OCP-IP
Functional Verification Working Group
Meeting Minutes
July 19, 2007
Attendees:
Steve McMaster – CHAIR – Synopsys
Ravi Venugopalan – Sonics
Brian Sage - VTM
Action Item Review:
Meeting Detail:
Steve called the meeting to order at 8:05 A.M. PST. Quorum was not established.
Steve asked Ravi if he had yet had a chance to review the performance chapter. Ravi replied
that he had been busy and therefore hadn’t done so, but would try to read it within the next two
weeks and get back to Steve afterwards. Steve asked Brian if the performance material had
been distributed to the Specification Working Group (SWG). Brian replied that he had sent the
material to Drew Wingard, chairman of the SWG. Steve said he had sent the performance
material to Jeroen, but had not yet heard back from him.
Steve said a discussion on throughput needed to take place as it had never been discussed as
much as latency. Ravi said that provisions are needed for measuring throughput and asked
Steve if he thought there were enough guidelines to measure throughput. Steve replied that he
did not think there were any guidelines yet. Steve described throughput as how many
transactions went through the system, where the delay lies between two measurement points.
Steve said they had covered latency or delay, but had not covered throughput or the transmission
of some quantity over some unit of time. Steve continued by saying they had not identified
anything to measure in terms of throughput. Steve said they could either decide not to cover
throughput and instead put a boiler plate paragraph in the write up or they needed to come with a
strategy on how to cover throughput. Ravi said he did not have much experience in this matter
and so therefore did not have much input and suggested that other people were needed to help
guide this issue.
Ravi said another topic that needed to be discussed was in regards to a transactional level
interface working between SystemVerilog and SystemC. Ravi said Vikas Grover (Synopsys) had
showed him an example of how this could be done using PLI . Ravi said it would be beneficial for
Vikas to work with the OCP-IP community who could use this type of PLI solution for verification.
Steve asked if Ravi was basically looking for a PLI interface on top of SystemVerilog for TL1 and
TL2. Ravi said they need some sort of interaction from the SystemC verification test bench to
high level abstract models possibly available in SystemVerilog. Ravi stated that Texas
Instruments thinks this is the way to go. Ravi suggested that VIkas could donate this to OCP-IP.
Ravi said Tom Borgstrom (Synopsys) had been pushing this type of PLI interface for some time.
Ravi wondered if they could leverage that experience in order to make these models available to
OCP-IP. Steve said this issue arose when he was on leave and it was pushed to his manager,
but he would check up on it. Steve said he did not completely understand what they have and this
could end up in the System Level Design Working Group (SLD WG) instead of the Functional
Verification Working Group (FVWG). Steve said if that happens the FVWG could offer some
assistance to the SLD WG.
OCP-IP CONFIDENTIAL
Functional Verification WG Meeting Minutes (7/19/2007)
Page 1 of 1
Ravi reiterated that he will review the performance material within the next two weeks. Steve said
he would continue to think about the throughput issue and then write a paragraph on it and send
it to the reflector for further discussion.
nd
The next Functional Verification WG meeting is scheduled for August 2 at 8:00 A.M. PST.
The meeting ended at 8:23 A.M. PST.
OCP-IP CONFIDENTIAL
Functional Verification WG Meeting Minutes (7/19/2007)
Page 2 of 1