First off, I ran some load testing. Behold, the site could now stand up to a few hundred requests coming in. It still bogged down, but it didn’t crash and didn’t start rejecting connections. The memory bog-down came from Apache, which was still running numerous instances. Apache is pretty poor in this regard and I think that if I had a site that was getting really popular I may switch to nginx or some such. Regardless, it’s Apache that’s the bottleneck now, not PHP.
I visited tools.pingdom.com and ran a page speed test. From New York my site loads in 3 seconds and a touch faster from Texas. Amsterdam is around 5 seconds. All of this is down a few seconds from before (although I don’t have the exact numbers). Additionally, these tests are all being done on a site that was already optimized just not cached or CDN’d. Very healthy improvements.