FX60 owners with coldbug read this !

**Major** · 02-24-2006, 10:48 PM

After testing my FX60 using a stock vapochill LS, even modest temps of -23c envoked the cold bug and I couldn't get past 260mhz without lockups. See my resluts HERE

But after some testing with the 233 and 250 memory dividers (available with RevE chips and DFI Expert & DFI NF4 SLI) I found that the couldbug has nothing to do with memory clocks but instead is a HTT limitation.

This has been confirmed by Ozan Baseski and his results with a FX60. He too was able to run the memory with the 233 divider keeping his memory speed at a higher level than HTT and encounted no problems with coldbug. His thread can be found HERE

This will work with any RevE chip that has coldbug issues but only the FX60 has the higher multis to make it practical. On the DFI NF4 SLI you can use the 704-2BTA bios, it has the 233 and 250 memory dividers enabled in bios.

I'd like to see others confirm this and post there results here

**eva2000** · 02-24-2006, 10:51 PM

interesting. will test it out when i get my phase change unit

**dinos22** · 02-24-2006, 10:56 PM

i think i suggested that to someone who was complaining of cold bug and low HTT but it wasn't working out either.....just can't find the post...............it's great it worked out for ya......who knows maybe this will be what many needed....arn't those dividers a bit buggy or is it just CPUz and other programs not picking them up correctly?

**Major** · 02-24-2006, 10:58 PM

Originally Posted by dinos22

i think i suggested that to someone who was complaining of cold bug and low HTT but it wasn't working out either.....just can't find the post...............it's great it worked out for ya......who knows maybe this will be what many needed....arn't those dividers a bit buggy or is it just CPUz and other programs not picking them up correctly?

Cpuz 1.32 reads them correctly so does Clockgen 1.0.4.1

I've tried both the 233 and the 250 dividers and they both worked flawlessly, no more coldbug

**metro.cl** · 02-24-2006, 11:24 PM

Originally Posted by Major

Cpuz 1.32 reads them correctly so does Clockgen 1.0.4.1

I've tried both the 233 and the 250 dividers and they both worked flawlessly, no more coldbug

great find, now i only need a FX or a miracle to get my opteron unlocked

another question this htt problem is mobo related or cpu related??

**Major** · 02-25-2006, 12:12 AM

Originally Posted by metro.cl

great find, now i only need a FX or a miracle to get my opteron unlocked

another question this htt problem is mobo related or cpu related??

Its a CPU issue

This will work with any RevE chip, if your using an Opteron or X2 you wont be able to increase your OC but atleast you can get the most from your memory again and improve performace for a given OC.

**Dumo** · 02-25-2006, 01:39 AM

Thanks Major

Nice find

**stone_cold_Jimi** · 02-25-2006, 05:09 AM

Yes Major, I think your premise is correct. Although I can now boot at a higher HTT after some memory tweaks, I can confirm that a high HTT with a lower multi and a mem divider (which should work, as the final CPU and mem frequencies have previously worked fine) fails.

What I fail to grasp is exactly why this should be so. How can the CPU "see" the HTT value? Once the multiplier is applied, one would be forgiven for thinking that it is only the multiplied clock that's presented to the CPU, and its origin shouldn't matter a jot. Any ideas?

Edit: brain in neutral this morning. Plainly the multiplier is part of the CPU, so the raw clock is present.

**mdzcpa** · 02-25-2006, 05:43 AM

Great find Major!

I'll have my FX60 under my Mach II today. I'll be sure to post back here for confirmation.

**Major** · 02-25-2006, 04:12 PM

Originally Posted by mdzcpa

Great find Major!

I'll have my FX60 under my Mach II today. I'll be sure to post back here for confirmation.

Looking forward to some results !

**mdzcpa** · 02-25-2006, 05:35 PM

Sorry, still struggle along here. Having a heck of time getting a good mount. I'm going to be putting the IHS back on soom if this keeps up.

BTW, just under modest testing so far, Majors theory seems correct. I was getting idle lock up at 260HTT at -20c. Lowering the HTT, upping the multi, and using the 233 divisor DOES work

As soon as I can get a good mount and decent temps, I'll be on my way. Gamed all day at 3.2ghz. But temps above 10c keep me limited. I'm trying another remount again.

**OldOne** · 02-25-2006, 05:41 PM

I think I will go ahead and buy a Phase unit now.Was going water because of the cold bug.This is what I have now. Thanks for the info.

