BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

[profJim]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

Glad to hear that running Prime95 as an admin fixed the problem with the Main Thread window.
On my systems I have only one user account which automatically makes it an admin account so I've never had the problem.

Will your system prove to be error free in all Prime95 tests if you drop the DRAM voltage back to the 2.00 volt setting? Maybe yes, maybe no.

HWiNFO reports the maximum DIMM voltage as 2.08 volts and this is well below the maximum recommended setting of 2.20 volts.
Bottom line, don't worry, be happy.

Here's what I recommend for a modest overclock:

Always press CTRL + F1 at the main bios window before making changes.

CPU Clock Ratio ---------------- : [[B][COLOR="#0000FF"]7x[/COLOR][/B]] 
Fine CPU Clock Ratio ----------- : [[B][COLOR="#0000FF"]+0.5[/COLOR][/B]] 
CPU Frequency -------------- : [B][COLOR="#0000FF"]3.008[/COLOR][/B] GHz ([B][COLOR="#0000FF"]402x7.5[/COLOR][/B])

Performance Enhance ------------ : [Turbo] 
(G)MCH Frequency Latch --------- : [[B][COLOR="#0000FF"]400MHz[/COLOR][/B]] 
System Memory Multiplier (SPD) - : [[B][COLOR="#0000FF"]2.66D[/COLOR][/B]]
Memory Frequency (Mhz) --------- : [[B][COLOR="#0000FF"]1068[/COLOR][/B]]

primary memory timings = 5-5-5-15

tRFC -------------------- 52 --- : [[B][COLOR="#0000FF"]66[/COLOR][/B]]

Static tRead Value -------- 7 -- : [7]  <-- 8 might be needed for both memory channels if there is a boot problem


CPU PLL ------------------------ : [[B][COLOR="#0000FF"]1.59[/COLOR][/B]] <-- 1.57 might be OK, but keep this below 1.60
MCH Core ------------ 1.100V --- : [AUTO] <-- 1.20 or slightly higher is OK, but should not be higher than about 1.32

Include an ET6 advanced voltages screen-shot so that we can confirm that there are no over voltage settings.

If these settings pass all Prime95 tests, we can increase the cpu multiplier to 8.0x and maybe even 8.5x.

These are my 3.41GHz settings with my 8.5x cpu multiplier:

[Bobolink1970]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

What does CTRL+F1 actually do in the main bios window?

402x7.5 doesn't equal 3008 but 3015 BTW.
Therefore I entered 400x7.5 which equals exactly 3000 Mhz.....ok?
This gets the RAM at exactly 1066 Mhz.
Static tRead Value I put on 8 just to be on the safe side.
All other settings like you proposed.

Only there is no CPU PLL for me to change....not with this mainboard.




So now I should Prime95 have its way with these settings for the next 3 nights...?

[profJim]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

Gigabyte has several hidden bios settings and pressing CTRL + F1 at the main bios window makes the hidden settings visible until you leave the bios.

ET6 shows much less information with your motherboard and you don't have access to many bios options that are available in the higher end EP45 boards.

In the past some users had problems with overclocking with Cpu Host Frequency = 400 MHz, this is why I suggested using 401 or 402 MHz. Using the (G)MCH Frequency Latch = 400 MHz with the Cpu Host Frequency less than 400 MHz, i.e. the D-memory strap shouldn't be used until the Cpu Host Frequency is 400MHz or higher. My math was a bit off because I was dealing with 401 or 402 MHz settings in my previous post.

Run each of the Prime95 tests overnight, but only go on to the next test if there aren't any worker errors.
With EIST and C-states enabled, it's possible that Prime95 might only have two of the four cpu threads running at 100%. Use HWiNFO's thread statistics to verify that all four threads are running at 100% during each test. I think that Task Manager's CPU Usage History will show if each cpu thread is running at 100%.

Post several of the incrementally higher Cpu Termination, Cpu Vcore and MCH Core settings so that we know the size of the next several voltage increments.

The voltages shown in ET6 are reasonable and it's possible that the Cpu Vcore or Cpu Termination settings might need to be slightly higher if there are any Prime95 worker errors.

[Chike]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

I wouldn't go any higher than 400 with this board, my EP43-UD3L walled at 411, there Is no reason to add chance of chipset limitation.
I would go with 7.5x400 to test bus setting, those 2MHz not going to make a difference..
Could be hidden settings in this board, but I don't believe there are any in M.I.T.
VTT has only 0.1v increment I believe as it on low end boards like the EP43-UD3L
I would also run the blend settings only, customized to use all available memory minus 100MB or so, possibly with min FFT 128k as the potential culprits would be memory and bus settings.
Note that test time on latest Prime95 is only 3 minutes when it was 15 on older versions

[Bobolink1970]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

7.5x400 looks like this in Prime95 @ Small FFT's:



I'm still a little insecure about the huge temp differences between the cores.
Although I noticed that my son's recently installed Q9550 has the same issue.
With both CPU's I was really meticulous about installing the heatsink, 'tinting' both heatsink and cpu cap with Arctic Silver 5 and applying Arctic Silver 5 for Core2 Quad cpu's as described on their website....

