Ryzen - Return of the Jedi

[chew*]

I already killed a $200 board lol. K7 is dead 46 post code...no clue...

[Formula350]

Give it the ol' "Chuck it in the Oven and Re-Flow" I say. Though I'm sure it's not that heh

Oddly, I've gotten code d2 a couple times when attempting 3466, which to me doesn't make sense (to me *shrug) as it pertains to "PCH Initialization Error"
Usually it's F9 for a second and a half before its power-off-wait3sec-power-on cycle. Thinking I need to download the updated manual, because F9 is "Recovery capsule is not found" (Unless this Recovery Capsule is supposed to be an AMI Module that attempts to correct whatever's problematic to establish stability?)

I tried your suggestion of higher ODT, even tried using all the timings in that ONE time 3466 worked, then tried relaxing all the Sub-Timings by one tick. Then gave 20-20-20-42 a shot, still no dice lol

Not sure if you're willing to comment (as per your previous statement regarding competition and this being a popular thread), or if you perhaps have a good resource to refer me to (like a guide someone has written), but this is such new territory for me that I basically have tons of questions haha Granted, I'm not against trial-and-error, to learn myself, but with how slow systems take to POST now... that method takes roughly 8 to 10 times longer than in the past
-Not sure if I should be trying a higher "Training" voltage or not? Currently it's automatic amount for (I think) speeds >3200, is 1.5V. No clue is 1.55V would make a difference, or if your past advice of "sometime less voltage" would apply to this...
-Or maybe manually set the DRAM VREF to 1/2 of the DRAM Voltage? I've been reluctant to touch that option because on the Titanium, and Ryzen Master, they both trigger there warning colors after only like 20mV (so .600V being default, .620V being red, and I'd need to set .680V for 1.36V, or .700V for 1.4V )
-For all I know, I need to raise tRC (though yours was an XMP profile level of low at 50, in the one SuperPi shot, but your main timings were also very tight), or maybe the tREF. There's a tRCPAGE option as well, with a range of 0~1028T, but it always reads 0 and nothing else seems to be able to report a value for it; thus, wouldn't even know where to start! heh
-Someone else has suggested to me (on [H]|Forums) "Since you are using single row modules, try disabling BankGroupSwap", which is a present option on the Titanium, at least in my modded 1.72 Beta.

I put a request out in the Memory Speed thread on there for BIOS shots from two guys, both I think who have CH6's, but one is running at 3466 and the other at 3600. Am hoping that, maybe, just through comparison I'll be able to glean a bit of info that'll point me in the right direction.
Thankfully, I find this enjoyable (at least when the BIOS is behaving), so will just have to keep my fingers crossed!

[Johan45]

I had something really odd with the CHVI on the weekend. Started benching saturday and things were going well. I have learned that this board is bad for the mem lifting after it gets cold. Second time I have froze it and both times about 30 minutes in I have to re-seat the ram. I was pushing it pretty hard trying to get .03 more pts in CB 11.5 really wanted to beat out Roman for first and while idle it just up and quit. Pushing power gave me zero activity. I figured it must be wet and I stopped for the day tore it down to dry. Sunday I put things back together, made sure it was dry but still getting the same thing. No activity when hitting power. That's when I noticed the USB BIOS flashback light blinking. That was the weird part I didn't have a thumb drive in while benching or when I re-assembled it. But the board was stuck in a loop. Putting a proper USB in the FB slot it would go through the flasback cycle but right when it's supposed to stop it had like a hiccup and the slow "searching" light started again.
I was not a happy camper, this board has no serial #s so RMA isn't possible and I wasn't done benching the 1600x. So after a bunch of messing around I found it would start if I held the power button down while switching on the PSU. Got into BIOS and flashed it to one of the new versions. Reboot and nothing stuck no restart just a blinking FB light. Then I used the flasbak to revert to ver 902 it did work. I restarted using the same trick and verified BIOS version. Set defaults, F10 and no reboot. I needed to do a few errands so I pulled all the power and CMOS battery and let it sit. When I returned it was fully functional and the FB light had gone out. Spent the next few hours benching LN2 and not one hiccup after that.
I'm just glad it sorted out

