AMD Motherboards - AMD response to AGESA/BIOS query.

Print Page | Close Window

AMD response to AGESA/BIOS query.

Printed From: ASRock.com
Category: Technical Support
Forum Name: AMD Motherboards
Forum Description: Question about ASRock AMD motherboards
URL: https://forum.asrock.com/forum_posts.asp?TID=6271
Printed Date: 17 May 2024 at 11:18am
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com

Topic: AMD response to AGESA/BIOS query.

Posted By: MisterJ
Subject: AMD response to AGESA/BIOS query.
Date Posted: 07 Oct 2017 at 2:50am

" rel="nofollow - Here is the response to the question below:

I understand that you would like to know what Agesa is and how the different versions compare. If this is incorrect, please let me know as the information provided may change.

Thank you for your email.

AGESA is an acronym that stands for ?�AMD Generic Encapsulated System Architecture.??As a brief primer, the AGESA is responsible for initializing AMD x86-64 processors during boot time, acting as something of a ?�nucleus??for the BIOS for your motherboard. Motherboard vendors take the core capabilities of our AGESA updates and build on them with their own ?�secret sauce??to create the BIOS files you download and flash.

Each updated version may contain small bug fixes, performance improvement and even new bios features and options.

For Ryzen, the latest version is 1.0.0.6b.

For Threadripper, the latest version is 1.0.0.3.

We recommend checking with your Motherboard manufacturer periodically to keep yourself updated with the latest bios and Agesa version.

And here is my original query:

I have done some research of X399 mainboards and several AGESA versions in use. One vendor has 0.0.6.0, several have 1.0.0.3a and my vendor (ASRock) has 1.0.0.3. I,of course, do not know what each contain. My X370 board has 1.0.0.6b. Do these version numbers compare directly? What is the latest you have released to board vendors for the 1950X and what does it contain. Some memory vendors are not selling high performance products now because they know they will not work and these AGESA versions. I think a number of problems exist here: -Users have no idea what is contained in a AGESA version. -We have no idea when a needed version will be available (eg,1.0.0.6b equivalent for TR) -We are currently stuck with low performing memory for lack of a up to date AGESA. Please help me understand the AGESA conundrum. Thanks and enjoy, John.

I have sent a response indicating how disappointed I was. Enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Replies:

Posted By: wardog
Date Posted: 07 Oct 2017 at 3:18am

" rel="nofollow - "Secret sauce" hahaha John. How'd that make ya feel after reading THAT? Yea. Me too........ Comforting, NOT.

John, which BIOS from what MB is claiming 3a?

Posted By: datonyb
Date Posted: 07 Oct 2017 at 4:16am

it would have been nice for amd to actually get someone technical to answer that

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold

Posted By: MisterJ
Date Posted: 07 Oct 2017 at 4:35am

Right, datonyb!

wardog, I posted this at the time, did not record the MB model, but remember these are TR and I would guess that a vendor has the same AGESA on all their boards.

Thanks and enjoy, John.

EDIT:Oops, Wardog:


AGESA!V9.ThreadRipperPI-SP3r2-1.0.0.3  ASRock
AGESA!V9.ThreadRipperPI-SP3r2-1.0.0.3A Gigabyte
AGESA!V9.ThreadRipperPI-SP3r2-1.0.0.3A MSI
AGESA!V9.ThreadRipperPI-SP3r2-0.0.6.0  ASUS

Posted By: nangu
Date Posted: 07 Oct 2017 at 1:25pm

wardog wrote:

" rel="nofollow - "Secret sauce" hahaha John. How'd that make ya feel after reading THAT? Yea. Me too........ Comforting, NOT.

John, which BIOS from what MB is claiming 3a?

:-))))))))

What's the english word for spanish "brujeria"? Witchcraft??? :-)))

Hey Asrock, you are all very old witches hahaha

-------------
R7 1700 @3.90 1.25v - GSkill TridentZ 3200c16 Hynix MFR @2933 14-16-16-32 - Fatal1ty Gaming X370 K4 - Gigabyte GTX 1070 G1 - WD 256 Black M2 Nvme as Windows 10 boot drive - EVGA Gold 650W

Posted By: parsec
Date Posted: 07 Oct 2017 at 9:07pm

" rel="nofollow - Wow, that reply was worse than I expected, without even knowing the questions.

Comparison between and any shared attributes between Ryzen and TR AGESA versions? Remains a mystery. Such a good question, an obvious one. Oh well...

