NEW!!! DRAM Calculator for Ryzen™ 1.6.0.3 (overclocking DRAM on AM4) + MEMbench 0.8 (DRAM bench) - Page 530 - Overclock.net - An Overclocking Community

Forum Jump: 

NEW!!! DRAM Calculator for Ryzen™ 1.6.0.3 (overclocking DRAM on AM4) + MEMbench 0.8 (DRAM bench)

Reply
 
Thread Tools
post #5291 of 5390 (permalink) Old 08-13-2019, 04:16 AM
yzy
New to Overclock.net
 
Join Date: Aug 2019
Posts: 4
Rep: 0
Quote: Originally Posted by cameronmc88 View Post
Good job, what ram kit and also what timings and voltages did you use? GearDown Mode on or off?
G.Skill Trident Z RGB F4-3200C14D-16GTZR
1,475v
Geardown mode enabled
Screenshot timings (did them in BIOS)

Also did a simple memtest86 (4 tests) and it passed.
yzy is offline  
Sponsored Links
Advertisement
 
post #5292 of 5390 (permalink) Old 08-13-2019, 08:39 AM
New to Overclock.net
 
Join Date: Jan 2019
Posts: 4
Rep: 0
Quote: Originally Posted by Spectre73 View Post
Ryzen 1st gen is incredibly picky and quite unstable with DR Ram at high frequencies.

I noticed that your SCL values (at 2) are probably a no go. Increase them first to around 5 and see, how it goes. tWR is also probably far to tight. Try 24.

ProcODT should probably lowered to at least 60, maybe 53.3 ohm. But that worked for my b-die. I have no experience with AFR.
I noticed on auto it would bring my tWR up to 24, but after snooping around i saw a post where 1usmus said its not a good idea to go above 12. However when it comes to 3400 this value is way better (for me)

SCL 2-2 I was told would be better for my bandwidth as im running lower than 3200mhz and it's working fine @3066 14-16-16-17-30-38 1.41v (in the screen shot those timings were tighter than my sig, so maybe this could be part of the issue)

I did notice a lot of B-Die users are able to get away with lower ProcODT compared to most other DR and SR kits running the same'ish mhz. On these I've never managed to get it to boot using anything lower than 60ProcODT. Theres like a weird relationship where it wont boot when im near the 1.4v with ProcODT, so thats why I'm on 68.6... That being said, dropping it on 3200 works better so thanks for the heads up, nearly most settings I've tried was on 68.6 so i kinda gave up with getting it lower, until recently



Quote: Originally Posted by possessed View Post
I have a 2400g apu and some afr. It’s really been a pain because fast means better FPS but it’s been real hard to get that. Last week it seems that I finally managed to get it pretty stable, but still not rock solid since it still crashes in one or 2 3D applications, but at least I’m no longer getting image corruption on the desktop.

So far I’m running them as 2866 with the setting from 3200 on the calculator, might try some more tweaking when I have the time.
Yeh I got these on a flash sale all i saw was 3200mhz 32gb, £110, free shipping. This was when RAM was costly, but it seems it's come down in price now
Hopefully some of these setting will get you closer to stability, you can see where I've managed to get and it helps you out somehow. From my recent experience I've found that setting the SOC closer to what my CPU defaults too is working out better for me, especially in games where the GPU is being accessed a lot. Having it lower like the calc suggests wasn't stable, im guessing the higher bandwidth was causing issues for the IMC and it wasn't keeping up?? Your guess is as good as mine, good luck though


Quote: Originally Posted by plazing View Post
My previous Ryzen 5 1600 with G.Skill F4-3200C16D-32GTZRX (2R-AFR) got to the XMP timing (3200C16-18-18-38) @ 1.37v DRAM & 1.1v SOC and stable 1000% memtest. XMP voltage 1.35v spewed out error few seconds on the memtest.

