The AMD Ryzen All In One Tread /Overclocking/Memory Speeds & Timings/Tweaking/Cooling

vcore

After many evenings of fiddling - I think I've got two stable OC's to use:

I believe I've got a stable 4Ghz and 4.025Ghz -

Manaul "static" voltages used (no offset option currently)

Temperatures range from 75 to 84 - but I believe this is not taking into account the offset.

4000Mhz OC:
vcore: 1.392 under load
NB voltage: ~1.1 or just under
Ram: 3200 1.35v 16-18-18-36

4025Mhz OC:
vcore: 1.416 under load
NB voltage: ~1.1 or just under
Ram: 3200 1.35v 16-18-18-36

Realbench stable on both - higher requires 1.43+ vcore (not tested enough)

CPU-Z: http://imgur.com/a/TTtdb

Speeds are lower than set due to bus speed which I can't do anything about currently :/

Is it worth running the slightly slower OC, or just use the best? - I wouldn't normally think about it but I would usually use offset vcore! - Dont think I'm going to bother trying for higher, as vcore goes up quite alot!

Cheers

 

personally id sit on 4ghz. try lowering vcore a tiny bit & increasing LLC, see if you can get it stable! good solid OC

try nb voltage & increase vpp a tick or two for good measure. if you have the option to modify vref, set it to 1/2 vdimm. id hesitate to pump 1.4v thru my ram tho idk

yup thats why the IC doesnt show up, but since its micron, the point is kind of moot. it all clocks about the same lol

 

I would wait for AGESA one. 1002 is tad better than 902 for sure. But the difference should be bigger with AGESA one.

 

After read through some user feedback in ASRock forum.
Seem like memory compatibility improvement for beta BIOS
with new AGESA is very minimum or totally none.

Anyway it's still too early to tell....

 

Wait. You two?! That also happened to chipsy and I don't understand how the heck is the OS affected by ram tweaking? Surely there is a way to fix the OS, I'd love to help out if that's possible, do you know more about the problem?

 

yes, i do hehe. memory instability can corrupt any & all data, including system files ive never seen it caused by a mobos shaky ram support, but here we are...with a new architecture, new chipset, & teething issues everywhere. cas14 @above stock volts was incredibly conservative for 2400, so im disappointed that i had to increase subvoltages at such a low speed...but im certain its not the ram itself, so im not tooooooo bothered; just patiently waiting for microcode updation from MSI

anyway i appreciate your offer but the issues been rectified! my repair tools were useless, system file checker only partially repaired the corrupt files, & even an external cmd prompt run from a windows install disk wouldnt cooperate (which ive never personally seen happen before in 10yrs of pc repair). i wiped the OS partition & reinstalled clean.

to everyone else, forgot to mention: the pro carbon also has that weird voltage bug, the auto VID seems quite off...sometimes above 1.5v running stock. setting vcore manually fixes it of course.

 

I have to admit one of my friends had the same problem with the OS when AMD released their K2 CPUs with a MSI motherboard. I'm sure it was bios issues but he kicked it to the curb right away and swapped it for a Pentium and never looked back.

Growing pains bound to happen with a new system, same for Intels first architecture products.

Funny in that all the test K2 systems we had never had problems or we didn't notice it since we had AMD to send constant updates back in the days.

I am sure AMD will sort it out. That is the price to be the "first" kid on the block with the shiny Ryzen system.

 

I'll be building my system in the next few days. It feels like I've been waiting ages then I look at the date and realise it's only a couple of weeks! I guess a good aspect of having to wait for the availability of certain parts is that I will be able to start off with a more mature bios for my tweaking enjoyment, bios 2.00 for the Fatal1ty Gaming K4 having just been released, with updated AGESA.

Asrock seem to have been super quick in supporting the new microcode, I hope the trend continues. The 'big' update in terms of memory compatibility is reportedly with AGESA 1.05 due in May.

 

Corrupted files are a typical thing to happen when your overclocks are shaky, especially with bad RAM stability.
Honestly I can't count how often that killed something like a RAID array back in the days...
Today I'd always use MemTest first.

