Question Case fans stays on after shutdown when I place my 6600xt in but doesnt stay on when I put it an rx 580?

[M.Zachary]

The case fans that I use are 4 molex fans connected in a daisy chain. Non of them have rgb lighting

When the 6600xt is plugged in, the case fans instantly start spinning without me turning on the system when I turn on the wall plug

With the rx580 on the other hand, the fans stay off, they dont move until I power on the system.

Cpu:i7 3770
Mobo:Asrock H77 Pro4
Ram: 16gb 1600mhz DDR3 kingston
Psu: Evga 700br
Gpu: Gigabyte rx6600 xt & Asus RX580

Does this mean that my 6600xt is faulty?

 

[Ralston18]

This:

"When the 6600xt is plugged in, the case fans instantly start spinning without me turning on the system when I turn on the wall plug" [ My underline.]

Do you mean plugging in the system to a switched wall plug (outlet) that would turn on a lamp.

Plugging in the computer with the 6600 XT installs starts the fans.

Plugging in the computer with the RX580 does nothing until the computer is physically turned on.

And the only difference/change is the physical presence of the GB RX6600 XT or the Asus RX580 - correct?

No other changes at all. And the computer does work no matter which GPU is installed - correct?

Just how are those molex connectors being connected? What is the physical path of the connections being made?

 

[M.Zachary]

This:

"When the 6600xt is plugged in, the case fans instantly start spinning without me turning on the system when I turn on the wall plug" [ My underline.]

Do you mean plugging in the system to a switched wall plug (outlet) that would turn on a lamp.

Plugging in the computer with the 6600 XT installs starts the fans.

Plugging in the computer with the RX580 does nothing until the computer is physically turned on.

And the only difference/change is the physical presence of the GB RX6600 XT or the Asus RX580 - correct?

No other changes at all. And the computer does work no matter which GPU is installed - correct?

Just how are those molex connectors being connected? What is the physical path of the connections being made?

Yes this,
Do you mean plugging in the system to a switched wall plug (outlet) that would turn on a lamp.

I shut off all power from the system before swapping the 6600xt/rx580, fans didnt start spinning when I installed them when the system was off

Yes the rx 6600xt makes the case fan start spinning when i switch on the wall plug but not the computer

While the rx580 does not make the case fan start when i switch on the wall plug and not the computer

Yep the difference is the physical presence of the 6600xt/rx580 and yes the computer does work regardless of the gpu

The molex fans are connected together with a daisy chain and powered by 1 molex adapter at the start of the chain
Molex adapter ->fan -> fan -> fan -> fan

 

[Ralston18]

I found the following motherboard link:

https://download.asrock.com/Manual/H77 Pro4MVP.pdf

[Do verify that I found the applicable user manual for the motherboard.]

If the computer/outlet is not turned on and the fans are still receiving power then the system is not truly off.

My thought is that "Sleep States" may be involved. FYI:

https://learn.microsoft.com/en-us/windows-hardware/drivers/kernel/system-sleeping-states

Question being is the computer really off with absolutely no power present.

As if the computer was completely unplugged. Power could be present at the computer any time the outlet wall switch is on even if the computer's power switch is off.

I do not know why the behavior is different between the two GPUs. Especially if nothing else is changed. Perhaps the RX 6600xt is being "sensed" and the RX580 is not sensed.

From the Microsoft sleeping states link:

"Some devices can wake the system from a sleeping state when certain events occur. In addition, on some computers, an external indicator tells the user that the system is merely sleeping."

What does the Power LED (Reference physically numbered Page 34 of the User Manual) indicate with wall switch off and on? Is the indication the same for both GPU's? If not, what is different?

 

[M.Zachary]

I found the following motherboard link:

https://download.asrock.com/Manual/H77 Pro4MVP.pdf

[Do verify that I found the applicable user manual for the motherboard.]

If the computer/outlet is not turned on and the fans are still receiving power then the system is not truly off.

My thought is that "Sleep States" may be involved. FYI:

https://learn.microsoft.com/en-us/windows-hardware/drivers/kernel/system-sleeping-states

Question being is the computer really off with absolutely no power present.

