llano overdose

[Soulkeeper]

I delidded all those 3870's and overclocked them (cept 1 which I broke)

seems they all overclock nearly exactly the same

3.3GHz 800MHz nb is the most common
one did 720 for the nb, but would do 3.4GHz core
also I noticed the default voltage for the newest one is 1.4v instead of 1.4125
and they stopped putting the foam square under the cpu in the packaging

[ShadowVVL]

someone loves llano

lol photo reminds me of the setup in my old appartment with the box spring and mattress on the floor.

Any plans on getting trinity? or is there no delidding trinity?

[VirtualLarry]

Originally Posted by Soulkeeper

and they stopped putting the foam square under the cpu in the packaging

that sounds like it could be dangerous to the pins during rough shipping.

[Centauri]

I doubt the CPU weighs enough to bend pins by bouncing around.

[Idontcare]

Originally Posted by Soulkeeper

I delidded all those 3870's and overclocked them (cept 1 which I broke)

seems they all overclock nearly exactly the same

Did you notice any pre vs. post delidding improvements? Cooler temps, higher OC's, lower power usage, etc? Just curious what kind of dividends you reaped from your efforts.

Originally Posted by VirtualLarry

that sounds like it could be dangerous to the pins during rough shipping.

My FX-8350 cpu came in basically the same type of cpu retaining clear plastic clamshell as my Intel 2600k and 3770k. Pins were fine, the cpu is really well immobilized by the package.

In handling the FX-8350 I had no cause for concern that the pins might get damaged. AMD knows what it is doing.

[Soulkeeper]

Originally Posted by Idontcare

Did you notice any pre vs. post delidding improvements? Cooler temps, higher OC's, lower power usage, etc? Just curious what kind of dividends you reaped from your efforts.



My FX-8350 cpu came in basically the same type of cpu retaining clear plastic clamshell as my Intel 2600k and 3770k. Pins were fine, the cpu is really well immobilized by the package.

In handling the FX-8350 I had no cause for concern that the pins might get damaged. AMD knows what it is doing.

I only noticed a few C lower temps, no higher overclocks.
I was read thru this and am starting to think the motherboard (or possibly any fm1 board) is just not up to the task over high overclocks. I can't help but also consider there might be a flaw in the whole fm1 design, possibly part of the reason they jumped to fm2 so fast.
I had to put scotch tape over the SMDs because there is maybe 1mm of space between them and the heatsink after removing the IHS, could possibly contact and short if the heatsink is heavy or pushed on.

I also noticed that the capacitors have different colors from llano to llano, as if they individually choose them. Also the default voltage for CPU/NB varies and is lower in the newest ones (1.4v core vs. 1.4125v).

[Soulkeeper]

Originally Posted by ShadowVVL

someone loves llano

lol photo reminds me of the setup in my old appartment with the box spring and mattress on the floor.

Any plans on getting trinity? or is there no delidding trinity?

No plans to get trinity, my next system will support ecc memory, more than 4 cores, and more than 4MB of cache. I'm hoping I can hold out for steam roller or even ddr4 just to see what AMD's next platform does. If I had to do it all over again i'd have gone socket 1155 or am3+.

[Soulkeeper]

I purchased another fm1 motherboard off the discount rack at the local fry's (my 4th fm1 board). The GA-A75M-UD2H died on me after a week of instability and problems booting it stopped seeing hard drives.

F1A75-V Pro
6+2 vrm likely the board I should have gotten in the first place.
It supports the samsung 30nm mem (GA-A75M-UD2H did not).
Amazing throttling ability for low power use, stable as a brick.
The only downside is the northbridge freq is locked so it's about 5% slower than the gigabyte board at the same clocks/settings.
Also it appears asus automatically sets the nb div by mem speed as per the following table I worked out. So with my chip only able to do about 800MHz nb that limits me to the lowest mem speed of 1064.
Chart of FM1 clocks
Almost certainly mapping 6.0 to 4.0 and advancing in lockstep.

Also the sata stops working at 133bclk, like all the other fm1 boards, (the asmedia chip likely works, but I didn't try it). So just like the other 3 fm1 boards 133bclk is a pipe dream.

This should tide me over untill haswell or further

[sm625]

So you delidded 5 cpu's, broke one, then posted a picture of the boxes. Tell me what sense is there in any of this? Why risk 20% damages? Instead of buying 5 and breaking one, why not just buy 4 intel and not mess with them?

[Haserath]

Originally Posted by sm625

So you delidded 5 cpu's, broke one, then posted a picture of the boxes. Tell me what sense is there in any of this? Why risk 20% damages? Instead of buying 5 and breaking one, why not just buy 4 intel and not mess with them?