Now with Ryzen 9 3900X I can get to 3466C16-19-19-40 @ 1.4v DRAM & 1.025v SOC (fast preset 2R-AFR). Not really confortable with voltage higher than that.
Dude, if you didnt suggest those values I'd more than likely still be doing this over the next few days
I thought i worked out a relationship between the SOC and vDIMM/VTTDDR but turns out the SOC was in the wrong place but now that the relationship is much tighter it's great, I was able to find a setting which was netting me much more consistent results. So from that point I then knew that adjusting my timings would be more noticeable.

Thank you all for your suggestions, it's shaved a lot of time off this process..

| RAM: CMK32GX4M2B3200C16 AFR @3066 14-16-16-17-30-38 1.41v | CPU: 1600 4ghz 1.39v SOC 1.075v | MOBO: Asus B350-F | PSU: RM850i
SSDs: 960EVO 500GB OS / 850EVO 250GB OS 2/ 860EVO 1TB Games | HDD: x2 WD Green 3TB Games/Storage |
ddabble is offline  
post #5293 of 5390 (permalink) Old 08-13-2019, 09:15 AM
New to Overclock.net
 
Join Date: Jan 2019
Posts: 4
Rep: 0
B350F 3200mhz DR Hynix AFR

Hopefully if anyones had the same headache as me with this board these settings might get you closer to victory. I'm yet to do a overnight test, but for now playing high FPS games are fine and its not crashing or rebooting w/or without BSOD and the last few tests I've done have been clean.



Little more info that isn't listed
CPU LLC - Med
SOC LLC - Regular

RCDWR 17 was giving errors so i used 18

Ryzen 5 1600 SOC Auto on this board is 1.0v. In these settings its set to 1.1v, dropping below is giving crashes in games where the GPU is being hit hard.

RDWR I noticed when i left it on auto from doing other tests CH_A was always 1 lower than CH_B, AFAIK its a T topology layout and the slots in use are A2 B2 (I know some of you like to leave some of these settings on auto just a heads up)

RFC 480 recommended by the calc was the culprit, that and 448 were not playing nice (this was after i found a voltage and SOC value that were giving me me reproducible results) i raised it by 10 on accident thinking that was "300ns" value from the XMP readout then BAM 4 successful runs in a row. After that i moved onto TM5 and that passed, then i played CS:GO and OW and non of them crashed after a few hours. So all is well... for now

Spread spectrum AUTO? (I don't know, i dont have this option)


Next step is to try apply the logic from these results and my stable 3066 and try to work out what 3400 wants. I suspect theres gonna be a lot of messing about with CLDO_VDDP at some point because I had to move to 945 to get memtest going without BSOD straight away

| RAM: CMK32GX4M2B3200C16 AFR @3066 14-16-16-17-30-38 1.41v | CPU: 1600 4ghz 1.39v SOC 1.075v | MOBO: Asus B350-F | PSU: RM850i
SSDs: 960EVO 500GB OS / 850EVO 250GB OS 2/ 860EVO 1TB Games | HDD: x2 WD Green 3TB Games/Storage |
ddabble is offline  
Sponsored Links
Advertisement
 
post #5294 of 5390 (permalink) Old 08-13-2019, 11:56 AM
Enthusiast
 
Krisztias's Avatar
 
Join Date: Dec 2017
Posts: 113
Rep: 1 (Unique: 1)
https://www.overclock.net/forum/13-a...l#post28085662

Anybody?

Reactor
(12 items)
CPU
Ryzen 7 3800X
Motherboard
ASUS Crosshair VIII Hero Wi-Fi
GPU
ROG-STRIX-RTX2080-O8G-GAMING
RAM
G.SKILL Flare X F4-3200C14D-16GFX
Hard Drive
PNY CS2030 M.2 NVMe SSD
Hard Drive
Samsung 860 EVO 1TB
Power Supply
Corsair RM750X White
Cooling
EK Custom Loop
Case
Phanteks Enthoo Luxe Tempered Glass
Operating System
Windows 10 Pro x64
Monitor
Samsung 49KU6450
Mouse
Logitech MX Master 2S
▲ hide details ▲
Krisztias is offline  
post #5295 of 5390 (permalink) Old 08-13-2019, 12:24 PM
New to Overclock.net
 
SalvorH's Avatar
 
Join Date: Mar 2013
Posts: 2
Rep: 0
I'm a noob to this DDR4 overclocking game but after some reading around I ended up with this result on my Ballistix 3000CL15 Micron E-die: 3533 CL16 1.36V. It passes Memtest86 & the Calculator Membench (500% per thread 6000% total) with zero errors.

