Ryzen - Return of the Jedi

[Formula350]

You can set PStates on that [+]Advanced menu, and then flash it again, it is not the way you wanted, but it works. I have set PSate to 1, and now Core Boost goes to 3750MHz. Further analyses needed.

Single Core boost goes to 3750? I assume you're running a 1700 and not a 1700X?
Because my 1700X boosts to 3900 just fine currently. Even still, I wouldn't quite know how to even configure it in AMIBCP. Some of these things confuse me since AMIBCP does not list them the same way as they show up in the BIOS.



That's how it is by default, only thing I changed was made them all "USER" (which didn't matter since I can't access that menu anyways).
I assume that there being 8 "PPC Adjustment" is one for each Core, but for all I know it might mean Max PState to run when that many cores are loaded (which I know it doesn't).

I wish PSCheck wasn't kept behind closed doors at AMD, and would offer it to us, or at least merge it's functionality into Ryzen Master. I only have v3.4.1 and even then it only supports up to Trinity CPUs (maybe Richland, not sure). I can understand not releasing it for Bulldozer, since there was a chance of killing CPUs with it, but then AOD could as well, and it's exactly why the big ass disclaimer we have to accept is there :P

Oh well heh Perhaps in time the Titanium's BIOS will mature enough to include it!

[aGeoM]

Yes, Ryzen 1700.

Just change OPTIMAL values, you can't make it (USER)show in BIOS, leave it at default.
PSTATE Adjustment (1st row of pstates) was the one and only changed.

[Formula350]

Yes, Ryzen 1700.

Just change OPTIMAL values, you can't make it (USER)show in BIOS, leave it at default.
PSTATE Adjustment (1st row of pstates) was the one and only changed.

Well PState 0 = Maximum as far as I know, so I'm not sure if it'd help me any. heh

[NEOAethyr]

cool thx.

Had to put my Rottweiler down today out of the blue.....was not expecting it, happened that quick so I'm going to step away from pc's for a couple days get my together as this was a crushing blow for me.

She has lived at my side literally glued to me for the past 7 years so.....yah anyway. to be continued.

My condolences dude .
I have my little cat, he's been around for probably 8 years or so.

Just to confirm what aGeoM has stated regarding modded BIOSes... I went ahead and flashed one of my modded ones to my MSI Titanium yesterday (used the original v1.10 BIOS that my board shipped with, which is the only BIOS I've been using since it worked fine, even with my RAM @ 3200).

Thankfully everything went just fine!

Points of interest:
- SubTimings indeed are currently non-Adjustable on that BIOS version, at least with how I went about making them visible (ticking them all as USER for Access), but I have one last idea... doubt it'll work though. They DO show up, and tell you what they're set to, which (probably unsurprisingly) are different than the Tertiary Timings shown in Mem TweakIt.
- CPU VDDP allows for entering values, even has a Help that indicates the Default (says 1.050V), Low and Max, but since I don't know how to verify that I have no idea if the 1.070V I set is has actually applied. (I think that HWiNFO's VIN6 under the motherboard might be the VDDP, but I have no evidence and it's a shaky theory at best. After fresh boot it hovers around only 1.000V, but after waking from Sleep it is sticking at 1.048V. Lots of odd stuff happens post-sleep so it's hard to determine things then just due to that.) If anyone knows of a way to figure out what CPU VDDP is running at, let me know.
- The thing that bummed me out the most was that due to these BIOSes full on GUI (at least MSI's specifically), the menu structure inside AMIBCP is very oddly laid out. Additionally, there is no way in the actual BIOS to navigate to these other 'un-hidden' menus (though, now after I've had sleep it just dawned on me that the "Search" feature may turn them up). There are plenty of additional menus listed as you can see, but only the highlighted menus are directly accessible (MFlash is as well of course, but that's not really a 'menu' seeing as it's an option that reboots the system):


In the old Award BIOS editor (the cmd prompt one), it allowed you to easily select a menu category and then literally cut-and-paste it elsewhere. The newer AMIBCP versions do not provide that ability at all, which is not only kinda sad given it's a full GUI, but also because v5.xx EXE size is 11MB >_> What's all that code for if it can't even provide drag-and-drop? v4.55 was only 500KB and it offers the exact same functionality *shrug* ANYways... the unfortunate nature of this means I'm unable to do the other primary thing I had wanted to accomplish through modding, which was to enable the PState overclocking options that are tucked inside the OTHER "Advanced" menu (the one with the [+] not the one above '[+]Overclocking'). Sadly, with my skillset (no hex editing know-how) I'm unable to shuffle options around like I had hoped. There are empty spaces, but they're all tied together, and I an't edit the "Handle". So when I changed one space, every empty space on that page mirrored it Oh well...

