Jump to content


3 question


  • You cannot reply to this topic
2 replies to this topic

#1 Robin

    Member

  • Members
  • PipPip
  • 7 posts

Posted 17 April 2009 - 12:42 AM

Hi everybody!

I have 3 question, 2 of which is about simulation in modelsim, the last one is about shared memory.
1: I have noticed that Codeveloper doesn't generate the testbench, have we write it myself handly?
If it does, any templates?
2: I am now using Altera's FPGA, and my project using float point, but the files with float library are encrypted by Altera
How to solve these problems if using third-part tools, like: modelsim, synplify?
3: In ImpulseC's document it said: The shared memory is accessed through the use of DMA.
but using function co_memory_ptr in hw files, is it still in the DMA way?
I notice that if use function co_memory_readblock and co_memory_writeblock, it will cost a lot of on-chip memory.

Regards,
Robin

#2 RalphBodenner

    Advanced Member

  • Admin
  • PipPipPip
  • 348 posts

Posted 17 April 2009 - 08:03 AM

Hi Robin,

QUOTE (Robin @ Apr 17 2009, 01:42 AM) <{POST_SNAPBACK}>
1: I have noticed that Codeveloper doesn't generate the testbench, have we write it myself handly?
If it does, any templates?

As of version 3.50, CoDeveloper can generate a ModelSim testbench, using the new CoValidator tool. These testbenches will currently only work with streams, though support for co_memory will be coming soon.

QUOTE (Robin @ Apr 17 2009, 01:42 AM) <{POST_SNAPBACK}>
2: I am now using Altera's FPGA, and my project using float point, but the files with float library are encrypted by Altera
How to solve these problems if using third-part tools, like: modelsim, synplify?

I believe the Altera edition of ModelSim (called ModelSim AE) includes behavioral HDL for the floating-point cores that can be simulated. This edition is available from Altera.

QUOTE (Robin @ Apr 17 2009, 01:42 AM) <{POST_SNAPBACK}>
3: In ImpulseC's document it said: The shared memory is accessed through the use of DMA.
but using function co_memory_ptr in hw files, is it still in the DMA way?
I notice that if use function co_memory_readblock and co_memory_writeblock, it will cost a lot of on-chip memory.

The co_memory_ptr function lets you do word-size transactions, which will most likely not be as efficient as block transfers. Block transfers, done with co_memory_*block, are generally more efficient, but as you recognize, they require an on-chip block RAM buffer.

Regards,
Ralph
Ralph Bodenner
Impulse Accelerated Technologies, Inc.

#3 Carlos

    Newbie

  • Members
  • Pip
  • 2 posts

Posted 11 November 2009 - 04:24 AM

How can i get the CoValidator tool??

I requested a 30-day software of that tool, but i have no answer yet





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users