News:

Attention: For security reasons,please choose a user name *different* from your login name.
Also make sure to choose a secure password and change it regularly.

Main Menu

Power consumption of xczu3eg-sfvc784-1-e Vs. xczu2cg-sfvc784-1-e

Started by narasapur, January 17, 2020, 06:17:41 PM

Previous topic - Next topic

narasapur

Dear Forum,

We use MPSoC-Module with Xilinx Zynq UltraScale+ for our application. We currently have two variants of the same board which have different SoC, namely xczu3eg-sfvc784-1-e and xczu2cg-sfvc784-1-e.

The variant xczu2cg has fewer FPGA resources in comparison to xczu3eg. Therefore, in theory, the xczu2cg is supposed to consume less power than the xczu3eg. The Xilinx Power Estimator tool is also suggesting the same.

However, this is not the case if we are measuring the power using a multimeter while the hardware is running. For our FPGA design, the xczu2cg is consuming more power than xczu3eg!

This does not make any sense. Any ideas so as to what could be happening?

Best regards,
Narahari

JH

Hi,
I would expect more or less the same as long as you did use manly default configurations. And in case you didn't use cooling solution, it will be also increase during the time....
Power consumption depends mainly on your design and your cooling solution.


On your power estimator, did you use the same conditions and logic resources? In this case I think, it should be not really change.


Can you tell me the exact article number of your both modules you try to compare?

There are a lot of mechanism to reduce power for ZynqMP, Xilinx has some white paper about this topic:
https://www.xilinx.com/support/documentation/white_papers/wp482-zu-pwr-perf.pdf

br
John



narasapur

Hi John,

Thank you very much for your reply.

For the power estimator, I exported the .xpe file from Vivado for both the designs(built for different SoC variants) and imported this file in the XPE (Excel-based tool). So, except for the design, other parameters are identical for both the boards.

About the cooling solution, we do have an identical cooling solution for both the boards.

According to XPE  (Excel-based tool), xczu2cg is consuming almost 0.8W less power than xczu3eg! The results are almost the same in the Vivado in-built power estimator.
But in reality, xczu2cg  seems to be consuming at least 0.5 W more than xczu3eg.

Below you will find the article number of the boards I am comparing
Article number for xczu3eg-sfvc784-1-e : TE0820-03-03EG-1EA
Article number for xczu2cg-sfvc784-1-e : TE0820-03-2AE21FA

If you need any further information, please let me know!

Thank you.

Best regards,
Narahari






narasapur

Hi There,

Is there any update on this issue?

Thank you very much!

Regards,
Narahari

JH

Hi,
can you tell me which carrier you use?



QuoteFor the power estimator, I exported the .xpe file from Vivado for both the designs(built for different SoC variants) and imported this file in the XPE (Excel-based tool). So, except for the design, other parameters are identical for both the boards.
Do you use the same PS configuration and also same PL design?

Quotexczu2cg  seems to be consuming at least 0.5 W more than xczu3eg.
always 0.5W more or only  if some design is running?

TE0820-03-03EG-1EA and TE0820-03-4AE21FA  has different eMMC (4GB (MTFC4GACAJCN-4M IT) vs 8GB (S21ES08G-JCLI))
https://shop.trenz-electronic.de/en/Download/?path=Trenz_Electronic/Modules_and_Module_Carriers/4x5/TE0820/REV03/Documents
--> www.trenz-electronic.de/fileadmin/docs/Trenz_Electronic/Modules_and_Module_Carriers/4x5/TE0820/REV03/Documents/SCH-TE0820-03-03EG-1EA.PDF
-- > http://www.trenz-electronic.de/fileadmin/docs/Trenz_Electronic/Modules_and_Module_Carriers/4x5/TE0820/REV03/Documents/SCH-TE0820-03-4AE21FA.PDF

maybe this makes the different.


br
John