AMD RYZEN Builders Thread

Page 194 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Apr 27, 2000
11,446
790
126
Looks like AMD messed up there! Their loss is your gain.
 

StinkyPinky

Diamond Member
Jul 6, 2002
6,369
50
126
That's crazy. Could be quite the collectors item. Does it detect 8 cores in the bios though?
 
Feb 19, 2017
77
19
61
That's crazy. Could be quite the collectors item. Does it detect 8 cores in the bios though?
Yes, it does! Before I gave this systemz away, I've tried CPU in ASUS B350 Prime board for sanity check and it also detected 8 cores and 16 threads. It's funny to read in BIOS System Information tab: AMD Ryzen 5 1600 Six-Core Processor; 8 cores running at 3.7GHz :D
 

spdfreak

Senior member
Mar 6, 2000
733
0
81
Yes, it does! Before I gave this systemz away, I've tried CPU in ASUS B350 Prime board for sanity check and it also detected 8 cores and 16 threads. It's funny to read in BIOS System Information tab: AMD Ryzen 5 1600 Six-Core Processor; 8 cores running at 3.7GHz :D
I wonder how many more are out there like yours... probably at least a batch of them.
 

plopke

Senior member
Jan 26, 2010
214
4
101
:p i just hope some poor folks that ordered a 1700 are not getting a 6 core xD , but damm let the the product number batch hunt begin.
 

msferre

Junior Member
Dec 4, 2017
4
0
1
Hi, all - I hope this is the right place to mention this. I recently built this computer:

GIGABYTE GA-AX370-Gaming 5 (rev. 1.0) AM4 AMD X370 SATA 6Gb/s USB 3.1 HDMI ATX AMD Motherboard
AMD RYZEN 7 1700X 8-Core 3.4 GHz (3.8 GHz Turbo) Socket AM4 95W YD170XBCAEWOF Desktop Processor
CORSAIR Vengeance LPX 16GB (2 x 8GB) 288-Pin DDR4 SDRAM DDR4 3000 (PC4 24000) Memory Kit Model CMK16GX4M2B3000C15

However - with Windows 10 installed, I'm having trouble with a BSOD. The error in question is Driver Power State Failure. Now, it happens in the following circumstance -

I turn on the computer.
When it reaches the login screen on W10, all I have to do is wait one minute or so (regardless of me logging in or not at all) before the above BSOD shows up. Then it restarts, then I have to wait a couple of minutes before the BIOS screen shows up again, then the W10 login screen appears. It's fine after that.

So it's every time I start up the computer after it powered down.

One of my friends thinks it has to do with Ryzen being too new for Windows 10 to adapt to. Is he right? What do you think is happening? Thank you all!
An update - I didn't do anything, and then this week, it seems to have corrected itself. I no longer get restarts. Might have been Windows 10 doing updates on its own. Fingers crossed that this continues!
 

IEC

Super Moderator
Super Moderator
Jun 10, 2004
13,553
357
136
I'd seen people stating that 2x4GB memory configurations on Ryzen were a waste of time because there is no equivalent of Samsung B-die for 4Gb ICs. Well, it appears that isn't entirely the case.

A well-binned set of Samsung E-die (in this case Team T-Force Xtreem 3733 CL17) is running 3600 16-18-18-38 1T on 1.38V (may require less, haven't finished long-term testing) on a MSI B350 Gaming Pro mATX motherboard using AGESA 1.0.0.4a. AIDA64 latency is ~70ns.

My next step will be to test a newer UEFI/AGESA and see if 3733 is possible (I doubt it, 3600 is probably max without BCLK adjustment) - or if I can tighten timings at 3600. So for people on a strict budget or people who don't need 16GB (I plan to turn this rig into a dedicated streaming box) it appears Samsung E-die may be a viable option for 8GB RAM.
 

IRobot23

Senior member
Jul 3, 2017
600
18
76
I agree, I have 2x4GB config with ryzen and my latency is around 69,3ns.

Exactly it is samsung E die.
 

IEC

Super Moderator
Super Moderator
Jun 10, 2004
13,553
357
136
So on the B350 board, it topped out at 3600 - it wouldn't boot 3733. Not bad for a $60 mobo.

As for what is possible with 4x8GB (32GB total) of Samsung B-die (3200 CL14 binned), I installed the F21 UEFI (AGESA 1.0.0.0a for PinnaclePI) on the Gigabyte AX370 Gaming K7 Board. It boots at 3466 and 3400 CL14 with 4 DIMMs but will error out eventually on memory testing. The highest stable clock I've achieved out of the box (i.e. using 3200 CL14 XMP profile, then changing the speed) is 3333 CL14, all 4 DIMMs populated:





This is high-end memory with a high-end X370 board and a reasonable CPU sample (1800X). I plan to compare against X470 and Ryzen 2700X once I purchase those.

Not bad for 1 year out from launch.
 

IRobot23

Senior member
Jul 3, 2017
600
18
76
So on the B350 board, it topped out at 3600 - it wouldn't boot 3733. Not bad for a $60 mobo.

