BIOS updates for AM4 motherboards

Well the new 7A34v14 "1.4" Tomahawk bios allowed me to cut voltage, but didn't do jack shit for ram lol. I went from 1.43v to 1.4v, maybe I can hit 4ghz now!


*edit nvm. As soon as I posted I crashed lol.
 
bobzdar said:
What about it? Jedec max is 2133 for ddr4, so you're technically overclocked above that which is why it's an xmp(intel) or docp(amd) profile.
Click to expand...
Top JEDEC speed on DDR4 is 2400, but I have a feeling it might technically be 2667 like Ryzen's max (however for the record, the JEDEC DDR4 Registering Clock Driver document has specs for everything up to 3200, so it may technically now be 3200 *shrug*)

os2wiz said:
Just installed beta bios 1.61 for the MSI Titanium a few hours ago. It allows 4 dimm configurations to run at 2667 mhz, a bump up from the previous 2400mhz. Things are looking good. I expect at least 2933mhz for 4 dimms after the big mid-May bios updates from AMD.
Click to expand...
I had just been running on v1.5 up until tonight when I went back to 1.10 Shipping BIOS as I had gotten tired of the lack of DDR4-3200 stability. I could trick it into booting with a lot of effort, and it'd run just fine, would sleep fine, right up until a reboot... and then I'd go through 20mins of getting it to boot again. (And I'm just on 2x 8GB!)

I might have to try 1.61 though... :) Thanks for the heads up.
 
Formula.350 said:
Top JEDEC speed on DDR4 is 2400, but I have a feeling it might technically be 2667 like Ryzen's max (however for the record, the JEDEC DDR4 Registering Clock Driver document has specs for everything up to 3200, so it may technically now be 3200 *shrug*)


I had just been running on v1.5 up until tonight when I went back to 1.10 Shipping BIOS as I had gotten tired of the lack of DDR4-3200 stability. I could trick it into booting with a lot of effort, and it'd run just fine, would sleep fine, right up until a reboot... and then I'd go through 20mins of getting it to boot again. (And I'm just on 2x 8GB!)

I might have to try 1.61 though... :) Thanks for the heads up.
Click to expand...


I am only on 2 dimms right now and I have zero problems with my ddr4 3200 Flare X dimms. Staying at 1.5 bios until there is something more promising.
 
JustReason said:
some times what you type makes no sense. So I will state this and maybe it will clear up whatever you were trying to convey to me.

The Superposition bench has little to no impact on the CPU so it would not likely show positive or negative results of a bios update.

Cinebench is a much better indicator of CPU performance and likely will show CPU increases as a result of a bios update if it does in fact increase CPU performance.
Click to expand...

What didy ou think I was talking about??? Cinebench 15.038 Open GL scores. With bios 1.50 for MS Titanium. I said zilch about Superposition and know nothing about superposition.
 
Alrighty! I'm liking Titanium's 1.61 Beta :D All the junk of the new BIOS, plus the No-Fuss "Shit Just Works" of the shipping version, for me.
Flashed it, set the couple of things that I need to (boot order, a couple other options unrelated to CPU/MEM), select DDR4-3200, set tCL to 14, set CPU-NB to 0.950V (since after selecting DDR4-3200m it shoots up to 1.028V), set DRAM Voltage to 1.35 (his is at least 1.36V operating, since 'auto' applies 1.376V [which is what setting it to 1.36V will operate at]), and that's it.

Previous version that'd result in no POST. Setting to tCL 16 would still result in a POST loop.
Setting A-XMP Profile 2 would work... SOMEtimes, but with a lot of effort. Also wasn't possible at tCL 14.

I also still am having DMI problems unfortunately, but might have found an AMI tool to fix that. Reason the wonky DMI is an 'issue' is because certain things either can't detect the string (AIDA64 says "Unknown Motherboard), or are expecting something different (MSI Command Center complains about unsupported motherboard). All I had done originally, was changed "X370 XPOWER GAMING TITANIUM (MS-7A31)" to be not screaming at you and removed the board model since it's present in OTHER DMI entries. I also corrected the "Baseboard Version" of 1.0, to read "rev1.1" since the shipping models all have PCB Revision 1.1. Also originally I had blew out the lines that read "To be set by O.E.M." since it was just ugly lol In the process of all of that, and I don't know if it's my fault or not, my Serial Number was also cleared, though I feel like that was the Betas to blame... but would need confirmation from someone else.

