[Gems-users] 答复: Average packet latency VS. Ruby cycles

["fubinzhang" <fubinzhang@xxxxxxxxx>]

Miray,

 

Thanks for your reply.

 

I insert MAGIC_BREAKPOINT after MEMSYS_ON and MEMSYS_OFF

 

At first breakpoint, I clear the ruby stats, and dump the stats at second breakpoint.

 

I think they run the same phase. Is it right?

 

发件人: gems-users-bounces@xxxxxxxxxxx [mailto:gems-users-bounces@xxxxxxxxxxx] 代表 Miray Kas
发送时间: 2010年8月25日 12:48
收件人: Gems Users
主题: Re: [Gems-users] Average packet latency VS. Ruby cycles

 

Are you running the application to completion? I mean, are you sure that you are running the same phase of the application, or how do you ensure that?

 

Miray

On Tue, Aug 24, 2010 at 11:04 PM, fubinzhang <fubinzhang@xxxxxxxxx> wrote:

Hi all,

       Recently, we simulate a new routing algorithm on GEMS with splash2,
and I get puzzled about the results.

For example, for application LU-non-contiguous, our routing algorithm
reduces the average packet latency by 7% compared with the default xy
routing.

However, the ruby cycles is reduced as much as 11%.

I remembered Dan has pointed out the performance uncertainty of parallel
applications. So, I run this application twice, but got the similar results.

How does this happen? Can 7% packet latency reduction lead to 11% execution
time reduction? Is it reasonable?

Thanks in advance.


P.S.
Network Topology: 4x4 mesh
Cache Protocol: MSI_MOSI_CMP_directory
VC per VN: 1
VC buffers: 4
Flow control: We modify the flow control to set an output unit free as soon
as the tail flit has been sent. Then, the VCallocator selects an output unit
once it is free, as well as it has credit.


Regards,

Fbz.

_______________________________________________
Gems-users mailing list
Gems-users@xxxxxxxxxxx
https://lists.cs.wisc.edu/mailman/listinfo/gems-users
Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/" to your search.