@Intaris wrote:
There's also a lot of disk caching going on int he background under Windows. This canlead to HUGELY modified benchmark results. Windows basically buffers everything you read from or wtite to disk (up to many GB) if you have the RAM available.
Actually clearing this buffer between runs can greatly help find a correct "baseline".
Indeed if your code or dev environment have such a seriously flawed and/or bloated memory management principles, for example; the whole shebang gets paged out to disk and needs a restart/reset/flush/clearing for the best results. Well, that certainly sounds like it could lead to some quite serious performance problems, measurement inconsistencies and shows a lack of robustness in the coding wherever that might be.
Fortunately, RAM is cheap nowadays and a 32bit address space hasn't been a limitation for more than 10 years. Most reasonably well built applications and dev environments have already migrated to transparently support both 32 and 64bit ages ago.
Roger
