processor question

[cobrabase]

Okay techies. I have always turned to you all for your excellent advice in the past and you have never let me down.

I’m playing ARMA 2 right now with low to mid settings and it’s time to get moving onto more power. I went to “can I run it” and had my system analyzed. The only thing that turned up low was my processor speed.

Current CPU is a simple duel core E6750 2.66GHZ. This is great for sims like DCS Black Shark but I need raw power with ARMA 2.

I have been doing some research and I was wondering what you would recommend. I am looking at the Intel Core 2 Duo E8500 Wolfdale 3.16GHz 6MB L2 Cache LGA 775 65W Dual-Core Processor. It has great reviews at new egg.com and a couple of awards. I have a motherboard ready for quad core but this thing looks to be just what the doctor ordered.

I’m asking for advice on my choice and on installation. I’ve installed CPUs before and did not find it very difficult.

Thanks for all of your help in advance.

[Froggy]

I have a 6750 to but OC @3.5 ghz with standart air cooling

[Ssnake]

You should check whether ArmA will actually profit from more cores (it may be that the engine is prepared for it - then again, maybe not). If the engine actually profits from multiple cores and can handle more than two, it makes sense to go quad. If it doesn't, you'd only be wasting money. It would then be better to look for a dual core CPU with a higher CPU clock setting.

The marketing drones say that 4 x 2.5GHz = 10GHz whereas 2 x 3.5GHz = "just" 7GHz (because they want you to buy the quad cores that make more profit for them). Yet the dual core processor can be 40% faster in practice as each core is clocked at a higher speed. If the application can use only a single core, the rest of the CPU will essentially idle most of the time.

Now, you already seem to have chosen the dual core which probably makes sense. You then must check whether the socket in your mainboard will actually accommodate the new processor. There have been numerous new socket designs in the past years which often prevent processor upgrades (unless you also replace the mainboard, of course).

[Enrage]

I would try overclocking first. I'm not sure how much additional performance you're going to see, considering that Wolfdale wasn't much other than a 65nm-45nm die shrink of Conroe, so trying to get "something for nothing" when you're already thinking about buying a new processor probably wouldn't hurt.

However, if your graphics card is the same generation as your processor, you may want to look there when upgrading first. You're more likely to see good gains there.

[Ssnake]

I'm skeptical with overclocking. Processors are tested at the factory for the "safe" speed at elevated room temperature, and only then grouped into one of the speed categories for which the current production batch is designed. That way pretty much every CPU is already sold close to its limit (quite naturally so, as they can be sold at higher prices, hence better profit). Therefore one can usually hope for maybe 10% performance gains at best, and that at reduced reliability and robustness on warm days. In this case, I'd be surprised to see the current CPU overclocked and working reliably beyond 2.8GHz. I somehow doubt that this will actually be noticeable. But if it makes you feel better...

[Enrage]

I'm skeptical with overclocking. Processors are tested at the factory for the "safe" speed at elevated room temperature, and only then grouped into one of the speed categories for which the current production batch is designed. That way pretty much every CPU is already sold close to its limit (quite naturally so, as they can be sold at higher prices, hence better profit). Therefore one can usually hope for maybe 10% performance gains at best, and that at reduced reliability and robustness on warm days. In this case, I'd be surprised to see the current CPU overclocked and working reliably beyond 2.8GHz. I somehow doubt that this will actually be noticeable. But if it makes you feel better...

I'd have to disagree.

Processors are "binned", but not at the maximum safe speed, but based on the percentile they wish to sell at a certain price point. (EDIT: Or more correctly, the amount they think the public is willing to buy at a certain price point. Just because Intel can make 90% of their dies produced into $1000 "Extreme Edition" CPUs doesn't mean the public will be willing to buy them.) Cheap CPUs get sold far more often than really expensive ones, so the vast majority of dies on a wafer are going to get put into the low end bin, even if they are capable of being put in the high end bin. Most of them are more than capable (especially on the modern processes) of reaching the advertised speeds of their higher priced brethren, they just can't match the extreme overclocking potential. In fact, the reason Intel and AMD don't offer unlocked multipliers on their low-end products is specifically because they know that their "low-end" processors are more than capable of running at the higher speeds, thus prevented the public from getting a free lunch.

I am running a Core i7 920, rated at 2.66 GHZ. I run it on air cooling at 3.8 GHZ at stock voltages. Intel has done a great job with their 45nm HKMG process, and did a great job on the 65nm too. I'd bet he could get a 30% increase in core speed without any loss of system stability, assuming he got a decent heatsink.