Happy Tuesday all. Last week we were delighted to share the additions coming along in the 1.24 Japan Update including new ideas, provinces and tags in the island of Japan and the Philippines. Today I want to talk about the much less sexy changes coming. Performance, bugfixes and balance changes
Let's start with performance. It's a tricky thing to address in a dev diary. I can't take an alluring screenshot of the game to show it off, and we can't say that this new performance comes with -20% core creation cost and +1 leader shock. No, it's not the prettiest thing to work on or talk about, but in the 1.24 Japan update we've put some real elbow grease into improving the performance of the game, with all thanks going to @mikesc and @Groogy for their efforts
If we look at how much time the game was chewing up for its daily tick actions, we are able to get an idea of who the real performance hogs in the game are. Here we have a graph with plots showing daily ticks accumulated for months, time is in ms:
Before:
As we can see, AI actors even though they are running in parallel, are drinking up the bulk of performance, with countries_serial not that far behind it. On top of it, as we go later into the game, performance was getting worse. This is to be expected to a degree: as the game progresses there are more and larger armies for the AI to consider, however it was reaching levels we were not happy with.
Following some real dedicated work, our performance logging now looks like this:
There are still some spikes at key parts that we're looking into, but on the whole, performance has been brought up to speed. I leave it to @mikesc who says the following on tackling the matter:
The tl:dr on the matter is that there has been a lot of performance work put into the 1.24 Japan Update, and we look forward to your feedback on how the game ends up running for you.
That's the bulk of what we have to talk about today, but we are looking to have 1.24 available to you as an early Christmas present, so let's fast forward. We were happy with the release of Cradle of Civilization, 1.23 and subsequent hotfix which tackled some of the more serious issues. Japan will, as all updates do, contain a lengthy assortment of bug fixes. Many thanks to the people who submitted reports through the bug forums to help us catch, reproduce and slaughter these issues. Japan will put to rest some rotters like the trade company merchants issues which we saw reported, as well as many others. Army Mintinance anyone?
Finally, there will be some Balance changes coming along with Japan.
For one, we were discussing the movement speed of an early modern army and comparing it to the overall speed of our units in game. Our in-game armies were moving oddly fast compared to historical estimations and as such, we have reduced the movement speed of units from 1.0 to 0.7, but not without ways of increasing it. For one, a historical strength of the Nomads was their ability to cover great stretches of land quickly and as such have tied faster unit movement speed to those governments. Additionally, Army Drill will increase an army's movement speed by up to 20% at full drill, so a well prepared army will have a much easier time moving into position compared to green units.
Other changes include Interest rates being capped at a minimum of 1.0. It is intended that nations will have to reasonably balance their debts, rather than it being treated as an infinite pool of money when modifiers were stacked heavily. To this end, we have also reduced the strength of many Interest reduction sources, as far more have been added since the launch of EU4 while the Interest and loans system have not.
Finally, a couple changes for Army Professionalism: the Supply Depot, which boosts reinforcement rate and supply limit, will last for 5 years instead of 2. The Reserves Morale Impact is now more effective due to increased daily morale damage and has switched places with the reduced General Costs.
Of course, there are many more fixes and changes along with the Japan update which we will talk about when the time comes. Until then, have a great day, which can be made greater by tuning into the Dev Clash at 1500 CET over at www.twitch.tv/paradoxinteractive
Let's start with performance. It's a tricky thing to address in a dev diary. I can't take an alluring screenshot of the game to show it off, and we can't say that this new performance comes with -20% core creation cost and +1 leader shock. No, it's not the prettiest thing to work on or talk about, but in the 1.24 Japan update we've put some real elbow grease into improving the performance of the game, with all thanks going to @mikesc and @Groogy for their efforts
If we look at how much time the game was chewing up for its daily tick actions, we are able to get an idea of who the real performance hogs in the game are. Here we have a graph with plots showing daily ticks accumulated for months, time is in ms:
Before:
As we can see, AI actors even though they are running in parallel, are drinking up the bulk of performance, with countries_serial not that far behind it. On top of it, as we go later into the game, performance was getting worse. This is to be expected to a degree: as the game progresses there are more and larger armies for the AI to consider, however it was reaching levels we were not happy with.
Following some real dedicated work, our performance logging now looks like this:
There are still some spikes at key parts that we're looking into, but on the whole, performance has been brought up to speed. I leave it to @mikesc who says the following on tackling the matter:
Thread utilisation of AI (Daily tick):
In the before image you can see a bunch of parallel AI updates that get started at the green marks and are done at the red marks. With an evenly distributed processing time need of all AIs this would be fine, however in the later game the AIs of the larger countries take much more time and break this. Empty space between the green and red marks was basically wasted time where some of the CPU cores were just idling.
What we do now is have each country measure the time it takes for its AI to process one day and distribute the AI updates across the threads in a better way for the next day, so less time gets wasted.
The tl:dr on the matter is that there has been a lot of performance work put into the 1.24 Japan Update, and we look forward to your feedback on how the game ends up running for you.
That's the bulk of what we have to talk about today, but we are looking to have 1.24 available to you as an early Christmas present, so let's fast forward. We were happy with the release of Cradle of Civilization, 1.23 and subsequent hotfix which tackled some of the more serious issues. Japan will, as all updates do, contain a lengthy assortment of bug fixes. Many thanks to the people who submitted reports through the bug forums to help us catch, reproduce and slaughter these issues. Japan will put to rest some rotters like the trade company merchants issues which we saw reported, as well as many others. Army Mintinance anyone?
Finally, there will be some Balance changes coming along with Japan.
For one, we were discussing the movement speed of an early modern army and comparing it to the overall speed of our units in game. Our in-game armies were moving oddly fast compared to historical estimations and as such, we have reduced the movement speed of units from 1.0 to 0.7, but not without ways of increasing it. For one, a historical strength of the Nomads was their ability to cover great stretches of land quickly and as such have tied faster unit movement speed to those governments. Additionally, Army Drill will increase an army's movement speed by up to 20% at full drill, so a well prepared army will have a much easier time moving into position compared to green units.
Other changes include Interest rates being capped at a minimum of 1.0. It is intended that nations will have to reasonably balance their debts, rather than it being treated as an infinite pool of money when modifiers were stacked heavily. To this end, we have also reduced the strength of many Interest reduction sources, as far more have been added since the launch of EU4 while the Interest and loans system have not.
Finally, a couple changes for Army Professionalism: the Supply Depot, which boosts reinforcement rate and supply limit, will last for 5 years instead of 2. The Reserves Morale Impact is now more effective due to increased daily morale damage and has switched places with the reduced General Costs.
Of course, there are many more fixes and changes along with the Japan update which we will talk about when the time comes. Until then, have a great day, which can be made greater by tuning into the Dev Clash at 1500 CET over at www.twitch.tv/paradoxinteractive
