Release CPU stress test

[ZXDunny]

My pandora is an original batch 1 256Mb CC model. As Notaz has said though, this is a stress test - PCSX-ReARMed runs effortlessly with no issues at way over 1GHz. Of course, there's no need to run PCSX at that speed now


D.

 

[_wb_]

Updating the aggregated info:


- original (CC) Pandora: 778 MHz (onpon4), 969 MHz (ZXDunny), 980 MHz (Takedown; was this the final number?)


- Rebirth (GC) Pandora: 999 MHz (_wb_), 985 MHz (TrashyMG)


- 1GHz Pandora: 1220 MHz (kamui_kun; not the final number)


- unknown Pandora: 956 MHz (apersonthinks), 956 MHz (Linux-SWAT; was this the final number?)


Note that these numbers represent the max overclock speed at extreme cpu stress - under normal load, you can typically go quite a bit further.


So far not a huge difference between CC and GC cpus, although onpon4 seems to have bad luck with his unit. More data please!

 

[Ian J]

Wow!


Original (CC I think) Pandora, stable at 1006 mhz.

 

[kamui_kun]

after a few hours, stable at 1214 MHZ


in practice I used my pandora for 1 week and worked perfectly at 1225 MHZ, but it hang from the start at 1250 MHZ and up

 

[Takedown]

final number is 967Mhz

 

[_wb_]

Updating the aggregated info:


- original (CC) Pandora: 778 MHz (onpon4), 967 MHz (Takedown), 969 MHz (ZXDunny), 1006 MHz (Ian J - are you sure it's the 256MB RAM model?)


- Rebirth (GC) Pandora: 985 MHz (TrashyMG), 999 MHz (_wb_)


- 1GHz Pandora: 1214 MHz (kamui_kun)


- unknown Pandora: 956 MHz (apersonthinks), 956 MHz (Linux-SWAT; was this the final number?)

 

[Lafazar]

Hm, during the stress test it suddenly said "sudo unknown user root" and now my XFCE menu is gone.


I managed to restore it, but running the test a second time my Pandora (1GHz Edition) keeps freezing the system at ~1160MHz @ OPP4. I guess my Pandora gets particularly quickly unstable somewhere above 1150.


Edit: Okay, it seems my limit is 1157MHz @ OPP4 (1GHz Pandora), going even one above that will quickly lead to weirdness and freezing.


Is it normal that it stops at OPP4? Wouldn't OPP5 allow higher clockspeeds? The limit I set in the CPU settings is 5.

 

[fahrstuhl]

Mine (256MB CC) has been at 957MHz for quite a while now. Does this test ever finish or will it just stay at the last stable clock?

 

[Linux-SWAT]

Forgot to plug back the power supply after a trip outside :/. Restarting at the last value i saw: 936MHz.


The board is CC.

 

[foft]

Looks good - I'd been using cpuburn to check stability since that uses heavy neon.


Be great to have a way of setting max mhz too, since it climbs until the Pandora crashes on both my 1GHz units. If I limit it in the OS then it complains about other things changing the speed.


I managed to get one going and so far its at 1178MHz (for about 20 mins so far). Will see how low it goes. I'd found 1150MHz to be stable with cpuburn and 1200MHz unstable and can't remember if I tested in-between.


Will update with the other one if I figure out how to run it... It climbs to 1246ish then crashes!


edit - an hour and a half later - at 1175MHz now

 

[Ian J]

Updating the aggregated info:


- original (CC) Pandora: 778 MHz (onpon4), 967 MHz (Takedown), 969 MHz (ZXDunny), 1006 MHz (Ian J - are you sure it's the 256MB RAM model?)


- Rebirth (GC) Pandora: 985 MHz (TrashyMG), 999 MHz (_wb_)


- 1GHz Pandora: 1214 MHz (kamui_kun)


- unknown Pandora: 956 MHz (apersonthinks), 956 MHz (Linux-SWAT; was this the final number?)

Mines definitely a 256mb model, I wish it was a 512 rebirth model but sadly not.

 

[PokeParadox]

MY CC 256MB is currently sat at 956MHz at OPP5.

 

[konqithedragon]

Pandora Rebirth


855MHz @ OPP3


933MHz @ OPP4


1000+ MHz @ OPP5


seems perfectly stable at 1000 MHz and it kept going up until 1015MHz where the program errors out..


with the message:


"something changed cpu config, aborting"


probably because I set the max limit to 1000MHz

 

[diligentcircle]

Be great to have a way of setting max mhz too, since it climbs until the Pandora crashes on both my 1GHz units. If I limit it in the OS then it complains about other things changing the speed.

I got that complaint myself, kept failing to climb to 815 because of the limit I'd set. but it eventually stopped complaining and started lowering because it detected errors at 800 MHz.


So just set the max allowed speed as high as you can overclock without crashing and you should be fine.

 

[dsleaf67]

...


with the message:


"something changed cpu config, aborting"

I get the same message with at 915 MHz @OPP5 on orig CC Pandora.

 

[Exophase]

Hm, during the stress test it suddenly said "sudo unknown user root" and now my XFCE menu is gone.


I managed to restore it, but running the test a second time my Pandora (1GHz Edition) keeps freezing the system at ~1160MHz @ OPP4. I guess my Pandora gets particularly quickly unstable somewhere above 1150.


Edit: Okay, it seems my limit is 1157MHz @ OPP4 (1GHz Pandora), going even one above that will quickly lead to weirdness and freezing.


Is it normal that it stops at OPP4? Wouldn't OPP5 allow higher clockspeeds? The limit I set in the CPU settings is 5.

This sounds like kind of a cautionary tale..


The goal here is to induce errors eventually. You have to accept some risk that those errors can cause something unexpected to happen.


If I had to guess I'd expect the 1GHz Pandoras to be more susceptible to this. Because the 600MHz units seem to have much less headroom with NEON (hence the nature of this test), meaning the errors are more likely to be isolated to the NEON units and therefore more isolated. But the 1GHz Pandoras seem to be limited by other things.


I'd say make backups and be prepared to reflash in the very worst case, probably very unlikely but still possible scenario.

 

[apersonthinks]

Updating the aggregated info:


- original (CC) Pandora: 778 MHz (onpon4), 967 MHz (Takedown), 969 MHz (ZXDunny), 1006 MHz (Ian J - are you sure it's the 256MB RAM model?)


- Rebirth (GC) Pandora: 985 MHz (TrashyMG), 999 MHz (_wb_)


- 1GHz Pandora: 1214 MHz (kamui_kun)


- unknown Pandora: 956 MHz (apersonthinks), 956 MHz (Linux-SWAT; was this the final number?)

Mine is a rebirth (GC) pandora.

 

[Googer]

Original 256MB CC Pandora here, currently testing at 946 MHz @ OPP5, slowly stepping down over the past day from a high of around 980 MHz (FWIW I had it start at 950 MHz). Its failure rate at a given speed has really slowed the last few hours so it's likely going to be stable very close to where it is now. I started at 950 since that was my guesstimate speed stable speed that I'd found in anecdotal testing of it and it's looking like that indeed is very close to being right.

 

[TrashyMG]

Another go at it, this time let it run for several hours on both my Pandoras.


Rebirth (GC) Pandora: 989MHz - odd this time it seems stable a little bit higher clock speed.


First Batch (CC) Pandora: 928MHz

 

[Linux-SWAT]

GC > CC