Anyways, the tool "Writes" to the DMI, updates it properly in Windows, programs read it correctly now and Cmd Center works as well. Only reason for "might have" is I've yet to restart and determine if it really DID write to the BIOS' DMI, or if it's just some temporary BS. So we'll see what happens after a reboot. In the end I might not bother posting the modded BIOS to flash, and instead just link to the program along with writing a tutorial (since it's stupid-easy), that way no one else catches any DMI issues as a result, since they seem to be persisting through updates, despite having left all the default entries when modding (as well as flashing back to my original dumped BIOS...)

So yea... TL;DR- If you're a Titanium user and having issues running DDR4-3200 on newer BIOS releases, when the older ones worked perfectly... Give Beta 1.61 a try! :) It still uses the same 1.0.0.4a microcode, so nothing 'cool' to be had in that regard, but memory is better (for me at least). :D EDIT: See followup below.
 
Last edited:
I have a CHVI and it worked wonderfully once I put the last bios on there. The current bios I think is the latest which I do not have. I may flash it. Not sure. I have to verify. I have a rock solid OC at 3.85 ghz 1700X and my system absolutely flies. Have 3200 ram running at 3200 round the clock. I just wished the Zen would clock throttle like Speedstep saving tons of electricty and heat.
 
tangoseal said:
I have a CHVI and it worked wonderfully once I put the last bios on there. The current bios I think is the latest which I do not have. I may flash it. Not sure. I have to verify. I have a rock solid OC at 3.85 ghz 1700X and my system absolutely flies. Have 3200 ram running at 3200 round the clock. I just wished the Zen would clock throttle like Speedstep saving tons of electricty and heat.
Click to expand...

As I posted in another thread, you need to be using PStates, and you'll get the throttling.
 
dook43 said:
As I posted in another thread, you need to be using PStates, and you'll get the throttling.
Click to expand...

Oh wow I just found your thread. So if I emulate your settings I should get a good overclock and have P State throttling... this is wonderful.

I think I have a project for tomorrow night :)
 
Followup to my previous MSI Titanium Beta v1.61 BIOS...

While the program AMIDEWIN does indeed 'fix' the DMI issues, that's unfortunately the only good news. 1.61 is still giving me hell when rebooting, as in, what settings run perfectly stable once initially applied... no longer function there-after. I literally can't seem to figure out what the deal is. Only difference NOW is that it makes it further into the diagnostic, hanging on D6 instead of instantly failing. But due to making it so far into the Pre-POST, it ends up needing a manual restart.

I dunno, it's crazy frustrating since after flashing and initially setting DDR4-3200, it works beautifully! Stable, no issues, wakes up from sleep no problem. It's only AFTER that first bootup where shit goes south and nothing works. :\ Doesn't matter if it's a successful shutdown, a restart, a crash... I'm not even sure if a fully cold boot would do the trick...

If there was an easy way (AGESA =/= MicroCode according to MMTool) to slip in the old AGESA, I'd do that in a heartbeat to see if that's the issue, but yea.... I've never worked with MMTool so it's a bit precarious for me to risk.

I guess back to v1.10 again with me... Which I love the fact it just works so well, but don't like the Sleep bug, since it screws with the fan temps. What a frustrating mess :p
 
I would not be surprised if the "mess" you encountered is just the same for the next few months , just hoping that most people can find a stable enough bios that can do a lot of things and with the coming months remaining issues get fixed sooner rather then later.
 
Formula.350 said:
Followup to my previous MSI Titanium Beta v1.61 BIOS...

While the program AMIDEWIN does indeed 'fix' the DMI issues, that's unfortunately the only good news. 1.61 is still giving me hell when rebooting, as in, what settings run perfectly stable once initially applied... no longer function there-after. I literally can't seem to figure out what the deal is. Only difference NOW is that it makes it further into the diagnostic, hanging on D6 instead of instantly failing. But due to making it so far into the Pre-POST, it ends up needing a manual restart.

I dunno, it's crazy frustrating since after flashing and initially setting DDR4-3200, it works beautifully! Stable, no issues, wakes up from sleep no problem. It's only AFTER that first bootup where shit goes south and nothing works. :\ Doesn't matter if it's a successful shutdown, a restart, a crash... I'm not even sure if a fully cold boot would do the trick...

If there was an easy way (AGESA =/= MicroCode according to MMTool) to slip in the old AGESA, I'd do that in a heartbeat to see if that's the issue, but yea.... I've never worked with MMTool so it's a bit precarious for me to risk.