Any suggestions or observations are welcome. Wondering if and what I could push more or just leave it alone as is.
Attached Thumbnails
Click image for larger version

Name:	Screenshot-(60).jpg
Views:	52
Size:	222.8 KB
ID:	288766  


My precious
(15 items)
CPU
R5 3600
Motherboard
MSI B450 GAMING PRO CARBON AC
GPU
MSI RX 470 Gaming X 4GB
RAM
Ballistix Sport LT BLS2K8G4D30AESBK 8GB x2
Hard Drive
SAMSUNG SSD 850 EVO
Hard Drive
Western Digital Caviar Green 2TB
Power Supply
CORSAIR TX550M
Cooling
Hyper 212 Black
Case
Phanteks Eclipse P350X
Operating System
Windows 10 pro 64
Monitor
LG IPS231 23"
Keyboard
Corsair K70
Mouse
Logitech G502
Audio
Asus Xonar DGX
Other
TP-Link WN951N Wireless N Adapter
▲ hide details ▲

Last edited by SalvorH; 08-13-2019 at 01:39 PM.
SalvorH is offline  
post #5296 of 5390 (permalink) Old 08-13-2019, 12:59 PM
New to Overclock.net
 
GTxFinish's Avatar
 
Join Date: Dec 2011
Posts: 9
Rep: 0
Hey all, I have a weird phenomena. I'm able to reliably reach 26,000-28000% karhu, but right around there I suddenly get 20-30 errors very quickly. The ram is a stable 52c for a good 12-16 hours straight before these errors start popping up. Im running 3800mhz 1:1 with tight subtimings. I also get random BSODs with WHEA_UNCORRECTABLE_ERROR citing my 1080ti (and I know that's a known issue with nvidia), but I only seem to get those BSODs running above 3600MHz.

I suppose I haven't confirmed that my IF can even run @1900 mhz stable on my 3900X, could that cause this weird behavior? You'd think if I were getting 30-40 errors they'd show up before nearing 30,000%. I'd expect maybe 1-2 errors after having gone that long during ramtesting.

Current vSOC 1.1v and VDDG 1.05v. Should I bump VDDG up to 1.07 (1.1-0.4)?
Attached Thumbnails
Click image for larger version

Name:	3800.jpg
Views:	44
Size:	625.7 KB
ID:	288774  

GTxFinish is offline  
post #5297 of 5390 (permalink) Old 08-13-2019, 01:25 PM
New to Overclock.net
 
umeng2002's Avatar
 
Join Date: Jul 2010
Location: Florida
Posts: 3,107
Rep: 167 (Unique: 105)
That's an extremely long time for Karhu. Maybe it's an issue with the program. That's WELL over 99.5% detection rate.

Run the test multiple times and see if it throws errors at the same point.

CPU
AMD Ryzen 2700X
Motherboard
Asus Prime X470-Pro
GPU
EVGA GeForce RTX 2070 XC Ultra
RAM
TeamGroup T-Force 16 GB (2x8) Pro Dark (B-die TDPGD416G3200HC14ADC01)
Hard Drive
Samsung 840 EVO 250 GB
Power Supply
Seasonic Focus Plus Platinum SSR-750PX
Cooling
Corsair H80i (not V2 or GT)
Monitor
LG 34UC80-B
Keyboard
Logitech G413
Mouse
Logitech G503 RGB
Audio
Creative SoundBlaster Z (OEM)
▲ hide details ▲
umeng2002 is offline  
post #5298 of 5390 (permalink) Old 08-13-2019, 03:48 PM
New to Overclock.net
 
