coldbug

[chew*]

Sounds like an honest mistake. Hope you didn't yell at her.

Nah she meant well, she felt bad afterward.....I't was my main rig but no biggie.......I've done worse intentionally to my rigs......

[dinos22]

sorry for not reading the entire thread in case someone mentioned this but in my experience with AMD the cold bug was directly related the HTT and you could "cheat" the level of bug by lowering HTT

i don't see how that would be a temp sensor issue on K8

[Aussie FX]

Can never be sure, but that's what the signs around now show. SOI.

If SOI is broken, why can Power hit 5ghz with seemingly no problems?

I've got no idea why AMD are having process issues but if IBM can get SOI to 5+ghz then it seems to me that isn't the problem.
Although I can't recall if IBM are using high K.

[chew*]

dino i kinda disproved the high htt theory on manta......350 x 4 multi.......and 300 x 5 multi..........

[KTE]

If SOI is broken, why can Power hit 5ghz with seemingly no problems?

I've got no idea why AMD are having process issues but if IBM can get SOI to 5+ghz then it seems to me that isn't the problem.
Although I can't recall if IBM are using high K.

IDK about that chips design and material selection because I haven't studied it, never needed to, so I can't speak for that and each IC is very different despite it using the same design process. However SOI isn't broken in itself, but it does have material and implementational limitations with the current properties of the chemicals used in K10. They can enhance these properties and get far lower current leakage in the "on" and "off" states, higher drive current and better transistor performance throughout, which will give you better MHz at lower power/temps/volts idle and load. One of these methods is to employ a relatively thin high-k dialectric/metal gate CMOS which is exactly what they've been working on with IBM and for the above specified reason.

There's a two things:
a) a design/material frequency and current limit
b) something which causes coldbugs

The first you can't address, only AMD can.
The second you maybe can address.. if you knew what it was. (hence the thread)

And yeah saaya, what you said is true. As it gets colder, there is much less resistance and more voltage actually gets to the cores, which also affects current leakage as it decreases drastically and more is then provided to the cores. There is a massive difference with the amount of current leakage in cold/hot temps, like 40x at 50C compared to 0C, if not more.

[dinos22]

dino i kinda disproved the high htt theory on manta......350 x 4 multi.......and 300 x 5 multi..........

BH chips were known not to coldbug at all chew*
you know that S7 did a lot of work on this and could not cold bug those sons of es even at -95C or whatever the temp was again

[chew*]

I know, remember I started the thread on them but if we compare a BH chip not known to coldbug to a non BH chip............I bet saaya could figure it out....

[saaya]

nah its 939 saaya......and i had 3 chips no coldbug.....all were LDBHE, although i did cherry pick these for that exact code....I wasn't picky once i found them with that code......

The one i still have was the best of them all....I lost my 3.0 gig 24/7 stable chip on grouper, wife shut the pc off and not the mach 1..........needless to say when she turned it back on the next day there was a fire in the PWM area and the chip screamed ohhhhhhhnoooooooooeeess.......

As you can see at -31c it was still cranking.......and cold boot temps were colder.......problem was my homemade DI tube just couldn't get any colder....http://www.xtremesystems.org/forums/...51&postcount=4

ouch!
no coldbug at all?
awesome!
did you notice anything else?
did the temperature probe work the same as on other cpus?
what about memory speed at low temps?

True.
Locked usually but for the unlocked Phenom BE models. 9500/9600 at 1.25VID draws 18.9A per core, lower is even lower and if you increase VID (P-States) you'll see a massive jump at the same VCore from 105W idling to 330W idling at the same MHz.

you think the current is "unlocked" for the BE cpus?
that... sounds weird... so basically there is an overcurrent protection but its... disabled for BE cpus?

your idle power jumped up to 330W????

sorry for not reading the entire thread in case someone mentioned this but in my experience with AMD the cold bug was directly related the HTT and you could "cheat" the level of bug by lowering HTT

i don't see how that would be a temp sensor issue on K8

for some people lowering the mem speed helped, for others it didnt afaik...