As if the computer was completely unplugged. Power could be present at the computer any time the outlet wall switch is on even if the computer's power switch is off.

I do not know why the behavior is different between the two GPUs. Especially if nothing else is changed. Perhaps the RX 6600xt is being "sensed" and the RX580 is not sensed.

From the Microsoft sleeping states link:

"Some devices can wake the system from a sleeping state when certain events occur. In addition, on some computers, an external indicator tells the user that the system is merely sleeping."

What does the Power LED (Reference physically numbered Page 34 of the User Manual) indicate with wall switch off and on? Is the indication the same for both GPU's? If not, what is different?

Ill will try it as soon as I can because I cant get it to run due to my last available ram stick getting killed again.

As of now 4 sets of my ram sticks have been killed in less than 2 months and this motherboard has killed 3 of the sets, the last set got killed yesterday and I have been trying to find the reason.

I thought that a faulty gpu keeping the system fans on after shut down could be the issue.

1st time it died was when I oc my cpu on my old motherboard.

2nd time it died was when I was in bios changing wake from keyboard and igpu shared memory size.

3rd time it died was when I was swapping ram sticks.

4th time which was yesterday, died when I was swapping too. Made sure I touched metal and the power was off but could it be that I switched too quickly before the system lost all of its power?

Similarities between the 1,2 and 3rd time was the cpu, gpu, motherboard.

similar item between the 4th time and the 1,2,3rd time was the cpu.

The Only Similarity between All, is windows 10 was plugged in.

1st time :
Cpu: i7 3770
Gpu: 6600xt
Ram: teamgroup 32gb 1600mhz DDR3
Mobo: Intel dh77kc
Psu: evga 700br
Reason ram died: oc cpu

2nd time:
Cpu: i7 3770
Gpu: 6600xt
Ram: kingston fury 16gb 1600mhz DDR3
Mobo: Asrock H77 Pro4 mvp
Psu: evga 700br
Reason ram died:changing bios setting, wake from keyboard and igpu shared memory size

3rd time:
Cpu: i7 3770
Gpu: 6600xt
Ram: kingston 4gb 1600mhz DDR3
Mobo: Asrock H77 Pro4 mvp
Psu: evga 700br
Reason ram died: swapping ram sticks from my cousin pc into my pc.(swapped too fast?)

4th time:
Cpu: i7 3770
Gpu: RX580
Ram: Kingston fury 16gb 1600mhz DDR3
Mobo: Asrock H77 Pro4 mvp
Psu: aigo gp550
Reason ram died: swapped ram from my 16gb 1600mhz DDR3 to an 8gb 1600mhz DDR3 and back to the16gb 1600mhz DDR3, died after I changed it back.(swapped too fast?)

I've been suspecting a few things and changing a few components but still can't seem to get the ram to stop dying.

Could it be a windows setting,esd that killed ram,a faulty cpu , faulty motherboard or following this thread which is the motherboard isnt fully asleep?

 

[Ralston18]

Windows 10 is not killing components.

Plus the overall situation is going from bad to worse as I understand the posts to date.

If anything there is some electrical problem - either in the power connections to the computer (wall outlet to PSU) or internally with respect to the cables and connectors being used. The power is believed to be "off" yet there is some existing voltage (may be low) that is just enough to surge through new components when the component is plugged in.

What other devices are plugged into the sockets serving the problem PC? Any surge protectors, power bars, TVs, audio equipment, etc. that could be creating some sort of electrical loop that routes power to the PC?

And, after reading back a bit I am starting to wonder about that 4 connector Molex daisy chain serving the four fans. How is that cable connected in - origin? Any adapters being used?

The motherboard would be a good suspect except that RAM has been killed on two different motherboards.

The PSU is a another likely suspect; however, two different PSUs have been tried. The evga not being modular. The aigo is a modular PSU provided I found the correct model. Are any cables being used from any other PSU?

