CCPs says they're making actually proper headway with 0.0 lag
This is complete lies, right?
The Hell
Moderator: Forum Moderators
1,000 people in M-0 (GBC vs NC + KIA, DAISHO and some GS) = 5 to 10 second module lag. Sounds like the hamsters were doing pretty well...until the multiple node crashes fucked it all up.
If what they say is true and they are able to resolve the node crashes, ghost ships and desynchs it'll be amazing.
Jita currently runs very smoothly with 800 to 900 people in system and only a slight delay on the market. There's already a tangible improvement there.
If what they say is true and they are able to resolve the node crashes, ghost ships and desynchs it'll be amazing.
Jita currently runs very smoothly with 800 to 900 people in system and only a slight delay on the market. There's already a tangible improvement there.
I had a quick read.
Basically, they've implemented a StacklessIO technology which rids them of all network delays. This was where the 'lag' was mostly generated as with high load random network packets would become heavily delayed, whilst others going straight through - giving module lag, grid load lag, de-sync, etc.
Now they've removed the 'lag' the servers running the nodes are able to handle more players, which lead to the problem of the servers running out of memory. It seems before the 'lag' acted to protect the server from running out of memory.
They are now moving from 32-bit to 64-bit server code giving them bugger-me-more addressable memory to use. Before they were limited by the 4GB/32-bit barrier.
I read all this in the dev-blogs whilst laughing at NC losing 50+ capitals and killing none in return.
Basically, they've implemented a StacklessIO technology which rids them of all network delays. This was where the 'lag' was mostly generated as with high load random network packets would become heavily delayed, whilst others going straight through - giving module lag, grid load lag, de-sync, etc.
Now they've removed the 'lag' the servers running the nodes are able to handle more players, which lead to the problem of the servers running out of memory. It seems before the 'lag' acted to protect the server from running out of memory.
They are now moving from 32-bit to 64-bit server code giving them bugger-me-more addressable memory to use. Before they were limited by the 4GB/32-bit barrier.
I read all this in the dev-blogs whilst laughing at NC losing 50+ capitals and killing none in return.
