shooting for 4.2 on air, need a little help!

Also noticed something else and one of the veterans here can chime in on this, but doesn't spread spectrum when enabled (and the op said it was on auto which means enabled i would guess) cause all the voltages to go wacky and give alot of voltage spikes, potentially causing already increased voltages to skyrocket and possibly damage his board????

EUREKA!

Velocity! If I Weren't straight - i'd kiss you!
Heck, considering you probably prevented a 3rd and 4th mobo from frying, i'd kiss you anyway!

I can't remember if it was the 880, or the 890, but one of them the stock NB was 1.25
so I figured 1.3 wasn't that much of a jump.

I believe, cpu/nb stock was 1.15....

i'll have to check old posts to see what was stock on the 890...

ty...

keep this thread alive guys, I should receive my new 890 on monday or tuesday - fresh, and ready to get cooked!!!

My cpu/nb stock was 1.175 and I haven't changed that one yet after reading the cautions here.

The temps on the cpu he was reading are ok though?

p.s.

forgot to mention that when I used the "rice grain" method in applying the paste, I was even generous, i've noticed that my cpu was only 75% covered

next time ill use the credit card method

You know I was messing with that option as well but I found it dried out the past it seemed. That's probably not the best was to put it but I didn't like how it turned out and it seemed to get a very weak contact.

well, no matter what method you use, the formula for success for any paste (don't know why they call it TIM) is thinist, even coat possible. It's not the paste that provides the thermal conductivity, all it does is fill in the microscopic holes in the metal.
Theoreticaly, if you could provide a layer of goop 0.1 mil thick, it would do the job.

credit card method is old school, and although i'm not a fan of it, it at least provides full coverage.



btw bug, 1.3 cpu/nb would be fine, I had to bump mine up to at least that to stabilize my high ram frequency.

magnus28 said:

well, no matter what method you use, the formula for success for any paste (don't know why they call it TIM) is thinist, even coat possible. It's not the paste that provides the thermal conductivity, all it does is fill in the microscopic holes in the metal.
Theoreticaly, if you could provide a layer of goop 0.1 mil thick, it would do the job.

credit card method is old school, and although i'm not a fan of it, it at least provides full coverage.



btw bug, 1.3 cpu/nb would be fine, I had to bump mine up to at least that to stabilize my high ram frequency.

Click to expand...

Im pretty new to this whole overclocking thing, but 1.3 cpu/nb is high. Quietice showed me the rated numbers and i believe for the 1090t 1.17 is where it should be.. 1.23-1.25 seems to be the max ppl go on overclock from what ive read. I have mine around 1.18 atm i believe with a 3.8 ghz OC, and so far its stable.

I could be wrong though, like i said im new to this.

magnus28 said:

... for any paste (don't know why they call it TIM) ...

Click to expand...

TIM = Thermal Interface Material but that's "Old School", I guess. I'm sure "paste" is waaay cooler and more accurate.

magnus28 said:

btw bug, 1.3 cpu/nb would be fine, I had to bump mine up to at least that to stabilize my high ram frequency.

Click to expand...

Are you on board #3 yet? You fried the first one and last I heard the second one was limping badly. If I were new to this I'm not sure I'd want to follow your path into Motherboard Cemetery ...

I like to call it Thermal Glop, personally
It's the most descriptive.

TIM conducts much better then air, it needs to be thick enough to bridge any gaps caused by uneven heat spreaders on the CPU and uneven bases on the cooler.
I've had the most consistent results by putting an X on the cpu from corner to corner, then a little dot in the middle of the triangles the X makes. These are small lines. (Sidenote: for heatpipe direct touch coolers, a line down each heatpipe and a small small line down each spacer between pipes has been best for me)
If you have good mounting pressure on the heatsink any excess will simply squeeze out the sides and hang out there.
If you use far, far, far too much you'll have issues.


cpuNB does need to be raised sometimes, especially if you're populating all the memory slots or OCing the cpuNB. That said, a little goes a long way, odds are 1.25 is enough (or more then enough) for anything a non-bencher is likely to do.

The NB voltage (not cpuNB!) stock is 1.1v for 700 chipsets with no graphics, and generally 1.3 for 700 chipsets with graphics. 800 chipsets i have no idea. The iGPU chipsets get pretty toasty, a fan or a cooler that puts air over 'em is a must IMO.

Bobnova said:

I've had the most consistent results by putting an X on the cpu from corner to corner, then a little dot in the middle of the triangles the X makes. These are small lines.

Click to expand...

Never heard of that one - I'll give it a try next time I mount my TRUE.

I still tin my sinks as well, do you ever do that - or did you ever?

Bobnova said:

I like to call it Thermal Glop, personally
It's the most descriptive.

Click to expand...

And I always thought "Glop" was a verb ...

I still tin my sinks as well, do you ever do that - or did you ever?

Click to expand...

I do that as well, I use the old wipe it on with a plastic bag then rub it off with a coffee filter method.

BugFreak said:

I do that as well, I use the old wipe it on with a plastic bag then rub it off with a coffee filter method.

Click to expand...

I've got an old CC that I've cut into TIM application tools so I use one of those (then the coffee filter) but it's the same thing. Good to see I'm not the only one using the old tricks ...!

QuietIce said:

Are you on board #3 yet? You fried the first one and last I heard the second one was limping badly. If I were new to this I'm not sure I'd want to follow your path into Motherboard Cemetery ...

Click to expand...

Lol. Thanks for the reminder! I was gunning for the "most boards fried in a month award." What I meant to add was I don't think 1.3 cpu/nb was too much on 800 series boards. I've never had a problem with temps or stability at that voltage... It was bumping it up to 1.35 that caused the problems.

