In this transform I am generating lots of result documents (essentially one result document for each input DITA topic, so about 1500 result documents) and I'm running on a quad-core MacBook.

So the performance difference is consistent.
Eliot Kimber
Austin, TX
On Wednesday, May 14, 2014 9:37 AM, Michael Kay <> wrote:
That's a bigger difference than we normally see, but by no means impossible. For example if the transform is generating a large number of output files using xsl:result-document, then it will benefit from the multi-threading in Saxon-EE which allows many result documents to be generated in parallel, especially if you have lots of (CPU) processors available. Also, if there's a join that's not supported by keys, then Saxon-EE will often reduce the performance from quadratic to linear which of course makes a dramatic difference as the source document becomes larger.

But with performance the devil is always in the detail and one can't really explain differences without seeing the detail.

Michael Kay

On 14 May 2014, at 15:13, Eliot Kimber <> wrote:

I'm doing some performance testing of DITA processing on a large document (1500 topics) using the XHTML transform (which means that all the relevant data processing is in XSLT, not some other processor).

When I run through Oxygen 15.2, which means I'm using Saxon EE, the process takes about 6 minutes to run. 

When I run on the command line, which means I'm using Saxon HE, the process takes about 12 minutes.

Is my analysis that Saxon EE is twice as fast as HE correct or should I look for some other explanation for the performance difference?


"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests across 300+ browser/OS combos.
Get unparalleled scalability from the best Selenium testing platform available
Simple to use. Nothing to install. Get started now for free."
saxon-help mailing list archived at