My possible solution to the coldbug

[alexio]

I was going through some AMD papers about the pinout of s939. Now my English isn't very good so I don really understand what they mean with "assert" but if I'm right after reading this, just by removing the PWROK pin or isolate it the thermal protection will be shut down and maybe fix the coldbug of the 90nm cores (if it is because of this protection).

3.7 THERMTRIP_L
The processor provides a hardware-enforced thermal protection mechanism. When the processor’s die temperature exceeds a specified temperature, the processor is designed to stop its internal clocks and assert the THERMTRIP_L output. THERMTRIP_L assertion is only valid when PWROK is asserted and RESET_L is deasserted.

THERMTRIP_L assertion indicates the processor die temperature has exceeded normal operating parameters. PWROK must be deasserted in response to a THERMTRIP_L assertion to enable proper
processor operation.Once asserted THERMTRIP_L remains asserted until RESET_L is asserted. If the processor’s die temperature still exceeds the thermal trip point when RESET_L is deasserted, THERMTRIP_L will immediately be reasserted and the processor’s internal clocks stop.


PWROK is pin E8 BTW


If this has nothing to do with the coldbug, then can someone explain to me what is exactly causing it?

regards,

alexio

[z24]

OPB just TEST on 4600+ no cold bug finally...........so i think its just a accident on some chip got cold bug and some chip doesn't.........

Look at OPB's Pcmark 10k break Lo

[alexio]

http://img.photobucket.com/albums/v342/AlexKok33/939pins.jpg

Someone with the coldbug problem willing to try this out ? If the thermal protection is indeed what's causing the coldbug, this might solve it.

If this works, this is a huge find :slap:

[NiCKE^]

If it works it's a great find! Hope someone can try it out.

[EMC2]

Hi Alexio :)

"Assert" means to take the signal to it's "True" or "Active" state, "Deassert" means to take it to it's "False" or "InActive state".

What you suggested about PWROK won't work... when PWROK is not asserted (not driven to it's "true" state, which is a '1') the CPU disconnects the memory voltages and control signals from its internal circuitry. No PWROK assertion = CPU shutdown.

Peace :toast:

[alexio]

Hi Alexio :)

"Assert" means to take the signal to it's "True" or "Active" state, "Deassert" means to take it to it's "False" or "InActive state".

What you suggested about PWROK won't work... when PWROK is not asserted (not driven to it's "true" state, which is a '1') the CPU disconnects the memory voltages and control signals from its internal circuitry. No PWROK assertion = CPU shutdown.

Peace :toast:

Then we have to make it think that it assertes don't we?

What you saying is that if it can't put a "1" or zero out of the cpu to the mainbord it will shut itself down, right?

Or does the mainbord shuts the cpu down when it don't get's a signal?

If we can make the cpu beleave the value of true is being transmitted, shouldn't it run fine then?

I'm sorry for being a total n00b :D

I will read some more about the s939 pinout and stuff.

[Rave]

So what happens if you insulate THERMTRIP_L ?:)

Edit: having re-read the original post it does look to me as if the processor itself is responsible for shutting itself down and the THERMTRIP_L assertion is just its way of informing the motherboard what has happened:(. Hopefully I'm wrong though:).

[alexio]

So what happens if you insulate THERMTRIP_L ?:)

Edit: having re-read the original post it does look to me as if the processor itself is responsible for shutting itself down and the THERMTRIP_L assertion is just its way of informing the motherboard what has happened:(. Hopefully I'm wrong though:).

Seems that is the case yes. If the complete area for shutting down the cpu is within the cpu itself there's not much chance that you can change anything by isolating pins.

Well if someone has the magic suggestion just post it in here.

[7he]-[0rr0r]

from what i've read on this very forum it is the memory controller that is sometimes affected by by sub zero cooling some of the controllers just won't operate (once they get below their breaking point which is different for each processor)
if emc2 says it's not going to work it may not be the word of god but its dang close i've been lurking here a while and his knowledge and the knowledge of many of the members here is humbling.

[i found nemo]

theretically the colder the better, it's good for fast electron movement, i really dunno why they mess up @ cool temps...

[eddieate]

intresting, where did you find these pin out diagrams?
thanks.
/Ed

[ozzimark]

holy thread bump!

and they can easily be had over at amd.. :D
http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_739_7203,00.html

[cdelong]

theretically the colder the better, it's good for fast electron movement, i really dunno why they mess up @ cool temps...

there's not many electrons moving in ice??

[Dumo]

Cold bug is our problem...collectively. Hope we can find a solution. Even a way to get around it:toast:

[eddieate]

holy thread bump!

and they can easily be had over at amd.. :D
http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_739_7203,00.html

Thanks, google wasnt turning much up :toast:
/Ed