And you call yourself an 'enthusiast'.

[Soulkeeper]

Originally Posted by sm625

So you delidded 5 cpu's, broke one, then posted a picture of the boxes. Tell me what sense is there in any of this? Why risk 20% damages? Instead of buying 5 and breaking one, why not just buy 4 intel and not mess with them?

Chances are i'd delid the intels too ...
the ivy bridges lack SMD caps under the IHS so it'd be easier.
It's not fun unless I push the limits.

[sm625]

Originally Posted by Haserath

And you call yourself an 'enthusiast'.

Well.... I might shake my head at this whole endeavor but I still want to see pics. It was the pics of nothing but boxes that made me

[Shellls]

I think it your MOBO was the problem. My early production 3850 got 3.5 ghz on stock voltage on my A75m-UD2H.

[Phynaz]

Originally Posted by sm625

So you delidded 5 cpu's, broke one, then posted a picture of the boxes. Tell me what sense is there in any of this? Why risk 20% damages? Instead of buying 5 and breaking one, why not just buy 4 intel and not mess with them?

Dude, chill.

[frozentundra123456]

Originally Posted by Haserath

And you call yourself an 'enthusiast'.

You have a point I guess, but to me there are 2 kinds of 'enthusiasts". One insists on tinkering and trying to get the most out of a lower performing part, and the one who just want to buy the best part to begin with. The first would be fun, but I am not very mechanically inclined, so it does not appeal to me. I would still consider myself an enthusiast though, simply because I am very interested in PC performance and gaming, and read everything I can lay my hands on about the topic.

[Centauri]

I believe the term which would apply to a nerd who's scared of tinkering is Armchair Enthusiast.

[Idontcare]

Originally Posted by Centauri

I believe the term which would apply to a nerd who's scared of tinkering is Armchair Enthusiast.

Are we so plentiful in numbers for our demographic that we have the luxury of further segmenting ourselves into denigrating sub-categories?

This stuff makes no sense to me, and not trying to pick on you. We see it all the time with the epeen quips for people who like to buy expensive hardware too.

To what end? What possible purpose is served in belittling or labeling our fellow enthusiasts with unflattering terms?

Our demographic is diverse, true, and none of us are going to identify ourselves with each and every aspect of the genre. But that is cause to celebrate how far (and accepted) our genre has come, not a reason to point and laugh at the people we might feel are the fringe elements of it.

[Jen]

my fm1 asus mini itx board would do 133 fsb as long as sata mode was set to ide . i think max was 137 fsb before things got strange so i just used it at 133 fsb


Jen

[spartan805]

Trying to get into LLANO, i just picked up an A6-3650

[Soulkeeper]

3.5GHz 1.391v - stable
3.6GHz 1.391v - unstable
3.6GHz 1.404v - unstable
3.6GHz 1.416v - stable

actual volts is -0.009 from what's set, not worth changing the vrm freq when I can just lift the voltage .0125
1.4125v is the max/default according to AMD, so this is excellent.

I got 3.6GHz stable
good motherboard/vrm makes all the difference as I had suspected
Maybe i'll try for more later.

[ctsoth]

Originally Posted by Soulkeeper

3.5GHz 1.391v - stable
3.6GHz 1.391v - unstable
3.6GHz 1.404v - unstable
3.6GHz 1.416v - stable

actual volts is -0.009 from what's set, not worth changing the vrm freq when I can just lift the voltage .0125
1.4125v is the max/default according to AMD, so this is excellent.

I got 3.6GHz stable
good motherboard/vrm makes all the difference as I had suspected
Maybe i'll try for more later.

That is an excellent overclock at that voltage. I've gone up to 1.6v in the past and my 3870k will not run stable at 3.6.

I have f1-a75pro/csm. I don't know what the 'csm' bit stands for.

[Soulkeeper]

Originally Posted by ctsoth

That is an excellent overclock at that voltage. I've gone up to 1.6v in the past and my 3870k will not run stable at 3.6.

I have f1-a75pro/csm. I don't know what the 'csm' bit stands for.

csm are office oriented, they underclock the memory controller 1 step for stability
identical otherwise
"Corporate Stable Model"

F1A75-M Pro is 4+2 vrm
Just like others, with that many phases or less, this seems to keep the quad llanos capped around 3.3-3.5GHz
Also I noticed a feature in the asus bios for cpu/nb "current capability", from what I gather there is a hard wall related to the over current protection and this must be raised or system will just not manage higher clocks.