 

Corsair has does some strange things with product names this last year or so.

Their current product, which is listed as H110i (I think it used to be called GT) does indeed work with AM4 brackets out of the box. Installing it on the C6H is super simple.

I think they use to call another product H110, which may not use the same mounting.

 

BIOS UPDATE :banana:

PRIME X370-PRO BIOS 0515
1.Improve system stability.
2.Enhance memory compatibility
3.Improve CPU temperature monitor function

ASUS EZ Update is showing AM4Chipset9.0.000.8_5.12.0.31?
Looks like a different bios 0512?

I'll try EZ Update first then the 0515 BIOS

Will test later

 

My Bad, EZ Update is not just for BIOS. So looks like we have a chipset update and a new bios!

 

CPU temps looks good in this BIOS update. Qfan profile on silent, now means SILENT, even after hours on full load.

Spoiler

Still can't get my RAM above 2933 MHz 18-42 settings.

 

http://www.guru3d.com/news-story/amd-fixes-more-ryzen-issues-with-new-microcode.html

 

So my 1700X and C6H finally play nice. Somewhat.

Currently 3.8ghz @ 1.35V (P-State 0 OC to retain idle downclock) with 3200 CL 14's running stable with the C6H's DOCP (sp?) preset(s).

Typically takes a handful of POST attempts per boot, though.

Anyway, anyone here familiar with P-State overlocking? I'd rather not lose the idle downclock capabilities.

The P-State OC works via HEX values from what I've figured. So running

translates into 3.8 @ 1.35V.

Whenever I push the VCore higher, though (i.e 1C => 1.375 VCore) I'm winding up with a 30x (=> 3ghz) mulitplier in Windows (CPU-Z, HWInfo) even though BIOS shows the expected clocks (i.e 3800/3900 and a multi of 38/39).

Did anyone experience something similiar?

I sort of can work around it by tinkering around with a positive VCore offset (~0.07 for 3.9ghz) leaving me with right under 1.4V Vcore under load. However, VCore will occasionally spike over 1.4V which I rather not have. I'm a fan of flat values insofar I avoid offsets - of course only if it means to retain some downclocking functionalities

Plus ~1.4 VCore sounds a bit highish. Could of course be, that I've lost the lottery.

Another thing I've noticed:

My Devil HDX isn't popping up. Unfortunately I've no other PCI x1 device. Does anyone have this combo working?

 

So true, my skylake build had hiccups, many a bios came after. But that has gone and ryzen is only a day or 2 away ( Waiting for heatsink bracket for my noctua 12s.) I am on tranquilizers as i type, to calm down my excitement :giggle2::giggle2:

 

Yeah it looks better, that 20c offset has been applied.

I've also managed to get all 4 banks of RAM running at 2666Mhz, up from 2133Mhz with the D.O.C.P profile. Any higher, the machine doesnt post. Def getting better though.

 

Hey guys, is there any way of knowing whether a memory is single or dual-rank? Is it indicated by part number?

Apparently SR and Samsung modules are preferable for Ryzen. In case I'd be upgrading soon...

 

Yeh, unfortunately the MSI boards (both the Pro Carbon and the Titainum) don't have "offset" voltages enabled - so I'm stuck with Manual - otherwise I'd push further - may push further next week anyway, impoving my case airflow first!

Using auto LLC is what I've settled on currently as it seems to be a happy medium.

Ryzen definately likes Samsung memory, and the higher the rated speed - the better. I have some 3200 which I tried first, that would refuse to go further than 2666 - but some 3600 rated C18, runs at 3200.

Note that the "A-XMP" profiles are useless, these do not work. I had to manually set the RAM speed and timings to post. I've even managed to run at C16 now

 

https://www.reddit.com/r/Amd/comments/62vp2g/clearing_up_any_samsung_bdie_confusion_eg_on/

( http://www.samsung.com/semiconducto...8/8G_B_DDR4_Samsung_Spec_Rev1.11_Mar.15-2.pdf )