I know, remember I started the thread on them but if we compare a BH chip not known to coldbug to a non BH chip............I bet saaya could figure it out....

pheew... dont think i could figure it out
id need compressor cooling dry ice or ln2, and i dont have the money for compressor cooling atm, i dont have a pot and i live at the edge of a rather small city in germany so getting dry ice or ln2 wont be easy... :/

i think KTE would be better at researching why your cpu doesnt coldbug.
maybe i can digg up some amd contacts to find out why this stepping doesnt coldbug.

how does it oc compared to other steppings? with aircooling and rather stock voltage i mean?

[dinos22]

for some people lowering the mem speed helped, for others it didnt afaik...

my tccd ram did not run well in cold
only bh5 sticks from corsair did
ocz ones were bad

[KTE]

^^Yep. 74A to 110A allowed choosable with VID values IMO.

330W AC would be like 277W DC on that system from 84W DC.

[saaya]

my tccd ram did not run well in cold
only bh5 sticks from corsair did
ocz ones were bad

hmmm so the drive strength changed under low temps and the mem didnt work that well anymore... or was it the actual mem getting cold and then not working well?

^^Yep. 74A to 110A allowed choosable with VID values IMO.

well when you up the vcore you the current increases as well right? are you sure you can actually select how much current can flow through the cpu and its not just a sideeffect of increased vcore?

330W AC would be like 277W DC on that system from 84W DC.

wha?
power consumption?

[dinos22]

hmmm so the drive strength changed under low temps and the mem didnt work that well anymore... or was it the actual mem getting cold and then not working well?

just couldn't get the rig stable with TCCD ram i had from G.Skill and CH-UTT from OCZ but older school corsair BH5 was find and benched just as it did on air

[KTE]

well when you up the vcore you the current increases as well right? are you sure you can actually select how much current can flow through the cpu and its not just a sideeffect of increased vcore?

Officially those are the allowed amp tolerances on released Phenom CPUs and you can find the amp values in various MSRs connected directly to VIDs. You can also increase the VID to 1.55 on Phenom BEs and that gives you 1.45xV-1.9xVcore possible.

For the above power test run, voltages were kept the same.

wha?
power consumption?

Yep.

[chew*]

memory speed was not impaired due to cold temps saaya, it was impaired period due to premature bios.....whether on stock cooler or cold.....

I'm sure on a dfi it would be more compatible with memory at high speeds cold.

Temps appeared to read the same as any other cpu.

For -31 I was using a 2 1/2" copper tube with an endcap soldered on and an old swiftech waterblock pin fin base soldered in the endcap......so -31C loaded at 1.6v seemed realistic........as my DI tube wasn't that efficient....

Here's some interesting info also.......My chip that is dead died once before.......Poor insulation and not enough dielectic grease resulted in a failed OC under phase and no post.....I tossed the chip under a stock heatsink and the memory controller was on the fritz and bios was reporting single channel only as well as memtest......on top of that it was taking 2 minutes to boot bios.......I hit it with 1.7v to end its misery and primed it.....

It crashed, I figured it was over. PC rebooted and memory controller was back online dual channel and all....My guess is whatever transistor or connection was severed inside the chip was reconnected during exposure to extremely high volts/temps.......

[saaya]

lol, freaky stuff we discuss here hahaha
literally torturing cpus until they die and then resurrect them with high volts

its AAALLLIIIIVEEEE

[chew*]

lol, freaky stuff we discuss here hahaha
literally torturing cpus until they die and then resurrect them with high volts

its AAALLLIIIIVEEEE

haha now that you mentioned it it reminded me of the gayol commercial where they had the yellow dude on the stretcher, we can rebuild him we can make it faster mwahahahaha.... anyway let me know if you want to borrow this chip for your studies.......I promise i didn't toture it.......well maybe just a little.......

I promise it wasn't any worse than my recent ventures pushing a Q6600 to 3350 on stock heatsink