Thinking bout flashing the 1.41beta (that was pulled). I don't now if the BCLK was visible by default since I never flashed their version, but I did mod it to ensure it's visible ^_^ We shall see how daring I feel. They pulled all their Betas due to a "flaw in the microcode", but what exactly that means I haven't a clue...

There's likely gonna be duplicate entries and useless ones.
You can't enable them all.
Some can be left overs from the ref bios, ie useless as well if they don't actually work or really do anything of importance.

Yes, Ryzen 1700.

Just change OPTIMAL values, you can't make it (USER)show in BIOS, leave it at default.
PSTATE Adjustment (1st row of pstates) was the one and only changed.

How are you guys modding your bios'es without messing them up?
I wanna beable to play with my r4be bios.
I've used both apps, the one that hides/unhides entries, and the one that does modules.
Both mess up the data, it's not 100% the same anymore.
Asus won't allow flashing it, however I can use the flashback and it works.
At that point, unhiding doesn't do much, they don't show anyways.
And not that it matters... I just need a way of modding things...
And the module thing works, tried with a raid rom, booted windows for a few secs and my cpu hardlocked, power off and on again and cpu is dead.
One time I noticed when playing around that my vcore wasn't setting and was on auto, at that time I had my pump hooked into the cpu fan header.
It makes me think there was some sort of checksum for the cmos or something.

I'de like to know what the heck I was doing wrong lol, or to find a guide that's 100% step by step correct.
Or something.

My goal is to lay out each of the modules in a folder, uncompressed in binary form.
I can do this with winhex, redo the format with winhex replace cmd and compile with fasm each module one by one into a binary.
From there I can format each file into there strings, terminations, and code, or at least tables of some sort.
And do checksums in fasm np, when I learn how they work...

Then the idea is to build a bios with a tool, one module at a time.
Here's where I'm stuck with ami, so I have yet to actually start...
I can't for the life of me figure out what compression scheme they use.
Otherwise I would compress them one by one and patch the small diffs with fasm.
them combined them and put int he boto strap code and decompress code etc, but it's all a mystery I can't figure crap out of the thing.

[chew*]

OK so 3400 @ 17-16-16 was ehhh well its rated kit at 16-16-16 3400 so not really exciting.

3200 @ 16-16-16 ehh well I would hope so.....once again not very exciting more like expected....

So a 3400 bin at 16-16-16 would downscale to about 15-15-15 at 3200 14-14-14 @ 3000.......realistic downscaling and goals.

That makes this sort of exciting @ stock volts.

2x 16g Dual Rank PC 3200 @ 14-14-14-34

Ok so its bench stable big deal right?


How about pretty damn brutal 13+ hours stress test stable?

[Formula350]

There's likely gonna be duplicate entries and useless ones.
You can't enable them all.
Some can be left overs from the ref bios, ie useless as well if they don't actually work or really do anything of importance.

How are you guys modding your bios'es without messing them up?
I wanna beable to play with my r4be bios.
I've used both apps, the one that hides/unhides entries, and the one that does modules.
Both mess up the data, it's not 100% the same anymore.
Asus won't allow flashing it, however I can use the flashback and it works.
At that point, unhiding doesn't do much, they don't show anyways.
And not that it matters... I just need a way of modding things...
And the module thing works, tried with a raid rom, booted windows for a few secs and my cpu hardlocked, power off and on again and cpu is dead.
One time I noticed when playing around that my vcore wasn't setting and was on auto, at that time I had my pump hooked into the cpu fan header.
It makes me think there was some sort of checksum for the cmos or something.

I'de like to know what the heck I was doing wrong lol, or to find a guide that's 100% step by step correct.
Or something.

My goal is to lay out each of the modules in a folder, uncompressed in binary form.
I can do this with winhex, redo the format with winhex replace cmd and compile with fasm each module one by one into a binary.
From there I can format each file into there strings, terminations, and code, or at least tables of some sort.
And do checksums in fasm np, when I learn how they work...

