Need some help

[madman7]

I've posted some time ago about some problems my Asus M4A79T deluxe. You guys helped me figure out it was memory causing the issue. Anyway after 3 or 4 boards from Asus I got one that working so far but I noticed something strange. I have only one memory stick in it now as I'm waiting for the replacement stick. I couldn't get the board to boot with the good memory in the A slots. I would boot in the B slots fine. I figure a bad board so Asus says they will replace it. I monkey around with it some and get it to boot in the A slot but it crashes and wouldn't reboot. So I put the memory in the B slot and it fires right up. I decided to go in and turn the CPUNB freq. back to auto from 2600 and set the CPUNB voltage to auto also. I put the memory back in the A slot and it fires right up. Okay, so i check AMD overdrive and it shows my CPUNB voltage in red and at 1.375 volts. What gives with this? Could the bad ram stick have damaged the IMC on the CPU and why is the CPUNB voltage showing so high? What should the CPUNB voltage actually be?
Is there a program that will read all the voltages?

BTW, I'm running the CPU at 4.1 GHZ with a CPU voltage of 1.4750v and the CPUNB voltage is set to auto.

 

[Johan45]

I don't know about your board speciffically but it should work in the A slot with just one stick, check your manual and see if that's what it recommends. If yes maybe try clearing the CMOS and then put it in slot 1. It might still be looking in slot b for the ram it "remembers"

 

[madman7]

According to the manual, the memory can go into any slot. I have been moving the memory to different slots. Sometimes it boots in Slot A and will start to load windows then crashes. The slots are laid out as A1-B1 A2-B2. If the memory is put in any slot A sometimes it works but most of the time it doesn't

 

[madman7]

I have it but don't know which of the 7 voltages it would be.

 

[trents]

What's the CPUNB voltage set to in bios? Also, do you have a bios setting for CPUNB LLC that may be set too agressively? Are you sure your PSU is putting out appropriate voltage on the various rails? You can check that in the first section of HWMonitor.

 

[RGone]

LLC was only spec'd by AMD for the 9XX chipset boards. Previous to that there was no such animal as LLC in the AMD world and almost certainly not on a 7XX chipset mobo.

 

[ATMINSIDE]

Any troubleshooting for bad components should be done at stock speeds, preferably after a CMOS clear. An unstable RAM speed/timing combination or NB/CPU-NB could easily cause the issue you see.

 

[madman7]

What's the CPUNB voltage set to in bios? Also, do you have a bios setting for CPUNB LLC that may be set too agressively? Are you sure your PSU is putting out appropriate voltage on the various rails? You can check that in the first section of HWMonitor.

All the rail voltages look good according to HWmonitor. I set the CPUNB voltage to auto and it shows 1.375 in AMD overdrive and is highlighted in red. I tried everything at stock and it will sometimes boot with the memory in any slot A. Most of the time it won't boot. The memory stick that went bad was hot when I found out it was actually bad. This was after a couple of times ASUS sending my board back saying it was good. The bad memory caused the BIOS to get corrupt. I have now had 2 other boards from ASUS. This one works but with 1 stick in slot B. I'm wondering if the ram damaged the IMC on the CPU. I can set the CPUNB voltage to what ever I want and it shows that voltage in AOD. I hope the CPU isn't damaged. Any ideas?

 

[Johan45]

Did you ever clear the CMOS as I suggested in the first post??? Same as ATM said put everything to default. These are our ideas at the moment what you do with them is up to you!

 

[madman7]

Did you ever clear the CMOS as I suggested in the first post??? Same as ATM said put everything to default. These are our ideas at the moment what you do with them is up to you!

I did all that. Like I said, sometimes it'll post sometimes not. It will even post, almost get to the Win splash screen and I lose video and it just sits there. If I do a reset, it won't post. Sometimes it will get to the Win splash screen and then shuts down. One time it loaded Windows and then froze. I then move the memory to any B slot and it will run fine. I curious if the bad ram damaged the CPU!!

 

[trents]

Have you tried manually setting the CPUNB voltage to something more normal, like 1.775 instead of Auto with the CPUNB frequency at stock 2000 mhz?