Perhaps TR AGESA 1.0.0.3A is currently a Beta version, that Gigabyte and MSI are using? Apparently AMD has no control over which AGESA code is used. Reading between the lines here. No response and more mysterious since TR AGESA 1.0.0.3 is the latest version.

But they did answer what each AGESA version contains: "Each updated version may contain small bug fixes, performance improvement and even new bios features and options.". Imagine that, who knew?

John, you are asking for information that will take effort and a good understanding of AGESA to explain, you are expecting too much! If you asked where the power button is located, you might get a good answer.

-------------
http://valid.x86.fr/48rujh" rel="nofollow">

Posted By: MisterJ
Date Posted: 07 Oct 2017 at 11:24pm

Thanks, nangu. Almost all of us here are users not ASRock employees. ASRock_TSD is the only employee I have seen post here.

Thanks, parsec, I always seem to expect too much. Weird that the logical is just too much to expect. Some one(s) in AMD designs, implements and maintains the AGESA. I bet that person(s) has at least some understanding. I suspect the response is political (CYA)!

Enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: SoniC
Date Posted: 08 Oct 2017 at 1:52am

Pity... your question John was very clear for a "techie" probably they employ students that just... copy something from the script that is given to them...

Maybe rephrasing the question and shorting the message would help... and indicate clearly that you would expect an tech-guy to reply and not just a blunt explanation for everyday user....

Pity... was eagerly waiting for the reply from AMD... what a letdown...

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: MisterJ
Date Posted: 08 Oct 2017 at 1:59am

Thanks, SoniC. I have responded and made my displeasure clear and ask my question in a different way. Enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: SoniC
Date Posted: 08 Oct 2017 at 2:04am

No need to thank me, I thank YOU for like... fighting for us (X399 users) rights to have a better mem compatibility ;-)

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: MisterJ
Date Posted: 13 Oct 2017 at 11:58pm

Here is the latest response - much better!

We appreciate your patience as we continue to work with our motherboard partners to deliver improved memory reliability and compatibility for the SocketTR4 platform. When platform improvements are available we will publish them on our AMD community blogs, as we have done for Ryzen in the past.

Regarding AGESA versions, there is not a shared code base between the AM4 platform and the SocketTR4 platform, i.e. Threadripper AGESA is not the same as Ryzen AGESA.

Our guidance for memory compatibility for Ryzen Threadripper when populating 2 DIMMs per channel, in all four channels, is to support DDR4-2133 speeds (for reference please see the attached picture). Speeds above that are achieved through overclocking, where manually tuning memory timing and voltages may be required to maintain stability.

I suggested AMD publish a list for various memory configurations with supported speed/latency. Enjoy, John.

Posted By: datonyb
Date Posted: 14 Oct 2017 at 12:24am

" rel="nofollow - reading inbetween the lines and making educated assumptions

is it in fact a better option to select ram with the jedec 2133 base/default

than to select ram with jedec 2400/2666 minimum ??

or in other words is this problem with 8 sticks 128gb limited to jedec minimum sticks carrying 2400/2666 and therefore the bios cant start to initialise ?

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold

Posted By: MisterJ
Date Posted: 14 Oct 2017 at 2:47am

datonyb, I would interpret it as saying use 8 sticks (single or double rank was not specified) with 2133MHz SPD or the user owns any difficulty. If they really publish the information I suggested (lots of work), we should see the effect of rank and other parameters. In my response I asked about my specific case - 4x8 GB. Perhaps users should open a ticket for their proposed configuration. I got no feel for initialization. Thanks and enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: MisterJ
Date Posted: 17 Oct 2017 at 2:05am

Here are a couple of pictures AMD support sent with the latest reply:

https://imgur.com/gLgCo7e" rel="nofollow - https://imgur.com/gLgCo7e

https://imgur.com/MUrhRcl" rel="nofollow - https://imgur.com/MUrhRcl

Enjoy, John.

Posted By: SoniC
Date Posted: 17 Oct 2017 at 2:21am

I assume these were all SS memory sticks.... and yeah... 64GB... where this platform should support 128GB max....

I have 4x 16GB DS and on X399 I'm reaching 90% of what they can on a X370 platform (2933 vs 3200).

It's definitely not what premium (or "pre-selected RyZEN modules") feels like LOL

...ohwell...

