• Skip to main content
  • Skip to search
  • Skip to footer
Cadence Home
  • This search text may be transcribed, used, stored, or accessed by our third-party service providers per our Cookie Policy and Privacy Policy.

  1. Community Forums
  2. Functional Verification
  3. pull mode vs push mode

Stats

  • Locked Locked
  • Replies 5
  • Subscribers 67
  • Views 18498
  • Members are here 0
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

pull mode vs push mode

onkarkk
onkarkk over 16 years ago

 Hi,

Could any help me out in understanding the difference between pull mode and push mode . I am really not understanding this theory . please help me out.

Thanks in advance,

regards,

Krishna Kishore 

  • Cancel
  • tpylant
    tpylant over 16 years ago

    I will assume you are referring to eRM or OVM generation of transactions.

    Push mode (recommended by other methodologies) refers to the generator creating transactions and pushing them to the BFM or Driver which then uses the transaction to drive the DUT. This has a problem that the generator is creating transactions before the driver requests them and therefore might not reflect constraints that the test is requiring, i.e. not reactive. This results in the possibility that those transactions will be discarded and new transactions having to being generated.

    Pull mode refers to the generator creating transactions in response to the driver requesting them. This is more efficient in that the driver can add additional constraints before requesting the transaction and therefore get only the transactions needed. This is recommended and built-in to eRM/OVM.

    Tim

    • Cancel
    • Vote Up 0 Vote Down
    • Cancel
  • onkarkk
    onkarkk over 16 years ago

     Thankyou Tim,

     Now i understood bit clearly..

    regards,

    Krishna Kishore 

    • Cancel
    • Vote Up 0 Vote Down
    • Cancel
  • rsl80
    rsl80 over 14 years ago

    Hi Tim,

    I was referring to the topic pull vs push mode, I am not clear that how constraint will not reflect that the test is requireng in push mode due to the transaction generated before to driver's request. Consider the sequence started body() will start then will call do action, so item will be generated according to the constraints given then date will be send_to_bfm(). The same case for pull_mode generator will create transaction to the drivers request and this also will call the do action on data as per constraint. So both cases doing the action on data item so how it will differe on constraint. Also where will be the push mode useful.

    Thanks,

    rsl

    • Cancel
    • Vote Up 0 Vote Down
    • Cancel
  • Efrat
    Efrat over 14 years ago
    hi, The main issue here is when does the sequence driver picks the next "do". In Pull mode, this decision is made right before injecting the item. In Push mode - this decision might be made many cycles before the injection. Consider, for example, the case when the driver picks the next "do" from sequence A. Cycles passes, and in the mean time - seq B proceeds (perhaps it was waiting for some event), and now it grabs the driver. But the driver already picked another "do", not the B's. This results with a scenario not matching user's intention. While developing layered sequences examples (some are included in uvm/uvm_examples), we came to realize several use models when Push mode just wouldn't work (or - more accurate - requires too much user code). Because in layered sequences, there is a delicate synchronization between high layer, low layer, their drivers and BFMs. To your question - we came to conclusion that Pull mode is the recommended mode. We yet have not seen a use model where Push is preferred.
    • Cancel
    • Vote Up 0 Vote Down
    • Cancel
  • jadec
    jadec over 14 years ago
    In UVM/OVM sequencers (uvm_push_sequencer vs the uvm_sequencer (pull)), the difference in the kind of TLM connection. A push sequencer has a "port" and pull sequencer has an "export". Because sequencers and drivers have a 2-way handshake, the sequencer will still wait for one transaction to finish before randomizing the next. The reason for using a push would be that if the driver doesn't need a dedicated thread (the sequencer calls the driver) it could be higher performance.
    • Cancel
    • Vote Up 0 Vote Down
    • Cancel

Community Guidelines

The Cadence Design Communities support Cadence users and technologists interacting to exchange ideas, news, technical information, and best practices to solve problems and get the most from Cadence technology. The community is open to everyone, and to provide the most value, we require participants to follow our Community Guidelines that facilitate a quality exchange of ideas and information. By accessing, contributing, using or downloading any materials from the site, you agree to be bound by the full Community Guidelines.

© 2025 Cadence Design Systems, Inc. All Rights Reserved.

  • Terms of Use
  • Privacy
  • Cookie Policy
  • US Trademarks
  • Do Not Sell or Share My Personal Information