If I had a dime for every time a single threaded process caused a performance issue I’d have enough for a few nice dinners.
It’s 2026, if you can’t effectively multi-thread and learn to parallelize processes i question your effectiveness as a dev. Especially in performance oriented and scalable apps.
It’s for gameplay programming, not engine subsystems. These events must eventually be serialized (in the serializability sense, not marshaling) and handled by a single thread for the net code to work. Basically every engine does this (or risk having unreliable networking). The goal of this (last time I looked was when it was announced) is to make it easier to have distributed networking for multiplayer games.
Also building this using c++ and LLVM will result in better performance than any blueprint system can achieve.
We can hate AI and the direction of the engine but don’t just assume the devs at Epic are dumb.
It’s 2026, if you can’t effectively multi-thread and learn to parallelize processes i question your effectiveness as a dev.
Especially when you’re architecting an engine targeting AAA development which needs to eek out every spec of performance they can. Like this isn’t them forgetting to use a feature of a language, this is them just not even adding it.
The LLVM also seems like a poor choice. Sure it’ll be fine for many but it’s still adding overhead and will make it harder to optimize that last 5% to nudge into okay performance. Hardware isn’t constantly evolving and being upgraded at the pace it used to.
And actually go look at the syntax of Verse. It’s painful to read, which explains why they’re pushing LLM driven development so they don’t have to see their own monster
…single thread
If I had a dime for every time a single threaded process caused a performance issue I’d have enough for a few nice dinners.
It’s 2026, if you can’t effectively multi-thread and learn to parallelize processes i question your effectiveness as a dev. Especially in performance oriented and scalable apps.
It’s for gameplay programming, not engine subsystems. These events must eventually be serialized (in the serializability sense, not marshaling) and handled by a single thread for the net code to work. Basically every engine does this (or risk having unreliable networking). The goal of this (last time I looked was when it was announced) is to make it easier to have distributed networking for multiplayer games.
Also building this using c++ and LLVM will result in better performance than any blueprint system can achieve.
We can hate AI and the direction of the engine but don’t just assume the devs at Epic are dumb.
Especially when you’re architecting an engine targeting AAA development which needs to eek out every spec of performance they can. Like this isn’t them forgetting to use a feature of a language, this is them just not even adding it.
The LLVM also seems like a poor choice. Sure it’ll be fine for many but it’s still adding overhead and will make it harder to optimize that last 5% to nudge into okay performance. Hardware isn’t constantly evolving and being upgraded at the pace it used to.
And actually go look at the syntax of Verse. It’s painful to read, which explains why they’re pushing LLM driven development so they don’t have to see their own monster