GPU Upgrade, Compatibility Questions

[sleepysheep]

Unless he plays alot of cpu intense games then that gc is just a money sink till the rest of his rig is upgraded

The problem is that you're using Crysis as the standard to how a "CPU intensive" game performs.  Go check the benchmarks for Crysis, you'll see that dualcore processors often deliver the same FPS as quadcore processors.  On games that are actually well optimized for multiple CPU cores, for example Bad Company 2, you'll notice that while the CPU may affect FPS it is not something that breaks the experience.  The 2500k may deliver twice the FPS but if the low number is 55-70 to start with does it really matter?

Upgrading your entire computer for just one game is stupid, especially if every other game you own plays just fine.  If you use that mentality with the rest of your purchasing decisions in life you'll die a very poor person.

Er he's using a really old cpu that will bottleneck the crap out of the gc no matter what game he's playing. The gc can only work as fast as the cpu can send data to it. Buying an almost top of the range gc with a 4+ year old cpu of that structure and thinking games are going to run well is stupid.

And the minimum fps wont be 70 in any game period with his cpu even in dow2 my minium fps is still 55fps and that game is five years old with my sandy cpu. Bfbc2 was around 50-55 minimum fps and this is with one gtx 580. Wtb a dual core cpu that's 4+ years old that can get 70+ minimum fps in all games please.

[Mitchpate]

Wtb a dual core cpu that's 4+ years old that can get 70+ minimum fps in all games please.

Troll all you want but seriously, go look at benchmarks.  A Core2 E8200 only has an 18FPS advantage over a Core i7 2500k.  In other games the comparison is absolutely terrible but in Crysis, not so much.  Going back as far as pre-Core i7, dualcores have always performed about the same as quadcores in Crysis.  It's old news.

I'm using a Nvidia GeForce 9500GT

In TF2, try setting AA /AF to 2x or 4x, disable high dynamic range, and make sure multicore rendering is enabled.  You should be getting in the 40-60 FPS in TF2 with that card and everything but AA/AF/HDR set to max.  I used to play it max settings as well until I went to a few competition events and saw that everyone else was running low settings on AA/AF and I've been using that myself ever since.  Visually, there isn't much of a difference, especially when there's alot going on.  FPS wise, hell of a difference.  You'll find that's true in most games, actually, when there's lots of action you don't really notice the visual improvements of AA and AF but you sure do notice the FPS slowdown they cause.

Unfortunately, 15-20FPS is about all you're going to get out of that card in MWLL.  It's not necessarily the age so much as the age coupled with the fact that it wasn't a high-end card to start with.

What's your average CPU usage when you play MWLL?

[Freeborn_Toad]

Wtb a dual core cpu that's 4+ years old that can get 70+ minimum fps in all games please.

Troll all you want but seriously, go look at benchmarks.  A Core2 E8200 only has an 18FPS advantage over a Core i7 2500k.  In other games the comparison is absolutely terrible but in Crysis, not so much.  Going back as far as pre-Core i7, dualcores have always performed about the same as quadcores in Crysis.  It's old news.

Dual-cores are notably weak in Crysis, and a properly configured MWLL mod only exacerbates this trend.

http://www.tomshardware.com/reviews/multi-core-cpu,2280-10.html

Here is a review from a few years back that compares the same CPU with 2, 3, and 4 cores enabled. As you can see, three cores is the sweet spot, while dual-core processors are notably weaker. These results mirror my own experiences with the game.

Have the taskmanager open on a 2nd screen while playing and see what your CPU usage is.  My overclocked Core2 Q6600 barely hits 50% while playing but some people here claim their Core i7s are getting maxed out so it really all depends on what game settings you're using.

Total CPU utilization is not really a good metric (though it is worth looking at) for determining whether a faster CPU, or one with more cores, would be beneficial. Even if if only one or two threads are doing the bulk of the work, its quite possible for much less intensive threads to still benefit considerably from being able to spread out over more cores.

[Steelbarrage]

All I can say is my 2500k at 4.4 ghz is a night and day difference from my stock q6600.
I kept the same gpu (gtx 460 768mb) for this build and I can comfortably set very high and have over 24 fps.
Before I could barely run high.
I think money spent towards a better proc, or at least a foray into OC what you have will perform better.

[DragonFire1170]

What's your average CPU usage when you play MWLL?

Just now playing a match, my CPU usage is bouncing between mid 30% and at times 60%, but it stays mostly in the 35% - 40% range.

[Mitchpate]

http://www.tomshardware.com/reviews/multi-core-cpu,2280-10.html

Here is a review from a few years back that compares the same CPU with 2, 3, and 4 cores enabled.

I don't suppose you actually looked at the computer they used, did you?  A Core2 Duo Q6600 (doesn't exist) at 2.7Ghz (overclocked?).  Regardless of the nonexistant processor model, the Core2 series isn't a true quadcore, it's a hybrid pairing of two dualcores.  The scaling is not going to be the same as a true dualcore, triplecore, or quadcore.

A much better comparison is conducted here: http://www.anandtech.com/bench/CPU/50

For years, the complaint against Crysis has been that it only has two threads.  If you compare the processing power per clock on the chart I linked you'll see that complaint is well founded.  The only reason that there is a such a large boost in performance between two and three cores in your toms hardware "benchmark" is because the game can be run on two dedicated cores, rather than having to share cores with the operating system.

it stays mostly in the 35% - 40% range.

You still have room to go upwards with that CPU.  There's no way to tell how much, though.  Could be alot, could be little.  If you have a decent motherboard you can try to overclock you processor.  Rumormill says that some of the newer Q8200s OC much better than their early-batch counterparts.

[Freeborn_Toad]

I don't suppose you actually looked at the computer they used, did you?  A Core2 Duo Q6600 (doesn't exist) at 2.7Ghz (overclocked?).  Regardless of the nonexistant processor model, the Core2 series isn't a true quadcore, it's a hybrid pairing of two dualcores.  The scaling is not going to be the same as a true dualcore, triplecore, or quadcore.

I was well aware of the system used when I made my post. I am also aware that the Core 2 Quads are a pair of dual-core dies connected via the FSB. Indeed, scaling would improve with the higher intercore bandwidth available to "native" multi-core chips.

The results of the Tom's test are very relevant.

A much better comparison is conducted here: http://www.anandtech.com/bench/CPU/50

Those tests were performed with a GTX 280 and a 5870, making the entire top third to half of that chart more GPU limited than anything.

For years, the complaint against Crysis has been that it only has two threads.  If you compare the processing power per clock on the chart I linked you'll see that complaint is well founded.  The only reason that there is a such a large boost in performance between two and three cores in your toms hardware "benchmark" is because the game can be run on two dedicated cores, rather than having to share cores with the operating system.

Crysis uses more than two threads, though it does use two threads more heavily than the rest.

I maintain my position that having more than two cores is of noticeable benefit to Mechwarrior Living Legends. I've seen it myself.