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: JUANNY
Date Posted: 17 Oct 2017 at 3:00am

SoniC wrote:

I assume these were all SS memory sticks.... and yeah... 64GB... where this platform should support 128GB max....

I have 4x 16GB DS and on X399 I'm reaching 90% of what they can on a X370 platform (2933 vs 3200).

It's definitely not what premium (or "pre-selected RyZEN modules") feels like LOL

...ohwell...

Why are the dimms running at 2133? Haha feels like buying a Mercedes then putting 20 dollar tires on it no pun intended.

Posted By: MisterJ
Date Posted: 17 Oct 2017 at 3:09am

SoniC, I agree, Double Sided (probably required for 16 GB sticks) and 8 sticks will, in effect, create quad rank memory load on the buss. I would suspect it may not make 2133 MHz.

JUANNY, did you see this in the AMD response:

Our guidance for memory compatibility for Ryzen Threadripper when populating 2 DIMMs per channel, in all four channels, is to support DDR4-2133 speeds

They are following their own guidance. As I suggested before, if you want to go faster and max GBs, you may need to wait.

I just noticed in the HW picture that the 4-pin CPU power connector is vacant - interesting.
Enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: JUANNY
Date Posted: 17 Oct 2017 at 3:16am

MisterJ wrote:

SoniC, I agree, Double Sided (probably required for 16 GB sticks) and 8 sticks will, in effect, create quad rank memory load on the buss. I would suspect it may not make 2133 MHz.

JUANNY, did you see this in the AMD response:

Our guidance for memory compatibility for Ryzen Threadripper when populating 2 DIMMs per channel, in all four channels, is to support DDR4-2133 speeds

Ya I know that is their guidance but I could imagine that they would try harder to not bog down their flagship platform with such low speed ram unless it was technically not possible to support higher speeds.

Posted By: datonyb
Date Posted: 17 Oct 2017 at 7:42pm

MisterJ wrote:

Here are a couple of pictures AMD support sent with the latest reply:

https://imgur.com/gLgCo7e" rel="nofollow - https://imgur.com/gLgCo7e

https://imgur.com/MUrhRcl" rel="nofollow - https://imgur.com/MUrhRcl

Enjoy, John.

tell the smart ass to send you a picture with 8 16gb dimms then installed and working

we already know 64 gb can be sorted

(i do presume without a doubt you have asked about the 128gb issue )

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold

Posted By: MisterJ
Date Posted: 17 Oct 2017 at 10:59pm

I have asked about 128GB, datonyb. We are trying to set up a phone conversation and I hopefully will ask many useful questions. Thanks and enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: MisterJ
Date Posted: 19 Oct 2017 at 8:04am

I had a phone conversation with AMD support today. They have tried 128 GB and it works at 2133 MHz - memory and board not stated. They made it clear this is really a main board vendor task. In our case ASRock should test and qualify memory to be used and I assume when it does not work for a user, ASRock should fix or explain why it does not work. As I expected, I got no information about future AGESA/BIOS or any ramifications of Ryzen 1.0.0.7. We are in a wait and see as we have been. Enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: datonyb
Date Posted: 19 Oct 2017 at 8:39pm

yesterday while borwsing youtube
i did in fact try and find some extra info/tips about the 128gb issue

wow !

it would seem most uploads would love to ONLY have an issue with 128gb, as theres so many complaining about even getting 4 channels single sticks working (eg 4 x 8 gb)

i did find one guy who managed 8 sticks of gskill tridentz and 128gb, who said he was stuck at 7 sticks for ages and it was a nightmare of epic magnatude to get the 8th working
alas he continued to spend the rest of the video showing off stupid rgb lights and poncy glass case rather than share his experiance with what he actually managed to get set to make it work

i do not have a you tube account so i couldnt ask him to detail some advice on a work around for the 8th stick

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold

Posted By: SoniC
Date Posted: 20 Oct 2017 at 5:44am

Share the YT link, I could ask him.

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: datonyb
Date Posted: 20 Oct 2017 at 7:33am

" rel="nofollow -

SoniC wrote:

Share the YT link, I could ask him.

it was this guy

although weirdly enough he has taken down the volg 3 and vlog4 videos that he was talking about the ram hes a link to the vlog 2 video

https://www.youtube.com/watch?v=SChTan3nZXo

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold

Posted By: parsec
Date Posted: 20 Oct 2017 at 11:24am

" rel="nofollow -

MisterJ wrote:

I had a phone conversation with AMD support today. They have tried 128 GB and it works at 2133 MHz - memory and board not stated. They made it clear this is really a main board vendor task. In our case ASRock should test and qualify memory to be used and I assume when it does not work for a user, ASRock should fix or explain why it does not work. As I expected, I got no information about future AGESA/BIOS or any ramifications of Ryzen 1.0.0.7. We are in a wait and see as we have been. Enjoy, John.

