Sun fudged Java benchmark results; NT is faster

Sun Microsystems revealed Thursday that it had, in fact, modified itscompilers to perform better in the CaffeineMark Java benchmark test, butthat had an "obligation" to do so."Our job is to provide the best performance on the benchmarks that

Paul Thurrott

November 5, 1997

1 Min Read
ITPro Today logo

Sun Microsystems revealed Thursday that it had, in fact, modified itscompilers to perform better in the CaffeineMark Java benchmark test, butthat had an "obligation" to do so.

"Our job is to provide the best performance on the benchmarks that are available and we did that," said Brian Croll, director of product marketingfor Solaris at Sun.

Translation: they cheated, and they did it on purpose.

As reported yesterday in WinInfo, Pendragon Software Corporation, makers ofthe CaffeineMark benchmark, accused Sun of purposefully building theirJava compilers so that they performed abnormally well on the benchmark.

"We have no reason to believe this was a master plan from Sun management, but the compiler was tweaked to look for our benchmark and that doesn't happen accidentally," said Pendragon president Ivan Phillips.

Phillips also verified that the fastest Java performance is on Windows NTrunning on a Pentium II 300, not Sun Solaris. Sun had fudged the results tomake Solaris come out ahead of NT, he said

About the Author(s)

Paul Thurrott

Paul Thurrott is senior technical analyst for Windows IT Pro. He writes the SuperSite for Windows, a weekly editorial for Windows IT Pro UPDATE, and a daily Windows news and information newsletter called WinInfo Daily UPDATE.

Sign up for the ITPro Today newsletter
Stay on top of the IT universe with commentary, news analysis, how-to's, and tips delivered to your inbox daily.

You May Also Like