[TECK]

DEBUG mode enabled, logged as Guest:
Index
* Page Generation 0.01844 seconds
* Memory Usage 553KB
* Queries Executed 4

Forumdisplay
* Page Generation 0.03062 seconds
* Memory Usage 635KB
* Queries Executed 9

Showthread
* Page Generation 0.03081 seconds
* Memory Usage 778KB
* Queries Executed 6

DEBUG mode enabled, logged as Admin:
Index
* Page Generation 0.02090 seconds
* Memory Usage 602KB
* Queries Executed 5

Forumdisplay
* Page Generation 0.03693 seconds
* Memory Usage 714KB
* Queries Executed 10

Showthread
* Page Generation 0.05298 seconds
* Memory Usage 921KB
* Queries Executed 7

I might work on the forumdisplay to lower it even more... but I really don't like to edit the vBulletin code. The above results were obtained while inserting code into hooks.
I think the best part is the memory usage. I lowered it's value from 3500KB to 600-900KB.

I use lighttpd as server software. All applications are set on bleeding edge.
Yes, I hate Apache... it eats memory like an elephant.

EDIT: I use both XCache and memcached, on my server.
Since we are here, I wonder if it's better to simply remove the queries or cache them with memcached. Somehow, I think 4 queries on the frontpage is better then 9, even with mighty memcached. Another issue is how to check if memcached is actually caching the queries. If I use -vv, it will give me all kind of huge numbers into the log files. However, memcached lacks on documentation so it's actually hard to know how and when the queries are processed/cached.

Let me know what you think about this matter. Thanks for your advice.

Originally Posted by mute

There are queries, but they hit memcache, and don't count.

Actually, they count very much. Even if memcached caches your datastore, it will still use bandwidth. Yes you save the processor, but your month bandwidth will increase. On the other hand, if you use XCache... it will chew the disk I/O. The real question is: What is more beneficial, performance wise?

Thanks for the input.