GA-MA785GMT-UD2H EES and other BIOS questions

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

I will check and post back tomorrow - are there any voltage issues (NB, DDR) that could possibly cause the USB issues?

I tried to locate the previous issue that made me get the PCI card - I thought it had to do with hubs and the VIA controller on the Shuttle AK38N mobo, but the closest thread I could find was this one and it says NEC controllers have issues and VIA controllers do not.

Not sure what I have or what the issue was at this point.

[artdrivers]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

We know that your memory voltage is still a little low. I would work on that first.

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

We know that your memory voltage is still a little low. I would work on that first.

Actually, I set it low based on your comments in Reply #14:

You could try 1333 @ 1.7v-1.8v Why put more voltage than is neccesary. CPU NB VID Control +0.2 for reference this will add voltage for stability for the memory controller.

What target should I be looking for?

Also, I significantly backed the NB Votage down from your recommendations b/c you mentioned heat issues - should it be raised higher?

Also - as I said, I have P95 running for 30 minutes with the lower CPU VID, but when I tried Visual Pinball, the system immediately re-booted, so perhaps P95 wasn't the greatest stress-tester - what should I be running and looking for?

Thanks again!

[artdrivers]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

P95 is good. I would set the DDR3 memory voltage to +2.5v=1.75+v or +3.0=1.8+v @ DDR1333Mhz
Your PC health status earlier suggested the memory voltage was 1.5v @ 1.6v. In my previous experience all desktop DDR 3 on auto is at 1.5v. The small% voltage increase from 1.5 @ 1.6v is only offered in the auto (which is like a generic or fail-safe) setting. When you change it to manual you override that small% increase. Your memory manufacturer suggest that the DDR3 memory voltage should be set to 1.9v @ DDR1600Mhz to 1800+. The memory modules would not need that much at DDR1333Mhz. This is why I suggested 1.7v to 1.8v. I believe if you increased DDR3 memory voltage to +.15v It is at 1.65+v. Please check that and let us know. We still must worry about that memory controller this is why we added voltage to the cpu NB VID controller. As I explained before I never used PMT, but if it makes some of these adjustments for you I would suggest you verify them. These are only suggestions. It seems you are getting things right for your use.

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

The RAM is at +0.15=1.75 and the NB is at +0.05=1.225 but these are the M-B Intelligent Tweaker Values, not PC Hardware Wizard or PC Health.

The timing is at 1600.

I guess my basic question is "If I seem to be working fine at the above settings, do I still want to go to 1.9 DDR or 1.375 NB just to have some cushion?"

P95 is good.

The reason I asked is b/c with the CPU VID, I had the system running P95 for half-hour to an hour with no issues and then it instantly re-booted when I launched a visual Pinball table ... (???)

As I explained before I never used PMT, but if it makes some of these adjustments for you I would suggest you verify them.

PMT allows you to set requested CPU VID, Multiplier, and NB VID for idle and high-power states. It and AMD Power Monitor only show the requested (not the actual values), but the CPU VID seemed about right or a bit low (+0.003-ish) in CPU-Z.

It seems you are getting things right for your use.

Yes, I'm happy with the voltage and memory settings (although I will investigate some more).

The USB flash drive problem is very irritating, and the mouse as well - although I might buy a $5 wired optical USB mouse and forget about that part.

Thanks again for all the assistance!!!

[artdrivers]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

The RAM is at +0.15=1.75 and the NB is at +0.05=1.225 but these are the M-B Intelligent Tweaker Values

This suggests to me you set them and did'nt go back into bios to verify. Because when you read current value and ad 1.5v you think done 1.75v. No problem I have done this numerous times as well. Please verify.
The voltage here does not ad near the heat as the increase in frequecy does. This is why I would find a happy medium and settle for the 1333Mhz which is very fast and at a good temp.

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

Correct - I did not verify (although I think the MBIT screen showed 1.75 "grayed out" when I selected +0.15.

Again, even if the system appears stable, you recommend increased to 1.9V measured?

(i.e. if HW Monitor shows the RAM is at 1.78 for example, I should bump it up 0.12 so it hits close to 1.9 (???))

Thanks in advance!!!

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

Temps are not a problem - I didn't notice they went up a bit when I went to 1600 Mhz, but even then, actual temps IMS were only around 112 F or so - they were enough to kick the HSF on high, but bothered me (made me think the system was working harder, but I reduced the CPU load state VID and it stopped kicking the fan into high speed (load temps dropped to around 103-105F).

Watch it raise cpu vid? With software in an OS? Are you sure it's cpu nb vid? Please verify this as well.

Didn't quite follow this part of your reply, but yes - MsrTweaker allows you to set CPU VID and NB VID and core multipliers inside the OS software (with a service, actually) - I did not verify NB VID, but I did verify CPU VID with CPU-Z.

Screenshot from their website:

[Tiger-Heli]

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

Google gave mixed results, but it looks like Everest Free Home can show NB Voltage, and someone said HW Monitor can, but I sure don't see it there.

I think it's more accurate than CPU Vcore, but I'll check it tonight anyway.

Thanks for all the assistance!

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

Hmmmmn, something now is making me nervous ....

Here's what I posted yesterday from Friday's tests:

That ended up not being stable, but that's a variation of 0.002 and 0.02 respectively between CPU-Z and MSRTweaker requested values.

Here's yesterday's results:

Thats a rough average variation of 0.2, not 0.002 or 0.02.

What bothers me is if MSRTweaker is not just consistently off a certain amout, than if MSRTweaker is only off by 0.002 the next time I run it, I am almost over-volting the CPU.

