Dear Choi,

I have the same results with you except from Runtime Dynamic. As Andreas said, you may need to check the stats.txt file and the GEM5toMcPAT.py python script.


Best regards,
Nikos


Quoting Andreas Brokalakis via gem5-users <gem5-users@gem5.org>:

Dear Choi,
    
there seems to be a problem with your results. If you see the Runtime Dynamic results, it mentions that they are either inf or nan. This means that in the xml file that was produced from the python script, the runtime statistics have not been written properly. As such McPAT does not compute the dynamic power consumption based on the runtime statistics produced by the gem5 simulation. It only computes a maximum power draw (something like a TDP value instead of a power value related to the program that you are actually running).
    
To correct this error you need to check the stats.txt file and the python script to see whether the version of gem5 that you are using somehow has modified the syntax or structure of the stats.txt file and therefore you need to account for these changes when you parse it with the python script.
    
   Best regards,
   Andreas
    
    

On Tue, May 9, 2023 at 7:35 PM 최순원 via gem5-users <gem5-users@gem5.org> wrote:

Dear Νικόλαος Ταμπουρατζής,
      
Your description is perfect. I was able to get "mcpatOutput.txt" without any problem like this.

      
     If you don't mind, can I ask you two more questions?
      
1) How can I get the total power consumption of the processor from the output file information?

2) Can I see the your "mcpatOutput.txt" file? I want to compare with my output file.
      
     Thanks for your time.
      
     Best regards,
     Choi

2023년 5월 10일 (수) 오전 12:56, Νικόλαος Ταμπουρατζής <ntampourat...@ece.auth.gr>님이 작성:

_Dear Soonwon,_
       _ _
       _I had used McPAT v1.3. Specifically, you can do the following:_
       _ _
       _1) git clone https://github.com/H2020-COSSIM/cMcPAT.git
 _
       _2) cd cMcPAT/mcpat_

        
       _3) make all
 _
       _4) cd cMcPAT/Scripts (here you can see the GEM5ToMcPAT.py).
Usage: GEM5toMcPAT.py [options] <gem5 stats file> <gem5 config file (json)> <mcpat template xml file> -o <output xml file>
 _
_Example GEM5ToMcPAT.py $GEM5/m5out/stats.txt $GEM5/m5out/config.json cMcPAT/mcpat/ProcessorDescriptionFiles/ARM_AtomicSimpleCPU_template.xml -o cMcPAT/mcpatNode.xml
 _
_To be noticed to replace the correct of the above paths and the mcpatNode.xml will be produced._

        
_5) Run McPAT using the mcpatNode.xml from step 4: cMcPAT/mcpat/mcpat -infile cMcPAT/mcpatNode.xml -print_level 1 > cMcPAT/mcpatOutput.txt_
       _ _
       _You may use print_level from 1 to 5 (more detailed output).

In the above example I have used a generic ARM_AtomicSimpleCPU_template.xml an I tested using the latest gem5 version (v22.1). You may change it according your ARM processor._

       _Best regards,
Nikos

Quoting 최순원 via gem5-users <gem5-users@gem5.org>:_

_Hello, I read your question, “RISCV Power/Energy Estimation”.

I’m sorry because It is not a answer to your question.

You said that “I got power/energy estimations of ARM and X86 ISAs through McPAT.”.
I want to measure a ARM core’s power consumption.
I got the gem5 simulations results, “stats.txt”, but I can’t see the power consumption of cpu.

Can you tell me how you got the CPU power consumption?

Best regards,
Soonwon Choi[1]_


 

________________________________________________
gem5-users mailing list -- gem5-users@gem5.org
To unsubscribe send an email to gem5-users-leave@gem5.org_



Links:
------
[1]
_______________________________________________
gem5-users mailing list -- gem5-users@gem5.org
To unsubscribe send an email to gem5-users-le...@gem5.org
  • [gem5-users] RISC... Νικόλαος Ταμπουρατζής via gem5-users
    • [gem5-users]... 최순원 via gem5-users
      • [gem5-us... Νικόλαος Ταμπουρατζής via gem5-users
        • [gem... 최순원 via gem5-users
          • ... Andreas Brokalakis via gem5-users
            • ... Νικόλαος Ταμπουρατζής via gem5-users
              • ... 최순원 via gem5-users

Reply via email to