Then the idea is to build a bios with a tool, one module at a time.
Here's where I'm stuck with ami, so I have yet to actually start...
I can't for the life of me figure out what compression scheme they use.
Otherwise I would compress them one by one and patch the small diffs with fasm.
them combined them and put int he boto strap code and decompress code etc, but it's all a mystery I can't figure crap out of the thing.

Yea, I know that there are going to be some duplicate entries. For example in the OC area there are two of each thing for RAM clocks and their readout (Adjusted Frequencies) lines, to which I've determined that it selects whichever one based on processor, as the Carrizo models only support up to DDR4-2400, which IIRC is the highest the first entry goes up to. BUT that's what makes it all the more confusing without being able to physically see what ends up showing up in the BIOS.

As for what we're using, I don't know if that is relevant in your case. Your system might be the Aptio IV (4) version, not the V (5) that ours is. Granted, I can only imagine that the AMIBCP v5 for Aptio V is backwards compat with IV. Nevertheless, we have, as mentioned, AMIBCP v5.02.0023, and there's also v4.55.007 (which I don't know if it'll SAVE correctly using that version of BCP). For kicks I opened mine in v4.55 and it showed up fine, but I definitely won't be tempting fate by using it on my system

Also I dunno if you had read my other post, or aGeoM's later on, but on the ASUS laptops the trick is to use AMIAFU-Win (flashing tool) and have it Save your current BIOS to a .rom file, and then edit said .rom instead of the .cap that ASUS package contains. Once editing has finished, go into the BIOS and flash that .rom.
Another tip I read was to save OVER the file you open, and not to use Save-As. As such, you'll want to backup your dumped one first before editing/saving (I just RAR it). That way everything stays the same and AMIBCP isn't trying to re-create a whole new package. Keep the filenames the exact same, too.

Example, using my system:
If I want to modify the official BIOS v1.30, I would first flash that to the system.
Next I'd boot Windows and load the AMI flash utility, and Save the BIOS (aka Dump it, for Googlers reading this), giving it the same name as that of what was downloaded. In this case it's E7A31AMS.130
NOTE: The .rom is replaced by .130. If your BIOS happens to say .cap then my suggestion is to leave it as .rom.
After renaming, I'd then ZIP it for backup.
Then I load it into AMIBCP (whether you need to use v4 or v5 might depend on your version of UEFI), make any changes I want, and click SAVE, overwritting that "E7A31AMS.130".
Now I get my FAT32 formatted USB drive and put it on there.
Reboot the computer, go into the BIOS, and load it's Flashing system.
Select that file and let it do its thing.
That's it.

After that my system reboots and I"m greeted with a CMOS warning about setting changes, which lets me either enter Setup, load an OC Profile, or just continue on. HIGHLY RECOMMENDED TO ENTER SETUP AND THEN LOAD DEFAULTS! After defaults are loaded, exit and save changes, and then go back into the BIOS to make your usual tweaks.


Personally I don't know enough knowledge to tinker with hex editing, thus I don't have a need for MMTool. However, the program UEFITool might be of use to you, and if you don't have it I can try and put up a copy for you. I mainly got all my stuff of VoltGround forums.

[chew*]

ok I am finally satisfied with the board enough to give it a big thumbs up.

Put the best the memory can do combined with known stable cpu oc and it just works....

Cpu ocs decent.

Memory ocs very good.

Capacity of memory with an oc on memory while overclocked is pretty stellar.

Runs very cool.



Couple complaints....lose wifi with this ref clock... ( i have a usb wifi in working fine )

No clue if 3d is realistic at this ref clock yet...but that pretty much applies to any ref clock of 120....

Workstation wise if no graphics needed...man this thing is going to rip..

[chew*]

Since this seems to be a big Question mark........

3d tackled @120 ref clock gen 1 on the ASROCK TAICHI.

borrowed launch day review timespy and firestrike results.





I like this board.......

24+ hours complete now its time to put the hammer down.......prime and heaven looping all night long.....

[techtrancer]

Why now 3,9ghz?

[chew*]

This chip is only good for 3900-3990 board dependant so...its kinda flaky chip.

Based on cpu test scores...looks like it is being driven harder at less speed...so preety understandable...

