An EPYC Threadripper: Der8auer Gets EPYC CPU Working on X399 Motherboard

[Raevenlord]

So, maybe it isn't really working - but at least the system boots up all the way to the BIOS memory checks, where it then stops emitting life signs. Der8auer went through a sort of blind process to discover that there is a particular ID pin on EPYC that when covered, allows the CPU to be booted up by a X399 motherboard (in this case, an ASUS X399 Zenith Extreme). ID pins are nothing new, and basically tell sockets whether or not they should be powering up a particular CPU.

So what exactly does this mean? Nothing much - only that the sockets and pinouts are the same. The approach towards detecting the ID pin was a crude, brute force one, appending a piece of electrical tape to different parts of the CPU, narrowing down the search for a single pin. When this particular pin was covered, standby power finally kept on, and the motherboard ran through some initial boot steps until stopping at the D0 memory boot code. Der8auer thinks that a "simple" BIOS switch on this TR4 motherboard to an EPYC motherboard's BIOS would suffice to get the EPYC CPU running on this Threadripper motherboard. Check out the full video after the break.

View at TechPowerUp Main Site

 

[RejZoR]

That's cool. Not so much for AMD which probably isn't too happy about it

 

[TheLostSwede]

Funny definition of working.
If it doesn't boot, it doesn't work imho.

 

[TheDeeGee]

Such a disgusting practise from both Intel and AMD... tricking the user to buy a new Motherboard while that's not needed.

 

[R-T-B]

Such a disgusting practise from both Intel and AMD... tricking the user to buy a new Motherboard while that's not needed.

Judging from the result, it may still be needed in actuality.

 

[GreiverBlade]

Funny definition of working.
If it doesn't boot, it doesn't work imho.

it does boot .... funny definition of booting .... if it does not boot it does not even get past POST and D0 memory boot code is after that , so, it's a boot, albeit incomplete.

so yep, it's working .... up to a certain point where it block

Such a disgusting practise from both Intel and AMD... tricking the user to buy a new Motherboard while that's not needed.

seriously... AMD is not even on the level of intel .... (although it would be logical to have EPYC and Threadripper on the same platform just like the i7/9 and Xeon on 2066) look at the AM4 prevision compared to the LGA 1151 actual trick (6700/7700 vs 8700 ) or ... just AMD and Intel sockets/chipset history ...

oh wait .... 0.5/10

you have a Threadripper CPU you don't actually need a EPYC CPU nor a new motherboard ... (also if you have a Zen then yep you need a new mobo ... tho you wouldn't even need a Threadripper CPU up to this point .... thus you wouldn't need a new mobo ... you could just wait till Zen refresh/2.0, they are not tricking users ... the users are tricking themselves well at last the user who don't think a bit before deciding something )

 

[R-T-B]

it does boot .... funny definition of booting .... if it does not boot it does not even get past POST and D0 memory boot code is after that , so, it's a boot, albeit incomplete.

That's a post. A boot would imply at least a kernel is loaded.

 

[GreiverBlade]

That's a post. A boot would imply at least a kernel is loaded.

(incomplete boot is a boot) a boot code error imply a boot.

 

[R-T-B]

a boot code error imply a boot.

What exactly is it booting if that's the case?

A boot implies something booted. This is usually implied to mean something beyond firmware, at least as I was taught. The term comes from the idea of pulling one up by their bootstraps.

 

[jabbadap]

Pretty pointless practice. EPYC has 8 memory channels and TR has 4. It would only work in UMA mode, but the memory latency will be awful.

 

[Vya Domus]

Such a disgusting practise from both Intel and AMD... tricking the user to buy a new Motherboard while that's not needed.

Except AMD isn't doing that at all. There are server boards and consumer ones , there's a big difference. They are not forcing consumers to buy server grade hardware , that would just be retarded.

 

[T1beriu]

Classic clickbait title. Good job TPU.

Stuck at boot translates in not working.

 

[infrared]

Interesting that it got that far in the POST sequence though, would be hella cool if he can get it to actually boot.

 

[newtekie1]

I'm guessing it gets stuck at the memory point in the POST because Epyc connects all 8 memory controllers to the motherboard for 8 Channel memory. Since the X399 board only has connections for 4 memory channels, it won't POST past initializing the memory.

This might be something that can't be fixed, or it could be a simple BIOS hack will allow the board to initialize the memory controllers even though there isn't anything physically on the motherboard to use them. IDK

 

[erek]

What I would like to see attempted is the unlocking of the other two dies on the Threadripper chip by matching the components configurations of the EPYC on the outside in terms of the resistors, capacitors, etc. In his first comparison video between TR and EPYC he made note of the components configurations differences and it seems the TR has the components solder pads ready to accept them.

 

[Patriot]

Bootloader is the ahem OS kernel loader... booting is OS, POST is making it through the power on self test.
This does neither.

 

[ZoneDymo]

Such a disgusting practise from both Intel and AMD... tricking the user to buy a new Motherboard while that's not needed.

ermm I think you dont really understand what is going on here.
There are server boards and there are consumer boards, they are different things made for different tasks as are the cpu's made for them.
Both are equally new...

Its like being mad that you can't buy a Ford F-150 with the engine and gearbox of a Ford GT in it even though if you are handy it COULD fit.

 

[OSdevr]

What I would like to see attempted is the unlocking of the other two dies on the Threadripper chip by matching the components configurations of the EPYC on the outside in terms of the resistors, capacitors, etc. In his first comparison video between TR and EPYC he made note of the components configurations differences and it seems the TR has the components solder pads ready to accept them.

That may not be too difficult to test. Just put a Threadripper in an EPYC motherboard and use a modded BIOS.

 

[danbert2000]

This is no surprise as Threadripper is basically just an EPYC processor with two of the Ryzen modules deactivated. I see why AMD would restrict its usage, as power consumption is much higher on an EPYC and may lead to some melted motherboard VRMs without validation and maybe design changes. We should be cheering AMD's attempts at reducing R&D for halo prosumer CPUs, not complaining about the lack of support for their server parts in a cut down server platform. I don't think having a full EPYC processor compared to "just" a Threadripper would affect anyone in the market for non-enterprise grade prosumer hardware. The performance for EPYC-level threads is really only needed in an enterprise scale. All 100 people who would actually attempt to buy and use such an expensive CPU in an X399 board can just bite the bullet and get a beefier motherboard to go with their mountain of threads.

 

[eidairaman1]

Interesting that it got that far in the POST sequence though, would be hella cool if he can get it to actually boot.

Minor change to the bios code itself should help, course it be nice if a multiplier unlock could happen on epyc.

 

[Jism]

Just put 3 volts on that chip. It will work. Trust me.

 

[lexluthermiester]

That's cool. Not so much for AMD which probably isn't too happy about it

Why would they care? No one is going to go through the hassle of doing this kind of procedure except the rare hobbyists.

 

[RejZoR]

Why would they care? No one is going to go through the hassle of doing this kind of procedure except the rare hobbyists.

Because easy to understand tutorials and ready to flash BIOS files have never been used by more advanced normies... right?

 

[lexluthermiester]

Because easy to understand tutorials and ready to flash BIOS files have never been used by more advanced normies... right?

You can't seriously think more than a token amount of people are going to try this do you? Especially when buying an actual EPYC mobo is just too easy..

 

[RejZoR]

Same was said for BIOS flashing of pretty much everything and almost everyone does it these days...