FX-60
XP-120 Delta fan
A8N32-SLI Deluxe
Two Gig G.Skill HZ
Two X eVGA 7800GTX KO in SLI
PC Power&Cooling 510 SLI

**Repoman** · 02-25-2006, 05:46 PM

You guys never thought of doing this before? I was wondering why no one did it with unlocked chips 6 months ago when I was a n00b, I figured there must have been some reason so I never asked

**Major** · 02-25-2006, 06:07 PM

@Mike

IMO its pretty hard getting a solid mount with the MachII head on a naked core. Keep trying I'm sure your FX has 3.4 in it

Originally Posted by Repoman

You guys never thought of doing this before? I was wondering why no one did it with unlocked chips 6 months ago when I was a n00b, I figured there must have been some reason so I never asked

Well the 233 and 250 multis only work with RevE chips to begin with, and it wasn't until the 704-2 bios that they were enabled.

My FX60 is the first chip I've had that the coldbug became an issue with my stock LS, so I've not had to deal with it b4.

**ozzimark** · 02-25-2006, 06:12 PM

Originally Posted by stone_cold_Jimi

What I fail to grasp is exactly why this should be so. How can the CPU "see" the HTT value? Once the multiplier is applied, one would be forgiven for thinking that it is only the multiplied clock that's presented to the CPU, and its origin shouldn't matter a jot. Any ideas?

Edit: brain in neutral this morning. Plainly the multiplier is part of the CPU, so the raw clock is present.

mentioned this a few days ago..

http://www.xtremesystems.org/forums/...1&postcount=19

**Major** · 02-25-2006, 06:14 PM

Originally Posted by ozzimark

mentioned this a few days ago..

http://www.xtremesystems.org/forums/...1&postcount=19

I think your right on the mark ozzimark

**Indyxc** · 02-25-2006, 06:23 PM

Great find!

Kind of a side note, but what is everyone's FX-60 doing on air/water?

So if HTT is infact the limitation, then the FX is infact really special due to the higher multiplier.

Glad to hear, maybe I'll have to invest/ build a single stage phase.

**ozzimark** · 02-25-2006, 06:26 PM

thanks major

i'd like to think i'm that lucky

the real question is of course.. why? what is it about the clockgen part of the cpu that doesn't like running at cold temps.. and only for 90nm? why isn't intel effected from what we can see? and what are the changes between the "memory controller revisions" that make such a big difference in how it acts in the cold?

maybe a difference in the transistor types, fast vs slow (thin gates vs thick gates iirc, but i'm probably getting way off base here

)

lastly.. something that could reallllly help the fx users.. options for higher than 250mhz ratios

**dinos22** · 02-25-2006, 06:29 PM

Originally Posted by ozzimark

thanks major

i'd like to think i'm that lucky

the real question is of course.. why? what is it about the clockgen part of the cpu that doesn't like running at cold temps.. and only for 90nm? why isn't intel effected from what we can see? and what are the changes between the "memory controller revisions" that make such a big difference in how it acts in the cold?

maybe a difference in the transistor types, fast vs slow (thin gates vs thick gates iirc, but i'm probably getting way off base here

)

lastly.. something that could reallllly help the fx users.. options for higher than 250mhz ratios

heheheh 150 HTT .....300/350 ratio hehehehhe

**Major** · 02-25-2006, 06:30 PM

Originally Posted by dinos22

heheheh 150 HTT .....300/350 ratio hehehehhe

LOL

**ozzimark** · 02-25-2006, 06:30 PM

Originally Posted by dinos22

heheheh 150 HTT .....300/350 ratio hehehehhe

well, not to get toooo far off topic, but one thing i've always wished for was a directly selectable memory divider, none of this "ratio" sillyness, but it doesn't seem that this is really all that possible from how amd set things up

**mdzcpa** · 02-25-2006, 06:44 PM

AHHHHHHHHHH!

Just when things were getting cooking when my PCP&C 510 SLI ASL took a dump

The rig just shut down. Thank god it didn't take anything else out with it. All I have is a 350 Antec Smartpower I use for testing. The rig fires up fine with that. Just not enough juice to continue tonight.

Sorry for the off topic Major. But your trick works nice!

**hawtrawkr** · 02-25-2006, 08:13 PM

great find here definantly something ill keep in mind in the coming weeks.

**~~FragTek~~** · 02-25-2006, 08:57 PM

Damn awesome find! Someone remind me of this thread when my cascade comes in... I'll definately be interested in trying this technique!

**stone_cold_Jimi** · 02-26-2006, 05:31 AM

Originally Posted by ozzimark

mentioned this a few days ago..

http://www.xtremesystems.org/forums/...1&postcount=19

Fantastic! I'd never realised that the clockgen was internal to the CPU. Nice1 oz.

Thread: FX60 owners with coldbug read this !

Thread Tools

Search Thread

Rate This Thread

Display

FX60 owners with coldbug read this !

Bookmarks

Bookmarks

Posting Permissions