Preliminary Performance Improvement Results

"

We found that users who have only two hardware threads (cores) indeed experience worse performance than before when using the new mode. Due to the nature of the improvement, it is probably necessary to only engage it if you have four or more hardware threads.

Wut? Can you provide an explanation a little more technical? I'm just curious. I'm a programmer as well, and have used threads plenty of times...there is no such thing like decrease the performance if there are only 2 cores...like wtf, it doesn't make any sense. The performance should boost with more than one thread, regardless of how many(up to the number of cores the user has).

Interesting stats. Thanks for sharing GGG.

Slightly concerned as I've seen a performance reduction with this option but have 4 cores :(. I hope I don't wind up in one of those small edge cases where I cant disable the option...

Great Performance update.

FPS in a burning ground sewer map before the update ~15-20 to >60 now.

No crashes for me, but a few lockstep problems which seems to be serverside. (as far as the global chats "anyone got the same lag?" is reliable )

System:
i5-4460
R9 390 Nitro
16GB RAM

"

poeFan2 wrote:

"

We found that users who have only two hardware threads (cores) indeed experience worse performance than before when using the new mode. Due to the nature of the improvement, it is probably necessary to only engage it if you have four or more hardware threads.

Wut? Can you provide an explanation a little more technical? I'm just curious. I'm a programmer as well, and have used threads plenty of times...there is no such thing like decrease the performance if there are only 2 cores...like wtf, it doesn't make any sense.

It would make sense if the main core now focuses more on delegation rather than action. In a two core set up, all this would do is add overhead?

"

Spanka6663 wrote:

"

poeFan2 wrote:

"

We found that users who have only two hardware threads (cores) indeed experience worse performance than before when using the new mode. Due to the nature of the improvement, it is probably necessary to only engage it if you have four or more hardware threads.

Wut? Can you provide an explanation a little more technical? I'm just curious. I'm a programmer as well, and have used threads plenty of times...there is no such thing like decrease the performance if there are only 2 cores...like wtf, it doesn't make any sense.

It would make sense if the main core now focuses more on delegation rather than action. In a two core set up, all this would do is add overhead?

Hm...it makes sense, but if that's the case their implementation is faulty. From my experience multi-threading should only be slower if the data is not big enough to compensate for the overhead(communication, thread creation). In this case if your data sample is too small, say, to sort an array, the multi-threading solution will be slower. But if you have big enough data to share between the cores...even if you have only 2, both should do roughly the same amount of work, aka load balance and therefore there should be a performance boost. Heck I've never seen this before. 2 cores is slower but 4 is faster? lmao seriously. GGG, I haven't seen your implementation but there's obviously something very wrong with it. Not trying to bash here. Just a heads up...review the implementation. Thanks for the hard work, even if it only works for more than 2 cores.

"

leonardorubin wrote:

I have only 1 core...:(

Time to upgrade, mate! The future is calling!!

40% of players having 8 core cpu? damn..that's nice

"

MusicalMind wrote:

40% of players having 8 core cpu? damn..that's nice

I have 4, what a relief lol. Even with GGG's faulty multi-threading implementation I'll still see the benefits hahaha.

Haven't test it yet personally but I'm glad to hear about performance updates. Group play is nearly impossible as it is.

my game just doesn't wanna work better with this, I guess I'll just wait for the patch :S