Intel Xeon vs Intel Core I

[Mehran Fazili]

Hello everyone, 

I am wondering about the effect of CPU architecture on the reliability of the results generated by optimizer. Some suggest when calculation process is extensively long, there might be some errors entered into the result if Intel core I5 , and I7 CPU are used. Apparently Intel Core I5 and I7 series  CPUs do not support Error Correction Core (ECC)  enabled Memory to be used.  On the other side, Xeon E series CPUs are capable of using ECC memories.  

 If there is an error to be entered into the calculation, how severely can it effect the calculation process? for example can not using ECC Memory prevent a MIP model from converging? 

For optimization then, the two alternative are to go either with higher clock Intel Core I7 with no ECC or to go with lower clock Xeon E5 with ECC enabled.  

So what CPU would you recommend for research purpose in less that a $1000 range? 

Thank you.

Mehran

[Ed Rothberg]

Google published an empirical study of DRAM error rates a few years back...

http://www.cs.toronto.edu/~bianca/papers/sigmetrics09.pdf

One conclusion: "...over 8% of DIMMs in our fleet saw at least one correctable error per year..."  A correctable error on a machine with ECC is basically equivalent to an unnoticed single-bit error on a machine without ECC.  Such an error can lead to a seg fault, a wrong answer, etc.  It's really up to you to decide whether you can live with this possibility.

Ed