While AMD apparently feels it is fine to throw the mother board manufacture "under the bus", and make it their responsibility, I know that ASRock will not do the same to AMD. ASRock may know why it won't work, but won't reveal it is AMD's problem, if it actually is. Of course, how could either party explain and prove why it can, should, or will work or not work without going into a highly technical explanation that is beyond most user's understanding.

Yes, a board should be able to work at the specs provided by the processor manufacture, which has never been a problem with Intel processor boards using DDR4 memory. DDR4 memory over clocks on Intel boards are simple compared to Ryzen.

When AMD themselves said 128GB works at 2133, while the TR processor stock guaranteed spec is 2400, that was an acceptable response?

-------------
http://valid.x86.fr/48rujh" rel="nofollow">

Posted By: SoniC
Date Posted: 07 Nov 2017 at 2:22pm

New AGESA out with BIOS 1.80 for the X399 boards, changelog:

Update AGESA to 1.0.0.4

Testing will follow today in the evening.

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: MisterJ
Date Posted: 08 Nov 2017 at 12:49am

Great! Thanks much SoniC! Enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: SoniC
Date Posted: 08 Nov 2017 at 4:08am

No need to thank me John...

I do it for myself and also doesn't harm to share the results with others "not completely satisfied customers" of AMD/MoBo Manufacturer = whoever is to blame here - I don't want to finger-point.

So the results are in after the testing of ver 1.80.:

1.80 BIOS :

Same old, same old....

Virtually nothing changed for me... nothing... nada.

On my 64GB G.Skill Ripjaws V kit the 3200 speed is still unreachable. 4 sticks, Sammy B-die.

The only stable speed is 2933.

Move along, nothing to see/cherish here...

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: JUANNY
Date Posted: 08 Nov 2017 at 4:19am

SoniC wrote:

No need to thank me John...

I do it for myself and also doesn't harm to share the results with others "not completely satisfied customers" of AMD/MoBo Manufacturer = whoever is to blame here - I don't want to finger-point.

So the results are in after the testing of ver 1.80.:

1.80 BIOS :

Same old, same old....

Virtually nothing changed for me... nothing... nada.

On my 64GB G.Skill Ripjaws V kit the 3200 speed is still unreachable. 4 sticks, Sammy B-die.

The only stable speed is 2933.

Move along, nothing to see/cherish here...

2933 is the same top speed for the theadripper optimized newly release flare x-wondering if maybe the theadripper IMC tops off at that frequenct regardless of the AGESA implementation

-------------
Asrock X399 Fatality Pro Gaming Bios 2.0 Threadripper 1950X Cpu cooler Enermax 360 mm TR4 GSkill 64 GB F4-2933C14Q-64GFX Windows 10 X64 PRO Version 1709 Build 16299.19 2X EVGA 1080 TI in SLI mode

Posted By: SoniC
Date Posted: 08 Nov 2017 at 4:52am

" rel="nofollow -

JUANNY wrote:

2933 is the same top speed for the theadripper optimized newly release flare x-wondering if maybe the theadripper IMC tops off at that frequenct regardless of the AGESA implementation

Or maybe they just certified it at that speed cause this is currently the max it can run stable...

We don't know without insiders knowledge...

I really wish one of the AMD gurus like R. Hallock would reply to our queries...

If we have reached the limit I would like to know it and not beeing kept in the dark...

I _honestly_ hope that the TR can do at least on par with RyZEN (especially due to the link between memory speed and infinity fabric)...

All this blabber about premium product, only pre-selected dies that go into the TRs blablabla...

Yes, cores overclock better with lower voltages, but memory controller? Goddamn...

I would gladly revert to dual-channel if it could do 3200+ on 4 dual sided sticks like RyZEN does.

P.S. I did another experiment and populated the other banks against ASRock recommendation.

With four sticks they recommend to populate D2 C2 A2 and B2. I populated the "1's".

It boots, mem compatybility even worse.

