Hi. I am the designer of the SS3602 discrete op amp. What questions did you have?
Discrete op amps can outperform AD, LT, etc in numerous ways, including output current (drive) capability, higher internal bias currents and higher class A bias in the output stage. These things are due to the fact that discretes can dissipate more power than their ICs counterparts. They also allow for far better compensation capacitors and compensation schemes. Capacitors are difficult to fabricate in ICs, and the dielectric they are made from is inferior to a discrete NPO or C0G capacitor. With discretes - one can use 2 pole compensation which allows for vastly higher loop gain (140dB out to 600Hz) that can be sustained deep into the audio band. IC op amps can not do this because they are mostly all single pole compensated due to the fact that they can not fit an additional large value capacitor inside the IC that is required for a 2 pole scheme. This comes with several drawbacks that are explained in the following white paper (if anyone is interested) -
http://sparkoslabs.com/audio-op-amps-ga ... bandwidth/Jfets typically have higher noise than BJTs. I would think putting a JFET input stage on the front of a BJT input op amp would worsen the noise performance. The SS3602 uses a BJT input and has a noise of 2.9nV / Root HZ in 20KHz bandwidth. This is superior to most "audio grade" IC op amps.