Quad core processors

[goodwood]

Just got a new computer today, and I can't find the thread on ways to get the processor to improve speed etc, can anyone help?

Ron

[TheVulture]

Just got a new computer today, and I can't find the thread on ways to get the processor to improve speed etc, can anyone help?

Ron

Only comment from BFC I know of is : http://www.battlefront.com/community/showthread.php?p=1094613&highlight=single+core#post1094613

That basically amounts to "CMx2 only uses a single core, and there are a lot more pressing things to do before we rewrite it to support multiple cores".

[goodwood]

ah it must have been another game I was thinking of.

[jcmil]

I was thinking of giving this a try -

CPU-Control

just start CM alt-tab out & assign cores to the exe. Seems to be working OK.

[Other Means]

I was thinking of giving this a try -

CPU-Control

just start CM alt-tab out & assign cores to the exe. Seems to be working OK.

Really? That works? Colour me shocked. I thought the only way would be to run CM in a VM and assign it to the 4 cores. I'd try it but I can't be arsed.

[Battlefront.com]

So if I understand this CPU-Control program correctly... what it allows you to do is assign CM to a specific core and therefore ensure that it isn't competing for processor time with other things assigned to the other cores? Neat! I'll be interested to know if anybody thinks this makes a tangible difference in terms of things like loading scenarios and framerate.

Steve

[MeatEtr]

I was thinking of giving this a try -

CPU-Control

just start CM alt-tab out & assign cores to the exe. Seems to be working OK.

Wow, this certainly deserves some testing/comparisons to see if it yields any performance improvements. Thanks for posting about it.

[MeatEtr]

Well initial tests with this indicate a big drop in loading times. A scenario loaded in about 23-24 seconds without it and then using all four CPUs my loading time was about 13-14 seconds. So for me anyway, it cuts load times in half.

I didn't notice any performance difference, was running FRAPs. Appeared unchanged to me.

[Sergei]

Well initial tests with this indicate a big drop in loading times. A scenario loaded in about 23-24 seconds without it and then using all four CPUs my loading time was about 13-14 seconds. So for me anyway, it cuts load times in half.

If you didn't reboot inbetween, it could be that the second load time was faster simply because a lot of the stuff was already in a buffer. I tried the same but the other way around - first with CPU-Control, then without, and the load time dropped from 65 seconds to 55 seconds on the second load.

[MeatEtr]

Damn it, your right Sergei, just did more testing and noticed the same. :mad:

Oh well, my brief joy was fun nonetheless.

[Erwin]

Re quad cores, I just got a new system and had to decide whether to get the brand new board that automatically shuts down the unused processors, but after much discussion with the tech folks it did not appear to me that speed would be noticeably increased (at least not for the apps that I use - including the CM series).

I decided to not bother paying the extra hundred bucks or so and go with the regular quad.

The feature appears to be more of a power saving feature (re more "green").

[jcmil]

Re quad cores, I just got a new system and had to decide whether to get the brand new board that automatically shuts down the unused processors, but after much discussion with the tech folks it did not appear to me that speed would be noticeably increased (at least not for the apps that I use - including the CM series).

I decided to not bother paying the extra hundred bucks or so and go with the regular quad.

The feature appears to be more of a power saving feature (re more "green").

Never heard of a quad core that doesn't allow power management of individual cores or a mobo chipset that doesn't allow that. I've seen a lot of 'entry' level mobos for under $75 which have core related PM enabled. That feature has been around for quite a while (I've been using it on quads for over 2 years.).

I believe even the really basic intel 'quad' (core duo? Which is really 2x 2 cores) allows core power management.

Don't see how shutting down cores would 'improve' performance of software in normal circumstances. Its coded to use x cores, single, double, multi, so the software is generally going to ignore what it doesn't use I think, active core or not.

***

That aside I found that using the CPU control app does impact my overall system performance. On auto it does seem to effectively pipe-line stuff to the less used cores. I think this is especially useful in ref to active background apps whilst CM is running.

However I am finding it difficult to definitively benchmark CMSF in the different CPU states.

e.g. running a saved wego turn in - auto; with specific cores assigned; or w/out core management.

I can see <70fps when theres a lot going on & in a relatively slow scene the gain is 10-20% according to Fraps.

Running the game in real-time I see no slow-downs & it seems to perform best with cpu control on auto rather than me forcing it to use only cores 3 & 4 for example.

Calculation time for turns seems quicker but I haven't put it to the stopwatch. Time to load a few more saves I guess.

I guess the main thing is it isn't 'hurting' it, I have maxed all settings & enabled vga driver side optimisation - quality optimized & multisampling at max Q etc.

[jcmil]

Having done a few timed tests:-

with CPU Control -

level loading - approx 20 secs (w/out- 23-4 secs)

WEGO turn calc - approx 7 secs (w/out 9-10 secs)

I think that is enuf to warrant it's continued use...