I always keep an eye on RC versions since they were available, looks like it will be early Christmas with these changes:
What's new in 6.24rc4 (2014-Dec-12 13:42):
*) fixed problem where some of ethernet cards do not work on x86;
*) improved CCR ethernet driver (less dropped packets);
*) improved queue tree parent=global performance (especially on SMP systems and CCRs);
*) eoip/eoipv6/gre/gre6/ipip/ipipv6/6to4 tunnels have improved per core balancing on CCRs;
*) fixed tx for 6to4 tunnels with unspecified dst address;
*) fixed vrrp - could sometimes not work properly because of advertising bad set of ip addresses;
After some testing this is what i noticed so far:
1) "Download" and "Upload" parent queues with small trees in HTB "Global", can now pass ~2 times more traffic with this RC over v6.23 on CCR1036
2) Having anything in HTB "Global" doesn't slow down rest of the traffic that doesn't have any queues in HTB "Global" - so basically queue tree bypass seems to be fixed.
3) looks like HTB Global performance scales with number of parent queues (similar like it was with Simple queues till now), but i didn't notice this on Interface HTB.
Looks like i will have to reconsider if many simple queue solution still is best for me, somehow i always liked queue tree solution better.