They have the same test with the 5D Mark II....see the difference....makes mee feel good.
Included RAW files....
After lookiing at the exif file and the exposure time /f stop, I was disappointed to find that the lighting conditions used would give low noise values not representative of actual use.
The images are really not suitable for evaluation of high ISO low light images, since they were taken under bright lighting of approx ev 10. This will result in lower noise than you would get in low light.
So far, only DPR gets it. Their sample images were taken in low light, about ev 3 which is a more realistic test for low light high iso shots.
True, this is a comparison under artificial conditions not likely to be encountered in a real-world shoot. We can't accurately judge the true high ISO performance of either camera based on the the Imaging Resource tests. However, since both cameras enjoy the same advantageous shooting conditions, might it still be possible to draw at least preliminary conclusions about their relative high ISO performance? I.e., neither camera is as good as the tests suggest, but the apparent 2+ stop difference in their performance might carry over into real-world conditions?
Are there optical/physical/electronic phenomena which would make this an invalid inference? If not, then given what we already know about the 5D2's performance, IR's "comparometer" should give us a rough handle on the low-light capabilities of the 5D3. (I think we're all hoping that's the case, because the difference in the images at comparable ISOs is fairly stunning.) Pity that DPR didn't elect to replicate their 5D3 low-light shots with a 5D2.