DSOGaming writes: "Lichdom impressed a lot of people when it was showcased at AMD's latest press event, and since this is a new game powered by CRYENGINE, we felt the need to ask the studio about some interesting things."
I talked about diminishing returns in hardware in a blog a while ago comparing next gen consoles and people were telling me I has full of it.
Lol what diminishing returns in console hardware? Devs will use all the power they can get from the hardware.
He is talking explicitly about the threading model in CryEngine 3 which might as well be true. However, Guerrilla for example might have a different opinion on that since they using a job based model (running on up to 6 cores) and not an arbitrary threading model - I have know idea how CryEngine spawns threads, though. But, yes, spawning random amounts of threads wont help much.
With more than 4 cores, engine needs some serious tasking capabilities in order to parallelize computation. Guerilla did a nice step towards that, that's true. Since I haven't got that far with CryEngine3 to reach multitasking, am unable to tell specifics, but since it's an consolified update to CryEngine2, we can assume it still uses older models for that purpose, therefore arbitrary one. But that's an assumption.
@angels3785 - You has full of it? I has cheeseburger. Diminishing returns. Meh... Necessary Evil. Technology moves forward.
I'd say it'd depend on the speed/capacity of the cores and the structure/nature of the code as to whether this was the case (particularly in the context of GPU bottlenecks, a la the article), not to mention the engine in question.
This title is a little misleading, no? Sure, they may be observed here, but it's diminishing returns for this particular use case - It's not diminishing returns w.r.t the benefits of multicore processing in general. The CryEngine threading model; the overall parallelizability of the code-base; gpu frame render times; the amount of general purpose calculations that need to be performed per frame. Now, all of these things bring about the dev's thinking on the matter. In the general case, however, a highly parallel code-base along with well thought out data access patterns will see speed-ups well into the 10's and even 100's of cores, but only if we ignore other external factors.
Edit: Sorry, had a submission error.
N4G is a community of gamers posting and discussing the latest game news. It’s part of NewsBoiler, a network of social news sites covering today’s pop culture.