There's nothing obviously wrong with your schematic. There could still be issues with layout, but I imagine they'd need to be pretty bad to prevent the PHY from coming up at all. The only thing I see that is very different from my board is the magnetics Vcc pin tied to PHY_AVDD0, where my board uses +1.8V for the magnetics.
Check the PHY_RESET signal. This is active low reset, so it should be high to allow the PHY to run.
Check the PHY_TXC_GTXCLK signal. It should have 125 MHz if the PHY's PLL is running. If not, you need to check that your 25 MHz is clean and that the power supplies are properly filtered.
Other things that I can't check are:
magnetics - make sure that your jack with integrated magnetics is similar to the one on the SP605 board, or at least uses the recommended magnetics topology for the PHY. I used a Tyco 5-6605463-6 on my board.
routing - make sure that the MDIP/MDIN pairs are routed with 100 ohm differential impedance and properly length matched. 47 Ohm resistors (I used 49.9 Ohm in my board) should be at the connector end of the signal traces.
If you suspect that the signal integrity may be an issue, try running at 100 Mbps instead of 1 Gbps. I usually grab an old 10/100 Mbps hub for this sort of testing. If you have a 1 Gbps switch you could try that as well, instead of connecting directly to a P.C.
-- Gabor