Bobnova said:

cpuNB does need to be raised sometimes, especially if you're populating all the memory slots or OCing the cpuNB. That said, a little goes a long way, odds are 1.25 is enough (or more then enough) for anything a non-bencher is likely to do

The NB voltage (not cpuNB!) stock is 1.1v for 700 chipsets with no graphics, and generally 1.3 for 700 chipsets with graphics. 800 chipsets i have no idea. The iGPU chipsets get pretty toasty, a fan or a cooler that puts air over 'em is a must IMO.

Click to expand...

The reason I bumped my CPU/NB up to 1.3 was because a GSKILL tech recomended it to stabilize my RAM @ 2000Mhz. I tried it, and it worked! Also 1.3Mhz helps keep my CPU/NB Frequency 2800-3000Mhz stable.

my last NB voltage was @ 1.2689 volts. That was enough to keep my 4.01Ghz stable, but I didn't have enough time testing different (lower voltages)

According to AMD the safe upper limit for the cpuNB is 1.175v for the 1090T. If you apply the same overvolt percent to the vCore of a processor with a safe limit of 1.45v (that's what most AMD's are) you'd be running it at 1.60v! I hope that puts it in better perspective for everybody.


And, yes, I'm aware of G.Skill's tech advise on this topic. Next time you talk to them you might ask them if they are aware of the safe cpuNB range as published by AMD and if they'll be replacing your CPU it if it fries. That might have been "OK" advise for the X4's (not great but OK) but for the X6's it's insane. I'd bet if they were liable for any CPU damage resulting from their "expert advise" they'd change their advise - or their advertising. How much did they suggest you overvolt their RAM past spec ...?

thx for the advice Ice... Every tidbit you offer makes me a little wiser..

the stock voltage for my ram was 1.65. They suggested 1.7.
I pushed it a little further to 1.71, and my ram ran as advertized 2000@ 9-9-9-24.
I even got it a little tighter - 2000@ 8-9-8-24.

I don't know if these stabilized because of BOTH the 1.71 and the 1.3 CPU/NB


My new board will arrive tomorrow...

After much debating and self-reflection, I've decided to not be greedy and stick with 4.01Ghz.

I will try and tighten my ram a little, and OC my NB...

My new Goal is :

4.01 Cpu
3000Mhz NB
2400Mhz HT
2000Mhz Ram @ 8-9-8-24

my proposed settings are:

1.45 Vcore
1.71 Dram
1.23 NB
1.25 Cpu/NB (has to be somewhat high to support my ram + NB Frequency)
LLC: Enabled
CPU Spread Spectrum: Disabled

what do you think?

I think you're going to need all the luck you can get to make the 3000 MHz cpuNB speed - but there's no reason you shouldn't try ...

magnus28 said:

Lol. Thanks for the reminder! I was gunning for the "most boards fried in a month award." What I meant to add was I don't think 1.3 cpu/nb was too much on 800 series boards. I've never had a problem with temps or stability at that voltage... It was bumping it up to 1.35 that caused the problems.



The reason I bumped my CPU/NB up to 1.3 was because a GSKILL tech recomended it to stabilize my RAM @ 2000Mhz. I tried it, and it worked! Also 1.3Mhz helps keep my CPU/NB Frequency 2800-3000Mhz stable.

my last NB voltage was @ 1.2689 volts. That was enough to keep my 4.01Ghz stable, but I didn't have enough time testing different (lower voltages)

Click to expand...

Don't feel bad, over the past 8 months that i have used my OCZ 1600mhz AMD special kit I have tried repeatedly to get it stable at 1600mhz, always ending up using it at 1333, and after repeated visit to their forums, time and time again I saw their staff repeatedly tell people to set the CPU-NB voltage to 1.35v, not lower, not even try 1.25 and work your way up..but straight to 1.35v.....do the people selling us equipment even care to research what it takes for that equipment to work properly in our system without destroying it???

well, i'm on the fence again.. lol.

i've decided that when my mobo gets here (it's friday now apparantly) i'm going to use zerotherm's reccomendations and apply the paste "evenly" using a credit card. i'm also going to mod y zerotherm cooler by supergluing another 92mm fan on the "pull" side. There is a plastic mount in the center of the cooler. If I can get the core temp 5 degrees cooler, I might try for 4.2Ghz clock speed. If that fails, i'm going to try and run my NB frequency to 3Ghz.


Ice can you provide a link so I can read up on AMD's safe limits? I tried googling it, but no success.

also if you don't mind explaining "tinning" your HSF.
never heard of it.

magnus28 said:

Ice can you provide a link so I can read up on AMD's safe limits? I tried googling it, but no success.

http://support.amd.com/us/Processor_TechDocs/43375.pdf

also if you don't mind explaining "tinning" your HSF.
never heard of it.

Click to expand...

Tinning a heatsink involves putting TIM on the surface of the sink then working it in with a plastic bag or CC. Use slow, circular motions pushing down on the sink as you move the TIM around. The idea is to fill the microscopic dents and scratches with TIM ((a "typical" application leaves air in these, which slightly reduces heat transfer)). After you've done that you take a coffee filter (it's lint-free) and lightly wipe the surface to remove the excess TIM. If you do it right the sink should look slightly discolored (gray) and dull looking with just small glimpses of the metal here and there.

The CC method of spreading tends to create a lot of little air pockets, that's why most TIM manufacturers don't recommend it. Bobnova uses an X pattern on his with small dots in each open space. I use the big dot in the middle method. Either way you also want to twist the sink back and forth once or twice (while pushing down) just before you tighten it ...

thx for the in depth info.. mobo just arrived.

going to keep it at 4.01Ghz (temps are not low enough to go past 4Ghz.) i'm going to try and push my ram either past 2000Mhz, or tighten it down to cl8.