GA-EP45T-UD3P: Old Board New Issue

[profJim]

Re: GA-EP45T-UD3P: Old Board New Issue

@MDDI

Linking to DropBox or other file sharing sites lets you include multiple images in one post, such as the screen-shots I included in post #2. To insert multiple images in one post you use the Insert Image icon for each file where you paste the web address of each image file and uncheck the Retrieve remote file and reference locally box. It's best to use the forum's advanced editor (Go Advanced) feature so that you can preview your post before you submit it.

You can also upload your dump files to your DropBox account and you can then include a link to one or more files in your Dropbox folder.



Some ram manufacturer's include a disclaimer on their webpage that basically says:

Rated memory speeds are guaranteed for one 2x2GB memory kit.
Installing a second identical memory kit might not be able to run a 4x2GB setup at full rated memory speeds.

In post #10 you reported that the SPD programming for each of your kits is not identical.

In post #4, I said:

You might have a problem with the auto-applied MCH Core voltage setting and EasyTune6 will display the current setting.
With my 2x2GB Mushkin Blackline 1066MHz memory, the optimal MCH Core voltage setting was 1.26 - 1.28 volts. YMMV

A higher MCH Core setting might eliminate your BSOD problem.

Right now your memory is running at 1333MHz (4.0 x 333MHz) using the 4.0 System Memory Multiplier. You can quickly and easily underclock your memory to 1066MHz for testing by manually setting the memory multiplier to 3.20 where 3.20 x 333 = 1066MHz. If your system is stable with this setup we can tweak the memory settings so that your ram runs at somewhat higher speeds such as:

• 1120MHz with 3.2 x 350MHz, and cpu = 2.975GHz using 8.5 x 350MHz
• 1171MHz with 3.2 x 375MHz, and cpu = 3.000GHz using 8.0 x 375MHz
• 1280MHz with 3.2 x 400MHz, and cpu = 3.000GHz using 7.5 x 400MHz

With any of these test setups your overall system performance will be somewhat higher and a very minor increase in Cpu Vcore voltage might be required.

FWIW, the WhoCrashed information from the first parameter refers to:

0x41790
A page table page has been corrupted. On a 64 bit version of Windows, parameter 2 contains the address of the PFN for the corrupted page table page. On a 32 bit version of Windows, parameter 2 contains a pointer to the number of used PTEs, and parameter 3 contains the number of used PTEs

[MDDI]

Re: GA-EP45T-UD3P: Old Board New Issue

profJim,

I have included a couple things you requested. I have not had time to dedicate hours a day to this. The tech at Mushkin requested I run a stress program called OCCT, it's like a amped up Prime 95 but runs faster. I have not run the full test yet, trying one thing at a time, slowly.

The screen shot is the start page of OCCT and the Who Crashed dump files. What's interesting is the CPU core temp and the +12V; it mirrors what HW Monitor reported. Wright or wrong I know HW Monitor reported the same way years ago when I put the system together. Core #0 temp (it's static) has never reported and the +12V appears to be reporting on the 1V scale? The +5V displays HW monitor but does not in PC Health. PC Health displays +12V, HW Monitor (and OCCT) does but not correctly. PC Health as does HW Monitor (and OCCT), shows only 1 CPU core temp. MOBO issue?

Looking back I believe there were two BSOD causes that I mistook for one. My system would crash within a minute of descktop and sometimes when browsing. The first I believe was addressed with raising the DRAM voltage and MCH Core voltage. Currently the DRAM voltage is 1.68V (BIOS), 1.664V (PC Health). MCH Core 1.2 (BIOS), actual? I think the DRAM V is set a little high and the MCH Core a little low as you indicated in your previous reply.

Right now I'm stable booting to desktop, it will stay there all day without a crash. I believe I should not mess with the RAM voltage yet until the second BSOD cause has been corrected?

WTFMAN suggested I uninstall AVAST, it does have a Web Shield. I have used AVAST for years and never had issues. I thought about what could cause crashes while browsing; I would receive "Internet Explorer Has Stopped Working" and crash after closing.

Win 7 installs Defender default, it's supposed to be inactive when a other party AV us installed; is it? Tuesday night I disabled Defender and AVAST before shutdown. All day Wednesday (shutdown night) and all day Thursday (24 hours running), carefully browsing without an active AV, I did not have a crash. So, was it AVAST or a conflict with AVAST & Defender?

Thursday night I enabled AVAST, Defender still disabled. So far today, Friday AM, I have been using my system for 3 hours, have browsed, has not crashed. I'm going to continue today and tomorrow, Saturday, before I pass judgment. If it crashes; AVAST is gone. If it doesn't, is two days (24 hours running) long enough to say all is good, was Defender & AVAST conflict, or continue testing?

If all goes well I will then run MemSet, HWiNFO and possibly the OCCT test. Tweaking the RAM voltage will be next.