What installation procedures are being followed when components are swapped? For example, some motherboards require that the first physically installed RAM stick be placed in a specific slot. As far as I am aware failing to do so does not "kill" the RAM stick. The system simply either does not work or only some of the installed RAM appears.

And I would also begin wondering about the source of the RAM being used. Noted use of cousin's RAM but where is RAM being purchased? It is also very important to verify that any installed RAM is indeed supported by the motherboard. Details/specs matter. And supported RAM can change so verify via the motherboard manufacturer's QVL (Qualified Vendors List).

But that does not include the other components that have been "killed".

All I can think of is an electrical problem or some sort. Either directly or by error of omission or commission.

Do you have a multi-meter and know how to use it? Or know someone who does? It may prove worthwhile to do some testing for voltages and shorts. That testing must be carefully done.

There may be other ideas and suggestions.

 

[M.Zachary]

Windows 10 is not killing components.

Plus, the overall situation is going from bad to worse as I understand the posts to date.

If anything, there is some electrical problem - either in the power connections to the computer (wall outlet to PSU) or internally with respect to the cables and connectors being used. The power is believed to be "off," yet there is some existing voltage (may be low) that is just enough to surge through new components when the component is plugged in.

What other devices are plugged into the sockets serving the problem PC? Any surge protectors, power bars, TVs, audio equipment, etc. that could be creating some sort of electrical loop that routes power to the PC?

And, after reading back a bit I am starting to wonder about that 4 connector Molex daisy chain serving the four fans. How is that cable connected in - origin? Any adapters being used?

The motherboard would be a good suspect except that RAM has been killed on two different motherboards.

The PSU is a another likely suspect; however, two different PSUs have been tried. The evga not being modular. The aigo is a modular PSU provided I found the correct model. Are any cables being used from any other PSU?

What installation procedures are being followed when components are swapped? For example, some motherboards require that the first physically installed RAM stick be placed in a specific slot. As far as I am aware failing to do so does not "kill" the RAM stick. The system simply either does not work or only some of the installed RAM appears.

And I would also begin wondering about the source of the RAM being used. Noted use of cousin's RAM but where is RAM being purchased? It is also very important to verify that any installed RAM is indeed supported by the motherboard. Details/specs matter. And supported RAM can change so verify via the motherboard manufacturer's QVL (Qualified Vendors List).

But that does not include the other components that have been "killed".

All I can think of is an electrical problem or some sort. Either directly or by error of omission or commission.

Do you have a multi-meter and know how to use it? Or know someone who does? It may prove worthwhile to do some testing for voltages and shorts. That testing must be carefully done.

There may be other ideas and suggestions.

All the times it died, the pc was connected to an avr that was connected to a wall plug. Only thing sharing the avr with the pc is the monitor

The daisy chain is power by 1 molex adapter and thats pretty much all the power it gets from. No other adapters were used

The aigo gp550 is also a non modular psu just like the evga 700br

The procedures to install the ram according to the manual should be to shut down the system, hold the power button to drain the remaining power. Then, place the 1st ram stick in A1, 2nd in B1, 3rd in A2, and 4th in B2

According to the qvl list of my asrock mobo, all the rams except my cousin's ram are compatible even though it did manage to work with it. If Im not mistaken, it is a 4gb kingston 1600mhz DDR3 ram

I do have a multimeter, but my sister took it to her college, and I'm not quite sure on how to use one on the mobo to test the voltages 😅

Do you suggest that I bring it to my local computer technician?

My only concerns about them is that they might steal my pc components and swap it with a broken one to scam me. Or either they overcharge me.

The first time the ram died was after I overclocked the cpu(the motherboard died, too). Could it possibly be that after the oc, the cpu became the silent killer that has been killing my ram?

Lastly, thank you so much for all the help you've given. Even though the problem isn't solved yet, it feels like you're giving me hope that it can actually be solved.

 

[Karadjgne]

There's stuff here that makes no sense, unless you are trying to run a BCLK OC. The i7-3770 is a factory locked cpu, you physically cannot multiplier OC it. The Intel dh77kc is a locked Bios mobo, almost all Intel boards are as Intel doesn't support OC by default. The ASRock H77 Pro4 is also a locked Bios mobo, all the B/H are by default.

