Supercomputing'95 Ptools Questionnaire on Parallel Tools
Q. 2. Least favorite parallel tool - what do you dislike about it?
General responses are given first, then specific tools in alphabetic order.
none / undecided (12 responses)
no particular tool named (7 responses)
- hard to say
- don't have a
- [8 blank]
programming languages (3 responses)
- Actually any parallel tools are far better than implementing my own.
- Not applicable. Most tools are of some use.
- Nearly everything I've used. They're too hard to learn, too complicated -
and change too frequently. It's almost always hard to do simple things
because the tools are oriented to complex needs.
- All complex graphics/color gadgets. They are too difficult to learn.
Return is small.
- Non-automated parallel tools.
- Lack of a good debugger...
- Parallel debuggers that do little more than bring up multiple dbx windows
for the different processes!
Convex's Partition Manager (1 response)
- C* - too much like C++.
- Linda - slow & machine dependent.
- MPI - just a library without any tool support.
Cray Research's Apprentice (2 responses)
- Not even the Convex tech rep could figure it out.
Cray Research's Tools (1 response)
- Processor aggregate information is nearly USELESS, especially
since you can't obtain load imbalance info, which is usually the dominant
- [no specific comments]
cvperf on the Intel Paragon (1 response)
- Different between T3D and CRO[?]. Why?
IBM's Load Leveler (1 response)
IBM's parallel debugger (1 response)
- [no specific comments] [Note: This is not SGI's cvperf.]
Intel Paragon Port of OSF/1 Operating System (1 response)
MQM - Message Queue Manager (1 response)
- Not as easy to use as something like xcdb.
PDM (1 response)
- Besides source and type also need size info.
Upshot Program Visualization System (1 response)
- My company (T.I.) ships a multiprocess debugger called PDM[?] that is truly
- singlestepping one process does not update other processes
- bug events can be counted but not associated w/ lines of code
- Nupshot without modifications.
Back to Ptools at Oregon State.
Back to Parallel Tools Consortium.
Last updated February 8, 1996.