Spring Training
SSCC's Spring training schedule includes the usual introductions to Stata and R (perfect for anyone starting out in research or looking to learn a second statistical language) plus some new classes:
- Getting More from Stata Graphics will teach you how to make better graphs, whether you're a graphics newbie or a veteran.
- Running Large Statistical Jobs on Linstat or HTCondor is designed for those who know they'd benefit from the computing power of SSCC's Linstat servers or HTCondor flock, but are intimidated by Linux. SSCC's integrated computing environment makes it a lot easier than you think.
- Working with String (Text) Data in Stata will teach you the most important tools for working with string variables in Stata and how to use them to carry out common tasks.
Reserve SSCC's Computer Labs for Your Class Now
Just a reminder that if you plan to use SSCC's Computer Classroom (3218 Sewell Social Sciences Building) or the Mobile Lab for a class you teach, you must reserve them through SSCC, and soon, or you're likely to find another class there when you arrive. Reservations made are on a first-come first-served basis.
Tip: Use the Right Number of Cores for Parallel Processing
Many programs that use parallel processing will default to creating one thread or process for each core it thinks your computer has. Unfortunately, they'll usually get the number of cores wrong on SSCC's servers because our servers use hyperthreading. It's not a big problem, but it will make your programs run a little slower than they could.
Here are the number of cores you should request for each of SSCC's servers:
| Winstat | 16 |
| Winstat Long Jobs | 16 |
| Linstat1-4 | 16 |
| Linstat5 | 36 |
| Condor1-4 | 8 |
More information about SSCC's servers can be found in Computing Resources at the SSCC. Note that Stata MP16 will always use a maximum of 16 cores, but Stata jobs run on the Condor servers will benefit from being told to use 8 cores.
If a server has more jobs asking for cores than are actually available, Windows or Linux will make sure they all get their fair share. Do not run more than one job per server, but feel free to have your job ask for all the server's cores—we don't want them sitting idle! If you notice a server running slower than usual it's almost always because people are using a lot of memory or disk I/O, not CPU time. Logging out and logging back in will move you to the least busy server.
RSS |