But what I've noticed that on the "1's" it restarts at least properly and reverts to stock mem settings (like it does on X370 FPG). On the LED display it goes in a loop, then restarts at the "d0" code three times (or x-times according to UEFI fail safe settings) and afterwards boot with stock mem. settings. All other settings remain in UEFI as set.

4 Sticks in "2's" and it never restarts properly & reverts to the stock mem speeds. Every time I try I need to reset CMOS. It goes through three or four loops (I can distinguish a loop when it starts to display "00" and the fans start to spin-up for a short sec.) and then it remains hung at some of the error codes, mostly 54-55* (mem-related).

For me it seems to be another bug... and no, it is not new.

It was always there with every UEFI version.

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: MisterJ
Date Posted: 08 Nov 2017 at 5:17am

" rel="nofollow - Thanks SoniC and JUANNY. How SAD! This really stinks. I hope some of the users of Beta 1.71 will try 1.80 and let us hear how they compare. Have either of you tried 1.71? Thanks and enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: SoniC
Date Posted: 08 Nov 2017 at 5:57am

MisterJ wrote:

" rel="nofollow - Thanks SoniC and JUANNY. How SAD! This really stinks. I hope some of the users of Beta 1.71 will try 1.80 and let us hear how they compare. Have either of you tried 1.71? Thanks and enjoy, John.

1.71?

There was no 1.71 beta ... at least not for the X399 FPG... or I overlooked it...

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: JUANNY
Date Posted: 08 Nov 2017 at 7:00am

SoniC wrote:

1.71?

There was no 1.71 beta ... at least not for the X399 FPG... or I overlooked it...

if youre interested in the beta I had it emailed to me from one of the forum members-it is older then todays releas bios so not sure if it is helpful or not-name of beta is X399PG1.71.rom-pm me if interested

-------------
Asrock X399 Fatality Pro Gaming Bios 2.0 Threadripper 1950X Cpu cooler Enermax 360 mm TR4 GSkill 64 GB F4-2933C14Q-64GFX Windows 10 X64 PRO Version 1709 Build 16299.19 2X EVGA 1080 TI in SLI mode

Posted By: MisterJ
Date Posted: 08 Nov 2017 at 8:25am

" rel="nofollow - SoniC, I just searched for 1.71 and found several threads. Here is one:
http://forum.asrock.com/forum_posts.asp?TID=6214&KW=1%2E71&PID=38417&title=fatal1ty-x399-wont-post-with-128gb-ram#38417
An ASRock support group member provided it to a user who provided it to other users via PM. Several users were Wowed by it, claiming it fixed all their memory problems and got their failing memory running fine (even 128 GB). I did not try it, but it also had AGESA 1.0.0.4 so I expected great things from 1.80. If you do try it, please tell us your views. Thanks and enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: SoniC
Date Posted: 08 Nov 2017 at 8:31am

Yeah.... overlooked it...

I am getting old Wink

Anyway, PMed JUANNY, not scared of downgrades ;-) Let's give 1.71 a spin...

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: SoniC
Date Posted: 08 Nov 2017 at 9:35am

Aaaand the results from 1.71 testing are in.... that was fast!

OMG!! IT IS WORKING! My memory problems are solved! I now remember even who I was in the last life and I can predict future now!

...

unfortunately - just joking.

Results as expected --- no difference between 1.70, 1.71 and 1.80. Maybe they have fixed a bug with 128GB RAM, maybe some users did some mistakes or reported just a placebo effect... dunno...

Max stable for me is... 2933 with 4 d-s sticks. Samsung B-Die. Pity.

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: JUANNY
Date Posted: 08 Nov 2017 at 9:51am

SoniC wrote:

Aaaand the results from 1.71 testing are in.... that was fast!

OMG!! IT IS WORKING! My memory problems are solved! I now remember even who I was in the last life and I can predict future now!

...

unfortunately - just joking.

Results as expected --- no difference between 1.70, 1.71 and 1.80. Maybe they have fixed a bug with 128GB RAM, maybe some users did some mistakes or reported just a placebo effect... dunno...

Max stable for me is... 2933 with 4 d-s sticks. Samsung B-Die. Pity.