I guess back to v1.10 again with me... Which I love the fact it just works so well, but don't like the Sleep bug, since it screws with the fan temps. What a frustrating mess :p
Click to expand...

On the Crosshair you need to set the dram boot voltage higher to fix that. I am not sure your Titanium has that option or not. Once I set my dram boot voltage 1.4 my system posted fine, when before a cold boot would always screw up and hang in the post process. You would think setting voltage for the dram would mean set this voltage at all times, but I guess not. Hopefully your board has that setting as well.
 
Gideon said:
On the Crosshair you need to set the dram boot voltage higher to fix that. I am not sure your Titanium has that option or not. Once I set my dram boot voltage 1.4 my system posted fine, when before a cold boot would always screw up and hang in the post process. You would think setting voltage for the dram would mean set this voltage at all times, but I guess not. Hopefully your board has that setting as well.
Click to expand...

Do you have a different boot voltage from normal dram voltage ?
 
I'm running 1.5 on the Titanium, everything is perfect (but I'm also not overclocking and only using approved memory, etc.).
 
I have been running 2/24 bios on my Asus B350 prime with 2933mhz cl15 memory -- the only qwerk i really had was on a cold boot it would save overclock failed, then press f1 to boot to windows -- all my overclock and memory settings would stick. Well, i updated to latest bios and it failed to boot 3 times, then started up with 2100 memory speed. i went into bios, turned on D.O.C.P. which set my memory back to 2933 cl15 i raised dram volts to 1.35 per spec of memory. and its running great -- my 720p CS:GO benchmark increased from 334fps to 350fps and im running 150mhz lower clock speed to boot. I do notice the timings of memory are more relaxed but i havent messed with them yet.
 
Been on A-xmp since bios 1.2 on this tomahawk. No issues with pc3200. Just works without doing anything else.

The Trident z pc3200 cl14 is apparently a good kit to have for this motherboard.
 
noko said:
For the ASUS CrossHair VI Hero, bios 1107 is now official:
http://dlcdnet.asus.com/pub/ASUS/mb/SocketAM4/CROSSHAIR-VI-HERO/CROSSHAIR-VI-HERO-ASUS-1107.zip

So far it is the best bios I've run on this board.
Click to expand...

They solved the cold boot problem in this way that it no longer fails to boot but it just puts the memory back to 2133, which is good but it requires another reboot to set it to 2933 then another to 3200. For me still have to find settings which bypass this.
My boot voltage is 1.395 and my normal dram voltage is at 1.365. At this point it would even make sense just to put the default speed before I switch off.
 
Pieter3dnow said:
They solved the cold boot problem in this way that it no longer fails to boot but it just puts the memory back to 2133, which is good but it requires another reboot to set it to 2933 then another to 3200. For me still have to find settings which bypass this.
My boot voltage is 1.395 and my normal dram voltage is at 1.365. At this point it would even make sense just to put the default speed before I switch off.
Click to expand...
What have you done to get your memory and the platform to work better together? What memory do you have?
 
noko said:
What have you done to get your memory and the platform to work better together? What memory do you have?
Click to expand...
Gskill TridentZ F4 3200CL14D 16GTZ

I have been cycling through voltages from 1.35 up to 1.38 normal voltage and boot voltage up to 1.4
It runs well but does not do cold boot.

I'm trying DOCP standard now.on the 1107 bios.
 
Pieter3dnow said:
Gskill TridentZ F4 3200CL14D 16GTZ

I have been cycling through voltages from 1.35 up to 1.38 normal voltage and boot voltage up to 1.4
It runs well but does not do cold boot.

I'm trying DOCP standard now.on the 1107 bios.
Click to expand...
XMP rated memory are rated/built to withstand 1.5v, I recommend you try 1.45v up to 1.5v. My Corsair memory on 2nd Ryzen rig which are also Hynix chips which might be the same as yours would not even boot to 2400 at 1.35v. Taking it up to 1.45v they go all the way up to rated speed of 3200 at 16-18-18-38(36 here is the rating but still working on that) and that is on a cheap Biostar motherboard. ASUS has taken DDR 4 modules up to 1.9v for testing. They are much more durable then most folks think. Hynix dims love voltage.

Recommend you watch this video with AMD engineer talking about memory overclocking, goes through what each voltage does, it is very good overall. AMD guy starts after 3min. ProcODT is also setting that you can fine tune your sticks at which he also covers. Outstanding information here.

 
I can run at 3200 fine it is just cold boot problems , I would even dare to say it is a problem with the bios rather then the voltage used.

