I think that I tried everything that I know and can do, please help.
Replacing CMOS battery didn't help, so as resetting BIOS to defaults.
Everything is cleaned. (I always have my computer cleaned, every 6 months).
BIOS version 0460 (can't upgrade to 0650, cause of THIS problem, it does not allow me to upgrade, I tried all the ways of upgrading (F7, iflash, EBU) presented on this site. Nothing is overclocked, tempratures are fine.
Tried to start without the other consumers (only MB, processor, RAM), nothing changes. Maybe there is a special option in BIOS, that can stop this mess and I don't know what it is, so please, help me. I remember that when I bought this MB, I didn't have this problem. Oh, and there is no swollen capacitors on it. This MB has a post-code display on it, and I noticed, that computer shuts down, when "2b" code is displayed... As I said, after that it turns on normally. What can I do in this case? I need any help!
Core i7 3820
4 x 4GB DDR3 Kingston HyperX Predator KHX24C11T2K2/8X (2400)
2 x 2 GTX 670 SLI (Gigabyte)
Chieftec aps-850c pfc
Zalman Z11 plus
SSD samsung EVO 860 1 TB
Everything works in stock, on default clocks, etc.
Al can be so terse sometimes. He's not incorrect, just terse...😉
When you see your system power on, run for a few seconds and then seemingly going through a full reset, what is happening is the Management Engine is resetting the system because it got into an unacceptable state and reset the system to recover. Typically, the ME gets into this state because it cannot access memory. It cannot access memory because the BIOS has been unable to initialize the memory in time. This is an indicator that the BIOS attempted to initialize memory and this initialization failed (memory bus locked up). The BIOS will retry this initialization, typically backing off the memory settings, and, if everything goes ok at these settings, the system preceeds through the remainder of POST. If not, the BIOS may retry once again, lowering the settings even further (typically to base line). Unfortunately, the ME will only wait so long for this initilaization to complete. Since the BIOS uses a timer to catch and recover from the bus lockups, this initialization can take time. Now, after the reset, it appears that the BIOS is getting the memory initialized in time (perhaps because of some knowledge of what was attempted previously) and everything proceeds from there. Ok, what can cause this? Well, first and foremost, the memory may only be borderline working. That is, at its XMP settings (i.e. 2400 speed), it is having problems initializing. Later, after retries, it may be configuring the memory at the base speed (i.e. 2133; you can verify this using runtime tools). The other possibility that Al refers to is that the power supply is not (or barely) meeting the peak power requirements at power on. This can occur because of an inadequate power supply or because of inadequate power generation when cold (due to power supply age).
BTW, another good indicator that the BIOS is going through the memory retries scenario is that the BIOS has a longstanding bug wherein, if memory initialization retries are occurring, it "forgets" about any BIOS upgrade operation that might be in progress. When the system reset occurs that will initiate the BIOS installation, if memory initialization retries are necessary, the BIOS' recovery from the memory initialization failure will (incorrectly) clear the pending operation.
Regardless of what or how the problem is occurring, Al is completely correct; the two things to try are going to be different memory and/or a different power supply.
Hope this helps,
Thank you for your answer, propably, you're right. I'll try another memory and power supply, when I'll get such an opportunity. I'll report on the results, if you're interested in it. =)
But I need to clarify details: if it is all because of memory, does it mean, that my current memory is malfunctioning or it has poor compatibility with my MB?
I don't think, that the problem is in the power supply, because the computer always stood in the room (temperature is 21-22 C) and there is no power surges or any problems with electricity in our region, but I do not exclude this option.
I forgot to say, that I don't use XMP settings now, RAM is running at 1333 (BIOS defaults), but when I had only 2 bars (8 GB RAM), I used dual channel, 2400 speed, XMP was on and everything was the same, the problem remained, so as at 1333 speed. Then I bought 2 more bars, now I'm using quad channel at 1333 speed, because 2400 speed "dissapeared" in BIOS (2133 remained). To say correctly, the 2400 speed XMP preset dissapeared, but the possibility of overclocking memory manually up to 2400 remained. But MB doesn't allow me to overclock memory even to 1600 speed. (this problem appears only when I use all of 4 memory modules). I also noticed, that first two bars, that I had had before I bought another ones, are 1-ranked, and the others are 2-ranked... And all modules are KHX24C11T2K2/8X. In the cpu-z I see, that the maximum bandwidth of first ones is pc3-12800 (800 MHz), and the second ones is pc3-10700 (667 MHz). But all of them can use XMP 2400 speed, linking to https://www.kingston.com/datasheets/KHX24C11T2K2_8X.pdf . So there is another question. This problem is not quite connected with my problem of "double start', but I need to solve this too, cause I need to know, shall I buy a new memory from another manufacturer (2x4 GB or 4x4 GB), if I want to use the quad channel at high speed (2133 or 2400), or I can do smth with mine?) Maybe updating the BIOS can solve this problem? What will be the best solution of this two problems in your opinion? I just want to get maximum perfomance from my current platform.