Thanks all,

MDDI

[profJim]

Re: GA-EP45T-UD3P: Old Board New Issue

Windows update prompted me today to download and install a new IE v11 security fix.
Have you installed this update yet?

Other tests you can run include:

• run your IE browser for a while, close it, reopen it and repeat this procedure several times
• using a different web browser
• using windows Defender alone or using a different A/V program

There might be a misbehaving driver or program that is corrupting Avast's memory.
If you run two or more anti-virus programs at the same time they might get in each other's way.
Only run one A/V program at a time.

A one hour OCCT or Prime95 is better than nothing, but a 3 - 4 hour or longer test will be a better stress test.
Why are you running OCCT with Number of Threads = 4 when you have a dual core processor?
The Number of Threads setting probably doesn't have a major effect on the test results.

HWiNFO has an optional logging option with a 2000ms (2 second) interval.
You can change the logging interval to every 5 or 10 seconds when running longer stress tests.
The default is to log all sensor readings, but you would be most concerned with the voltage and temperature sensor readings.

It looks like the Core1 temperature sensor is stuck.
In HWiNFO you can monitor the Core #1 and Core #2 Usage sensors during your stress test.
Both usage sensors should be at 100% during the stress tests.

In your bios what are the displayed PC Health Status voltages?

This is an example of an EP45-UD3P PC Health Status bios screen-shot:

[profJim]

Re: GA-EP45T-UD3P: Old Board New Issue

Your current bios version is F7, dated 2009/04/16.

The Gigabyte download website has a newer F8c beta bios, dated 2009/09/03.
Gigabyte removes beta bios versions when there is a newer final bios version or there are reported problems with the specific beta bios version.

http://forums.tweaktown.com/gigabyte...beta-bios.html has a newer GA-EP45T-UD3P beta bios: F8l.
Stasio posted the following information at ** Official ** Gigabyte UD3 P45 Series -- EP45-UD3 / EP45-UD3R / EP45-UD3P - Page 291, post #7275:

GA-EP45-UD3P - F11d
GA-EP45T-UD3P - F8l

-Fix SMART Command Fail for OCZ Sandforce FW 1.11
-28.Oct 10

I didn't check the rest of the thread to see if there was any feedback about the F8l bios.

It's premature right now to flash to a newer bios, but a newer beta bios might help with stability.

[MDDI]

Re: GA-EP45T-UD3P: Old Board New Issue

profJim,

Yes, installed the latest IE security update last night.

I'm 90% certain that my browsing crash issue was due to Avast & Defender conflict. I disabled Defender Thursday evening, I have not had a BSOD for 48 hours with Avast enabled. I'm not admitting this is resolved yet; I have been surprised before.

I do feel that I should adjust the MCH Core V, currently set at 1.2V. The DRAM V is probably higher than needed (1.86V in BIOS, 1.664V reported in PC Health). I kept stepping up the DRAM V trying to stop the browsing crash. Maybe 1.24V MCH and 1.62V (1.6V report in PC Health) to start?

I accept your suggestion to update the BIOS. My board is EP45T-UD3P (rev 1.0). Like you stated it may not be needed yet. When I received the board the BIOS version was F5 or F6; I did flash to F7. I never flashed the latest because it was BETA.

I did not run OCCT yet, that was a screen shot of the start page for temp/voltage reference to what HW Monitor reported. The Four Core test I believe is the default setting, I did not change it. That utility was suggested by the Mushkin tech.

I have received advice not just from this forum, others and sources also; TMI. I'm here because I trust TT the most. I have tried to stay focused on one procedure at a time. The Avast test was easy to perform and appears to have been a problem and possibly solved. Only time will tell.

I have to leave next week and will return May 13th. I have much work and prep before I go. My system appears to be stable now, at least the most stable since I loaded Win 7 over three weeks ago. I can actually use it without screaming and anxiety attacks. I need to work with it for awhile, not on it.

I am aware other issues exist; core #0 temp reading, +12V reading; they have always been that way so I don't believe they are related to my current major problem. I plan to install/run HWiNFO and Memset at some point. For the time being I would like to be as productive as I can with my PC and quit fiddling with it hours at a time.

I will continue to update if I have another BSOD. Could we pickup the remaining tests after I return on the 13th?

As always, appreciate all the help,

MDDI

[profJim]

Re: GA-EP45T-UD3P: Old Board New Issue

It looks like you're making progress.

Lsdme recommended HCI MemTest: RAM reliability tester for Windows for memory testing in windows.
They have FAQ and user manual links on the web page.
Right now my system has 5400 MB of free memory, so I would run four instances of HCI Memtest with each test using about 1200 MB of memory and let the test run until each test completed 200 - 300%.

Prime95 and OCCT are also excellent stress tests but they need to run for a much longer time.

We can pick up later when you get back.

Have a safe trip.