It makes little sense that I can run the same voltage and boot after I cleared everything with load optimized defaults.
 
DRAM Boot Voltage is still buggy on Ryzen. Hopefully the next AGESA update will improve it. For example, on "Auto" it's meant to track the DRAM voltage, but it always defaults to 1.2V. Wouldn't surprise me if it isn't reliable when set manually, either. At least you C6H master race overlords have the option. Us Prime peasants have to live without that option (so always 1.2V boot voltage with no way to change it).
 
JimmiG said:
DRAM Boot Voltage is still buggy on Ryzen. Hopefully the next AGESA update will improve it. For example, on "Auto" it's meant to track the DRAM voltage, but it always defaults to 1.2V. Wouldn't surprise me if it isn't reliable when set manually, either. At least you C6H master race overlords have the option. Us Prime peasants have to live without that option (so always 1.2V boot voltage with no way to change it).
Click to expand...

In the end the bios is unstable for certain configurations across the board means that the whole AM4 platform suffers. People with the same ram or different ram get mixed results if only there was more coherent function and that as you said will come with a few more bios upgrades. Even the one in May is prolly not the last that will improve things for DDR4 ram.

The only benefit C6H crowd has is more options , Taichi owners can run dual rank dimm , Before the end of the year I hope all of these problems are solved, sadly to late but you can't do anything about that.
 
Last edited:
Pieter3dnow said:
In the end the bios is unstable for certain configurations across the board means that the whole AM4 platform suffers. People with the same ram of different ram get mixed results if only there was more coherent function and that as you said will come with a few more bios upgrades. Even the one in May is prolly not the last that will improve things for DDR4 ram.

The only benefit C6H crowd has is more options , Taichi owners can run dual rank dimm , Before the end of the year I hope all of these problems are solved, sadly to late but you can't do anything about that.
Click to expand...

I have an MSI x370 Titanium. They released a new bios 1.6 this morning. I was hoping for agesa code 1.06 which has the major memory enhancements in it. But according to hwinfo64 the agesa code is 1.04a which is very old hat. It had no new settings or features. I have been stuck using two 8GB dimms instead of 4 since I can only get to 2667mhz on 4 dimms versus 3200mhz on 2 dimms with my Samsung B die chips. I am hoping the major updates will arrive in the next 7 days as promised by AMD. They had indicated mid May.
 
Last edited:
Just updated to latest bios on msi b350 tomahawk arctic, running Geil Evo X 3200 (Hynix memory) at 3200 with 1.5v. No luck at 1.4v or less. Processor is at 3.8Ghz 1.3v. Womp.

Currently stress testing at 3.9Ghz 1.3v 3200 16-16-16-16 1.45v.
 
Last edited:
Pieter3dnow said:
I can run at 3200 fine it is just cold boot problems , I would even dare to say it is a problem with the bios rather then the voltage used.

It makes little sense that I can run the same voltage and boot after I cleared everything with load optimized defaults.
Click to expand...

I also have the CHVI with a kit of Gskill TridentZ 3600CL17D 16GTZ and it runs at XMP 3200 with no cold boot problems at all with 1107 where with the 1002 I couldn't even POST above 2133 without getting an F1 error. This BIOS dropped my CPU OC from 4075mhz to 4000 but thats OK with me for having much higher RAM speed. One thing I noticed right away was the SOC voltage defaulted to the "normal" AMD required 1.100V instead of 0.750V (and I'm not sure increasing it manually actually changed it on the old BIOS).
 
LigTasm said:
I also have the CHVI with a kit of Gskill TridentZ 3600CL17D 16GTZ and it runs at XMP 3200 with no cold boot problems at all with 1107 where with the 1002 I couldn't even POST above 2133 without getting an F1 error. This BIOS dropped my CPU OC from 4075mhz to 4000 but thats OK with me for having much higher RAM speed. One thing I noticed right away was the SOC voltage defaulted to the "normal" AMD required 1.100V instead of 0.750V (and I'm not sure increasing it manually actually changed it on the old BIOS).
Click to expand...

I think I need to check settings what is done by DOCP and manual , but there are a few values I change which I checked that DOCP standard does not do (boot voltage for one). I have severe problems with how flaky this works , booting fine all afternoon (with power on and off but not removed from the power socket) then in the evening same stuff happening again. failing to post. I have had successful operations at normal 1.35V (boot and normal bios 1002) up to as high as 1.385V normal and 1.4V boot.
 