 

[madman7]

I set it manually to 1.250v and AOD reads it as 1.250v. I wondering why in auto it's so high and if the CPU is damaged. I'm not sure what the CPUNB voltage is supposed to be. With CPUNB in auto which is 2000 the voltage is still showing 1.375 v in auto. ASUS will replace this board but if the CPU is hosed a new board won't help.

 

[RGone]

You >> Anyway after 3 or 4 boards from Asus I got one that working so far but I noticed something strange.
Question: What was the issue that had you getting 3 or 4 boards from Asus?

You >> I couldn't get the board to boot with the good memory in the A slots. I would boot in the B slots fine.
Suggestion: Use the B Slots then and certainly so while you wait on the mate to your current memory stick. By the way, when I RMA ram, I do so in pairs so that the ram 'set' should be made up of the same type of ram stick.

You >> AMD overdrive and it shows my CPUNB voltage in red and at 1.375 volts. What gives with this?
Answer: That is an inbuilt setting chosen by the mobo according to Bios programmers. Asus may have decided to get DDR3-1600 speeds the CPU_NB needed to be X voltage.

You >> Could the bad ram stick have damaged the IMC on the CPU?
Answer: Possible one may suppose but not so likely. Is the memory in bios set to run "ganged" or "unganged"? Run it "unganged".

By the way at this time, only swapping out the cpu would verify a wonky memory controller (IMC)

You >> What should the CPUNB voltage actually be?
Answer: Generally it would be 1.175-ish for Deneb type cpu.

You >> Is there a program that will read all the voltages?
Answer: Almost without fail we use Free HWMonitor in the AMD Forum Sections. How much information is given is really board dependent though. A chip on the board is read by the reporting software and what 'is' reported depends on what the board engineers routed to the 'reporting' chip.

I went back to your first post and made my responses from it. That is where the crucial information may have been hidden before we began a big trouble-shooting journey, which by the way, there is nothing wrong with doing.

I will suggest this to you though. Check the "ganged/unganged" thing. Adjust as suggested. Check for the one stick of ram you are left with working in A or B type slot then. If it does not work, you have parts and pieces that are going to have to go thru a process of elimination again. In your first post Asus has already agreed to swap out the 4th or 5th mobo based on actual RMA count. No way to really verify suspect IMC without swapping out the cpu. You going to do that? No I did not think so. Why not? Well you are only likely to run 2 sticks of ram anyway and that 790Chipset board you have is getting long in the tooth.

Because of what I just mentioned in that paragraph above, I would put the one good stick in a working slot and use my computer. When a replacement stick comes I would put it in the correct memory slot to produce dual channel mode and rokkon. You are not likely to need more than two sticks of ram as mentioned earlier and you are driving yourself nuts with actual swap out of parts needed to be done to really troubleshoot much further. It seems time enough to have a little fun out of your system. IMO. For sanity sake with an older 790 chipset mobo.
RGone...

 

[madman7]

You >> Anyway after 3 or 4 boards from Asus I got one that working so far but I noticed something strange.
Question: What was the issue that had you getting 3 or 4 boards from Asus?

You >> I couldn't get the board to boot with the good memory in the A slots. I would boot in the B slots fine.
Suggestion: Use the B Slots then and certainly so while you wait on the mate to your current memory stick. By the way, when I RMA ram, I do so in pairs so that the ram 'set' should be made up of the same type of ram stick.

You >> AMD overdrive and it shows my CPUNB voltage in red and at 1.375 volts. What gives with this?
Answer: That is an inbuilt setting chosen by the mobo according to Bios programmers. Asus may have decided to get DDR3-1600 speeds the CPU_NB needed to be X voltage.

You >> Could the bad ram stick have damaged the IMC on the CPU?
Answer: Possible one may suppose but not so likely. Is the memory in bios set to run "ganged" or "unganged"? Run it "unganged".

By the way at this time, only swapping out the cpu would verify a wonky memory controller (IMC)

You >> What should the CPUNB voltage actually be?
Answer: Generally it would be 1.175-ish for Deneb type cpu.

You >> Is there a program that will read all the voltages?
Answer: Almost without fail we use Free HWMonitor in the AMD Forum Sections. How much information is given is really board dependent though. A chip on the board is read by the reporting software and what 'is' reported depends on what the board engineers routed to the 'reporting' chip.

I went back to your first post and made my responses from it. That is where the crucial information may have been hidden before we began a big trouble-shooting journey, which by the way, there is nothing wrong with doing.