It all points to a total inability to OC unless you are using software, which will use BCLK. And is impossible to totally remove from a system once activated without a full reinstall of OS, factory default in bios etc as some of the voltage settings are technically within spec so don't get reversed or reset by simply uninstalling the software or turning it off.

As bad as the Evga 700BR is viewed as, it's actually a better psu than the Aigo.

QVL has nothing to do with ram compatibility. It's a qualified vendor list, not a qualified ram list. It's only purpose is to state that the mobo vendor used several different ram OEM's and the mobo worked, not that only the ram used works. There's only a handful of ram OEMs like Samsung, SkHynix, Nanya etc and they manufacture ram for every ram vendor like Corsair, G.Skill, Team etc. So you'll get G.Skill Trident-Z manufactured by both Samsung and SkHynix, depending on the model or speed.

You seriously cannot 'swap ram too fast', but the ram Can be bricked by a BCLK overclock or at a minimum can seem like it's bricked if under a BCLK overclock and not stable at those changed speeds with other factors remaining factory.

The first time the ram died was after I overclocked the cpu(the motherboard died, too). Could it possibly be that after the oc, the cpu became the silent killer that has been killing my ram?

No. The ram deaths, cpu issues, motherboard issues/death are a direct result of another source, the initial cause of the problems, You.

If you walked past a car, holding your keys out and running them down the length of the car, you cannot then put blame on the keys as being responsible for the scratches in the paint. Forcing an OC on a locked Bios, locked cpu on a mobo that physically does not and was not designed for an OC is no different.

 

[M.Zachary]

There's stuff here that makes no sense, unless you are trying to run a BCLK OC. The i7-3770 is a factory locked cpu, you physically cannot multiplier OC it. The Intel dh77kc is a locked Bios mobo, almost all Intel boards are as Intel doesn't support OC by default. The ASRock H77 Pro4 is also a locked Bios mobo, all the B/H are by default.

It all points to a total inability to OC unless you are using software, which will use BCLK. And is impossible to totally remove from a system once activated without a full reinstall of OS, factory default in bios etc as some of the voltage settings are technically within spec so don't get reversed or reset by simply uninstalling the software or turning it off.

As bad as the Evga 700BR is viewed as, it's actually a better psu than the Aigo.

QVL has nothing to do with ram compatibility. It's a qualified vendor list, not a qualified ram list. It's only purpose is to state that the mobo vendor used several different ram OEM's and the mobo worked, not that only the ram used works. There's only a handful of ram OEMs like Samsung, SkHynix, Nanya etc and they manufacture ram for every ram vendor like Corsair, G.Skill, Team etc. So you'll get G.Skill Trident-Z manufactured by both Samsung and SkHynix, depending on the model or speed.

You seriously cannot 'swap ram too fast', but the ram Can be bricked by a BCLK overclock or at a minimum can seem like it's bricked if under a BCLK overclock and not stable at those changed speeds with other factors remaining factory.
No. The ram deaths, cpu issues, motherboard issues/death are a direct result of another source, the initial cause of the problems, You.

The last time i told someone i increased the cpu multiplyer, a person corrected me and said it was the same thing as overclocking. From then on out I kept thinking they're the same thing. I only changed the multiplier on the intel mobo but not the asrock as i thought i might kill the ram and mobo again

 

[Karadjgne]

The last time i told someone i increased the cpu multiplyer, a person corrected me and said it was the same thing as overclocking. From then on out I kept thinking they're the same thing. I only changed the multiplier on the intel mobo but not the asrock as i thought i might kill the ram and mobo again

But you then used the same ram, same cpu, same gpu, same hard drives and initially the same OS on the new mobo.

Reason ram died: oc cpu

No. Reason ram died (if it died), is indeterminate, could have been weak, over-volted, bad bios settings, any number of possibilities. Bumping the 3rd Gen multiplier available on some motherboards a possible 400MHz (specific to Ivy-Bridge cpus) isn't any guarantee of ram death.