Im not surprised we will have to wait until AMD os Asrock clarifies whether TR is physically limited to 2933ish or thereabout. Odd that the board is advertised to support 3600+ in the product page

-------------
Asrock X399 Fatality Pro Gaming Bios 2.0 Threadripper 1950X Cpu cooler Enermax 360 mm TR4 GSkill 64 GB F4-2933C14Q-64GFX Windows 10 X64 PRO Version 1709 Build 16299.19 2X EVGA 1080 TI in SLI mode

Posted By: MisterJ
Date Posted: 08 Nov 2017 at 10:07am

" rel="nofollow - Thanks much, SoniC.
JUANNY, please see my thread here:
http://forum.asrock.com/forum_posts.asp?TID=6534&title=memory-vendor-comment-on-x399-agesa
Enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: parsec
Date Posted: 08 Nov 2017 at 1:34pm

" rel="nofollow -

JUANNY wrote:

SoniC wrote:

Aaaand the results from 1.71 testing are in.... that was fast!

OMG!! IT IS WORKING! My memory problems are solved! I now remember even who I was in the last life and I can predict future now!

...

unfortunately - just joking.

Results as expected --- no difference between 1.70, 1.71 and 1.80. Maybe they have fixed a bug with 128GB RAM, maybe some users did some mistakes or reported just a placebo effect... dunno...

Max stable for me is... 2933 with 4 d-s sticks. Samsung B-Die. Pity.

Im not surprised we will have to wait until AMD os Asrock clarifies whether TR is physically limited to 2933ish or thereabout. Odd that the board is advertised to support 3600+ in the product page

TR and AM4 memory speed and memory OC specs are from AMD, mother board manufactures simply pass them on. Some low end AM4 boards won't reach the same memory OC speeds of better boards, but that should not apply to X399.

TR currently not meeting its memory speed specs? Welcome to Ryzen and Ryzen derivative processors! The memory controller is part of the processor, and memory compatibility and speed is primarily determined by the AMD AGESA processor microcode that is part of the UEFI/BIOS.

AM4 board owners lived through the same type of fiasco with memory compatibility and speed. We had several AM4 AGESA updates, and our memory OC speeds slowly improved. But AMD is not done yet with the AM4 AGESA updates. The upcoming AM4 1.0.0.7 update is such a big change that it requires major rewrites of the UEFI/BIOS code. Memory compatibility and OC speed is still a work in progress for AM4 and TR. But AM4 and TR is AMD's first time with DDR4, Intel is on its fifth platform using DDR4.

AM4 memory compatibility and OC capability with existing DDR4 memory, and even DDR4 designed for AM4 is easily described in one word. Random. The best we know is Samsung B-die, but there is more to it than that, not that we know what it is. Some memory released for the first two (Intel) DDR4 platforms can OC over 3200 on X370 boards, but most won't. My G.SKILL FlareX 3200 memory, "designed for Ryzen", finally reached 3200 with my X370 board's latest UEFI/BIOS/AGESA version, but with manual entry of the secondary and tertiary memory timings with settings found on the Internet.

Comparing the history of Intel's and AMD's use of DDR4 is an interesting story. Starting with Intel's X99 system, we got a couple memory divider OCs over the max stock specs. But with nothing to compare it with, who could complain? No one did. Then with each new Intel system, the OC speeds got a little better. No one did better, so nothing to compare it with.

Now Intel's X299 and Z370 systems have DDR4 OC specs of speeds over 4000. AM4 and TR users see that and wonder what the problem is. AMD is still learning, and their first generation DDR4 memory controllers cannot be expected to be the equal of Intel's fourth and fifth generation DDR4 memory controllers. That is the reality, in contrast to what some users expect and want from AMD.

Yes, AMD over-specified their memory OCs early on, and maybe they won't be reached. But all OCs are not guaranteed, right? How soon we forget. Or do we acknowledge that?

-------------
http://valid.x86.fr/48rujh" rel="nofollow">

Posted By: SoniC
Date Posted: 08 Nov 2017 at 3:31pm

Thanks for your words parsec.

That gives me a little hope that there are still things that can (and hopefully - will) be done.

(I try to read between the lines here a little).

I am well aware that DDR4 is a new mine-field for AMD, but I hope that they will finally pull through (like they did more-or-less with RyZEN as it finally reaches some nice OCs on the DDR4. Not there yet compared to intel, but at least in an acceptable area).

Anyway, still hoping that the AGESA rewrite will also apply to TR and not only to the AM4 platform as - thanks to MisterJ - we know they are independent branches (the big question is - how much of the same code they share AM4/SR4).