As for what is possible with 4x8GB (32GB total) of Samsung B-die (3200 CL14 binned), I installed the F21 UEFI (AGESA 1.0.0.0a for PinnaclePI) on the Gigabyte AX370 Gaming K7 Board. It boots at 3466 and 3400 CL14 with 4 DIMMs but will error out eventually on memory testing. The highest stable clock I've achieved out of the box (i.e. using 3200 CL14 XMP profile, then changing the speed) is 3333 CL14, all 4 DIMMs populated:





This is high-end memory with a high-end X370 board and a reasonable CPU sample (1800X). I plan to compare against X470 and Ryzen 2700X once I purchase those.

Not bad for 1 year out from launch.
Are you sure that your system is completely stable? Those number are really bad for that speed, I get better results with 3333 CL16
 

IEC

Super Moderator
Super Moderator
Jun 10, 2004
13,553
357
136
Are you sure that your system is completely stable? Those number are really bad for that speed, I get better results with 3333 CL16
With 4 DIMMs populated?

It passed an overnight AIDA stress test on the memory. Maybe I'll try something else for verification.
 

.vodka

Golden Member
Dec 5, 2014
1,039
30
136
With 4 DIMMs populated?

It passed an overnight AIDA stress test on the memory. Maybe I'll try something else for verification.
See if you can get one of The Stilt's 3200-3333-3466MHz profiles stable, as these have proper secondary and tertiary timings. Don't leave these up to the motherboard since resulting performance is lackluster. These are to be used with at least 1.4v, stock 1.35v XMP isn't enough.


If 1T is hard to stabilize, then 1T+Geardown+even CL or 2T+no geardown should make things easier.


These will require more voltage to be stable vs only tight primaries, but the results are more than worth it:



This requires 1.425v on my particular kit.

----------------------------------------

1usmus over at OCN has dual rank sticks (similar load to 4 SR sticks), and he's been getting better and better results as AMD improves their AGESA code.

http://www.overclock.net/forum/11-a...vi-overclocking-thread-3448.html#post26972489

It seems he's stabilized 3333 14-14-14-28 with proper subtimings..... you'd usually only see stuff like this on single rank sticks.
 
Last edited:

B-Riz

Senior member
Feb 15, 2011
934
21
91
Has anyone run CL16 2x16GB 3200 kits at CL14 and been stable? I really don't want to shell out for CL14 if CL16 can be tweaked.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
17,507
931
136
Has anyone run CL16 2x16GB 3200 kits at CL14 and been stable? I really don't want to shell out for CL14 if CL16 can be tweaked.
You want 4 x something. Get the 3600 cl15 4x8 gig (they don't make cl14 at that speed) and run them at 3466. $450 (about)
 

IEC

Super Moderator
Super Moderator
Jun 10, 2004
13,553
357
136

Hans Gruber

Senior member
Dec 23, 2006
480
23
91
Was working on my R3 1200 build. Updated the bios for the new gen Ryzen CPU's coming out. I noticed the machine was a bit lethargic and I think my vcore may be low for 3.85ghz. I was 1.2875 and I bumped it to 1.3v and then 1.31v. It's watercooled by a 240mm AIO. What is the base Vcore for an OC'd Ryzen CPU? My ram runs XMP settings 3200mhz perfectly with all the correct timings since the last bios update before this new bios.
 

Hans Gruber

Senior member
Dec 23, 2006
480
23
91
That is a great chart. Looks pretty accurate up and down. I knocked the R3 1200 down to 3800mhz and 1.285v. I will leave it there for a few days. It's running without the lag now. Most people would not notice a slightly lethargic response to their system. Sometimes computers will eventually hang. The best way to fix them is to run the system at a lower clock rate for a few days.
 

IRobot23

Senior member
Jul 3, 2017
600
18
76
Quick help here. I am overclocking some ryzen builds and some friends are to careful.

Safe voltage
Vcore : 1.4V - LLC2 (max 1.4V, under load ~1.356V)

SOC : 1.2 LLC Auto (max 1.181V, under load ~ 1.175-1.155V)
- with 1.125V some did crash

Safe enough?

- This configuration runs different core clock (depends on lotery), but 3466MHz memory.
 

Hans Gruber

Senior member
Dec 23, 2006
480
23
91
Quick help here. I am overclocking some ryzen builds and some friends are to careful.

Safe voltage
Vcore : 1.4V - LLC2 (max 1.4V, under load ~1.356V)

SOC : 1.2 LLC Auto (max 1.181V, under load ~ 1.175-1.155V)
- with 1.125V some did crash

Safe enough?

- This configuration runs different core clock (depends on lotery), but 3466MHz memory.
They are all safe voltages. When you consider the cost of 1.4v vs. the benefits of increased performance. You can run low voltages on ryzen but over time the machines could become sluggish which is directly attributed to an unstable OC. The benchmarks look good but the system hangs ever so slightly at times. You may have a machine hang while in sleep mode. Requiring a hard reset of the system.
 

IRobot23

Senior member
Jul 3, 2017
600
18
76
Thanks,

yeah I do testing and when i get stable 1.375V for 6-10H still go for 1.4V at same settings. Same goes for SOC 1.05-1.15V then I add 0.05V for maximum stability (mem test for 8 hours).
 


ASK THE COMMUNITY