[os2wiz]

Give it the ol' "Chuck it in the Oven and Re-Flow" I say. Though I'm sure it's not that heh

Oddly, I've gotten code d2 a couple times when attempting 3466, which to me doesn't make sense (to me *shrug) as it pertains to "PCH Initialization Error"
Usually it's F9 for a second and a half before its power-off-wait3sec-power-on cycle. Thinking I need to download the updated manual, because F9 is "Recovery capsule is not found" (Unless this Recovery Capsule is supposed to be an AMI Module that attempts to correct whatever's problematic to establish stability?)

I tried your suggestion of higher ODT, even tried using all the timings in that ONE time 3466 worked, then tried relaxing all the Sub-Timings by one tick. Then gave 20-20-20-42 a shot, still no dice lol

Not sure if you're willing to comment (as per your previous statement regarding competition and this being a popular thread), or if you perhaps have a good resource to refer me to (like a guide someone has written), but this is such new territory for me that I basically have tons of questions haha Granted, I'm not against trial-and-error, to learn myself, but with how slow systems take to POST now... that method takes roughly 8 to 10 times longer than in the past
-Not sure if I should be trying a higher "Training" voltage or not? Currently it's automatic amount for (I think) speeds >3200, is 1.5V. No clue is 1.55V would make a difference, or if your past advice of "sometime less voltage" would apply to this...
-Or maybe manually set the DRAM VREF to 1/2 of the DRAM Voltage? I've been reluctant to touch that option because on the Titanium, and Ryzen Master, they both trigger there warning colors after only like 20mV (so .600V being default, .620V being red, and I'd need to set .680V for 1.36V, or .700V for 1.4V )
-For all I know, I need to raise tRC (though yours was an XMP profile level of low at 50, in the one SuperPi shot, but your main timings were also very tight), or maybe the tREF. There's a tRCPAGE option as well, with a range of 0~1028T, but it always reads 0 and nothing else seems to be able to report a value for it; thus, wouldn't even know where to start! heh
-Someone else has suggested to me (on [H]|Forums) "Since you are using single row modules, try disabling BankGroupSwap", which is a present option on the Titanium, at least in my modded 1.72 Beta.

I put a request out in the Memory Speed thread on there for BIOS shots from two guys, both I think who have CH6's, but one is running at 3466 and the other at 3600. Am hoping that, maybe, just through comparison I'll be able to glean a bit of info that'll point me in the right direction.
Thankfully, I find this enjoyable (at least when the BIOS is behaving), so will just have to keep my fingers crossed!

Disabling BGS helped me get above 3200mhz on my Titanium. Not much of a blip but I got to 3333mhz stable at 14-14-14-14-35 CR T2 with SOC 1.075 and dram voltage 1.39 volts

[chew*]

Me personally and i may video it have found that ODT is huge and very very chip dependant. Good chips can post @ 40 ohm over 3200 ram...bad chips need 48 ohm to post @ 3200.. Normal chips need 43 ohm to post @ 3200.

I can actually swap chips on video live and pop one chip after another showing chips post/not post at a fixed odt setting.

My 1800x for example posts @ 3200 like a normal chip should...but to exceed 3200 require 48 ohm for 3333,3466 and needs to step up to 53 ohm for 3600-3733.

This is all on taichi...but i doubt boards will vary much.

Basically if you can boot @ 40 ohm over 3200 you have a damn good imc...

Msi is still or was still on 1004. Fairly certain without a unicorn imc...3466 and up...not happening at least not stable.

[Formula350]

Well speaking of Titanium... new 1.73 Beta went up. I grabbed it. Paged through...
New setting in CPU Features called "Relaxed EDC throttling".
Help string reads:

Disable - Part-specific EDC throttling protection enabled.
Enable - Reduce the amount of time the processor will throttle.
Auto - AMD's recommendation (Disable)​

(It's actually one long line, so I'm interpreting the line breaks)

New RAM Timing options...
MISC ITEM

BankGroupSwapAlt (this is in addition to the other BankGroupSwap option; no help string)​

CAD BUS TIMING CONFIGURATION:

AddCmdSetup - Setup time on CAD bus signals Range 0~63
CsOdtSetup - (same help string)
CkeSetup - (same help string)​

CAD BUS DRIVE STRENGTH CONFIG:

ClkDrvStrength - 120Ohm/60/40/30/24/20 - (no help string)
AddrCmdDrvStren- 120Ohm/60/40/30/24/20 - (no help string)
CsOdtDrvStren- 120Ohm/60/40/30/24/20 - (no help string)
CkeDrvStren- 120Ohm/60/40/30/24/20 - (no help string)​

I made a very special chew* version with the "SB Spread Spectrum" and "Chipset Power Saving Features" options set to Disabled (There seems to be an HPET for the SB as well, but I didn't touch that)
̶W̶i̶l̶l̶ ̶e̶d̶i̶t̶ ̶w̶i̶t̶h̶ ̶D̶L̶ ̶l̶i̶n̶k̶s̶ ̶f̶o̶r̶ ̶u̶n̶h̶i̶d̶d̶e̶n̶-̶o̶p̶t̶i̶o̶n̶s̶ ̶m̶o̶d̶d̶e̶d̶ ̶(̶l̶i̶k̶e̶ ̶b̶e̶f̶o̶r̶e̶)̶ ̶a̶n̶d̶ ̶a̶l̶s̶o̶ ̶t̶h̶e̶ ̶"̶c̶h̶e̶w̶*̶ ̶e̶d̶i̶t̶i̶o̶n̶"̶ ̶m̶o̶d̶d̶e̶d̶.̶
"Standard" Modded v1.73 Beta (TITANIUM ONLY!)
"Chew* Edition" Modded v1.73 Beta (Same, but with the above two additional changes)
[Same disclaimer as previous post. I hold no liability for anything that happens to your motherboard after using this, yadda yadda.]
(I do use my own modded BIOSes, so you can take solace in that, at least)

[chew*]

Thanks man. Are they on 1006 yet or still fooling around on 1004?

Heres your help string on drive strength. Weaker usually gives better compatibility.

Based on higher odt ohms offering compatibility weaker could be "higher" but that is a better question for the stilt as i am not sure . previous gen chips bios just said weaker, normal, medium, stronger etc etc...

Bank group swap sounds like a "compatibility" option...i would compare bandwidth before/after....compatibility is just in simple terms...we turned your performance down so it "works"

So you gain "speed" but probably perform identical to your slower speed...dirty tricks is a term i like to use...but hey its a bigger number so it must be faster lol.

I am pretty anal when it comes to performance...tony from ocz once said use whats fast not what you think is fast...if tight as a virgin but lower speeed beats omg over 9000 then screw over 9000.

[Johan45]

Thanks man. Are they on 1006 yet or still fooling around on 1004?

Heres your help string on drive strength. Weaker usually gives better compatibility.

Based on higher odt ohms offering compatibility weaker could be "higher" but that is a better question for the stilt as i am not sure . previous gen chips bios just said weaker, normal, medium, stronger etc etc...

Bank group swap sounds like a "compatibility" option...i would compare bandwidth before/after....compatibility is just in simple terms...we turned your performance down so it "works"

So you gain "speed" but probably perform identical to your slower speed...dirty tricks is a term i like to use...but hey its a bigger number so it must be faster lol.

I am pretty anal when it comes to performance...tony from ocz once said use whats fast not what you think is fast...if tight as a virgin but lower speeed beats omg over 9000 then screw over 9000.

According to the Stilt the bank group swap "should" be disabled for single side 1DPC you will see a drop of ~ 2k in bandwidth but ~ 4% decrease in latency. I tried it and he was spot o. Any SW that benefits from latency will show slight improvement.

[Formula350]

Thanks for that info, Chew! I'm always happy to have more RAM knowledge! Sucks, especially, since AMD had fallen so far out of favor that no one wants to write up any tutorials or tips articles. As such, and I mean no offense since I do understand your reasoning, those that do know are really tight lipped about what settings to run or where to start when getting things working, as it means giving the competition a leg up. It's a shame that a fun hobby with free-flowing info has turned into this industry, where competitions are won and lost by sharing the knowledge Granted, it's meant the delivery of systems capable of SO much more than once upon a time, where we used to have to rely on awesome user-made programs like WPCREDIT or SetFSB. While that's all in the BIOS now, we're still lacking the knowledge to utilize it all. At least in plain english lol Datasheets are obtainable at least, provided you can make sense of them :P

Anyways, "ranting" aside, I had no luck with BGS set to Disabled when it came to getting 3446 running. Went so far as to try the timings I got that ONE time, and increased them. Tried ODT's up to 60Ohm, but since MSI doesn't provide what things are at while running on Auto, I have no clue what some of those settings were during that magic moment lol After Thought: Would changing the CPU NB's LLC or Switching Frequency, or the DRAM Switching Frequency make any difference, in the DigitALL Pwoer menu?

Alas, I think I'll try to get my 3200 tighter instead. However, I tried that yesterday, and ended up hitting some weird issue. Posting, it'd pause at 99 for quite awhile, during USB I assume since keyboard would flash, but my mouse connected via keyboard wouldn't, untill it picked up after maybe 10 seconds (which is when code 64 displays). IF I let it boot, Windows would take eons to load. The one time it was stuck at the Login Screen... more like the screen you get with C+A+D but minus the options in the middle... but wouldn't do anything further. The Power options worked at least :P So I dunno what that was all about. Thought it was my timings, but even resetting defaults didn't resolve it, so I shut down. It's fine today though *shrug* Back to the grind I guess heh

Have you, when using that K17TK program, noticed any performance quirks? The FIRST time I used it, things worked fine, performance increased like expected, but now it seems to cause a DEcrease in performance after changing just the P0 multi a quarter higher than default. I almost feel like it, somehow, has something to do with me having messed around with CPB and C6 options, despite seeing no reason why this should persist beyond a restart. I'm not utilizing the Save Profile feature, but I'm wondering if it's wrote something in the registry... I unno.

As far as Microcode, I don't know. BIOS reads 8001126, which I don't know how to decipher the Patch ID into a verison number heh I'll have to check with something next time I'm in Windows.

[os2wiz]

Me personally and i may video it have found that ODT is huge and very very chip dependant. Good chips can post @ 40 ohm over 3200 ram...bad chips need 48 ohm to post @ 3200.. Normal chips need 43 ohm to post @ 3200.

I can actually swap chips on video live and pop one chip after another showing chips post/not post at a fixed odt setting.

My 1800x for example posts @ 3200 like a normal chip should...but to exceed 3200 require 48 ohm for 3333,3466 and needs to step up to 53 ohm for 3600-3733.

This is all on taichi...but i doubt boards will vary much.

Basically if you can boot @ 40 ohm over 3200 you have a damn good imc...

Msi is still or was still on 1004. Fairly certain without a unicorn imc...3466 and up...not happening at least not stable.

Chew, have you seen bios 1.73? It has an additional 8 or 9 memory settings. As you said I can not get over 3333mhz. On 1.73 bios I got there without disabling BGS. Of course that was on 2 dimms. I have yet to figure a way to get above 2667 on my 4 dimms of Flares X DDR4 3200. I think it is time for me two put the second 16GB dimm kit on sale on Ebay. At least I'll have a down payment on my Vega card that I'll buy in 2 months.

[chew*]

How much you want for them?

Btw i just did some SPD profile snooping...I need to do some testing to verify but....they look like they were flashed with a very "loose" profile.

I just compared vs Trident -z 3200,3600 bins vets flare X....plug and play load xmp...3200 tridents should be fastest if you set 3200 14-14-14-34 and just go.

[Formula350]

Chew, have you seen bios 1.73? It has an additional 8 or 9 memory settings. As you said I can not get over 3333mhz. On 1.73 bios I got there without disabling BGS. Of course that was on 2 dimms. I have yet to figure a way to get above 2667 on my 4 dimms of Flares X DDR4 3200. I think it is time for me two put the second 16GB dimm kit on sale on Ebay. At least I'll have a down payment on my Vega card that I'll buy in 2 months.

Hey bud, fancy seeing you here! LOL
Yea I posted and told him about all the new mem settings on the previous page and linked him one of my modded BIOSes with the SB SpreadSpectrum and PowerSaving disabled :P

I hate to see folks struggling with RAM, since at this point (in the industry, not Ryzen's lifecycle) you'd expect things to be working far better. However, as people have pointed out, even Intel had teething issues with their first DDR4 :\ I'm on the fence about what I'd do in your shoes. Part of me thinks that yea, might be better to offload the other kit.... But, part of me thinks that kit prices may only go up due to demand once more Ryzens get into consumer hands, nevermind when Threadripper lands on the market. As such, perhaps it'd be better to hold onto them? I dunno <_>

Btw i just did some SPD profile snooping...I need to do some testing to verify but....they look like they were flashed with a very "loose" profile.

I just compared vs Trident -z 3200,3600 bins vets flare X....plug and play load xmp...3200 tridents should be fastest if you set 3200 14-14-14-34 and just go.

Here's something I've been wondering about SPD and memory on these systems, and had actually asked os2wiz about on [H] forum heh
If you take a TridentZ CL14 B-Die kit, and a FlareX CL14 B-Die kit, ran them both on the same system with all timings set to Auto, not running any profiles at all, would their timings differ at all?

I ask because when I set 3200 manually all the subtimings etc, are the exact same as what they are if I apply A-XMP. Even the tRC, which on my CL15 TridentZ kit's XMP SPD it's tRC 50 and tFAW 39, but Auto still shows the usual of 75 and 34 (respectively). So is the BIOS actually calculating timings based on timing tabled in the DIMM, or is it all pre-programmed?
If it'd help, I can upload another BIOS with the DRAM Patch ID option unhidden, but I just figured it wasn't of much use and leave it off.

Also I feel like L3 speed in AIDA is an indicator of RAM instability. No background task is running but Read speed is wwwwway low at 257gb/s...
EDIT: and maybe Latency? The usual for me lately has been 71ns, but this run it's been consistently 75ns +-0.3, which I'm running 3200 at 14-14-14, timings I've already wondered the stability at.

[os2wiz]

Thanks for that info, Chew! I'm always happy to have more RAM knowledge! Sucks, especially, since AMD had fallen so far out of favor that no one wants to write up any tutorials or tips articles. As such, and I mean no offense since I do understand your reasoning, those that do know are really tight lipped about what settings to run or where to start when getting things working, as it means giving the competition a leg up. It's a shame that a fun hobby with free-flowing info has turned into this industry, where competitions are won and lost by sharing the knowledge Granted, it's meant the delivery of systems capable of SO much more than once upon a time, where we used to have to rely on awesome user-made programs like WPCREDIT or SetFSB. While that's all in the BIOS now, we're still lacking the knowledge to utilize it all. At least in plain english lol Datasheets are obtainable at least, provided you can make sense of them :P

Anyways, "ranting" aside, I had no luck with BGS set to Disabled when it came to getting 3446 running. Went so far as to try the timings I got that ONE time, and increased them. Tried ODT's up to 60Ohm, but since MSI doesn't provide what things are at while running on Auto, I have no clue what some of those settings were during that magic moment lol After Thought: Would changing the CPU NB's LLC or Switching Frequency, or the DRAM Switching Frequency make any difference, in the DigitALL Pwoer menu?

Alas, I think I'll try to get my 3200 tighter instead. However, I tried that yesterday, and ended up hitting some weird issue. Posting, it'd pause at 99 for quite awhile, during USB I assume since keyboard would flash, but my mouse connected via keyboard wouldn't, untill it picked up after maybe 10 seconds (which is when code 64 displays). IF I let it boot, Windows would take eons to load. The one time it was stuck at the Login Screen... more like the screen you get with C+A+D but minus the options in the middle... but wouldn't do anything further. The Power options worked at least :P So I dunno what that was all about. Thought it was my timings, but even resetting defaults didn't resolve it, so I shut down. It's fine today though *shrug* Back to the grind I guess heh

Have you, when using that K17TK program, noticed any performance quirks? The FIRST time I used it, things worked fine, performance increased like expected, but now it seems to cause a DEcrease in performance after changing just the P0 multi a quarter higher than default. I almost feel like it, somehow, has something to do with me having messed around with CPB and C6 options, despite seeing no reason why this should persist beyond a restart. I'm not utilizing the Save Profile feature, but I'm wondering if it's wrote something in the registry... I unno.

As far as Microcode, I don't know. BIOS reads 8001126, which I don't know how to decipher the Patch ID into a verison number heh I'll have to check with something next time I'm in Windows.

Anyone that f-d up to not share info when so many are struggling is prima facia a-hole

[chew*]

The info i do not share will benefit you none at all 99% of the time for 24/7 use anyway

Unless of course you want to run 1.8v+ to your memory while sub 0 which if your doing that you are competing which btw cost me a set of ram last week.

Should i teach you how to kill ram like me? Probably not

[Lightman]

Hey guys!
I will be joining Ryzen club with my new R7 1700 on Asus Prime X370 and some RipJaws 3200 CL14-14-14 2x8GB sets. Hope to get system up and running over this weekend. Any initial suggestions as to RAM settings? Should I start from XMP auto and then make my way up?
My target is to either run them above 3333Mhz or stick to 3200 and go low on CL. BTW can't wait to bench my first 8 core build

[chew*]

Have not used pro on latest bios but previous bios spec was hard enough.

[ZenEffect]

The info i do not share will benefit you none at all 99% of the time for 24/7 use anyway

Unless of course you want to run 1.8v+ to your memory while sub 0 which if your doing that you are competing which btw cost me a set of ram last week.

Should i teach you how to kill ram like me? Probably not

Yes, I've been torturing my ram since day 0 and no degredation. Show me how to kill a trident lol

You know I won't mind

[Lightman]

Have not used pro on latest bios but previous bios spec was hard enough.

Well, at least there now is unofficial BETA BIOS with AGESA 1006 support bringing more RAM options to this board. I had to choose between Gigabyte Gaming 5 PWM or Prime's weaker BIOS support and concluded that I have better chance of ASUS slowly fixing BIOS than PWM circuit miraculously upgrading itself
This is a stopgap setup to play with and enjoy without too much of a serious overclocking, before I can jump to Threadripper late this year.
Tomorrow starts the fun, so don't be surprised if I come up with some stupid question (to any helpful XS member here!)

[chew*]

Np ask away. Yep pwm sinks can not be fixxed with a bios. The asus run a tad warm to full load but not as warm but bios should be able to fix it...did wonders for my b350 and that is not even on 1006 yet.

[Formula350]

The info i do not share will benefit you none at all 99% of the time for 24/7 use anyway

Unless of course you want to run 1.8v+ to your memory while sub 0 which if your doing that you are competing which btw cost me a set of ram last week.

Should i teach you how to kill ram like me? Probably not

While that is definitely understandable, since you're running your phase change setup, the issue I have with it is I learn better by examples and pictures. My brain is just wired to better figure out how to accomplish something by seeing the end result and working backwards from there with the explanation of why it is that way or works like that. In a sense I guess my brain "decompiles" in order to understand, rather than compiling the answer from the text.

As an example... You've explained to me what quite a few settings do and how they help. The ProcODT being one of them, but for the longest time I wasn't really able to make any use of that because I had no idea where to start, so it wasn't until you offered the baseline of "for 3200 use 43.3Ohm, and for 3333 use 48Ohm" that I was able to make some sense of it all. Same goes for those Drive Strengths. While, granted, you also said you're not entirely sure yourself how they'll function on this platform, your understanding of it as a concept is way more understood due to your experience. You've relayed that to me, but it's a generalization, which while I find helpful, I don't really know where to start with my tinkering heh

While typing that I realized it basically goes back to what I've said earlier (a post from the last couple days), that it's tough for me to get started because on my Titanium it doesn't tell me what Auto is applying. I have no baseline to work from, and that's what the examples from you experienced users are providing me with. That way I'll at least have a data point to base everything on, so I know that I'm not ultimately working against myself if I set one drive strength and things work, then set another to various other points which all fail... Have I actually set the first one wrong (non-ideal) or are the other ones not working because of some other setting that isn't being set as it should due to whatever Auto is applying?

That's also where even other user's auto-settings at higher speeds help me out a bit. Sure, I might not be able to run at their speeds with their timings, but now I have a direction and more importantly a distance. If (arbitrary, unrealistic example) at 3333 my tRDRD is 10 and at 3466 someone's is 14, now I know that it's going to likely take more than just setting mine to 11 in order for there to be a chance.

Lastly, but equally important, the fact that I have no clue what settings carry the greatest impact on stability. Admittedly, I know it sounds like hand holding, but gone are the days of quickly being able to tell if a setting is working or not. This all went way way wwwaaayyyy faster back when we had A64Info with the ability to adjust all the timings IMC's from within Windows :P Set an overclock, if you get into Windows, run a test. Test fails? Adjust a timing, test again. Pass? Try tightening something else, repeat tests. Etc etc. Even ignoring that, before, we'd press the power button and instantly be at the POST screen. Now we have to wait 8 seconds, or 20 should a setting fail multiple times and enter failsafe mode. On mine, the first restart after saving a change might post an F9, but the second attempt passes. So even tweaking via BIOS has became slow and arduous... but painfully so when you don't know what you're doing, like some folk! *raises hand* lol I love doing it, I accept the risks, but not having a clue is a buzzkill

[chew*]

Well heres a tip..

Xmp is intel specific and "unsupported" by AMD officially.

Jedec is industry standard and supported by all and 2666 is the highest officially supported speed by amd.

Drop to 2666 @ auto on all but primary timings....look at the subs 2666 strap sets...start from there as a guide line.

As far as the bclk question previous guy asked...

There was a rookie "rumble" comp at hwbot. One part of it was highest ref clock...money prizes etc...if i narrowed it down to which board and exact settings to use...it might have well been me doing the result.

If i did tell that one guy and he got the leg up on everyone with my info....think he would share info with competitors with cash or prizes on the line?

Anyway that said outside interference has compromised comps in past....be it legal or illegal. I will not partake in influencing outcomes.

If i influence an outcome in a comp it will be with my own result.

[os2wiz]

While that is definitely understandable, since you're running your phase change setup, the issue I have with it is I learn better by examples and pictures. My brain is just wired to better figure out how to accomplish something by seeing the end result and working backwards from there with the explanation of why it is that way or works like that. In a sense I guess my brain "decompiles" in order to understand, rather than compiling the answer from the text.

As an example... You've explained to me what quite a few settings do and how they help. The ProcODT being one of them, but for the longest time I wasn't really able to make any use of that because I had no idea where to start, so it wasn't until you offered the baseline of "for 3200 use 43.3Ohm, and for 3333 use 48Ohm" that I was able to make some sense of it all. Same goes for those Drive Strengths. While, granted, you also said you're not entirely sure yourself how they'll function on this platform, your understanding of it as a concept is way more understood due to your experience. You've relayed that to me, but it's a generalization, which while I find helpful, I don't really know where to start with my tinkering heh

While typing that I realized it basically goes back to what I've said earlier (a post from the last couple days), that it's tough for me to get started because on my Titanium it doesn't tell me what Auto is applying. I have no baseline to work from, and that's what the examples from you experienced users are providing me with. That way I'll at least have a data point to base everything on, so I know that I'm not ultimately working against myself if I set one drive strength and things work, then set another to various other points which all fail... Have I actually set the first one wrong (non-ideal) or are the other ones not working because of some other setting that isn't being set as it should due to whatever Auto is applying?

That's also where even other user's auto-settings at higher speeds help me out a bit. Sure, I might not be able to run at their speeds with their timings, but now I have a direction and more importantly a distance. If (arbitrary, unrealistic example) at 3333 my tRDRD is 10 and at 3466 someone's is 14, now I know that it's going to likely take more than just setting mine to 11 in order for there to be a chance.

Lastly, but equally important, the fact that I have no clue what settings carry the greatest impact on stability. Admittedly, I know it sounds like hand holding, but gone are the days of quickly being able to tell if a setting is working or not. This all went way way wwwaaayyyy faster back when we had A64Info with the ability to adjust all the timings IMC's from within Windows :P Set an overclock, if you get into Windows, run a test. Test fails? Adjust a timing, test again. Pass? Try tightening something else, repeat tests. Etc etc. Even ignoring that, before, we'd press the power button and instantly be at the POST screen. Now we have to wait 8 seconds, or 20 should a setting fail multiple times and enter failsafe mode. On mine, the first restart after saving a change might post an F9, but the second attempt passes. So even tweaking via BIOS has became slow and arduous... but painfully so when you don't know what you're doing, like some folk! *raises hand* lol I love doing it, I accept the risks, but not having a clue is a buzzkill

I have some good news to report on 4dimm 32GB memory overclock on my FlareX DDR4 3200 memory on Titanium bios 1.73. With CL14-14-14-48 at 2800mhz with SOC 1.15v Proc 60ohm, dram voltage 1.40 v. at 3.875GHZ at 1.40 v cpu voltage. I have booted successfully and run several benchmarks. I will run stability test later and I plan on saving the oc profile. I will attempt to refine voltages later. Here are attached benchmarks.

I have an AIDA64 benchmark on this memory overclock but it is 235kb and the attachment loader indicates there is a 200kb maximum.

[os2wiz]

That was a general statement about some people who are so competitive that they have forgotten their humanity. Not applicable to you. You go out of your way to help.

[os2wiz]

New progress on 2 dimm overclocking. 3466mhz achieved at 3.875 GHZ . I am uploading validations. I have screen shots of the bios but they a re too large a file to upload the image. I just added AIDA64 memory report only.

Bios settings core voltage 1.3625 LLC profie 1 actual 1.408 v. , SOC voltage 1.152 actual CLDO_VDDP .99 v , dram voltage 1.44 v actual
memory timings 2T CR CL 16-16-16-16 42 tRC 72 BGS disabled ProcODT 63.3 OHMS. BIOS 1.73 beta MSI Titanium


Adding AIDA64 to the mix. Only memory benchmark ATTACH]132346[/ATTACH]

Aster several hours of use I had to abandon 3466mhz. I had 2 incidents where the system blanked out on me. My temps are great that is not an issue. Since I had soc voltage at 1.175 and dram voltage at 1.43 I did not feel comfortable jacking them up higher for stability. I see now why that bclock generator comes in handy. So I am back down to 3333mhz where I am completely stable. I played a little bit with my timings there. II noticed at default it sets trow cycle at 76 with cr2. I felt it was a bit too high and it did affect frame rates. I i dropped it to 64 cycles and it booted fne but performance in FPS was not what I expected. I played until I set it at 67 and my FPS jumped over 3 frames per second on cinebench 15, so I left it at 67. Ny CL is 14-14-14-36 CR2 and t row cycle 67.

[Lightman]

Quick update from me - Asus Prime X370 with BIOS 0801 booted without a problem with DOCP profile for 3200MHz CL14-14-14 RAM. So far in my limited time I've been able to get my R7 1700 up to 3.8GHz with 1.3Vcore on boxed cooler stable. Don't want to push much more as I need to wait for my AM4 mounting bracket to be able to fit my AIO
Regardless, I'm pleased with how easy it was to get the baseline up where I wanted, once decent cooler is in place, I can start my journey to 24/7 stable settings. I target realistic 4GHz all cores and some fun runs above that

BTW Ryzen feels really fast, especially in productivity applications!