CompuTronix's temp guide - I take issue with your concept of "delta"

First off, very thoughtful and well written guide, CompuTronix. (I'm referring to this post.) I have a problem with the concept 'delta' which I understand you to be terming the difference between what you're calling tcase and tjunction. I believe this delta isn't a constant nor is does it exist on my system under p95 stress. Let me explain:

On my lapped Q6600 (B3 stepping), stressing the CPU with x264.exe encoding video, your published 10 °C offset is more or less accurate for my system (screenshot shows a difference of about 12 °C between the tcase and the hottest core); however, when I stress the CPU using prime95 v25.5, the tcase and tjunction as you call them approach the same value after only 10 seconds of stress. The screenshot below is from HWMonitor 1.07 but I confirmed this using coretemp 0.96.1 and speedfan 4.34 beta 40.



To me this is suggesting two things:

1) That p95 does a much better job stressing the cores equally, and if you look at my screenshots, you'll see that x264 leaves a bit of a temp gap between cores 0+1 and cores 2+3. Prime95 on the other hand keeps them more or less even.

2) That the physical location of what you're calling the tcase sensor is near the cores 2+3 in my chip since it approaches the core temps when properly stressed, that is, when the temps of cores 2+3 get elevated, the tcase gets elevated nearly 1:1.

If I'm mistaken about this, can you explain my observations?

I can provide additional info about this system (board, cooling, BIOS revision, etc.) upon request.