Infrared Imaging Spectrograph (IRIS) is the first light instrument for the Thirty Meter Telescope (TMT) that consists of a near-infrared (0.84 to 2.4 micron) imager and integral field spectrograph (IFS) which operates at the diffraction-limit utilizing the Narrow-Field Infrared Adaptive Optics System (NFIRAOS). The imager will have a 34 arcsec x 34 arcsec field of view with 4 milliarcsecond (mas) pixels. The IFS consists of a lenslet array and slicer, enabling four plate scales from 4 mas to 50 mas, multiple gratings and filters, which in turn will operate hundreds of individual modes. IRIS, operating in concert with NFIRAOS will pose many challenges for the data reduction system (DRS). Here we present the updated design of the real-time and post-processing DRS. The DRS will support two modes of operation of IRIS: (1) writing the raw readouts sent from the detectors and performing the sampling on all of the readouts for a given exposure to create a raw science frame; and (2) reduction of data from the imager, lenslet array and slicer IFS. IRIS is planning to save the raw readouts for a given exposure to enable sophisticated processing capabilities to the end users, such as the ability to remove individual poor seeing readouts to improve signal-to-noise, or from advanced knowledge of the point spread function (PSF). The readout processor (ROP) is a key part of the IRIS DRS design for writing and sampling of the raw readouts into a raw science frame, which will be passed to the TMT data archive. We discuss the use of sub-arrays on the imager detectors for saturation/persistence mitigation, on-detector guide windows, and fast readout science cases (< 1 second).