[Bobolink1970]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

Large FFT's:

[Bobolink1970]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

Blend:

[profJim]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

Looking good!!

[profJim]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

I just noticed that you posted results for all three Prime95 tests.

When stress testing, my average and maximum core temperatures have a 4^o (C) temperature difference with my Mugen 2 rev:B cpu cooler.

I haven't used Arctic Silver 5 for several years but when I did my core temperature range was reasonable.

If your TIM is old it the contents in the tube might have settled to a point where you don't have an optimal mixture.

It might be worthwhile to buy a new tube of thermal grease and possibly rotate the heatsink's orientation 180^o to see if this helps.

Because of my memory's tall heat spreaders I have a 2000 rpm cooling fan pulling air through the cooling fins and exhausting the air out the back of my computer case.

[Chike]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

For core 2 Intel specs are +5 -10 for temp accuracy I think or +5 -5.
CPU shows 55c and it's always lower than ay core temperature, but is not too accurate as well.
You can choose to go by the highest temp, or maybe a little lower, don't matter much while you are under 75c.
Or just go by the CPU temp which is OK up to 76.3 according to Intel specs.

[profJim]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

As far as I know, Real Temp's Sensor Movement Test is the best way to determine if there are stuck cpu core sensors and if there is a problem.

Real Temp download page: https://www.techpowerup.com/realtemp/

Real Temp documentation: https://www.techpowerup.com/realtemp/docs.php

Real Temp support forum: RealTemp | TechPowerUp Forums

Originally my Q9550 core temperature range was not as close as it is now but I was able to narrow the core temperature range by using a different TIM application method but some thermal grease brands don't do well with some of the techniques.

An oldie but goodie article: Benchmark Reviews: Best Thermal Paste Application Methods (April 2008)

Spreading a very thin TIM application by using your finger in a plastic bag does not work well with some or many TIM brands but it worked well for me with one of the somewhat more expensive Xigmatek offerings and I think that it might have been their PTI-G4512 "flavor": XIGMATEK PTI-G4512 thermal grease with gold grade thermal matrix filler particles - Newegg.com

My advice is to leave your TIM alone for now and run the RealTemp sensor movement test first.



Sorry, this is the best picture that I found on the web.

[Bobolink1970]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

When put together this system 5 years ago the small tube of Arctic Silver 5 wasn't old, I bought it together with Scythe Mugen heatsink.
The Scythe Mugen has both a pushing fan and a pulling fan to the back of the computer case.
I thought I was very precise in applying the thermal grease, I followed this guideline, including the so called 'tinting' of both cpu cap and heatsink, using an old bank card and a coffee filter....

http://www.arcticsilver.com/pdf/appmeth/int/hl/intel_app_method_horizontal_line_v1.1.pdf

As you know, I've put together a simular system with a Q9550 for my 13 year old son a couple of weeks ago. I used new Arctic Silver 5 for this and followed the same official guideline.
Although I have not oc-ed my son's system yet, I noticed exactly the same core temperature range issue there.
The general cpu temp is slightly higher as wel because it only has a Arctic Freezer 7 Pro Rev.2 heatsink installed.
But the temperature difference between the cores is simular to my system.

[Bobolink1970]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

Anyway...I cheated....
I skipped the 8x400=3.2 GHz and went for the 8.5x400=3.4 GHz straight away.
All other settings remained as with 7.5x400=3.00 GHz.
I followed Chike's advice this time, using only the Prime95 blend test.
No errors after almost 10 hours:




BTW I noticed that the standard blend test uses only 2000 MB of RAM. Since you didn't advise me to change that, I guess that is ok....?

[profJim]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

You didn't post an ET6 advanced voltage screen-shot so we don't know what the applied values are for the 3.40GHz AUTO settings.

The variance in the number of tests completed for each of the Prime95 worker threads is much lower and better than my results and this tells me that the reported maximum cpu core temperatures are evenly balanced in spite of what HWiNFO shows.

I almost always use the default Prime95 torture test settings in in each Prime95 test and I see no reason to run the Blend test with more than 2000MB of ram being tested.

With your 8.5x cpu multiplier, increasing the Bus Clock by 12 will raise the cpu speed by 102MHz and it's unlikely that your system will feel faster with 50 - 100MHz increases. If you choose overclock higher you will shortly need to deal with changing the memory strap multiplier and the primary memory timings. My 8GB of ram is stable with 1080MHz (450MHz x 2.40B) with 5-5-5-15 timings but as Chike said your motherboard will probably not let you overclock much higher.

Given the age of our hardware I think that it's a bad idea to stress our aging systems to the highest possible overclocked speeds. Right now you have a 20% overclock and I wouldn't go any higher for everyday usage. If you had a higher end motherboard you could safely go higher.

[Bobolink1970]

Re: BIOS tweaking or best BIOS settings GA-P43-ES3G (rev. 1.1)

3.4 GHz is fine.
I'm more than happy with that result.
So I can leave it like this?