"does it mean, that my current memory is malfunctioning or it has poor compatibility with my MB?" Yes. After all, it is kingston memory.
Power supplies do not last forever and can go bad.
Sure, update the bios, just to eliminate that as a possible cause.
So Al was trying to make a joke, but I have to agree that Kingston's memory is crap. I won't use it.
Ok, that statement aside, no, it isn't necessarily because the memory is malfunctioning or has poor compatibility with the board. The issue is age. All components generate some kind of noise. As these components age, they generate more noise. The processor's memory controllers generate noise. Components around the memory buses generate noise. The DIMMs and the ICs on these DIMMs generate noise. All of this noise combines to disaffect data being transmitted across the memory buses. If the amount of noise exceeds the tolerance of the bus interfaces in the processor or the bus interfaces on the DIMMs, then data errors, bus lockups, etc. will occur. The motherboard is designed to try to limit noise - using components like termination resistors, etc. - there is only so much that can be handled. When the board, processor and DIMMs are fairly new, certain transmission rates can be sustained. As the components age, however, the transmission rates that can be sustained will diminish. That memory that works at 2400MHz initially may only be able to operate at 2133 or lower speeds after three or four years. The BIOS will attempt to compensate for this, using lower and lower transmission rates, but you could eventually reach a point where even the lowest clock speed (1333) cannot be sustained and components will need to be replaced in order to continue.
The same thing goes for the Power Supply. As it ages, its power generation capability will degrade. Eventually, this will impact its ability to handle the peak load at power on time. It can also generate more noise and this noise will eventually disaffect the operation of the board.
My final comments - First of all, 2400 and 2133 are awfully high speeds for older system designs to try and sustain. We're not talking a problem in design or anything to do with the age of the system; we are talking about limitations in the DDR3 specification. DDR4 was (better) designed to handle these speeds and incorporates much better noise tolerance capabilities.
Ok, I understand, that my system is no longer a fresh one, and I assembled it in 2011-2012, of corse everything has it's end. But I would never have thought, that my system would have troubles with components degradation. For example, PSU, when I bought it, it was almost on the top of the market, with different protection methods (surge protection and so on).
Going back to my problem of double start: I tried two different memories, as you recommended. It didn't help, the system starts only from the second time. (BIOS default settings). I inserted it in different slots, swapped them, left only one module. Nothing helps. So we can make the conclusion, that the problem is in the PSU.
And what about the second problem, I have a good news. Yesterday I managed to cheat on my system. After I understood, that my PSU is worn out, and memory is fine, I decided to deal with frequencies. I inserted only 2 of my modules (kingston 2400), as it had been, before I bought 2 more the same modules. I enabled in BIOS XMP 1.3 2400 GHz and took a photo of XMP presets (timings, voltage, etc.). Then I started Windows, made sure, that clocks are active, everything works well, then I shut down the computer, inserted 2 of my remaining modules, turned it on. When I opened the BIOS, of corse, XMP settings had already dissapeared. 1333 speed was set automatically for my 4 kingston modules. And as you know, even with a slight overclocking (1600 for example), the system always refused to start. So I simply entered values from photo, that I had taken before, in manual mode. I didn't know, would they fit 4 memory modules. I saved BIOS profile, and tried to start the system. I was pleasantly surprised, when I understood, that it had worked. Then I made a few stability tests, nothing interferes with good computer performance. Now it is running on 2400 speed :D
But the problem with double start and psu still remains. I think, I have to buy a new one, this is a bad news.
"For example, PSU, when I bought it, it was almost on the top of the market"
You could purchase a new PSU today, and it could fail tomorrow, regardless of price.
"good" is a relative term. I have used thermaltake, cooler master, PC power & cooling, etc, and have been pleased with the quality. A fan went out on my cooler master, which I replaced.
Just search a store (like newegg, for example) to narrow your parameters.
First of all, I only said that I suspected an issue; I did not say that you absolutely have to replace your power supply. What I am suggesting is that you test with an alternate power supply to see if this is indeed an issue. If it is, then (and only then) should you spend the money to replace the power supply. In the meantime, try to borrow one.
A number of sites provide calculators that allow you to determine power supply requirements. Do an internet search for "power supply calculator" and you should see hits for a number of these.