As many of you know, server memory is capped at 6.4GB today. While this enables a wide range of content, we want to give you the tools to bring larger, more complex works to life with more players.
Starting today and over the next few weeks, we will be running experiments that will increase the amount of server memory your experiences can use using a dynamic formula based on multiple signals. We’re running these experiments to ensure we have the global capacity to successfully give you more memory without any side effects.
Once the experiments are finished – assuming they are successful – we’ll share more detail about how much more memory you will have along with the rollout timeline.
There is nothing you need to do, and the only impact we expect to your experiences is that servers should live longer (memory leaks will leak longer before the instance runs out of memory). It is possible you will see your P90 server memory usage increase within the Creator Hub Monitoring dashboards: this is normal and expected.
Please let us know if you have any questions or observe any issues beyond the above!
Thank you.
this is just amazing, I can’t wait what more things we’ll be able to do now with the increase of the server memory. thank you Roblox engineers, if this is successful this is gonna help us with a lot of stuff for sure
This is less motivation for me to find my memory leaks let’s go! On a real though, nice change haha, curious to see if the Dynamic formula can cause the memory for specific games to be lower than 6.4GB
Out of curiosity, will it be guaranteed that the memory cap will not go below 6.4GB? Since it’s been written that the amount of server memory will dynamically update, I’m curious as to whether it will dynamically update to a value above 6.4GB, or could potentially go below also.
This is definitely very handy for large games. Thanks for this!
Does this come for free? I can imagine games with high memory usage cost roblox a lot more to run.
Will games be penalized (or rewarded) for their memory usage?
Interested in hearing what metrics are used to determine memory increase.
Awesome, more memory allowance will be very beneficial towards the creation of more complex experiences. On the topic of memory leaks, could we possibly see a more explicit description of memory categories? Currently dealing with a leak that is resulting in “CoreMemory/default”, “CoreMemory/replicationCoalescing”, and “UntrackedMemory” constantly increasing (until the server crashes), and so far I’ve been unable to determine what types of actions are contributing to each category.
I think it is crucial for Roblox to modernise the tools developers have to trace bugs, especially the memory categories and the ability to see what specific processes and instances are contributing to them and to see a breakdown of usage into those specific processes and instances rather than a combined total.
And crash breakdown reports. Similar to minecraft, make it so that whenever a client crashes then a local log will be sent to the developers which can outline which script may have caused it, % script process usage and so on and so fourth.
This would make development & debugging a whole lot easier and I think it’s exactly what we need.
OH MY GOODNESS YES IVE BEEN BEGGING THIS FOR A WHILE BRO THANK YOU! seriously this is amazing. i have a big realistic game and our assets are hefty, this is a life saver.
AWESOME!!! I never knew the size of server memory before, but having access to more server memory opens up more possibilities like having more enemy AIs spawned at once, more detailed map builds and larger maps! Thanks Roblox!
Where can I find the documentation on server memory size, and how it’s calculated? I never knew server memory was capped at 6.4 GB originally, so documentation will be good information for me to have going forward.
Yeah I’ve seen this post, this isn’t the only situation that causes UntrackedMemory to increase. It generally increases as your game increases its total allocation of system resources, however the engine either isn’t redistributing the resources, or there is some other factor resulting in the engine opting to acquire new resources instead of utilizing its current memory pool. Hoping we can get more transparency from ROBLOX here.