Pieter3dnow said:
I think I need to check settings what is done by DOCP and manual , but there are a few values I change which I checked that DOCP standard does not do (boot voltage for one). I have severe problems with how flaky this works , booting fine all afternoon (with power on and off but not removed from the power socket) then in the evening same stuff happening again. failing to post. I have had successful operations at normal 1.35V (boot and normal bios 1002) up to as high as 1.385V normal and 1.4V boot.
Click to expand...

Yeah that sucks. Definitely post what you figure out. The memory issues on this platform are really irritating but it does seem to be improving rapidly.
 
LigTasm said:
I also have the CHVI with a kit of Gskill TridentZ 3600CL17D 16GTZ and it runs at XMP 3200 with no cold boot problems at all with 1107 where with the 1002 I couldn't even POST above 2133 without getting an F1 error. This BIOS dropped my CPU OC from 4075mhz to 4000 but thats OK with me for having much higher RAM speed. One thing I noticed right away was the SOC voltage defaulted to the "normal" AMD required 1.100V instead of 0.750V (and I'm not sure increasing it manually actually changed it on the old BIOS).
Click to expand...

Try giving itan extra 50 ohms in the advanced dram settings in bios.
 
os2wiz said:
Try giving itan extra 50 ohms in the advanced dram settings in bios.
Click to expand...

The default value for bios 1107 is 53.3 ohm . you can lower it as well , what I saw from the video Noko posted is that Robert (AMD) is saying that you can move it from 40 to 80 , but I'm not to sure that he has a good idea on what he is talking about (does he mean for every ram kit or just the "high end" ones).
 
Terminating a transmission line properly involves killing the reflection, and may not be a specific impedace that does that perfectly.
Usually, termination is 50 ohms per line end. But across differential pairs, thats often 100. To which situation does this BIOS refer?
"Give it an extra 50" could make perfect sense, or be complete nonsense. I fault the keepers of unnecessary secrets. Just give
us the full specification already, and the community could figure it their own damn selves.

Honestly, I think both ringing and overdamping will simultaneously exist at several different frequencies, no matter the "match".
These aren't necessarily a problem unless they cause a problem. Experiment may be the only way to find optimum mismatch.
That sort of better than half-ass works over a brief interval the device might actually care.

What ever happened to Forced Perfect Termination, if we might remember back to our old scuzzy daze???
 

Attachments

  • 91-037r0.pdf
    1.5 MB · Views: 26
Last edited:
Pieter3dnow said:
The default value for bios 1107 is 53.3 ohm . you can lower it as well , what I saw from the video Noko posted is that Robert (AMD) is saying that you can move it from 40 to 80 , but I'm not to sure that he has a good idea on what he is talking about (does he mean for every ram kit or just the "high end" ones).
Click to expand...
http://www.overclock.net/t/1624603/rog-crosshair-vi-overclocking-thread/14150#post_26080357
elmor said:
Depends on your specific sticks/ICs, CPU, DIMMs per channel, single/dual rank. A few numbers that worked well for me:

Samsung B (SR) 2x8 = 53.3 ohms
Samsung B (DR) 2x16 = 80 ohms
Samsung B (DR) 4x16 = 43.6 ohms

Hynix A (DR) 2x8 = 53.3 ohms
Hynix A (DR) 4x8 = 40 ohms
Click to expand...

Looks like experimentation, trial and error with good testing methods is what is needed. Next AGESA is going to be fun with a hell a lot of memory timings exposed :LOL:. I am just in a weird position where my TridentZ will go up to 3500mhz at 1.35v without issue and a Hynix set does 3200mhz without issue at 1.45v - took all the fun out of it for me.
 
noko said:
http://www.overclock.net/t/1624603/rog-crosshair-vi-overclocking-thread/14150#post_26080357


Looks like experimentation, trial and error with good testing methods is what is needed. Next AGESA is going to be fun with a hell a lot of memory timings exposed :LOL:. I am just in a weird position where my TridentZ will go up to 3500mhz at 1.35v without issue and a Hynix set does 3200mhz without issue at 1.45v - took all the fun out of it for me.
Click to expand...

At this moment in time (1107) I'm crying my eyes out , I just spend a good time (more then one hour) dealing with a bios that is so flaky that even suggests when changing (only change) dram voltage to 1.35 that on the next reboot my overclocking has failed or is stuck at 54 .
I'm a little lost for words on this crosshair VI hero the bios upgrade is a downgrade.
 
You must log in or register to reply here.
Back
Top