It.... would be just nice to know that we are not there yet, but they are working on it.... or... to hear at least - guys, sorry, we overpromised. This is your max, we are sorry. This is soooo darn frustrating...

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: MisterJ
Date Posted: 08 Nov 2017 at 11:41pm

" rel="nofollow - Then next year we get DDR5 - 2x speed of DDR4. Will there be MBs next year using DDR5? I guess we'll see. Ultimately I think AMD will produce on Ryzen and TR. AMD tests with much later AGESA versions and I suspect memory vendors do too via NDA. Enjoy, John.

-------------
Fat1 X399 Pro Gaming, TR 1950X, RAID0 3xSamsung SSD 960 EVO, G.SKILL FlareX F4-3200C14Q-32GFX, Win 10 x64 Pro, Enermx Platimax 850, Enermx Liqtech TR4 CPU Cooler, Radeon RX580, BIOS 2.00, 2xHDDs WD

Posted By: datonyb
Date Posted: 09 Nov 2017 at 3:06am

guys
may i suggest something

there is a thread/topic over on overclcokers specifically with ram timings achieved

a great deal of time has been spent testing ram on x399 by a guy called chew (brian maclochlan)

he has a taichi but the findings will apply well enough to add to your skills and efforts im sure

he has also proven that like ram chips and like cpus not all imc's are created equal

compounded by swapping out exact same spec cpus in a system and finding one just cant run the ram at the same settings as the other

@ sonic
with your ram ,and your skills and time
i would expect you can find the missing ingrediant after reading though that thread and making a few notes to gain 3200 stable

chew has even posted exact screenshots of samsung b dies with safe timings and fast timings

its really worth the read mate

http://www.overclock.net/t/1638140/asrock-x399-taichi-fatality-overclocking-and-unofficial-support" rel="nofollow - http://www.overclock.net/t/1638140/asrock-x399-taichi-fatality-overclocking-and-unofficial-support

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold

Posted By: SoniC
Date Posted: 09 Nov 2017 at 6:13am

Thanks datonyb, appreciate.

But given the amount of time I've already invested (and that this is not my first platform based on ZEN) I have my doubts if I reach higher...

Maybe (just maybe) with some timings from chew I might be able to squeeze another step stable (3066 that is) but this is really a hair-thin thing.

Anyway - I'll give it a shot on the weekend where I will need to stay home anyway (on-call duties in the IT sometimes really suck ;-) ).

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: datonyb
Date Posted: 09 Nov 2017 at 7:38am

SoniC wrote:

Thanks datonyb, appreciate.

But given the amount of time I've already invested (and that this is not my first platform based on ZEN) I have my doubts if I reach higher...

Maybe (just maybe) with some timings from chew I might be able to squeeze another step stable (3066 that is) but this is really a hair-thin thing.

Anyway - I'll give it a shot on the weekend where I will need to stay home anyway (on-call duties in the IT sometimes really suck ;-) ).

i dont know how much help it is

but

there is a lot of happy people also using the ryzen ram calculator program

ive got high hopes it may also help with x399 settings (seeing as at deep levels its just two ryzens)

there is a lot of help by the author as well
maybe say hello to 1usmus over on his support thread and ask how suitable it is for assisting with TR
http://www.overclock.net/t/1640919/ryzen-dram-calculator-overclocking-dram" rel="nofollow - http://www.overclock.net/t/1640919/ryzen-dram-calculator-overclocking-dram

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold

Posted By: SoniC
Date Posted: 09 Nov 2017 at 8:13am

datonyb wrote:

SoniC wrote:

Thanks datonyb, appreciate.

But given the amount of time I've already invested (and that this is not my first platform based on ZEN) I have my doubts if I reach higher...

Maybe (just maybe) with some timings from chew I might be able to squeeze another step stable (3066 that is) but this is really a hair-thin thing.

Anyway - I'll give it a shot on the weekend where I will need to stay home anyway (on-call duties in the IT sometimes really suck ;-) ).

Just checked the RyZEN DRAM calc, and it more-or-less is also viable solution to calculate TR DRAM timings (once the platform hopefully gets an update for Hynix and other stuff).

The profiles/timings provided my chew are however tighter and more fine-tuned.