Join Date: Aug 2019
Posts: 6
Rep: 0
Quote: Originally Posted by Krisztias View Post
why are the read and copy speeds so low? (...)The speeds I get seems like 3200C14 read/copy speed not 3600C14... please help Thank you.
Quote: Originally Posted by Krisztias View Post
Anybody?
I'm in the same boat.
3600 with 14-15-14-14-21-35 (tCL-tRCDRD-tRCDWR-tRP-tRAS-tRC) and tight subtimings, like tFAW 16, tRFC 252 - please check below.
Latency is OK (66.4ns), but read bandwidth 49xxx MB/s only. What went wrong?
G.Skill b-die 2x16GB F4-3200C14D-32GTZ, MSI X470 Gaming Pro Carbon, 3700X
Any advice?
Attached Thumbnails
Click image for larger version

Name:	3600 tight low read.png
Views:	32
Size:	258.5 KB
ID:	288808  


Last edited by Piotrucha; 08-13-2019 at 04:40 PM.
Piotrucha is offline  
post #5299 of 5390 (permalink) Old 08-13-2019, 04:49 PM
New to Overclock.net
 
lordzed83's Avatar
 
Join Date: Mar 2013
Location: Nottingham
Posts: 1,559
Rep: 38 (Unique: 30)
Quote: Originally Posted by GTxFinish View Post
Hey all, I have a weird phenomena. I'm able to reliably reach 26,000-28000% karhu, but right around there I suddenly get 20-30 errors very quickly. The ram is a stable 52c for a good 12-16 hours straight before these errors start popping up. Im running 3800mhz 1:1 with tight subtimings. I also get random BSODs with WHEA_UNCORRECTABLE_ERROR citing my 1080ti (and I know that's a known issue with nvidia), but I only seem to get those BSODs running above 3600MHz.

I suppose I haven't confirmed that my IF can even run @1900 mhz stable on my 3900X, could that cause this weird behavior? You'd think if I were getting 30-40 errors they'd show up before nearing 30,000%. I'd expect maybe 1-2 errors after having gone that long during ramtesting.

Current vSOC 1.1v and VDDG 1.05v. Should I bump VDDG up to 1.07 (1.1-0.4)?
Maybe leave HCI memtest overnight ??
lordzed83 is offline  
post #5300 of 5390 (permalink) Old 08-13-2019, 05:17 PM
DRAM Obsessor
 
CJMitsuki's Avatar
 
Join Date: Mar 2018
Posts: 1,016
Rep: 7 (Unique: 6)
Quote: Originally Posted by GTxFinish View Post
Hey all, I have a weird phenomena. I'm able to reliably reach 26,000-28000% karhu, but right around there I suddenly get 20-30 errors very quickly. The ram is a stable 52c for a good 12-16 hours straight before these errors start popping up. Im running 3800mhz 1:1 with tight subtimings. I also get random BSODs with WHEA_UNCORRECTABLE_ERROR citing my 1080ti (and I know that's a known issue with nvidia), but I only seem to get those BSODs running above 3600MHz.

I suppose I haven't confirmed that my IF can even run @1900 mhz stable on my 3900X, could that cause this weird behavior? You'd think if I were getting 30-40 errors they'd show up before nearing 30,000%. I'd expect maybe 1-2 errors after having gone that long during ramtesting.

Current vSOC 1.1v and VDDG 1.05v. Should I bump VDDG up to 1.07 (1.1-0.4)?
Sounds to me like possible corruption of display driver and/or system files. I have a 1080ti and I dont get any WHEA errors. Run DDU firstly and clear out every bit of the old display driver then restart and reinstall the display driver. After that then open a command prompt as administrator and run "SFC /scannow" and see if any error messages are displayed at the end. Then run these commands in this order and reboot (without quotes) "Dism /Online /Cleanup-Image /CheckHealth" "Dism /Online /Cleanup-Image /ScanHealth" "DISM /Online /Cleanup-Image /RestoreHealth" Most of the time those commands will repair corrupted system files that can result from the process of memory overclocking after booting into windows with tons of errors. Windows will be accessing the system files constantly and writing to them. You can get lucky and not get any corrupted files after spending an hour in the OS or you can get corruption immediately as the system writes a bad value to a file. Ive completely screwed an install in no time from this. Sometimes its hard to tell it has even happened for awhile until odd things start to happen.


