Horrendous optimization

[Steve The Cynic.3217]

> @"Maze.3825" said:

> > @"Steve The Cynic.3217" said:

> > > @"Maze.3825" said:

> > > > @"Steve The Cynic.3217" said:

> > > > > @"wanya.1697" said:

> > > > > RAM Speed does not matter in GW2 case its the single core performace of the cpu that matters bring it up to 5 ghz and you will see results but big fights will still not be fluid unless you limit player models

> > > >

> > > > RAM speed matters because the CPU must fetch data from memory and write it back, unless the CPU cache is more monstrously huge than any current CPU has. If the RAM is slow, the CPU will sit around waiting for it. If you boost the CPU's clock speed, all that will happens is that the CPU waits quickly, but it still waits.

> > >

> > > You are being subjective especially when mentioning this slow ram concept.

> > > Oh! Being subjective does not help much. :)

> >

> > Do you have a justification for why you think I'm wrong?

>

> You're being very unclear is my point. You say fast ram, right? What does fast ram constitute for you and furthermore 'the client gw2.exe' that you claim can take advantage of fast ram & specifically?

 

"Subjective" and "unclear" are not remotely related. I notice that you haven't explained why you think I'm wrong.

 

In any objective sense, all we can say is, "The faster, the better." If all the data a program manipulates (including its own code) fits in the CPU's cache, then RAM speed doesn't matter once it's all been loaded at least once. No game on the market today will meet this condition, however. If there is code and/or data that's too much to fit in the cache (there *is* too much, as noted), then the CPU must spend some time going to main system memory, and main system memory ("RAM") is *much* slower than the CPU's cache (that's *why* CPUs have cache, so that wherever possible, data will be fetched from the copy in the very fast cache rather than the relatively much slower RAM...).

 

Although RAM is slow compared to the CPU, its speed can still affect CPU performance. Imagine two otherwise identical systems, where one has 3000MHz RAM (faster) and the other has 2400MHz RAM (slower). In all other respects, the machines are the same. With a "large" (i.e. game-sized) workload, the one with the 3000MHz RAM will be slightly faster, since one "RAM clock cycle" is a bit shorter. (MHz is a measure of frequency, how many times a second things happen, so the more MHz you have, the shorter the cycles are.)

 

Overall, it probably doesn't make a *lot* of difference - 3000MHz is only 25% faster than 2400MHz - but the difference is not zero.

[crepuscular.9047]

> @"maddoctor.2738" said:> > @"crepuscular.9047" said:> > However, there are dependencies on CPU because the CPU gotta tell the GPU what to render, and this is where GW2's engine, and most MMO's engine fell behind because multi-threading is difficult due to the nature of real-world gaming instructions are non-linear> > To be fair the game engine is from ~2004 (updated GW1 engine) and during that time we were expecting faster and faster CPUs instead of more and more cores. Even game companies with graphically impressive games fell in that trap, like for example the original Crysis which is a horribly CPU bound game and makes even an amazing 8700K cry in terror when playing it. However, those other engine companies managed to fix these issues by adding worker threads to the engine, so parts of the process go to different cores. As an example, Crysis 2 is a full multi-threaded game that eliminates the CPU bottleneck of its predecessor. To make the example even better, a third party company ported the original Crysis to the engine of Crysis 2, in order for it to be playable on consoles of the era (X360 and PS3) and that version of the game was indeed multi-threaded. It had the same assets as the PC version, but used the updated (multi-threaded) engine.> > This all means having a multi-threaded is possible, and further that porting a game from one engine to another is also possible and doable. Arenanet could make another engine and port the entire GW2 game to the new one. Assets will be the same, textures, models, sounds, music, even materials and particle effects can remain the same. All they need to do is to look at Crytek's example and find out how they managed to port their entire game from one engine to the next version and do something similar.unlike first person shooter, MMOs' engine needs to deal with a lot more unknowns, especially when you got to turn that 50+ unknownsinstructions to process which calculation on which thread needs to come from a single thread, that's why gaming benefits from higher clock speed over more corethe additional cores are just under utilised if the core that's directing instructions is unable take on more loads; hence if you look at CPU's core speed in BIOS, only a single core has the higher frequencyhttps://en.wikipedia.org/wiki/Multithreading_(computer_architecture)

Yes, they can port the game to another enginebut, the question remains is 'is anet willing to use a game engine from a 3rd party like Unreal, or will continue with in-house engine?'an in-house engine takes years to develop (and could turn out to be a nightmare that dragged out for a decade like FF15); probably the most economical path for an in-house engine development is to get NCSoft to approve for pooling the technical experts from all branches under its umbrella to work on a single engine that can be used for all games going forth.