For Richardd43

[richardd43]

Vista was a driver problem. Both systems are up and running at our last OC and looking good. Will get the new cooler sometime today and we will be ready to Rock & Roll.

[Nick N]

Excellent Richard. I thought it was software issue.

Sorry you had to deal with that, but it is one of the pitfalls to setting up a clock. Typically on new hardware I will run a blank copy of the OS and just install the drivers and supporting software I need for testing. During high clock tests it is possible for O/C file damage to occur so after I find the stable clock I will wipe it out and start clean. From that point on I know all the limits and do not exceed them so corrupt file issues from clocking are not a problem.

Lets hold now until you get the 9700 installed. At the very least you will have a much better thermal conductivity point with that unit. We will need to go backwards a bit to recheck CPU and memory limits after that is installed. Since the northbridge memory controller is stabilized by Vcore.. you may have some surprises in the memory coming we could not reach before.


One thing I would go ahead and do is drop the NB Core Voltage from 1.5 back down to 1.4

If you see any instability try 1.45 but I think it will not be needed since that was boosted to try and hit a lower memory timing.



EDIT: Pick up Arctic Silver thermal compound. Don't use the stuff they supply with the 9700. It gets a dab in the middle of the processor about the L x W of 3 grains of very fine rice and do not apply it to the 9700. Make sure to verify the 9700 does not have a plastic cover sheet on the base to protect the mirror finish. The smell from such thin clear plastic covers being missed can induce the need for a nitro tab under the tounge. You may get some minor oder from the 9700 fin and vein finish treatment but it will pass quickly.

[richardd43]

LOL on the plastic. I have not done that yet but have seen it. I bought an Arctic Silver mounting kit when I first started working on computers that I use. It has all of the required cleaner and the paste itself. I carry it with me anytime I have to go somewhere and fix a computer.

This was with the last bios profile we used.

Here is the before picture:



And the after:

[richardd43]

Here is the run with the corrected voltages:



Looks like we gained a few degrees

[Nick N]

Here is the run with the corrected voltages:



Looks like we gained a few degrees

Thats sweet

You were easily hitting 58c before and I am sure after a period of "up" time that would have creeped up to 60 with some spikes being higher

Now you are hitting 50c and I would estimate the same thermal condutivity as the AMD block running 42-44, perhaps lower.

Nice job


I am still hacking up a lung so give me a day or so to recover. I may get an energy spurt and post some things to start. Right now I think that issue you were seeing is directly related to a dual card install. The error has always shown up with twin card rigs and it is not Nv or ATi specific I will continue to look into that one.


If you would like to play... here is something you can do


Using the current O/clock settings, change the following:

9x Mult
360 CPU Frequency
Memory Clock Frequency [DDR2 400]
CPU<-> NB HT Speed 2xCPU Frequency



Advanced Memory Settings

[Nick N]

Since the above was not stable I suspect the memory may not like the divider -or- the CPU has simply reached it's limits.

Let

[richardd43]

Overclocking    
CPU Frequency 200  
CPU Multiplier 15x  
Memory Clock Frequency DDR2 800  
PCIEX_1 Clock 100    
PCIEX_2 Clock 100    
CPU<-> NB HT Speed 5xCPU Frequency  
NB to SB HT Frequency 200    
NB --> SB HT Speed 5x NB to SB HT Frequency    
SB --> NB HT Speed 5x SB to NB HT Frequency    
CPU<->NB HT Width [ch8595]16 [ch8593]16    
NB<->SB HT Width [ch8595]16 [ch8593]16    
 
CPU Voltage Range 1.55-1.65    
CPU Voltage 1.55v    
DDR2 Voltage Control 2.1    
HT Link Voltage 1.4    
NB Core Voltage 1.4  
SB Core Voltage 1.6    
SB PLL Voltage 1.65    
 
DRAM Configuration    
Tcl 5    
Tcrd 5  
Trp 5  
Tras 15  
Width of DRAM Interface Auto    
1T/2T Memory Timing 2T    
AI Clock Skew Auto      
 
Advanced Memory Settings    
Trc 22    
Twr 6    
Max Async Latency 8ns    
Idle Cycle Limit AUTO  
Dynamic Idle Cycle Counter AUTO    
 
Advanced    
PEG Link Mode    
PEG Link Mode DISABLED    
PCIPnP    
Plug & Play O/S YES    
 
Output Driver Control    
MEMCLK Drive Strength AUTO  
Data Drive Strength AUTO  
====================

F10-Save and exit.. go back into the BIOS and change the following:


Overclocking    
CPU Frequency 246  
CPU Multiplier 13x
Memory Clock Frequency DDR2 [800]
CPU<-> NB HT Speed 4xCPU Frequency

CPU Voltage 1.55v  

I need to know if the memory speed shows as: 457.1Mhz -or- 492.0Mhz

[Nick N]

[quote][quote]Overclocking

[richardd43]

Well, my guess is the 9700 is getting some run time on it, the temps are a lot better.

Overclocking    
CPU Frequency 270  
CPU Multiplier 12x
Memory Clock Frequency DDR2 [400]
CPU<-> NB HT Speed 3xCPU Frequency

CPU Voltage 1.625v

 

[richardd43]

[quote]Overclocking

[Nick N]

It appears we are making progress. With those temps it looks to me like you may be able to go to 1.65v Vcore and push further up, but that may also place it in a postion to get warm after running for a long period of time.

Once you hit the highest stable point using 10x 3xx, raise the Vcore to 1.65 and see if if will go a bit higher. If it will, run a load test for temps

[richardd43]

It looks like 325 might be the limit. I can boot and run stable at 327 but 329 refused to boot.

[Nick N]

It looks like 325 might be the limit. I can boot and run stable at 327 but 329 refused to boot.

Did you try 1.65 Vcore?

[richardd43]

Well............. no

But I will right now

[Nick N]

Well............. no

But I will right now

LOL

To tell you the truth, it may not net very much in CPU but it might help push the memory contoller. I would be interested in seeing the temps on the highest stable clock @ 1.65v

That will tell me if we should bother working with that voltage.


It looks to me like you are probably correct and 3250 is the CPU top end. If you manage to get it to 3280 stable at 1.65v but no higher, them I will start running calculations on the new top end of the processor and Vcore we think will work best.


At that point I think even if you end up at 1.62v, the NB memory controller will be much more stable and then we will start pushing the memory up on even variables so it stays on the 590 chipset