All over the internet, I can read that every hop in a daisy chained WDS network will be halved.
Basicly they claim that the speed will be cut in half, cut in half, cut in half, cut in half, cut in half and cut in half, until there is no bandwith left in the last hop. 
Please, can anyone sheer some light here? 
Maybe. WDS is old. It has been around for a while. It is a protocol that is the answer to āI have two networks that I cannot connect with CAT5 cable, but I want it to act like it is.ā
Every hop, it must distribute (the āDā in WDS) all packets on all interfaces, as if it were connected by a CAT5 cable. So it takes exactly twice the time to ādistributeā the packets to all interfaces if there are twice as many interfacesā¦and on, and onā¦
So if I have six 108 Mbps nodes, the speed will be like this if I transmit data from WDS Node 1 to WDS Node 6? Sorry, but I donāt understand why the speed would be cut after WDS Node 3.
First hop = 108 Mbps
Second hop = 54 Mbps
Third hop = 27 Mbps
Fourth hop = 13.5 Mbps
Fifth hop = 6.75 Mbps
Every Node receives data.
It has to buffer this data.
Then it will transmit the previously received data with the SAME antenna it has received the data.
This results in a theoretical 50% loss in throughput.
I donāt believe that the speed will be cut to half after ever hop. Iām quite sure that the speed will be the same from WDS Node 1 to WDS Node 6 (at least if you only transmit data in one direction, otherwise it wil be cut to 25%).
Can anyone confirm this?
I have a six hop PTP system that consisted of 12 RBs each with one XR5 connected back to back over about 300 miles. When I used WDS to bridge them all together speeds were much lower than when I used straight up bridge/client and routed them.
I cannot explain the problem in technical terms like you are requesting but my throughput for each hop (e.g., RB1 to RB2) was perfect but my throughput from RB1 to RB12 was abysmal even though taken individually each hop had perfect stats (CCQ, good fade margins, appropriate settings, etc.)
Best-practice is to minimize your use of WDS. It is the simple beginner way to get wireless links working in Mikrotik but you quickly run into problems when you try to scale it up like you are doing (and I did.)
Not to disagree Colebertā¦
Against our advice, we have deployed multiple 4-6 hop paths for a customer, using WDS, CM9s, 433s (3.30), and using 20mhz channels get 36-40mb consistently. When using 40mhz channels we typically run about 62mb. We do not use separate RBs, we place 2-3 radios (1 ap, 2 bh) on each RB.
Recently we have made some inroads into having them segment their network, but for now, it is what is is.
Our longest string of WDS hops is 8. This one even has a 12,000 foot copper segment just to make it more interesting.
WDS with multiple cards should not have the same issues as a single card WDS repeater.
You would still have network broadcasts going however far just to get to the guy literally next door, butā¦