But lower clocked memory. I wouldn't expect memory bandwidth to be the limiting factor in typical usage scenarios for these chips anyway. I believe the added shader power of the 8400M GS (note: not 8400M G) should win out most of the time.
It's all speculation at this point, since no one cares to test mobile GPUs as thoroughly as their desktop brethren.
Here's the specs I was able to find for each GPU after lots of searching:
HD 2300:
API support: DX9c, OGL2.0
GPU clock: 480MHz
pipeline configuration: 2 vs 4 ps 4 rop 4 tmu
Pixel fillrate: 1.92GP/s
Texture fillrate: 1.92GT/s
Mem clock: 550(1100)MHz
memory interface: 128-bit
Memory bandwidth: 17.6GB/s
8400M GS:
API support: DX10, OGL2.1
GPU clock: 400MHz
shader clock: 800MHz
pipeline configuration: 16 scalar alus 4 rop 8 tmu
Pixel fillrate: 1.6GP/s
Texture fillrate: 3.2GT/s
Mem clock: 600(1200)MHz
memory interface: 64-bit
memory bandwidth: 9.6GB/s
So HD 2300 has a slight pixel fillrate advantage and a massive memory bandwidth advantage whereas 8400M GS has large shader throughput and texture fillrate advantages. Also, 8400M GS is DX10 whereas HD 2300 is "only" DX9c. HD 2300 has UVD support and 8400M GS has Purevideo2 support.
Looks like performance would vary wildly between the two depending on the bottleneck of the game being played. Too bad there's absolutely no performance numbers out there for either...