1075T and GA-890GPA-UD3H Overclock Attempt

DaddyTinman · Aug 3, 2011

Trents and crew,

The system in my sig has been bending, creaking, spewing fire, and snorting under the weight of Prime95 testing. I will begin with the Base info below and then highlight the journey from 3.0Ghz to 3.675Ghz. My Super Pi 32MB run decreased by about 3 seconds from Base to highest overclock.

However, as most do on these OC posts, would like to checkout 4.0Ghz, if possible.

BTW, exit fan in back of the case, side fan pouring air right onto the TRUE, and one puller fan actually on the TRUE.

These are my first image uploads, and I read the other posts for the howdy dooes.

Here are the Base data points:

DaddyTinman · Aug 3, 2011

This was the second or third time through the tests.....so that I could gain some perspective from the toil and trouble.

I began to increment the FSB, allowing the HT and NB multiplers to climb uncontained....until HT was just way up there. Here is the 220:

DaddyTinman · Aug 3, 2011

The 225:

DaddyTinman · Aug 3, 2011

Skipping 230 to show 235 (as you can see, temps were up there but held steady under 55 degrees Celsius):

DaddyTinman · Aug 3, 2011

SKipping up to 245:

DaddyTinman · Aug 3, 2011

Got to 500 but couldn't get stability. Tried upping my VCore, and adjusting memory timing to no avail. Brought the FSB back down to 240.

While I did get BSOD on my earlier journies, this latest trip up to 500 began freezing the OS on loadup. The whirling sparles around the Windows logo stopped in midarc. Had to reset machine with a suspended-power button activation.

Thoughts? Thanks ahead for burning some human thought cycles on this.

trents · Aug 4, 2011

500? Do you mean 250?

Your images are so small I can't read the numbers. Please upload them one at a time if necessary. Use Snipping Tool in Windows accessories to 1. crop and 2. capture the images.

What is your CPUNB voltage set to now? Have you given it any increase? You aren't going to get where you want to go unless you bump it up to about 1.25.

Also, your ram frequency may be too high. It may be rated for 1600 when overclocked ("XMP") but the AMD integrated memory controller is only rated for 1333 mhz. Many folks find they just can't get things stable at a ram speed of 1600 mhz and you have even exceeded that. You are up to 816.7 mhz, which exceeds even the XMP rating of the ram. I would move it back to 1333 in bios. and give it a voltage bump from 1.5 to 1.55.

DaddyTinman · Aug 4, 2011

Trents, yes, 250. Guess I had my Double Core Brain going.

BTW, I use another tool in IE, under settings. I can zoom in and see small stuff. I will clip the data pods since your zoom in IE may be unavailable.

I have changed the mem freq to 1333Mhz, mem volts to 1.50, and NB to 1.20. I added .035 volts to the CPU voltage, too, now at 1.35.

Now to start from 200 and work toward 500....er....250 again. I will report back after the climb to Mt. Quadgigeverest.

trents · Aug 4, 2011

I would still bump your ram voltage to 1.55. Just a little extra for insurance when overclocking.

DaddyTinman · Aug 4, 2011

Aye, make it so, number one!

Can't give you 1.55, BIOS hardwired for either 1.545 or 1.60 on the DRAM volts. I used 1.545, changed the DRAM multiplier to bring mem freq back just under 1333, and am running my FSB 230 Prime95 stress test. I will post the 230 here in about 20 mikes.

DaddyTinman · Aug 4, 2011

Here are the CPU-z and HWMonitor data sets for 230. DRAM voltage at 1.545.

trents · Aug 4, 2011

I would lower the multiplier in bios for the HT Link frequency. Its at 2300 now. You want to keep it between 1800 and 2000 for stability sake. Other things look appropriate. Your core temps are nice and cool at 42 C. Bump your CPUNB from 1.2 to 1.25.

Now, start increasing the CPU frequency by 5 mhz increments and run 20 minute Prime95 blend tests after each increase, always with HWMonitor open to watch core temps. When the latest 5 mhz CPU frequency increase causes you to fail the Prime95 blend test, bump the CPU core voltage up a notch and retest. Repeat this cycle of 5 mhz CPU frequency increases, stress testing and core temp monitoring alternated with CPU core voltage increases as necessary to restore stability until either:

1. You hit 55-57 C core temp

or

2. You have raised the core voltage to 1.5

or

3. You can no longer get it stable (the "wall")

Any questions? Report back with results.

DaddyTinman · Aug 4, 2011

trents said:
I would lower the multiplier in bios for the HT Link frequency. Its at 2300 now. You want to keep it between 1800 and 2000 for stability sake. Other things look appropriate. Your core temps are nice and cool at 42 C. Bump your CPUNB from 1.2 to 1.25.

Trents,

Froze at OS loading (during the dancing sprites, before they formed the Windows logo) during first load at 250.

I will be offline but will get back with increased VCore as you mentioned to get past the non-moving sprites.

My NB Voltage is now at 1.30 (no 1.25 option), HT is at 2000, and mem freq is 1333.

trents · Aug 4, 2011

Don't confuse the NB chipset voltage with the CPUNB voltage. Many bioses don't do a good job of distinguishing them in the terminology they choose to use.

DaddyTinman · Aug 27, 2011

More research done....and completed....on the NB ball of confusion. Got cracking on driving past 3.675 Ghz last night.

Pleased to report 4.05Ghz was achieved and maintained under SuperPi stress for 20 minutes. Left the computer on over night (sans SuperPi) and no restarts/BSDODs/lock-em-ups.

My CPU Volt (Vcore) is 1.5, CPU NB Vid Control is 1.30, FSB is 270, mullti set at x15.
HT Link freq is 1890, NB Freq x9 (2700Mhz), Memclock is at x5.33 (1420Mhz).

Do I follow the same procedure to probe upper boundary of the Danger Zone(inc FSB, SuperPi test, if incomplete use move Vcore, add CPUNB Vid when needed)? Seems I am at the top end of Vcore.

Thanks, Trents and crew. Working to establish some benchmarks in HWBot from this and my new 1155 i5 bencher.

trents · Aug 27, 2011

You are at the boundary, core temp wise. No room for more vcore or CPUNB VID. Looks great to me! Good job! All frequencies and voltages look appropriate. Have you stress tested it with Prime95 blend for 2+ hours?

DaddyTinman · Aug 27, 2011

Running that today while away from the keybd on errand missions.

1075T and GA-890GPA-UD3H Overclock Attempt

Registered

Attachments

Registered

Attachments

Registered

Attachments

Registered

Attachments

Registered

Attachments

Registered

Senior Member

Registered

Senior Member

Registered

Registered

Attachments

Senior Member

Registered

Senior Member

Registered

Attachments

Senior Member

Registered

Similar threads