(offtopic: Purify and Valgrind) Re: [Coin-discuss] Purify UMR CoinFactorization::pivotOneOtherRow

[Brian Borchers]

In short, Valgrind can detect many errors that Purify can't detect, but
it's also vastly slower in doing so.  
 
I regularly use valgrind to test new versions of CSDP, but it slows
computations down by several orders of magnitude, so I can only run
the code on small test problems.  Purify and other malloc debugging
libraries are much faster.