[chew*]

Ryzen - myths busted

https://youtu.be/aXHlTMKyse8

[WR-HW95]

Hi.
Long time since posted here last time, but had to come here wonder this.
I?m running 4x16Gb 3200 TridentZ c14 on my C6H fine on 3200 divider 18-14-14-38-T1 with 1002 bios, but when I tried to run 2x16Gb there was no way to get boot up on same settings or even 18-18-18.

[chew*]

Which channels closest or furthest from cpu when 2x16g..

[Formula350]

Which channels closest or furthest from cpu when 2x16g..

I think I speak for WR-HW95 when I --completely genuinely I might add-- ask the question of: Why would that matter in this instance?
If 4x16GB works at 3200MHz (which is awesome to hear!), that means every slot is populated, closest AND furthest.

I'm just rather curious what difference it makes in case I come across someone else having a similar issue, so that I'm able to better explain to them. heh

[chew*]

Closest solo just with 2x are usually less stability..

Sort of physically = a weaker drive strength

[WR-HW95]

Which channels closest or furthest from cpu when 2x16g..

Furthest. I think it was written somewhere you should use channel B with 2 sticks.
There is still some voltages what i could try to set manually correct and try, but i would like to try out new bios versions too.
What i have ready about them is pretty mixed thoughts.
Oh. And i didnt tell that only way to get 4x16Gb run on 3200 divider is to set first Cas at 18 with this 1002 bios.
I did try on 2666 divider and it worked with 14-14-14, but stopped testing around 2880 when one of hdd?s messed up.

[I.nfraR.ed]

Yes, you can't go lower than CAS18 with dual rank or 4x8 single rank at 3200 divider. If you want lower CAS, you have to clock bclk.
But this will probably change when May update gets rolled out, when we're supposed to have higher dividers and control over subtimings.
I hope 3600 would then be possible at 100bclk and tight timings (set manually).

Because benching is one thing, but I also try to help people who use their systems 24/7.

[I.nfraR.ed]

0082 is the best so far for me.

Mem could not go so high before. Also don't see the Win10 bugs, everything seems fine in terms of scores.
Also this latest Aida version seems to be fine, too. I can't reproduce the bug with abnormally high scores anymore.

[Don_Dan]

It's great to see you're still alive and kicking, Brian. Good work on the tests, as always!

[Formula350]

lol DDR4-385 @ 12s. That is indeed a cherry picked kit!
Also, Infrared, can you confirm a suspicion I've had, since I've not bothered with OCing yet until I get my water cooler on (bracket came today!)...
The increases in AIDA's Cache speeds, those are indeed increasing due to CPU clock and are otherwise unaffected by RAM speeds, yes?
Which, on that note, nice job on breaking 1TB/s :P

EDIT: Oh, right, reason I came was to post my 32M speed. I think at defaults, 2933's tighter timings may lead to improve speeds in certain regards. My time with memory at 3200 (same timings) is 9mins flat, so this is substantially quicker... Which also worth noting is my system has been up for almost 4days, so this is not a fresh system boot! heh

[charged3800z24]

Ryzen - myths busted

https://youtu.be/aXHlTMKyse8

Awesomeness Chew* !!!

[ZenEffect]

0082 is the best so far for me.

Mem could not go so high before. Also don't see the Win10 bugs, everything seems fine in terms of scores.
Also this latest Aida version seems to be fine, too. I can't reproduce the bug with abnormally high scores anymore.

are you noticing a significant slow down in 32m? 2x boards on the same agesa code and same result. 6-12 second loss depending on board.

[I.nfraR.ed]

Haven't tested. Had to change from MBR to GPT in order to install Win 10, so I have no Win 7 to compare at the moment.

[chew*]

I noticed a regression to. Magically clocking easier usually means something got slacked. Its not asus...its the agesa.

[chew*]

Ok a couple updates.

Someone said I should update prime 95.

I am on and have been on version 29.1 for awhile now........if there is newer feel free to link me.....

4x8gb on x370 taichi = fail 1.94a and 2.0 bios......

By fail I mean if they can not run 3200 14-14-14-34. Could be due to the fact that I am mismatched with Flare X and Trident-Z but I do not believe that to be the case....

Both have same specs although I have not looked into the SPD table in mem tweak it yet to look for differences.

2x8g pc3200 14-14-14-34 Flare X works great on taichi. Load XMP and go....that simple.

I will be reviewing it soon on multiple boards as it seems to matter more board then cpu in my testing......