Dear Pl group,
This week we have Nilesh Mahajan talking to us about optimization techniques
for reducing communication overheads. Details follow:
Roshan
------------
Title:
Communication overhead reduction using static analysis
Speaker:
Nilesh Mahajan
When and Where:
Friday, LH101, 4:00pm
Abstract:
Communication related optimizations form an interesting class of
compiler optimizations. Data-flow analysis can be used to determine
efficient placement of communication primitives (e.g. send, receive).
In this talk we will see how data flow equations can be set up to
perform various optimizations like communication-computation overlap,
message coalescing, vector message pipelining. I will start with a
brief introduction to data-flow analysis, proceed to explain the
data-flow equations required for the optimizations. I will also talk
about different approaches to verify these analyses. Finally I would
like to discuss the efficacy of these optimizations and whether
similar analyses can be applied to memory hierarchy.
This sounds interesting, but I'm away. Could someone record it so I can
view it later?
On 09/30/2010 11:29 AM, Roshan James wrote:
> Dear Pl group,
>
> This week we have Nilesh Mahajan talking to us
> about optimization techniques for reducing communication overheads.
> Details follow:
>
> Roshan
>
> ------------
>
>
>
> Title:
> Communication overhead reduction using static analysis
>
> Speaker:
> Nilesh Mahajan
>
> When and Where:
> Friday, LH101, 4:00pm
>
> Abstract:
> Communication related optimizations form an interesting class of
> compiler optimizations. Data-flow analysis can be used to determine
> efficient placement of communication primitives (e.g. send, receive).
> In this talk we will see how data flow equations can be set up to
> perform various optimizations like communication-computation overlap,
> message coalescing, vector message pipelining. I will start with a
> brief introduction to data-flow analysis, proceed to explain the
> data-flow equations required for the optimizations. I will also talk
> about different approaches to verify these analyses. Finally I would
> like to discuss the efficacy of these optimizations and whether
> similar analyses can be applied to memory hierarchy.
>
>
>
> _______________________________________________
___________________________________________________
Posted on the PL-wonks mailing list. Go to http://www.cs.indiana.edu/mailman/listinfo/pl-wonks to subscribe.
Sorry Oct 1st, Friday (not the 2nd).
Roshan
On Thu, Sep 30, 2010 at 11:29 AM, Roshan James <> wrote:
> Dear Pl group,
>
> This week we have Nilesh Mahajan talking to us
> about optimization techniques for reducing communication overheads. Details
> follow:
>
> Roshan
>
> ------------
>
>
>
> Title:
> Communication overhead reduction using static analysis
>
> Speaker:
> Nilesh Mahajan
>
> When and Where:
> Friday, LH101, 4:00pm
>
> Abstract:
> Communication related optimizations form an interesting class of
> compiler optimizations. Data-flow analysis can be used to determine
> efficient placement of communication primitives (e.g. send, receive).
> In this talk we will see how data flow equations can be set up to
> perform various optimizations like communication-computation overlap,
> message coalescing, vector message pipelining. I will start with a
> brief introduction to data-flow analysis, proceed to explain the
> data-flow equations required for the optimizations. I will also talk
> about different approaches to verify these analyses. Finally I would
> like to discuss the efficacy of these optimizations and whether
> similar analyses can be applied to memory hierarchy.
>
>
I have recorded the talk. The link to the recording can be found here:
http://lambda.soic.indiana.edu/talks
-m
On Thu, Sep 30, 2010 at 11:31, Michael D. Adams <> wrote:
> This sounds interesting, but I'm away. Could someone record it so I can
> view it later?
>
> On 09/30/2010 11:29 AM, Roshan James wrote:
>> Dear Pl group,
>>
>> This week we have Nilesh Mahajan talking to us
>> about optimization techniques for reducing communication overheads.
>> Details follow:
>>
>> Roshan
>>
>> ------------
>>
>>
>>
>> Title:
>> Communication overhead reduction using static analysis
>>
>> Speaker:
>> Nilesh Mahajan
>>
>> When and Where:
>> Friday, LH101, 4:00pm
>>
>> Abstract:
>> Communication related optimizations form an interesting class of
>> compiler optimizations. Data-flow analysis can be used to determine
>> efficient placement of communication primitives (e.g. send, receive).
>> In this talk we will see how data flow equations can be set up to
>> perform various optimizations like communication-computation overlap,
>> message coalescing, vector message pipelining. I will start with a
>> brief introduction to data-flow analysis, proceed to explain the
>> data-flow equations required for the optimizations. I will also talk
>> about different approaches to verify these analyses. Finally I would
>> like to discuss the efficacy of these optimizations and whether
>> similar analyses can be applied to memory hierarchy.
>>
>>
>>
>> _______________________________________________
___________________________________________________
Posted on the PL-wonks mailing list. Go to http://www.cs.indiana.edu/mailman/listinfo/pl-wonks to subscribe.
