AB350 Pro 4 - Ram Voltage

wardog wrote: ket wrote: Corsair RAM truly is horrible, ...... I've never understood the rabid following Cursair maintains.

That makes two of us. Whenever I've had the displeasure of using a Corsair kit it has always been any order of magnitude inferior to any other RAM kits I've tested, wether that be G.Skill, Mushkin, when they were about Cellshock, and more often than not even generic kits. Unless you are buying Corsairs most expensive RAM it's a real grab bag of utter dung and absolutely not worth it. The kits that are £/$10 more are so for a reason, theres actually at least some quality in those kits.

datonyb wrote: availabilty for consumer mark ups for stockist price for consumer i agree i dont think it excells as ram myself but i did used to favour it with ddr2 ( xms2 used to be a safe bet for the money) i still feel blessed by the gods that i managed to overclock some ddr4 2400 corsair lpx to 2933 on a ryzen 1600 system a week or two ago ,and on the same brand motherboard last weekend got another system to click xmp directly at 2933 with the 3000 lpx sticks (it is what a consumer expects but dosnt often see on ryzen at the moment) what choice do we have ? gskill tridentz has seen prices skyrocket in the UK since ryzen launched (well the samsung sticks at least)

Honestly, buy any other brand or even generic and you'll have less problems getting it to run with Ryzen than a Corsair kit. Thats coming from someone who has had quite a few chats about how problematic Corsair kits are on Ryzen and numerous users coming to me with Corsair kits not working on Ryzen hoping I have some timings for them. It's generally agreed to stay away from Corsair if you have or plan to get a Ryzen system. In my case I'd take it further and say don't bother with Corsair at all go with any other brand, pay the bit extra and get a kit that isn't made with bottom of the barrel ICs. I've heard a lot of stories over the years about Corsair overclocking ICs out of the box so to speak and selling them at higher labelled frequencies. Make of that what you will.


Edited by ket - 08 Sep 2017 at 9:38am

gabrimor wrote: Its really odd behavior, I know. RAM is CMK8GX4M2B3000C15I5, and i'm using the banks at slots a2 and b2. As the RAM works with 1.21v, we can say 1.35v its not enough. As I said, I tried everything, but the problem still. With 1 bank, I can set 1.35v, with two, 1.23v is the limit. The new fact is I can set 1.35v with A-,Tuning software.. I tested RAM with Memtest, Windows Memory Diagnostic, Prime Blend and Asus RealBench. PASSED. But fails with 10 seconds with AIDA64 stability. Sorry for poor text, Um using mobile phone to right....

Have you cleared the UEFI/BIOS with the jumper on the board at any time while testing your memory? If not, you need to do that now. Also do a Load UEFI defaults when you get into UEFI after it is cleared.

When over clocking memory, we know we need to increase the DRAM voltage to allow the memory to complete POST. If increasing the DRAM voltage causes a POST failure, which we still don't have any clue about the cause, not even POST beeps that I asked about, I would only dismiss the board as bad if it did that with more than one model of memory.

 I'm very surprised that this memory runs at 2666 at 1.21V at all. For the two Intel systems this memory was designed for, X99 and the 100 series chipset boards, any memory speed above 2133 is a memory OC. There is no chance this memory could operate a 2666 at 1.21V on any of those Intel systems.

What are you using to check the DRAM voltage when in Windows?

When you set the DRAM voltage to 1.35V with A-Tuning, did you check what the DRAM voltage is with another program, like Aida64?

parsec wrote: gabrimor wrote: Its really odd behavior, I know. RAM is CMK8GX4M2B3000C15I5, and i'm using the banks at slots a2 and b2. As the RAM works with 1.21v, we can say 1.35v its not enough. As I said, I tried everything, but the problem still. With 1 bank, I can set 1.35v, with two, 1.23v is the limit. The new fact is I can set 1.35v with A-,Tuning software.. I tested RAM with Memtest, Windows Memory Diagnostic, Prime Blend and Asus RealBench. PASSED. But fails with 10 seconds with AIDA64 stability. Sorry for poor text, Um using mobile phone to right....       Have you cleared the UEFI/BIOS with the jumper on the board at any time while testing your memory? If not, you need to do that now. Also do a Load UEFI defaults when you get into UEFI after it is cleared. When over clocking memory, we know we need to increase the DRAM voltage to allow the memory to complete POST. If increasing the DRAM voltage causes a POST failure, which we still don't have any clue about the cause, not even POST beeps that I asked about, I would only dismiss the board as bad if it did that with more than one model of memory.  I'm very surprised that this memory runs at 2666 at 1.21V at all. For the two Intel systems this memory was designed for, X99 and the 100 series chipset boards, any memory speed above 2133 is a memory OC. There is no chance this memory could operate a 2666 at 1.21V on any of those Intel systems. What are you using to check the DRAM voltage when in Windows? When you set the DRAM voltage to 1.35V with A-Tuning, did you check what the DRAM voltage is with another program, like Aida64?

This is definitely one of the oddest things I've heard in 22 years of being involved with technology I think we are just going to have to keep taking baby steps until the problem can be isolated. The next thing I would suggest is to run HCI Memtest, opening enough instances to fully utilise your RAM if you use the free version, and let it run for at least 2-3 hours then see if it reports any memory errors. As you say 2133Mhz is even problematic right now my prime suspect would be the memory itself just not playing nicely.

Hard troubleshooting....