Well.... I tested these back to back, with two different WZR-HP-G300NH routers (both firmwares tried on both), same cable modem, same wireless client, same wireless channel, same encryption. As much as I hate to say it because I love DD-WRT and was excited when Buffalo went to them, the Buffalo firmware performs much better.....
Both WZR-HP-G300NH routers with the DD-WRT firmware showed packet loss on wireless... This is verified with two routers both running the DD-WRT and Buffalo firmware just to test. Load DD-WRT on, wireless errors, back to Buffalo, none.
On the DD-WRT most wireless N clients would connect as 54 mb/s although they would show as high as 104 mb/s on the router display page. With the Buffalo firmware clients display 300 mb/s. This isn't true throughput though, see below.
Running the Lan Speed test program the Buffalo firmware had over 3x actual wireless network read speed vs the DD-WRT firmware. On network write they were close, with the Buffalo firmware just a bit higher.
Add to this that the NAS function is not nearly as easy to use, if at all, on the DD-WRT.
I hate to say it as I love DD-WRT, but the Buffalo firmware is better. I tried two routers in case mine was an anomaly...
And as to the sarcasm above, "As for your bugs; yes the TX settings do some thing; select a lower TX power and it will out put less power (WOW!)" I think what he was trying to say was increasing the TX power in the DD-WRT firmware does not increase the transmit power when you go verify the changes on the wireless status screen. I verified this on both WZR-HP-G300NH's I had as well.
One thing, on the Buffalo firmware internet MTU may be too high for your ISP, it's set to 1500, most internet provider limits are just below that, resulting in packet fragmentation to the internet and possible slowdowns. You can check that in Windows by typing "ping -l 1500 -f yahoo.com" and keep decreasing the number until it's no longer fragmented. Changing MTU from the default 1500 to 1472 jumped my download speedtest.net up by 4 mbps.