OMAP L138 PRU sleep

Nigel Paton

Other Parts Discussed in Thread: OMAPL138

Hi,

From what I've read in other forum posts the PRU silicon version in the OMAP L138 is 1 which means the sleep instruction cannot be used. Does this mean that sleep in this version of the PRU is not supported even by directly writing to the 'SLEEPING' bit in the Control register (doesn't seem to work when I try it)?

If PRU sleep is not possible will halting it (HALT instruction or ENABLE bit in Control register) have any affect on power consumption?

Thanks

Nigel

over 9 years ago

0 Sivaraj Kuppuraj over 9 years ago

TI__Mastermind 35645 points

Hi,

Thanks for your post.

The PRU information is given at this wiki site:

http://wiki.davincidsp.com/index.php/Programmable_Realtime_Unit_Subsystem

Please walkthrough the below E2E post which would provide you more insight on deep sleep:

https://e2e.ti.com/support/dsp/omap_applications_processors/f/42/t/103695

Also, the training slides are available at the below:

http://processors.wiki.ti.com/images/1/18/PRUSS_Training_Slides.pdf

To my knowledge, only way to wake from deep sleep is via deep sleep pin or RTC ( mutually exclusive)

Thanks & regards,

Sivaraj K

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

Please click the Verify Answer button on this post if it answers your question.

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

0 Mukul Bhatnagar over 9 years ago in reply to Sivaraj Kuppuraj

TI__Guru* 83965 points

Hi Nigel
I do think SLP instruction had an issue , but also thought there was a PASM release that had a workaround for it. It has been long since this came up. W If I may ask, where did you see the issue with SLP instruction documented.

I am guess HALT should be fine too. If your concern is on power consumption , I don't think the overall PRUSS power consumption is that high to begin with

processors.wiki.ti.com/.../OMAP-L138_Power_Consumption_Summaryl
You can look at the PRUSS row, the internal module clock runs CPU/2 , and if you use 456 MHz, you will see that the power saving for pru idle vs pru active doing nothing is ~ 16 mW. I think the IDLE to sleep power will be very negligible.

Regards
Mukul

0 Nigel Paton over 9 years ago in reply to Mukul Bhatnagar

Intellectual 320 points

Hi Mukul,

Thanks for the response.

I've tried setting the SLEEPING bit in the control register of the PRU directly and it seems to have no effect.

I've also tried using the 'SLP' assembler command as documented here:

processors.wiki.ti.com/.../PRU_Assembly_Instructions

but the assembler just complains about unknown instruction (we have silicon version set to 1). I think this post:

e2e.ti.com/.../403649

indicates that the OMAPL138 has version 1 of the PRU and SLP is only supported from version 2 onwards.

Do you know what PRU IDLE is defined as in the power spreadsheet (is this the HALT state)?

Thanks

Nigel

0 Mukul Bhatnagar over 9 years ago in reply to Nigel Paton

TI__Guru* 83965 points

Thanks. What versions of the tools are you using? I do believe SLP is broken in the OMAPL138 , but I remember there was a workaround, but I don't have it handy.
The PRU IDLE is being profiled using PRU NOP instructions , the difference between NOP vs PRU clock cut off via PSC was around 2mA at 150 MHz PRU at 1.2V core - so again, the power saving should be negligible for NOP vs SLP instruction.
Hope this helps.
Regards
Mukul

0 Nigel Paton over 9 years ago in reply to Mukul Bhatnagar

Intellectual 320 points

Hi Mukul,

We are using PRU C/C++ Compiler v2.1.1.

Thanks

Nigel

Processors

Processors forum

OMAP L138 PRU sleep