Seems another weekend will be gone quickly for me ;-)

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: datonyb
Date Posted: 10 Nov 2017 at 2:58am

ive just read a reply from the calculator author (1usmus)

he advised the calculator will work with threadripper

may i suggest trying them at 'safe' suggestions first check its stable then trimming them down to either chews findings or the calculators fast settings step by step

chew has put a fantastic amont of effort and real world testing into amd platforms

but its hard to get all data from his findings in one place
(eg he'll post a ryzen timing checker screenshot but may lack procodt or volts that he used)

a great mind with too little time to babysit all of us i presume Smile

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold

Posted By: SoniC
Date Posted: 10 Nov 2017 at 7:33am

That's what I've also found after comparing the results.

My suggestion is to take chew's timings (safe ones with 4 DR modules) and the rest from the safe settings in the calculator.

Will test a few of the combinations during the weekend and will drop here some results.

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: SoniC
Date Posted: 11 Nov 2017 at 12:50am

Ok guys... and gals... and girls (in case there are any on this forum - drop me a PM ! :-P ).

Now to the point -- I couldn't wait any longer and have checked virtually all the combinations --- the proposed by chew (for 4 DR 16GB Sticks) plus combinations done with the RyZEN DDR Calculator based on a mix of chew's settings, safe settings from calc, with some viariances as suggested.

Results:

None of the combinations was able to get me past 3066 boot (unstable), which I already archieved earlier on.

3200 is still unbootable.

Still stable is 2933 for the 64GB 4x DR Ripjaws V Sticks. These exact sticks run without a problem with full speed when put into a X370 X1800 Platform. So I know they are perfectly fine.

After so many hours spent setting up the BIOS I think my next challenge will be to set it up from scratch with my eyes closed.... LoL.

Anyway... let's hope that the complete rewrite of AGESA will bring some positive effects to those who have been unlucky...

P.S. Chew provides the settings for the 1.50 BIOS, I've tested it with the latest (1.80) and on a X399 FPG platform, not Taichi. So the results may vary for others....

Ohwell....

-------------
-=SoniC=-
TR x1950, ASRock X399 FPG (v. 3.33A), G.Skill 3200 CL14 64GB, Enermax LiqTech 280,
AMD Vega 64 LC, 10x HDDs (mostly Hitachi), 2x M.2 (970EVO,960Pro), Xonar DX, AX1200i PSU

Posted By: JUANNY
Date Posted: 11 Nov 2017 at 1:09am

" rel="nofollow -

SoniC wrote:

Ok guys... and gals... and girls (in case there are any on this forum - drop me a PM ! :-P ).

Results:

None of the combinations was able to get me past 3066 boot (unstable), which I already archieved earlier on.

3200 is still unbootable.

Still stable is 2933 for the 64GB 4x DR Ripjaws V Sticks. These exact sticks run without a problem with full speed when put into a X370 X1800 Platform. So I know they are perfectly fine.

After so many hours spent setting up the BIOS I think my next challenge will be to set it up from scratch with my eyes closed.... LoL.

Anyway... let's hope that the complete rewrite of AGESA will bring some positive effects to those who have been unlucky...

P.S. Chew provides the settings for the 1.50 BIOS, I've tested it with the latest (1.80) and on a X399 FPG platform, not Taichi. So the results may vary for others....

Ohwell....

AMD definitely got their work cut out for them in terms of trying to get higher speed ram to work reliability in threadripper. I just ordered 64gb of flare x that I should get next week gskill F4-2933C14Q-64GFX https://www.newegg.com/Product/Product.aspx?Item=N82E16820232644 Going to see if this kit that is optimized for threadripper has potential to OC past 2933

-------------
Asrock X399 Fatality Pro Gaming Bios 2.0 Threadripper 1950X Cpu cooler Enermax 360 mm TR4 GSkill 64 GB F4-2933C14Q-64GFX Windows 10 X64 PRO Version 1709 Build 16299.19 2X EVGA 1080 TI in SLI mode

Posted By: datonyb
Date Posted: 11 Nov 2017 at 1:25am

shame to hear you didnt get any further sonic

i really had high hopes with your attention and the help of chew and 1usmus that you could be a frontrunner with the x399 regarding ram

it seems
the threadripper is in infancy much the same along its timeline as ryzen was with its own timeline

(you would assume lessons learnt from ryzen by both amd and the board partners would have assisted greatly with threadripper at launch........it appears not )

-------------
[url=https://valid.x86.fr/jpg250]

[/url]

3800X, powercolor reddevil vega64, gskill tridentz3866, taichix370, evga750watt gold