I'm programming an ATmega328p using avrdude, and to program it with a program size of 3074 bytes, I am waiting 13 seconds. That seems pretty dang long to me.
I have tried using both the internal oscillator at 8MHz, and an external oscillator at 8MHz. CKDIV8 is unprogrammed. I'm not sure why 3kb is taking so long. I remember it being much shorter for some reason.