 but when I am trying to run it from flash memory,it is not load cpu1 application.Is it possible to run xapp 1079 from flash?
A application on cpu0 is running from flash but from cpu1 it can't.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Former Member over 11 years ago

Hi,

CPU 0 willboot up first,then it will awak cpu 1 by SEV
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel