Performance doubts

[01XUAN]

1.21.1 - 0.9版本
在加载的时候（如刚进游戏，加载短时间内没有加载过的区块），区块越多，卡顿越明显。刚进游戏会持续这样卡顿十几秒，仅一个模组时也是如此

[joe-vettek]

tps? 那需要看spark日志

tps? which needs the spark profiler log

[joe-vettek]

哦，我大概知道了，这个其实是一个优化，主要是在区块首次加载并发送给玩家的时候影响。后续我看能不能优化吧。不过目前的话，其实也不必过度担忧，对于老区块而言就会优化模组逻辑了。

Oh, I roughly understand. This is actually an optimization, which mainly affects the first time a chunk is loaded. I will see if it can be optimized later. But for now, there is actually no need to worry too much, as the impact on old blocks is very small.

[01XUAN]

我的问题（优化模组不禁用完就下结论了），c^2m干的，前者有c^2m，后者没有
https://spark.lucko.me/yt6VQV8lzQ
https://spark.lucko.me/dyiQeWB5BN

[joe-vettek]

简单来说，就是节气首次加载区块时，会预先计算和缓存群系信息，因为Minecraft查询速度非常慢，做了这一步优化的话，群系查询速度可以提高20倍，也就是减少了95%的耗时。一般来说，正常跑图速度的话，其实不必担心。

In simple terms, when a chunk is loaded for the first time, biome information will be pre-calculated and cached. Since Minecraft query speed is very slow, this optimization can increase biome query speed by 20 times, which means 95% of the time consumption is reduced. Generally speaking, if the map runs at normal speed, there is no need to worry.

[01XUAN]

好的

[joe-vettek]

C^2M 似乎利用了多核性能来加载，这放大了这个工作的影响，一下子太多了应该是，发送数据的这个步骤应该还是单线程的。

C^2M seems to take advantage of multi-core performance to load, which amplifies the impact of this work. There is probably too much at once. The step of sending data should still be single-threaded.