Quote: Originally Posted by Krisztias View Post
Hi Guys,

what do you think, why are the read and copy speeds so low? I dialed in everything from Calculator, seems stable too (RamTest 3x 3k), I didn't had the time to test it good. The speeds I get seems like 3200C14 read/copy speed not 3600C14... please help
Thank you.
What is the model of your ram sticks for starters. Also, you cant really expect to just dial in the numbers from the calculator and everything be perfect. Memory overclocking doesnt work that way, its dependent on tons of factors such as silicon quality of CPU, IC and PCB quality of the DRAM and Motherboard etc. The calculator is pretty much just a base and to get the optimal configuration is up to you. Memory OC is probably the most complicated part of configuring a computer as its not totally understood and there are no set timings to "plug in". Sure, you may get lucky and get the optimal timings from the calculator on the first try but I guarantee you that is a rarity. So, before knowing what may be causing your low bandwidth according to AIDA64 which is a good program but not for showing you if a change you have made to the timings gave more performance. Aida64 is horrible in that aspect as people will associate bigger numbers on the bandwidth and lower numbers on the latency as performance and its just not the case a great deal of the time. Sure, once you get dialed in correctly AIda64 will show higher numbers in that regard but performance must be tested with other benchmarks such as Geekbench or watching how fast memory testing will complete a certain percentage when nothing but memory configuration has changed. In my opinion, SiSoft Sandra, Geekbench, and Passmark Performance Test are the best. For something quick, Geekbench 3 and 4 are probably going to be the best bet. It will drop your memory score if you tighten timings too much or have errors or just plain, old bad timings. SiSoft is long and pretty taxing on the system but very in depth and Passmark is probably in between. The thing about SiSoft and Passmark Perf test is that they show the effective latency of the setup. Even though Ryzen 2nd gen has similar latency scores in Aida64 to Ryzen+ (2700x, etc) the effective latency is miles apart. 2700x best effective latency i saw was 48ns but on 3700x it is around 27ns with Intel floating around 13-15ns. Either way, using Aida64 to gauge performance is a bad habit to take up. Use Geekbench and youll be surprised at how youll begin to see the difference.

Enough rambling, just post your model number of your DRAM or post a screen of CPUZ showing the SPD tab but make sure you select a slot that is occupied by a stick of ram and tag me and ill see whats going on with your setup.


Quote: Originally Posted by Piotrucha View Post
I'm in the same boat.
3600 with 14-15-14-14-21-35 (tCL-tRCDRD-tRCDWR-tRP-tRAS-tRC) and tight subtimings, like tFAW 16, tRFC 252 - please check below.
Latency is OK (66.4ns), but read bandwidth 49xxx MB/s only. What went wrong?
G.Skill b-die 2x16GB F4-3200C14D-32GTZ, MSI X470 Gaming Pro Carbon, 3700X PBO enabled
Any advice?

For one, you are probably over tightening your timings...there is a point where tightening further will hurt performance but still be stable...back the tRFC off to something like 280 and loosen to 22-36 on RAS and RC...Post your full timings too
This is what I run on the same kit as yours but a 2x8gb setup and at 1.5v DRAM so youll more than likely need to be loosened up much more than what im running on your 2x16gb kit.
Spoiler!


Benchmark scores for reference

CPU- AMD Ryzen [email protected] *
MOTHERBOARD- ROG Crosshair VII Hero
GPU- EVGA GTX 1080ti SC2 w/ XOC [email protected] (Morpheus Cooler)
RAM- G.SKILL TridentZ F4-3200C14D-16GTZR @ 3800mhz (14-15-15-22-36-1T)
COOLING- CORSAIR Hydro Series H115i Pro w/ Noctua NF-A14 (x7)
CASE- Silverstone RL06 Pro (Modified)
MONITORS- VIOTEK GN32C (x2)



Last edited by CJMitsuki; 08-13-2019 at 05:40 PM.
CJMitsuki is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 13 (1 members and 12 guests)
SaccoSVD
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off