Difference between revisions of "Staff Priorities - David"

From SCECpedia
Jump to navigationJump to search
Line 1: Line 1:
 
== Yesterday ==
 
== Yesterday ==
  
- CVMS4.22 Evaluation<br />
+
- CVM-S4.22 Evaluation<br />
&nbsp;&nbsp;&nbsp;&nbsp;- Tried perftools, but requires use of Cray compiler<br />
+
&nbsp;&nbsp;&nbsp;&nbsp;- Worked on locating source of slowdown in CVM-S<br />
&nbsp;&nbsp;&nbsp;&nbsp;- Found, sort of, source of slow down<br />
+
&nbsp;&nbsp;&nbsp;&nbsp;- Issue is checking to see if lat, lon is within surface boundaries<br />
&nbsp;&nbsp;&nbsp;&nbsp;- cvms_query given array of 10000 points is 1 second<br />
+
&nbsp;&nbsp;&nbsp;&nbsp;- If we query depth-first, then we don't need this re-check<br />
&nbsp;&nbsp;&nbsp;&nbsp;- for loop calling cvms_query on 10000 points is 5 seconds<br />
+
&nbsp;&nbsp;&nbsp;&nbsp;- If we query different lat, longs this is necessary and time-consuming<br />
&nbsp;&nbsp;&nbsp;&nbsp;- question is why?<br />
+
&nbsp;&nbsp;&nbsp;&nbsp;- Met briefly with PM, decided need to modify UCVM<br />
&nbsp;&nbsp;&nbsp;&nbsp;- Two options - modify UCVM code or block in CVM-S?<br />
+
&nbsp;&nbsp;&nbsp;&nbsp;- Began work on structuring CVM-S4.22 to allow UCVM to call with many pts<br />
- CVM-H Fortran API<br />
 
- CyberShake call<br />
 
  
 
== Today ==
 
== Today ==

Revision as of 22:56, 14 June 2013

Yesterday

- CVM-S4.22 Evaluation
    - Worked on locating source of slowdown in CVM-S
    - Issue is checking to see if lat, lon is within surface boundaries
    - If we query depth-first, then we don't need this re-check
    - If we query different lat, longs this is necessary and time-consuming
    - Met briefly with PM, decided need to modify UCVM
    - Began work on structuring CVM-S4.22 to allow UCVM to call with many pts

Today

- CVM-S4.22 Evaluation
    - Attempt to delve into Fortran
    - See if there are initialization routines left in the cvms_query code
    - If so can we remove those?

Blocked

- No

Followups

- No