Those are still fairly beefy processors by some standards, and I'd expect for a discussion forum that in general beefing up disk or memory would be of a greater advantage than bigger processors. Past experience with MySQL says that it will happily use as much memory as you can throw at it for caches.

How much memory is in it now? Most machines to that spec can handle at least 64gb, many of them as much as 288gb. If that machine is low-memory, on the order of 16gb or less, for far less than the cost of a new server (or even of a CPU upgrade) you can probably bring it up to 64gb (between $600-$800, assuming the existing memory has to be thrown out; possibly less if they can be combined.)

What is the disk system like? Enterprise-class SSDs are ridiculously expensive, and whether using RAIDed consumer SSDs makes sense depends on your workload, your access to the machine, and your uptime requirements. If it's on old 3.5" disks and it needs more speed than space, going to 2.5" disks is a big win -- it typically saves money, and it usually both performs better and uses less power (and puts out less heat, and needs a smaller case.)

That said, rather than speculate, you've got a workload which is highly amenable to instrumentation. Are your CPUs maxed out? Assuming that is running Linux, you should be able to run instrumentation (SAR, Nagios, or even just a vmstat running in a screen) to see how the CPUs and memory are doing, given the present workload. Some tips on SAR/sysstat if you're not familiar:
http://www.thegeekstuff.com/2011/03/sar-examples/
(*BSD or OpenSolaris should be similar.)

There is similar instrumentation on Windows, although I have only a very limited idea how to get at the performance monitors involved.

Have you run mysqltuner.pl or something similar to check on how the mysql is being handled? Are you running an up-to-date version (and have you considered moving to MariaDB instead?)

The 1st-generation Core-i processors (Nehalems) as in that case are reasonably power-efficient by modern standards; depending on what you pay for colo and power, it might not matter. You could probably match the performance of that machine with a 65-80W single-socket CPU and with a simpler chipset (etc) pull half the power or so, but your maximum memory would be much, much lower (only 32gb!) which is likely a bigger limit.

Warranty is another issue, especially on disks; I'd expect that machine of that age to either be out of warranty or going out soon... and if you are not comfortable maintaining it on a pay-as-you-go basis, a new machine might be more cost-effective than extending the warranty on the current one (then again, it might not.)

Going to a newer machine would get you a higher theoretical memory capacity (because of the use of 32gb DIMMs, although they're not cost effective) and might get you a higher practical memory capacity (from more slots; some dual-socket systems go up to 24, although the base capacity in cheaper servers starts at 8 same as the generation you've got.) The processors you'd be looking at are probably dual-6-core, relatively low clocked, but still a bit faster per core.

If your CPUs are a bottleneck, that kind of workload tends to do very well with additional cores. OTOH, if CPU isn't the bottleneck, cheaper upgrades to the current server are likely to help more.