Alright, more cross-testing results. Seems like 3/5 57x chips were false positives: they rebooted to 57x in the office, but when re-testing at home they did ~53x. One chip is doing 5500 6x12t and 5610 2c, which is in line with Viss' findings that there seems to be a large variance between max frequency depending on the amount of cores you've enabled. Truth be told, this comes as a suprise to me as, just like Hiwa, I found most SBE chips to be quite balanced (similar to SB). I assume it comes down to particular chips as the 4 others are behaving quite consistent (hitting the same frequency on 2c or on 6c).
One of the chips is a truely interesting chip. This one could reboot all the way upto 5843MHz in the office (tested multiple times) and seems to reboot at 57x here at home as well. When cooling down the chip, however, the reboot process gets a lot more complication. Here's what I found:
- 5100MHz, 6c12t, +21°C => OS
- 5300MHz, 6c12t, +8°C => OS
- 5500MHz, 6c12t, +3°C => reboot, but no OS
- 5500MHz, 6c12t, -7°C => no reboot!
- 5500MHz, 2c2t, -15°C => reboot, but no OS
It's the first Sandy Bridge(-E) I've seen behave like this. Really odd.
Today's tests were on the XPower-II, which seemed to give more accurate results than the R4E I tested before. Not higher OC results, just more accurate following the method. This weekend, I'll re-try the method on the R4E to do another cross-check.
Disappointing for sure ... but interesting nonetheless.