1800 @ CL14 :shocked:
Printable View
1800 @ CL14 :shocked:
Yah thats normal for ref clock boards :)
Prime 95 custom blend with 13312 allocated out of 16g shows ryzens weakness though with 3200 divider.
http://i1253.photobucket.com/albums/...psw3fvx9nu.jpg
Thats not throttling...thats rounding errors. Drop to 2933...fine..back up to 3200 not fine.
Sure it wont effect benching but im helping 24/7 guys atm.
I think they will still go lower I am not even pushing voltage yet that was like 1.45v on the ram. They are some good damn sticks. It is actually set to 14-14-14, but the bios didn't save correctly yet again.
That's curious I hate p95 (it's an old hatred don't ask) for me 24/7 use is all the effort I am willing to put into it. Honestly the best ram killer I have dealt with is chrome with YouTube streaming. I have had memtest pass and that still have errors.
Maybe normal?
I habe 4x8gb Dual ranked DDR3 in My AM3+ Board
Can run them by ref clock with up to ~970 in Prime Blend Custom
Also Boot into win around 1000
https://abload.de/thumb/968mhz11-11-11-23-56-vwslt.png
But can't boot with the 1866 multiplier
For those of you performing bclk OC, are you using any SATA devices at all? What are you using for storage?
Yes it would be very helpful to find out how the system behaves with different bclk`s, what voltages to raise in order to stabilize etc.
Also , im with chew on the stability front.I ve found out through empirical testing, if it "seems" stable, even with everything youre doing, but isnt stable in prime blend, it will surely crap out on you sooner or later, or for example, write or copy a file with a crc in it, and you are screwed, and dont even know about it :-/ . I run my pc 24/7, sometimes encoding AND playing a game and running a vm at the same time.Stability and most of all ram/IMC stability is very important.Theres also a trap, even tested and confirmed setting with fixed frequency, CAN become unstable when p states are active.
Agreed.
I will give another example.
"My pc is stable for gaming so i don't need to stress test"
Ok so a new game comes out or a new driver.
You install. You think an error when writing to hd will not happen?
http://img.techpowerup.org/170326/170326101259.jpg
http://img.techpowerup.org/170326/170326101308.jpg
http://img.techpowerup.org/170326/170326101313.jpg
http://img.techpowerup.org/170326/170326101318.jpg
http://img.techpowerup.org/170326/170326101324.jpg
These aren't necessarily the exact settings I ran for the runs above, but the voltages are the same as what I have kept for all of the runs. I do not recommend using that much SOC voltage, but my chip seems to eat it up when going for high ram clocks.
Progress with 32G dual rank on a non ref clock board. 3200mhz booted and validated :D
https://valid.x86.fr/cache/banner/1dngd6-5.png
My Friend Post
3733 CL14 on C6H (Bios 1002)
http://upic.me/i/bl/11capture.jpg
Yep once again thats faily easy on a ref clock board.
We are discussing dual rank on non ref clock boards.
Totally different animal.
Plenty of people are doing it if you check the net or bot.
Just saying.
Infrared who has done it tried 4x8 g ( mimic of dual rank ) struggled to hit 3200 and only clocking in windows not at 3200 divider.
I struggled to boot it 3200 @ 3200 divider ( dual rank ) since i can't bump ref clock in windows.
Try doing that and in comparison SR 2x8g 3733 on ref clock board = easy.
Perspective matters.
i'm seeing weird stuff myself. 3200 has been a pain in prime blend but I think I may have it sorted. less is more type scenario for whatever reason. Maybe my chip is dumb.
New bios out for msi boards with new agesa.
I have a habit of going a bit high with voltages. This particular chip and board isn't mine nor do I really care if it died so it got what made it stable.
Voltages is one of those tweaks chew* really makes me look silly. I am a bit rushed and brute force with them he definitely spends more time getting it just right.
Guys, silly question
What do you mean by 'ref clock board' in the context of Zen ?
Cheers
A ref clock board has fine grain ref clock tuning in .00 grain.
A non ref clock board will have depending on bios adjustable bclock in 1.+ resolution.
So non ref clock 100>101>102
Ref clock 99.80>100.00>100.20
Currently the majority of non ref clock are "locked" at 99.80.
Have never seen this before, are these production boards or engineering samples?
If production, have not come across any sort of list that defines which AM4 boards are ref or non ref.
Never saw such feature in a production board in all my time as an overclocker, from epox, abit, dfi, gigabyte, asrock, asus, something in the back of my mind re MSI not sure if its was to do with ref clocks.
Learn something new everyday!
Production and its kinda new to ryzen am4.
Seeing as a .25 multi gives you 25 mhz and that can make or break stability on an oc with these chips ( yes its that fine of a line) 101x40 gets you 4040 mhz.. So even a multiplier is finer grain than Zens ref clock.
So 100.20 is much finer grain control and one point of having a chip.