I will suggest this to you though. Check the "ganged/unganged" thing. Adjust as suggested. Check for the one stick of ram you are left with working in A or B type slot then. If it does not work, you have parts and pieces that are going to have to go thru a process of elimination again. In your first post Asus has already agreed to swap out the 4th or 5th mobo based on actual RMA count. No way to really verify suspect IMC without swapping out the cpu. You going to do that? No I did not think so. Why not? Well you are only likely to run 2 sticks of ram anyway and that 790Chipset board you have is getting long in the tooth.

Because of what I just mentioned in that paragraph above, I would put the one good stick in a working slot and use my computer. When a replacement stick comes I would put it in the correct memory slot to produce dual channel mode and rokkon. You are not likely to need more than two sticks of ram as mentioned earlier and you are driving yourself nuts with actual swap out of parts needed to be done to really troubleshoot much further. It seems time enough to have a little fun out of your system. IMO. For sanity sake with an older 790 chipset mobo.
RGone...

I'll try to answer all your questions.
#1- I sent my original board to Asus because the BIOS got corrupt. I was using the computer when it suddenly blue screened. I restarted it and it went to the splash screen and then quickly blue screened again and restarted and did it again. I shut it down and restarted and the BIOS was corrupted. I tried to flash it but it wouldn't take. I sent it to Asus. It came back with the BIOS reflashed. I put it together and as soon as it started I got the corrupted BIOS message again. I tried to flash it again but no go. I sent it back to Asus and they returned it with a reflashed BIOS chip. It did once again. I sent it back and they sent a repaired board. I installed everything and booted it up. You may have guessed it. Corrupt BIOS again. I call Asus and we discussed about bad memory or bad CPU. They eliminated those and focused on the HD and a virus. He had unhook the HD and flash the BIOS. It took. I then restarted and the BIOS was corrupt again. I sent this board back. In the meantime I login in here and ask about what could be corrupting the BIOS. I get some replies and it's determined a memory problem. I get another board which is the one I have now. I use it and determine one memory stick is bad and also that the A slots won't work.
#2- When I get the stick back, I don't think I can run memory in the B slots. Asus suggests using the orange slots which are A1B1. That would put my memory in the B1B2 slots. I asked about sending both sticks back and was told by Adata that they could match the stick.
#3- On my original board, I remember from AOD that the CPUNB voltage was never that high. I used it to look at voltages while I OCed the CPU.
#4- I 'll check in the BIOS for ganged etc. but I don't think there is a setting for that. I have been all thru the BIOS and don't remember seeing that choice.
#5- The reason I ask about the CPU IMC being bad is the commonality of the A slots not working. I haven't found anything showing the interconnections of the CPU and the memory. From what I can find is the memory is connected to the CPU which contains the IMC directly and was wondering if the memory shorted and damaged the IMC on the CPU.
I ran HWmonitor and changed the CPUNB voltage but never saw a corresponding change in any voltage in HWmonitor. It could be the board is indeed flaky and not reporting voltages right. IDK
I plan on keeping the computer for a while longer. I don't want to upgrade mobo and cpu at this time. I don't want to loose everything on my HD at this time. I'll get this board replaced or what ever and see what happens with a different board. If the next one does the same thing, I can figure the IMC got hosed. I would like to run 2 sticks in the correct slots to get the max out of it while I can.
Thanks for your thoughtful reply. I hope I answered everything.
I found this in the BIOS.
DCT unganged mode
Auto-by dram settings
enable-unganged mode
disable-ganged mode
I'm set to auto.

 

[trents]

I am suspicious when motherboard companies say they checked out something or fixed something as to whether or not much was really done. It's not very cost effective for them to put in much time on trouble shooting and repairs. RGone used to work for DFI and maybe he has some insight into how RMAs are really handled. If I were you I would be tempted to buy a cheap board and check out the RAM and CPU with a new hardware base as it were. Then you can turn around and sell the thing if you have no use for it.

 

[RGone]

Not overly cost effective on a just about obsolete 790 Chipset board.

DFI handled most RMAs a little differently than most companies do as I see it. DFI handled RMAs from Taiwan and two other locations. One of which was the Usa. My zone was Canada to South America. Big yes. All of those went for repairs in CA, Usa.

