Taking a look at DDR4 voltage regulation with a 5950X and dual rank Samsung B-die

[@corsairsloop3234]

This must be the first ddr4 oscilloscope video youtube. Pretty cool to see. Y-cruncher really pushes the imc to the extremes as seen on the scope. God bless that stress test.

[@winebartender6653]

For tm5, if you create a custom test sequence using 1usmus that throws in a bunch of test 0's, set test time % lower and increase total cycles, you can create a lot of high->low->high load situations to create more transients during the test.

[@stigi3849]

Takk!

[@KG_BM]

ahh ryzen content. sign me UP

[@gmt-yt]

Neat! More questions than answers! I guess the memory is like the dish washing machine, the CPU is like the person who sometimes wants to use various dishes, and the memory controller is like the pile of dirty dishes next to the sink.

[@Michael-OBrien]

I'd wager that voltage stability doesn't affect DDR4 OC because the current load of isn't that high so idle power vs laod power doesn't vary that significantly.

[@ActuallyHardcoreOverclocking]

first

[@waitwhaaat7949]

Hey you cheated

[@stevofinj]

Hahahaha 🤣 BZ you are the man

[@-FOXX]

2nd? Yeah. 2nd. Dibs.

[@christopherjackson2157]

Do you get a spikier result with single threaded versus multithreaded workloads? I'm wondering if measuring it at the memory stick there's a bunch of tasks all averaging out. Might explain issues at the end of tests as some threads are finishing before other ones I think the issue with the mouse in Windows is that it generates a bunch of interrupts. I think they are generated 200x per second.

[@samiraperi467]

If it's a USB mouse there are no interrupts. It's all polling.

[@christopherjackson2157]

@@samiraperi467 ahhh ur right. Thanks for correcting me!

[@certain_media_outlets]

@buildzoid hi you know 9E code is for d-p board. cant figure it out. some memory issue or acknowledgment? and thanks for this vid so much. still loving the regular d version. mods are sweet on yours

[@operator2618]

ty

[@Bobis32]

if i had a guess the reason whey memtest is realitively flat is its mass reading and writing data sequentially while y-chruncher is reading and writing data off the memory in non-sequential blocks with inconsistent intervals

[@FunKaYxxD1sCO]

I wanted to ask when I'm doing work with my 5950x, I get like a random slow down, and then it comes right back. The mouse cursor would move in slow motion and the audio gets slowed down.

[@fleurdewin7958]

That is usually because your Infinity Fabric (FCLK) is not stable if you are sure that the CPU is not overheating , thus throttling.

[@GameBacardi]

Wierd behaviour... never happened to me. 5950x owner

[@thanatoast4]

HOLY SHIT SOMEBODY ELSE THAT HAS THIS. Dude, I've been dealing with this since last year. On TWO different CPU's. 2700 and a 5600X. Both on an MSI B450 Carbon. Nothing is or was defective, everything is rock solid stable. Friends PC also started doing it too, on a 3600X. For me personally, its become much less frequent, maybe I'll get it once a week, but it used to be once a day. Absolutely ZERO idea what it is. Has nothing to do with unstable settings because it does it even with PBO disabled and no XMP.

[@thatguy7595]

Isn't this the fTPM stutter issue?

[@thanatoast4]

@@thatguy7595 I've heard lots about that on the sidelines but never made the connection, thanks for pointing it out!

[@wardexgo]

Does anyone know, why DDR voltage regulator on B550 TOMAHAWK increase switching frequency based on load, or it's a common things for single phase regulator?

[@Basement-Science]

Dont know about specific boards, but generally, you can _reduce_ the frequency to save power, and there's basically no disadvantage while not at full load because you only get more ripple at high load.

[@ActuallyHardcoreOverclocking]

pretty common feature for voltage regulators to have.

[@ole7736]

Maybe voltage in memtest is not a limit problem but a signal integrity (eye test) thing showing under constant substantial load?

[@Ph42oN]

I think this may explain some of the weird behavior i see on my system. I can find settings that can pass hours of memory testing, but as soon as i stop test whole pc crashes.

[@thatguy7595]

Anyone know what's safe daily voltage for 16 Gbit DDR4 Samsung A-die? It's hard to find info about this IC.

[@Arokhantos]

Whats easier to stabilize on a daisy chain board like Asus x570-E 4x 8 gb ddr4 3600 single rank or 2x 16 gb dual rank ? and which overclocks the furthest or are all Samsung B die 2x 16 gb kits also single rank ?

[@AA-kx5es]

All 16gb sticks of b-die are dual rank

[@tadasnanartonis5139]

Hi, buildzoid. Could you mod freq of memory voltage supply? Maybe that can make at least a little difference between MoBos with various phase count for memory voltage or maybe compensate (or make it worse) for transient mess from cpu?

[@jannegrey593]

Would it be safe to run 5950X on Gigabyte X370 Gaming K5? You said it has good VRM, but that was for 8 cores. Would it be dangerous without active VRM cooling? No overclocking except normal Ryzen Boost.