Reason ram died:changing bios setting, wake from keyboard and igpu shared memory size

No. Wake from keyboard is a Windows timer. Igpu is a physical part of the cpu, has absolutely nothing to do with how bios/windows organizes ram hardware reserved settings.

Reason ram died: swapping ram sticks from my cousin pc into my pc.(swapped too fast?)

Nonsense.

Reason ram died: swapped ram from my 16gb 1600mhz DDR3 to an 8gb 1600mhz DDR3 and back to the16gb 1600mhz DDR3, died after I changed it back.(swapped too fast?)

Nonsense.

 

[M.Zachary]

But you then used the same ram, same cpu, same gpu, same hard drives and initially the same OS on the new mobo.

Yes.

No. Wake from keyboard is a Windows timer. Igpu is a physical part of the cpu, has absolutely nothing to do with how bios/windows organizes ram hardware reserved settings.

Nonsense.

Nonsense.

The asrock did automatically increase my cpu multiplyer without me doing it myself.

Could that possibly be the root of the
issue?

 

[Karadjgne]

Sandy/Ivy-Bridge cpus were a one-off experiment by Intel. They would allow (on certain motherboards) the multiplier (generally that's turbo rated speed multiplier, 32-38) to be raised upto 400MHz. But afaik that only applied to a locked cpu (non-K) on a P67/Z68/Z77 overclockable motherboard, not H77 chipset boards, and definitely nothing built by Intel.

So the only way to OC a locked cpu, on a locked Bios H77 mobo is by BCLK, (default 100.00) which affects Everything attached to the front side buss, namely the cpu speeds, ram speeds, memory controller speeds, igpu speeds, gpu speeds, communication speeds, storage speeds etc. Essentially everything that has a MHz or GHz bandwidth.

With a BCLK OC comes not only instability, but if the mobo is not specifically built to handle the higher MHz, a seriously higher heat output. A 103.5 setting is significantly warmer, a 107.3 BCLK will fry components.

 

[M.Zachary]

Sandy/Ivy-Bridge cpus were a one-off experiment by Intel. They would allow (on certain motherboards) the multiplier (generally that's turbo rated speed multiplier, 32-38) to be raised upto 400MHz. But afaik that only applied to a locked cpu (non-K) on a P67/Z68/Z77 overclockable motherboard, not H77 chipset boards, and definitely nothing built by Intel.

So the only way to OC a locked cpu, on a locked Bios H77 mobo is by BCLK, (default 100.00) which affects Everything attached to the front side buss, namely the cpu speeds, ram speeds, memory controller speeds, igpu speeds, gpu speeds, communication speeds, storage speeds etc. Essentially everything that has a MHz or GHz bandwidth.

With a BCLK OC comes not only instability, but if the mobo is not specifically built to handle the higher MHz, a seriously higher heat output. A 103.5 setting is significantly warmer, a 107.3 BCLK will fry components.

First time the ram died was when i changed the multiplier on the cpu with the intel motherboard.

Second time onwards I used the asrock motherboard which automatically increased the cpu multiplyer from 3.4 ghz to around 3.9 ghz but never 4.0 ghz

Could it be that all the ram died because the motherboard automatically changed the cpu multiplyer and increased the voltage by too much?

 

[Karadjgne]

One doesn't affect the other. If the cpu was rated for 3.4GHz, the most you could get from it would be 3.4GHz because the mobo doesn't allow multiplier changes above what the cpu turbo speed is. The multiplier being a 34 or lower.

A cpu speed is BLCK x multiplier. So if the BCLK is 100.00MHz x 34 you get 3.4GHz. If you wanted 3.9GHz (K cpu on Z board) you either raise the BCLK from 100.00 to 114.70 or raise the multiplier from 34 to 39.

At 114.70MHz, your effective ram speed, memory controller speed, igpu etc set at 1600MHz would be 1832MHz, and if not supported by both voltage and timings changes would likely be unstable and refuse to boot.

