Over the years i've done a lot of different rigs and overclocks. is this game? A:going to want higher frequencies as well as more cores and more ram. B:going to like stable overclocking, like most games including forged alliance.
You can't really code for overclocking. Either your system is stable and works, or it doesn't. More cores and more memory are always your best bet. More ghz has incremental improvement in most cases, and for my own home rig, it's not worth how much extra cooling is required (increasing noise with fans or expense with liquid cooling) and even then, it generally reduces chip life. That said, if you have a stable overclocked machine, it should be faster. But there's no explicit work that can be done. This is entirely determined by how CPU limited the game is, and if cycle time is the limit, or if instruction pipelines are. Overclocking will only really help on cycle time issues.
theres no skill in overclocking anymore the boards just do it for you. I noticed the new 2011 pin Ivybridge will go to 5Ghz no problem.
The current Intel CPUs also do it for you; the Turbo Boost thingy is just a shorter name for "self overclocking on demand as long as you're within the accepted thermal envelope". Die yields are also so much better now that there is often no thermal difference between an i7 3820 and a 3930K, and the prevalence (not before time) of thermal sensors both on-chip, and on-board, mean that all the info the firmware needs to be able to dynamically adjust clock speed for optimal performance under current thermal load is right there for it. May as well take advantage of it Unless the client has masses of CPU intensive threads though, the main advantage of a faster clock is going to be for the server running the sim. Beyond a certain point (TBD), the client is going to be GPU bound rather than CPU bound. If you can comfortably encompass all client calculations and still have idle cycles, throwing more at it isn't going to make a difference, and the client has considerably less to do compared to the server. A headless (dedicated) server on the other hand is entirely CPU bound (since it doesn't require or use a GPU) so more cycles directly translates to smoother and/or bigger games. I doubt todays top-end i5s or FX series will break a sweat running just the client, so a carefully and stably overclocked box is prime server material, not a requirement just to play the game
Eh, I remember some obscure game I think BC2 but not sure reportedly had problems with non power of 2 frequencies even if it was otherwise completely stable. But that game had a ton of obscure problems so they probably mucked something up assuming a clock cycle was a certain length of time or something.
you could always link something important inside the engine to the cpu freq and let the fun begin. It gets really interesting if you use the CPU temp sensor as input variable.
That tells me that the devs who wrote BC2 did a bad job. An application should never care what the CPU frequency is. Competent devs stopped using CPU frequency to measure time in late 80's.
BC2 is really quite an amazing CPU test because of how crazy poorly optimized it is. You can stable a burn test for hours but BC2 might crash in 10 minutes.
