Re: [Gems-users] question about the interconnection network part of ruby

[Niket <niketa@xxxxxxxxxxxxx>]

If you want to create throttles from your Node(L1Cache, etc) into the
network, you can link it to a Switch that has one outport and that is
linked to the actual network switch. But, you will incur additional
latency for this extra hop.

An example for a 2 node is -

ext_node:L1Cache:0 int_node:0 link_latency:1 bw_multiplier:64
ext_node:L1Cache:1 int_node:1 link_latency:1 bw_multiplier:64

int_node:0 int_node:2 link_latency:1 bw_multiplier:64
int_node:1 int_node:3 link_latency:1 bw_multiplier:64
int_node:2 int_node:3 link_latency:1 bw_multiplier:64

-Niket

Mike Marty wrote:
> ?国? wrote:
>   
>>  Hi everyone,
>>
>> I set my simulation system as a CMP with 4 processors on a single chip, the
>> interconnection network topology as file-specified using the default file.
>>
>> Now, I have the following questions about the GEMS system.
>>
>> 1. I found that there is a throttle from switch to L1 cache or whatever
>> components, but there is no throttle from L1cache(L2cache and Directory) to
>> switch. Any reason or explanation about this?
>>
>>   
>>     
> Throttles are only created on out-ports
>
>   
>> 2. Now, ruby considers infinite message buffer for throttle and switch, can
>> we set a certain value for the messagebuffer by calling
>> MessageBuffer::setSize(newsize)？
>>
>>   
>>     
> There is some experimental support for finite buffers. See the
> FINITE_BUFFERING variable in the rubyconfig.defaults file (and elsewhere).
>
> --Mike
>
> _______________________________________________
> 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.
>
>