114.70MHz BCLK would also likely cook several components, including the cpu and VRM's and other unshielded and unheatsinked power delivery components or at a very minimum would be so unstable as to constantly crash, bsod or simply refuse to boot.

The only ppl who have any business with jacking the BCLK that high are running LN2, liquid nitrogen cooling.

 

[M.Zachary]

One doesn't affect the other. If the cpu was rated for 3.4GHz, the most you could get from it would be 3.4GHz because the mobo doesn't allow multiplier changes above what the cpu turbo speed is. The multiplier being a 34 or lower.

A cpu speed is BLCK x multiplier. So if the BCLK is 100.00MHz x 34 you get 3.4GHz. If you wanted 3.9GHz (K cpu on Z board) you either raise the BCLK from 100.00 to 114.70 or raise the multiplier from 34 to 39.

At 114.70MHz, your effective ram speed, memory controller speed, igpu etc set at 1600MHz would be 1832MHz, and if not supported by both voltage and timings changes would likely be unstable and refuse to boot.

114.70MHz BCLK would also likely cook several components, including the cpu and VRM's and other unshielded and unheatsinked power delivery components or at a very minimum would be so unstable as to constantly crash, bsod or simply refuse to boot.

The only ppl who have any business with jacking the BCLK that high are running LN2, liquid nitrogen cooling.

Ohh that might be why the rams keep dying.

Even though I wasnt able to change the base clock on the cpu, what the intel motherboard and asrock motherboard did allow me to do was change the max turbo boost multiplyer for the cpu, the turbo boost base clock and the duration of the turbo boost.

Intel motherboard and ram died after I changed the turbo boost multiplyer from 34 to 40 but its works fine at 39, the turbo boost base clock at 100 and the option for turbo boost duration was (!), instead of seconds the option was (!)

Asrock motherboard had turbo boost multiplyer set to auto, turbo boost base clock set to auto and duration set to auto.

I left everything in bios at the asrock motherboard at default. When I was playing games or running benchmarks with it, it mostly stayed at 3.8 ghz and sometimes went up to 3.9 ghz.
Most of the time it runs at 3.3 - 3.6ghz during idle.

Cpu temps at max are around 70°C and idles around 35°C

Never really went out of my way to check ram speeds but according to msi afterburner, it stayed at 1600mhz in games.

 

[Karadjgne]

Most of the time it runs at 3.3 - 3.6ghz during idle.

Idle speeds should be 1.6GHz....

So what software did you install that is supposed to boost cpu performance, because that software isn't doing you any favors. If it's asrock XTU, that program is garbage, which is why it wasn't updated, it was replaced in later gens.

 

[M.Zachary]

Idle speeds should be 1.6GHz....

So what software did you install that is supposed to boost cpu performance, because that software isn't doing you any favors. If it's asrock XTU, that program is garbage, which is why it wasn't updated, it was replaced in later gens.

Only apps I installed that might be related is Prime 95, 3D Mark, MSI Afterburner, Crystaldiskmark, and heaven benchmark.
Other than those, nothing related to enhancing performance.

I checked back the motherboard and found out a few things though


OC Tweaker Screen:
Host Clock Override (BCLK)
: The default is [Auto]

DRAM Timing control:
Load XMP Settings
DRAM tCL
DRAM tRCD
DRAM tRP
DRAM tRAS
Command Rate
DRAM tWR
DRAM tRFC
DRAM tRRD
DRAM tWTR
DRAM tRTP
DRAM tFAW
DRAM tCWL
ODT WR (CH A)
ODT WR (CH B)
ODT NOM (CH A)
ODT NOM (CH B)
: The default is [Auto]

Voltage configuration:
CPU Core Voltage
DRAM Voltage
VTT Voltage
PCH voltage
CPU PLL Voltage
VCCSA Voltage
: The default is [Auto]

Could either one of these settings be the reason why the ram modules have been dying?

 

[M.Zachary]

I swapped a motherboard and now the fans dont stay on after shutdown

 

[M.Zachary]

The problem turned out to be a faulty cpu, seems to kill the ram whenever I change any settings in bios or when swapping ram.