And I don't think MSRTweaker has any option to NOT adjust Vcore but still lock the multipliers, and I don't know another tool that will do anything similar - K10Stat looks about the same, but not sure it works any better than MSRTweaker since it seems to be developed from K10Stat.

Suggestions?

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

Answering my own questions, but ....

Stock Vcore for the Regor seems to be 1.425 as reporting by the BIOS and several web forums (Not 1.35 like SSS reported).

So - with yesterday's settings - if the actual values stay close to requested, than I am still (slightly) below stock voltage - and if they are 0.2 lower, then I am pretty much where I want to be.

Far from ideal, but at least I can continue to use MSRTweaker and not worry about stablility or overvolting the processor.

(Another thought - this could also easily explain why the processor was stable in Prime95 (when the voltage was only 0.002 off from requested) and crashed loading Vpinball and running Word (when the voltage was 0.2 below requested).

[artdrivers]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

You have 2 different NB's CPU NB VID controls voltage supplied to the IMC (on die memory controller).
NB Voltage controls voltage supplied to the chipset (on the motherboard).
If you are increasing NB speed, then increase CPU NB VID as necessary, and vice versa when decreasing speed decrease CPU NB VID as necessary.
You will need to find the voltages to speed that best suites your needs.

System Voltage Control [Auto]
DDR3 Voltage Control [Auto]
NorthBridge Volt Control [Auto] [COLOR="DarkRed"]voltage for the Northbridge chip [/COLOR]
SouthBridge Volt Control [Auto] [COLOR="darkred"]voltage for the Southbridge chip[/COLOR]
SidePort Mem Volt Control [N/A on my board]
CPU NB VID Control [Auto] [COLOR="darkred"]voltage for the memory controller on cpu [/COLOR] 
CPU Voltage Control [Auto] [COLOR="darkred"]voltage or vcore for the cpu[/COLOR]
Normal CPU Vcore is 1.4250V

This is voltage initially. When you use the AMD drivers and PMT overrides some of these voltages. They change. PMT is showing the CPU VCORE voltage and the CPU NB VID, although they are calling it CPU VID and NB VID. This was confusing to me so I tried the program out to confirm this. It's a good program I like it too. When I open PMT and activate a setting then close PMT. The cpu continues at what ever the setting was when I closed PMT. To change it you only need to open the program and activate a different setting.
Hopefully when I access bios the voltages go to the settings in bios because the Windows drivers will not be present. But when I boot into Windows the drivers load again and make the changes. This is why I have different voltage readings compaired to the windows programs and bios.
Some programs will only have a specific set of numbers as to what it normally is. Some read the bios settings. Some read the sensors. I prefer HWmonitor. It works for my system, cpu cores, hard drives, and GPU. It reads temperatures and voltages from the sensors all on one quick accessed program.

[Tiger-Heli]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

Hi ArtDrivers!

Okay, I sorta followed all of that.

I'm assuming I should bump the DDR Voltage by 0.08-ish to hit 1.9 Volts - agreed?

CPU NB VID control is what I bumped up in BIOS to increase the memory controller. I think I left it alone now - but I'll double check that.

If I understood what you posted above, NB Volt control is what PhenomMSRTweaker is setting and it shouldn't affect NB VID control, and I can reduce set it back to stock or bump it up and it shouldn't affect the memory controller (which is what I am trying to stabilize).

Errrmn, no wait, you said it is setting NB VID - so I do need it increased in PMT, correct?

I like PMT, but the main thing I need it for is keeping both cores at the same multiplier - that's a 1.5-2X frame rate increase for me in some games.

Hopefully when I access bios the voltages go to the settings in bios because the Windows drivers will not be present. But when I boot into Windows the drivers load again and make the changes. This is why I have different voltage readings compaired to the windows programs and bios.

I'm not sure you understood what I meant. CnQ is disabled in BIOS b/c PMT recommends that. PMT service and tray icon are always running when I am in Windows. On one day, measured CPU-Z values were within 0.002 to 0.02 V of PMT requested values (which I consider acceptable) and yesterday they were around 0.2 below - which I don't. BTW, AMD Power Monitor only shows the requested values - to verify them you need HW monitor or CPU-Z. (I assume you knew that, but just in case ...)

HW Monitor is fine, but I don't see that it displays NB VID - and since PMT is changing that, I would like to verify that it is what I think it should be. I think Everest Free Home Version does show this.

If you have a spare moment, I would be curious how closely your HWMonitor Vcore values match PMT requested values.

Thanks again for all the help!!!

[artdrivers]

Re: GA-MA785GMT-UD2H EES and other BIOS questions

Yes it increases CPU NB VID.
Here is some more verification.
PhenomMsrTweaker: Cool & Quiet customized
Some background infos
What about these P-states?
Each core of a K10 CPU may run in one of up to 5 performance states (P-states). P0 is the highest-performing (default) state, the others ascend to the max-powersaving state. Phenom CPUs use only 2 states, P0 (high performance) and P1 (powersaving). Phenom II CPUs specify 4 states, P0 to P3.

Each core boots in the P0 state. If Cool & Quiet is enabled, a driver may put a core later in a powersaving state if its load is low and then back to a higher-performing state when its load increases.

Each one of these per-core P-states consists in:

a multiplicator controlling the clock frequency (4x to max. multi),
a voltage ID for the CPU core (0 to 1.55V), and
a voltage ID for the integrated Northbridge/memory controller (0 to 1.55V).
PhenomMsrTweaker allows you to customize these settings for all cores and the first two P-states, P0 and P1. Therefore you may choose your optimal settings for P0 (overclocked state) and P1 (undervolted state).