speed questions about the re-instated purge rule
This question derives from my questions in https://hashcat.net/forum/thread-5661-page-2.html, and the slightly incorrect point made in Matt's blog post at http://reusablesec.blogspot.com/2016/08/...-rule.html

For as long as there was CPU hashcat and GPU hashcat the purge rule was only in the CPU version, never the GPU version.

From what I recall, the reason was a speed issue, the the CPU was more suited than the GPU.

Now, in hashcat v3, since the purge rule has been restored to the CPU platform and added to the GPU platform, do the old understanding still apply?  That is, would there be a speed penalty for running the purge rule on the GPU platform, in which case we should run it on the CPU platform.  Or, does the OpenCL device not matter?

In other words, should I still divide rule sets up into ones for the CPU platform and GPU platform, or will the rule sets run equally fast on the GPU alone?

(And if I keep asking questions about the purge rule, that is what the moderators are likely to do with me.)
Not sure what you're talking about, but dividing rules for cpu and gpu makes no sense to me. Also, this was never a direct performance issue. It's more like that rules have been applied to vector datatypes in the past and now we use scalar datatypes, because all modern GPUs run on scalar datatypes as efficient as on vector datatypes plus the vendors dropped support for old GPU that required vector datatypes.