For Richardd43

[richardd43]

She is very understanding of guns so I told her it is like having a 375 Magnum that will not take the barrel pressure of the 357 and can only fire 38 Specials. Have to beef up the barrel to take the pressure of the heavier loads.

[Nick N]

She is very understanding of guns so I told her it is like having a 375 Magnum that will not take the barrel pressure of the 357 and can only fire 38 Specials. Have to beef up the barrel to take the pressure of the heavier loads.

A good cross reference always works!

You ARE lucky! I always had to use jewelry or credit card references to make a point

[Nick N]

OK, one already came up..


This test will tell us if the Vcore increase will stabize the CPU -and- the memory at a higher speed. You just ran a low speed memory and HT test to find the highest CPU. Now we are going to see if the Vcore of 1.625v -and- using the 590 chipset math will run the CPU -and- memory boost together



CPU Voltage to 1.625v

Overclocking
CPU Frequency 232
CPU Multiplier 14x
Memory Clock Frequency DDR2 [800]
CPU<-> NB HT Speed 4xCPU Frequency

If it is unstable, raise Vcore to 1.65v and retest... no higher. You are correct that 1.7v is over the top

That should net the following numbers:


CPU = 3248
FSB = 232
HT = 928
Mem = 464


If you can not run that, the we have to lower the CPU speed until we find the balance between that speed and the memory stability. I will run different sets of numbers to take that into consideration but lets see if it will run the above. If it makes it into Windows but blows out during 3DMark... then we are close.

Here is the order to try if that first run won

[richardd43]

CPU Voltage to 1.625v

Overclocking
CPU Frequency 232
CPU Multiplier 14x
Memory Clock Frequency DDR2 [800]
CPU<-> NB HT Speed 4xCPU Frequency

I did not take a picture of the temp but there was 1 spike at 54 Deg

[Nick N]

CPU Voltage to 1.625v

Overclocking
CPU Frequency 232
CPU Multiplier 14x
Memory Clock Frequency DDR2 [800]
CPU<-> NB HT Speed 4xCPU Frequency

I did not take a picture of the temp but there was 1 spike at 54 Deg

Just for giggles, bump up the CPU<-> NB HT Speed 4xCPU Frequency

[richardd43]

almost forgot... how did the 3DMark score turn out?

I was in multimonitor to watch the temps so am not sure how much it varied but it was up some. Will run it again before I make any changes.

[Nick N]

almost forgot... how did the 3DMark score turn out?

I was in multimonitor to watch the temps so am not sure how much it varied but it was up some. Will run it again before I make any changes.

I think I may have found the sweet-spot... and I am thinking instead of using a high HT like I suggested above, we may want to go this route:


Based on what you accomplished stable there is one more set of numbers we can try and I think they will work better because the HT remains at spec and it may give us a bit of room to work memory skew harder...


CPU Voltage to 1.625v

[richardd43]

Here are the results for 5X:









Am off to run the next set

[richardd43]

3D scores for Default Bios was 4850

[Nick N]

3D scores for Default Bios was 4850

You have to be kidding


I admit the clock we got was not anything to go down in the record books but 8.3% CPU, 16% FSB (25% FSB if 250 works) and 16% Memory should have yeilded a heII of a lot more than the value in performance of 40pts

3DMark06 is so freaking CPU bound to sell Intel processors, its not funny. The CPU type alone will drive that program up. One of the reasons I dont use it for any of the older cards and AMD systems. Its rigged to favor the sale of target hardware, in my opinion.

3DMark03 and 05 would be a better test to run between the default and new clocked BIOS. I would use 03 because it has a fair balance of VC and CPU in it. They started the CPU trend in 05 and expanded that in 06.

[richardd43]

I have 03 and 05 will run 03 next time and do the comparison.







06 this run was 4891

[Nick N]

I have 03 and 05 will run 03 next time and do the comparison.







06 this run was 4891

I think that is where we want to work this now... 13x 250. Save the post above for your information in reducing FSB and trying other settings in case of some type of instability after we get finished with the clock project. You can use it as a just in case fallback to trim it out.



I have 06 pro here.. I will run it on a 7900GTX and see what it scores on a system that is similar in O/C to what you have but much slower components. I can do a conversion of the clocks and see if it follows the curve out to your score. If it does, what you are getting is probably a direct result of the test software, not your hardware. It would be nice to see what 03 and 05 show.

I assume when you say the default BIOS you mean it is running 3000 CPU, 1000HT, 400Mem on 4-4-3 10 23

[richardd43]

Well, 03 would not run at the current settings so ran default bios to make sure.

Which way do we want to go now....

[Nick N]

Well, 03 would not run at the current settings so ran default bios to make sure.

Which way do we want to go now....

3Dmark 03 crashed on the O/c setting and ran on the default, but 3DMark 06 ran on O/c?

Thats strange...


Ok

drop it down... 1Mhz

[richardd43]

LOL..... OK