Release CPU stress test


^ TI says it will last at least 100000 hours if you run OPP5 no longer than 23000 hours out of them.
 
Last edited by a moderator:
Need more data at OPP3!


At max OPP (OPP5 for the 600 MHz CC/GC units, OPP4 for the 1GHz units), these are the statistics so far:


Average: CC: 960, GC: 988, 1GHz: 1205


Minimum: CC: 917, GC: 947, 1GHz: 1157


90% of the units can go higher than ... MHz: CC: 930, GC: 953, 1GHz: 1160


Median (50% of the units can reach this speed): CC: 967, GC: 999, 1GHz: 1214


Maximum: CC: 1006, GC: 1017, 1GHz: 1249


This information could be useful for the overview table in ED's shop.
 
To put 50000 hours into perspective. If you ran at OPP5 and averaged 8 hours usage a day for the rest of the Pandoras lifetime, it would last over 17 years :)
 
Rebirth edition reporting in: 837mhz @ opp3 stable. Not tried opp4 or opp5.edit: 838mhz reported errors after 11.5 hours
 
Last edited by a moderator:
It was at 969 for 7 hours last night. Ran at 970 for about 2 hours before that, so I'm going to assume 969 is pretty close to safe.


This is strange because I remember doing tests before and having N64 lock up consistently at anything over 920. Maybe I'm crazy.
 
This test specifically tests NEON (I suppose this is the most demanding part of the CPU). Maybe your lockups are related to the GPU? Or maybe even weird bugs in the emulator that cause it to hang? Maybe the screen froze and everything became unresponsive, but you could still ssh into your Pandora?
 
Tested friend's units:


CC no-wifi : 945@OPP5


CC noisy no-wifi : 945@OPP5


CC noisy no-wifi : 944@OPP5
 
My own unit was reported lower, and tested in the same house (but not the same exact place).
 
Last edited by a moderator:
Tested friend's units:


CC no-wifi : 945@OPP5


CC noisy no-wifi : 945@OPP5


CC noisy no-wifi : 944@OPP5

Forget about these, i think the test didn't run enough. Had some surprise when running longer on another unit.


Will continue testing later.
 
Well as it's getting colder outside where I am I re-ran the test outside in 12 C temperatures.


I managed to get a stable 1014 MHz this time compared to my 1009 MHz last time with indoor temperatures.


will try again when temps start getting into minuses.
 
Last edited by a moderator:
Yay, playing n64 in the fridge :)

Sadly the light turns off when I close the door in the fridge.


It is interesting tho to see how temperature can effect clock speeds tho. If you live in a colder climate you probably will get better clock speeds.
 
On my rebirth unit, I ran stress test from Thursday night till Friday evening (about 18 Hours). Results 993@OPP5


One thing I did note was that it would have continued to run, the pandora stopped charging. Not sure why. When I noted it, it was down to 9% battery life and giving me warning messages. I stopped the test and rebooted and the Pandora started charging again.
 
Back
Top