[@ActuallyHardcoreOverclocking]

At stock it'll be fine but the VRM will get really hot if the CPU pulls more than 170W.

[@jammetortiz808]

@@ActuallyHardcoreOverclocking how about something like crosshair vi hero? you think it will be fine? I also wonder how high can the memory go as for example if someone wants to put an APU in.

[@jannegrey593]

@@ActuallyHardcoreOverclocking Thank you very much! Still a bit afraid to put it in this mobo, since temperatures in my attic room in the summer are insane. But at least now I know that if I put a fan near VRM (have to figure out how to do this) it might not blow up.

[@Airwave2k2]

@@jammetortiz808 If a crosshair doesn't handle a 5950X Asus would have screwed up royally. ROG Crosshair Boards are top of the line what Asus offers, so the question makes itself redundant. And "like" is like asking, do the high-end boards of other vendors handle a 5950X, which comes down to the same answer.

[@Arokhantos]

Whatsup with AMD bioses after agesa 1203c they all handicap vcore with edc above 140 im almost certain it also causes instability.

[@whatistruth_1]

Honestly the whole AGESA situation really makes no sense. The EDC limit as well as the VDDP and VDDG limits make anything above 1203c worse. Definitely not sticking to amd for systems that i overclock in the future as in my opinion Intel is just superior for overclocking.

[@Drumaier]

I have to buy soon 2x16gb ram for a 5950x on a asus tuf b550m, here in Argentina they sell a kit of kingston fury beast 3600mhz cl18 that is my first choice so far because of price mainly. Sounds like a decent option? The pc is meat to run 3d programs, is not a gamer pc and the GPU is a 3080 12gb.

[@isubiez]

Hey bro, could u advise me on what would be the most optimal ram for the MSI am4 b550 gaming plus, 7 5800x, and a 3080 ti. I'm looking for something like these G.SKILL Trident Z Royal Series 16GB (2 x 8GB) 288-Pin PC RAM DDR4 4000 (PC4 32000) Intel XMP 2.0 Desktop Memory Model F4-4000C14D-16GTRG. If you had anything better in mind, would love your advice. My goal is to get the most fps in warzone, something RGB is preferable, the g, skills trident gold I want but will take something if it's gonna be the better choice, I don't mind tuning it but would like it to be able to run its xmp to fall back on if anything goes wrong. If I could take some of your time to help me out would really appreciate it and would find a way to make it up to you. Thanks

[@WillFuI]

Wow no heat spreader 1.65v and no fan. How would u stop at for 4 sticks with heat spreaders no direct fan

[@GoodOlKuro]

I get this whine too on a B550 A-Pro

[@egeozgultekin4331]

Hello all, I ve just discovered b die meaning and I upgraded my ram with corsair vangance 3200 mhz cl14 b die ram 8X4. Huge performance jump for me with 5600X and b550 aorus pro oc. BUT I only enable xmp profile and no more overclocking is done. Is this ok or should I overclock ?

[@whatistruth_1]

B die is only really superior when overclocked as that's when it can be pushed to its full extent as it unlike every other ram die scales linearly with voltage(falls off eventually but that's at water cooling ram level of voltages so I wouldn't worry about it), allowing for nearly all timings to be tuned tight.

[@Greez1337]

For sure. Expirement and see what you can get away with.

[@jake20479]

theres a lot to ram overclocking that i wouldnt recommend to beginners without extensive research. a few wrong timings and inproper stability testing can result in corrupted programs, and windows operating system.

[@egeozgultekin4331]

Thank you all for your answers. I have one more begginer question. When I do the trials for oc on ram, is it possible to increase only one parameter and keep the rest same ? For example my ram stock xmp stats are 3200 mhz with 14-14-14-34. Can I keep the cas values volt etc and only increase the speed like 3433 or something ? OR all the paramaters should be increase at the same time ? OR I want the latency as small as possible, therefore may be I can bumb the voltage directly 1.4-45 and try to increase mhz with cl14 values as much as possible ? Last one is my aim I can say

[@THISLOVETHISHATE99]

i watch u run far higher voltages then me and im the one that vastly degrades my memory controller and bdie i can no longer run XMP on my R5 3600 if i load up a playlist pf every episode of hogans heros go to sleep il wake up to a frozen pc hell if i load up that playlist launch chrome(5gb used) and then close it, it freezes the pc run hci memtest errors after 140%

[@MultiKokonutz]

What is the lowest voltage you have killed a ddr4 stick with?

[@ActuallyHardcoreOverclocking]

I haven't killed any DDR4 sticks.

[@MultiKokonutz]

@@ActuallyHardcoreOverclocking That is super reassuring to hear!

[@-FOXX]

@@ActuallyHardcoreOverclocking that's shocking.

[@goldnoob6191]

Voltage regulation is crap on this motherboard, clearly missing capacitance.