The majority of RMAs for that large zone went to RMA with an enclosed message as to the likely failure in the mobo from me. I only saw the repair facility in CA once and it was extensive.

Board hooked to power supply and checked when received. Yes they had test ram and cpu and video card and cooler for testing. Tech could tell quickly if bios was hosed. REflash bios and check operation. Good. REturn to customer.

Now at this juncture is where it gets sticky all around the world. The customer I spoke with at or before beginning the RMA had my recommend following his board to the repair facility. That "recommend" had within what suggestions applied to the 'user' and often they reflected that he (the customer) needed to do certain things as well.

At one point there were two of us handling the "large zone" I spoke of and at any given time he might have a "problem" board on his test bench at his location not in CA, Usa. We kept multiple cpus and various brands of ram for testing the real problem "child" board from a customer and then we sent to RMA if so needed.

I can only actually remember but three (3) instances where a board I had customer send to RMA and was returned to him and he had the REcurring issue. All were REsolved. One of those in particular I had sent to me since the customer was using Kingston Value Ram and it was one double wadded pain in the arse to get setup and running with that Value shett. I had Kingston send me same kit he used and setup the stuff and saved in a bios profile and sent it off to the happy customer. NO mobo companies that I know of had that sort of customer service. By the way that scenario and one news blurb kingston made in public about 6 years ago is why I never buy Kingston ram. My personal move for my own sanity.

So what I have described is how a mobo company that I was once employed by and that no longer participates in the DIY market did their RMAs in the zone I was employed to deal with. I was American, dealing with a large consumer audience that spoke english and I ran DFI boards, participated in a few forums and managed the DFI forum and pushed the crap out of DFI motherboards. That push the crap out of our own product at the time, was likely the difference. Run what you brung and then you know what the heck is likely to be the issue with most customer boards. Nope you will not find that RMA process any longer, anywhere that I know about.
RGone...ster.

 

[Johan45]

Nope you will not find that RMA process any longer, anywhere that I know about.
RGone...ster.

I know they just don't care about us lowly customers once we've passed the checkout line.

 

[madman7]

Update about the board issue. I got another board from Asus. This one will boot with memory in A1-B1 memory slots. Now I find some odd things. With the Vcore voltage set to auto, the Vcore voltage is 1.5160v in HWmonitor. In the BIOS hardware monitor it shows 1.4667v. It seems there is some offset between what the BIOS says is true and what HWmonitor says. Even CPUZ says the same as HWmonitor. It would seem that I got another flaky board. The BIOS rev was something like 5 revs ago. I updated it to the latest BIOS rev. I'm about ready to give up on Asus repair. I set the the Vcore manually to 1.4375v and HWmonitor reads the vcore as 1.4780v. Also the CPUNB voltage was still showing as 1.375 in AOD. This is the only way I have to read that voltage. Oh, Asus PC probe II reads the Vcore the same as HWmonitor and CPUZ. I'm going to contact Asus again about this board. It looks like I might have to upgrade after all if I can't get a good board.
I just got off the phone with MS. I had to reactivate my Windows due to a hardware change. I'm not sure what the hardware change was. All parts are the same except for the replacement MB.

 

[madman7]

This is the second time I had to reactivate Windows. The first was a quick internet activation. I'm not so worried about the activation. I'll tell them I've had MB problems and that I couldn't get a good stable board. I just wonder if I should just run this board or try to get another so called repaired board. I don't really want to upgrade yet. If I do upgrade, I'm going to try to do it without reinstalling Windows. Asus said I might be able to install a new board and do a repair with the installation disc as long as I stay with all AMD.

 

[RGone]

I just wonder if I should just run this board or try to get another so called repaired board.
I thought you said that the ram worked now in all the slots. If it does there is really nothing wrong with the board. Software is just that software and is prone to error. If you want to know what the Vcore is then get a DMM and locate the test point for Cpu Vcore and test it. Often software is off in reading voltages. Heck it is software reading a 7 cent chip.

I don't really want to upgrade yet.
Well if you don't want to upgrade then you will have set the voltages manually and use the board. Correct the voltage offset in your head and set bios low or high to get what voltage you think you want. We cannot fix the problems you write about having with voltages. Don't think Asus will either. Get a good board or use the one you have that now has all memory slots working. Long and short of it.
RGone...