well maybe the extended maintenance will fix vivic?.......... rofl sorry I had to

Today is bump day

Off the top suggestions that have probably already been stated (if not in this thread, at some point in time)

These are just my opinions based on my “feelings” so take them with a grain of salt.

1) it’s possible the optimizations they are doing to “normalize values” will improve things due to simplification of the calculations. Caching some of this data around combat values, rather than any extra effect data or map data may be of more importance.
Example: if i log in directly to Cyrodiil, the game has less data cached or in memory than if I jump to 5 or 6 different wayshrines/zones & then go into Cyrodiil.
Ie...if you’re not already doing it, assign memory to heap using malloc instead of the stack & ensure it’s being freed properly.

2) client side validation of key press per millisecond could help, as spamming a single key is not what I would call “preferred” & macros, while they are allowed (to my understanding) should not be setup with key press timings faster than what is humanly possible. I believe pressing 1 button to start your macro rotation & it trying to tell the game “all of these abilities attempting to happen simultaneously” can cause further issues.

3) certain abilities, possibly when performed in certain sequence or from multiple players, appear to create reproducible results such as health desync. I believe that ZOS should simulate lag in their test environment whether it’s with open source tools or otherwise. It’s pretty easy to simulate tcp/udp floods, broadcast storms, etc.. & then attempt to play in that environment.

4) review of existing security protections should be checked for inefficiencies to see if there is a better way to achieve the intended result.

5) placing rate limiting or caps on server transmission rates kind of exist but needs further tuning to extend outside of its normal checking. For instance it seems some items are being checked or rate limited but not others?

6) input validation should be done everywhere for testing & production should be limited/only where needed to reduce overhead, yet logs of values/variables should exist to be parsed for where users have taken advantage of these items to see if a check should be moved from testing to prod.

7) different hardware is good for different types of math. CPU’s are more efficient at some types, ASICS are better at another, & GPU’s at another. If some calculations can be offloaded to another type of hardware this could improve things but you still have to account for any latency around the connectivity such as pci/pcie/pci-x/usb/cable/etc.. This could even be offloaded to something like ibm’s Watson or a more scalable architecture

8) I wonder if there can/should be configuration items normalized on client pc’s such as max retransmission rate, mtu, or other network settings? There are formulas for determining some of these numbers.

9) reduce redundancy in calculations & checks where possible, this ties back to item 1 however should be extended to all aspects of the game: server side calculations, client side calculations, security checks, & data transmissions.

10) I don’t know what the backend looks like but ensure the usage of relational databases such as Postgres rather than stuff like nosql, for those who haven’t seen it:
https://youtu.be/b2F-DItXtZs

That’s all I can think of for now, good luck!

-Kai

Game running the worst it has in 6 months. OP update is OP