Fatal1ty 990FX Killer - Trident X 2400 RAM

thlayli wrote: This is my 2nd set of this RAM and Mobo G.Skill replaced the first set of 2400 Trident X and Asrock replaced the first 990FX Killer mobo Now, after a few solid months with RAM @ 1866MHz and CPUNB @ 2600MHz, I tried bumping it up above those and one or both are fried again. And unfortunately, I spoke too soon. It's still turning off. Just off. Not a Windows shutdown. Instant power off.

You're saying the memory and/or North Bridge is fried? Do you mean just not working with your setting in the BIOS, or literally damaged?

If the speed of the communications link between the North Bridge chipset and the processor is beyond a stable speed that one or the other or both can work with, IMO an instant power off is not a bad reaction.

That may be one of the few things that could be done in the situation where a basic link between two interdependent pieces of hardware fails.

That seems to be unbelievable to you. If so, why?

You may need to clear the BIOS before the PC will start again with the unusable frequency you selected.

thlayli wrote: Thanks for the reply. That doesn't account for the same behavior happening with http://www.gskill.com/en/product/f3-10666cl9q-16gbxl - which IS on the QVL. The IMC is crap on the AMD's, I know. Originally my goal was to hit CPUNB of 2800 and 1866 with the RAM. I'd been at 2600/1866 and running fine for the most part. It takes CPU/NB of 1.4V for me to do 2600. That's worried me since it would seem everyone else on the internet only needs +/- 1.3V to hit that speed. On a whim last night, I put the 2400 TridentX back in and loaded the 2600/1866 UEFI config that was known-good, that I spent forever dialing in. I know I'd done this already, but here I am, PC not shutting off when I open a browser or video game. Temperamental beasts. As for WHY I thought they'd be compatible, in my 30 years I'd never known RAM to be exclusive to a chipset/IMC. Sure, it may have said "Optimized for Z97" on it, but all I really cared was the top speed of the RAM was 2400, and the top speed of the Mobo was 2400. This was supposed to be a build that would last me a good while. I'm just happy it's not shutting off at this point.

thlayli, firstly, what exactly is your problem with this MB/RAM combo? Your system suddenly shuts off or does not shut down? I don't see through this.

Secondly, where in the QVL of this board do you find your memory? I can't.
http://asrock.com/mb/AMD/Fatal1ty%20990FX%20Killer/index.asp?cat=Memory&Model=Fatal1ty

There are so many types of memory modules, just one letter different, it is another product! Those brands buy a cheap bunch of chips, opt them for high speeds on whatever CPU/board combo while testing is second priority - AMD is highly possible not testet much, thanks to small maket share.

I'm not a friend of these passing fad memory kits anyway, so hip and hype with their funky heat spreaders, not knowing which memory chips are beneath. I recommend regular "OEM" modules from Kingston ValueRAM, Samsung, Crucial (Micron) and so on. These are testet for widespread compatibility and availability is given much longer.

Just because you were lucky the past 30 years, it doesn't mean you ran into a compatibility issue here. 20 years ago, there were much viewer kits available, with nearly no "OC sticker" on them. It is far more complicated these days.

parsec wrote: Your thread's title includes "Trident X 2400 RAM". I assume that is at least the 2400 memory you are trying to use with the Fatal1ty 990 FX Killer board. If this is the Trident X 2400 memory you were trying to use (whose default speed is 1333): http://www.gskill.com/en/product/f3-2400c10d-16gtx There are zero AMD boards listed in the G.SKILL QVL list of this memory. That includes seven mother board manufactures, among them ASRock. ASRock does not include this memory in the Memory Support List of the Fatal1ty 990 FX Killer board. Why did you think you could use this memory with this board? What caused you to believe this memory was compatible with an AMD PC system? This memory is not compatible with the memory controller specs of AMD processors. That would specifically be the memory chips used in this model of memory. Simply a reality of the differences between AMD and Intel processors. It's not that the board is a bad design or defective. The board itself is the least contributing factor regarding the incompatibility of this model of memory when used on an AMD system. Could this board be made to work with this memory and still use AMD processors? No. DDR3 memory is not a generic commodity. We cannot simply purchase any model of DDR3 memory we find attractive and use it in a random PC system. We will get lucky sometimes doing this, perhaps even 50% of the time with inexpensive, standard speed memory. But not every time. It's not the board.

Thanks for the reply. That doesn't account for the same behavior happening with http://www.gskill.com/en/product/f3-10666cl9q-16gbxl - which IS on the QVL.

The IMC is crap on the AMD's, I know. Originally my goal was to hit CPUNB of 2800 and 1866 with the RAM. I'd been at 2600/1866 and running fine for the most part.

It takes CPU/NB of 1.4V for me to do 2600. That's worried me since it would seem everyone else on the internet only needs +/- 1.3V to hit that speed.

On a whim last night, I put the 2400 TridentX back in and loaded the 2600/1866 UEFI config that was known-good, that I spent forever dialing in. I know I'd done this already, but here I am, PC not shutting off when I open a browser or video game. Temperamental beasts.

As for WHY I thought they'd be compatible, in my 30 years I'd never known RAM to be exclusive to a chipset/IMC. Sure, it may have said "Optimized for Z97" on it, but all I really cared was the top speed of the RAM was 2400, and the top speed of the Mobo was 2400. This was supposed to be a build that would last me a good while.

I'm just happy it's not shutting off at this point.

thlayli wrote: They're incompatible. RMA'd both once already. Replacements have now succumb to the same fate. I've put over 100+ hours into this on the test bench.  I'm just gonna get a different motherboard.  I'm posting this from the PC in question, but with Ripjaws 1333 in. I guess in both instances it zapped the RAM. So be forewarned.

thlayli wrote: So I replaced the RAM with known-good 1333MHz. Last time, before the board and 2400 RAM were both RMA'd, this would resolve the problem. But it's still shutting down on me. Gotta be the board. Was a bridge too far to try to get there, I know. Reaching CPU-NB of 2600MHz was tough enough, had to take it over 1.4V. I should've just been happy with that and some tighter timings at 1866 with the Trident X. But then the new Doom came out and I got greedy.

Your thread's title includes "Trident X 2400 RAM". I assume that is at least the 2400 memory you are trying to use with the Fatal1ty 990 FX Killer board.

If this is the Trident X 2400 memory you were trying to use (whose default speed is 1333):

http://www.gskill.com/en/product/f3-2400c10d-16gtx

There are zero AMD boards listed in the G.SKILL QVL list of this memory. That includes seven mother board manufactures, among them ASRock.

ASRock does not include this memory in the Memory Support List of the Fatal1ty 990 FX Killer board. Why did you think you could use this memory with this board? What caused you to believe this memory was compatible with an AMD PC system?

This memory is not compatible with the memory controller specs of AMD processors. That would specifically be the memory chips used in this model of memory. Simply a reality of the differences between AMD and Intel processors.

It's not that the board is a bad design or defective. The board itself is the least contributing factor regarding the incompatibility of this model of memory when used on an AMD system. Could this board be made to work with this memory and still use AMD processors? No.

DDR3 memory is not a generic commodity. We cannot simply purchase any model of DDR3 memory we find attractive and use it in a random PC system. We will get lucky sometimes doing this, perhaps even 50% of the time with inexpensive, standard speed memory. But not every time.

It's not the board.

So I replaced the RAM with known-good 1333MHz. Last time, before the board and 2400 RAM were both RMA'd, this would resolve the problem.