Introduction
In this page you'll find a technical test. Please refer here for an introduction, details and testing method.
NOTE: this test is not done with the true RAW format as usual in all my testing, but with sRAW, a reduced format with nearly 50% of the linear image dimensions. In principle it should retain the flexibility of a RAW format but with smaller file size. Unfortunately this is absolutely NOT a RAW format because in-camera the data is heavily processed (mainly de-bayering, pixel binning and Lab conversion). However I've tried this kind of testing anyway to see for any other "behind the scene" properties. Nothing significant has been found, confirming that Canon firmware is doing only minor processing to the data, as desirable.
Available sensitivities:

• standard: from 100 to 6400 ISO with 1/3 EV steps
• extended: 12800 ISO
  

Main results

• Gain. The measured gains follow quite accurately the power law.
  
• Unity gain at 846 ISO
  
• Read noise. A constant decrease is found up to 1600 ISO. Using 3200 ISO or more produces no significant improvements on the read noise. For best results in low light circumstances, mainly use 1600 ISO.
  
• Dynamic range. The full well capacity has been found between 29 and 32kADU, so the dynamic range is computed and reported both in the table and graph below. The best dynamic range is obviously at 100 ISO, but also 200 and 400 ISO has a similar dynamic range (in truth the best value is at 200 ISO, but the difference is meaningless and below the measurement uncertainty). Above that sensitivity the dynamic range start to decrease noticeably. For best results on bright subjects, use 400 ISO or less.
  

True sensitivities
To distinguish between true and software interpolated sensitivities, the best method is to look at the histogram of a nearly uniform image (like a flat field). The result is shown in the figure below, and the conclusion is:

• all ISO settings do not show any gap, and this is clearly explainable with the firmware processing done on the RAW data that is removing any eventual gap in the histogram.

Gain and read noise