AMD Ryzen 7 7800X3D CPU Burns Up

[GoogalyMoogaly]

As I recall AM4 had a bunch of issues at release too (maybe not this bad) and now AM5 has.
Makes you appreciate the fact that Intel have to get things right (or closer) right out the gate because they don't stay on a socket/chipset as long as AMD so don't have as long to work out all the issues.

Like some other here I've just bought a new AMD setup and now I'm concerned, in part that the CPU may go pop but also that part of my reasoning behind it was that I could hopefully keep the motherboard for a while and just update the CPU, but now it's looking like it might be advisable to buy a new motherboard if they release new ones to fix these issues (whether that be revisions of current boards or the X770 & B750 boards).

 

[jigger]

As I recall AM4 had a bunch of issues at release too (maybe not this bad) and now AM5 has.
Makes you appreciate the fact that Intel have to get things right (or closer) right out the gate because they don't stay on a socket/chipset as long as AMD so don't have as long to work out all the issues.

Like some other here I've just bought a new AMD setup and now I'm concerned, in part that the CPU may go pop but also that part of my reasoning behind it was that I could hopefully keep the motherboard for a while and just update the CPU, but now it's looking like it might be advisable to buy a new motherboard if they release new ones to fix these issues (whether that be revisions of current boards or the X770 & B750 boards).

Zero issues with all but one AM4 systems and we’ve used many for years. The offending individual was an MSI board that run fine with 1000 series chips but refused to run a 200 series chip.

 

[Awahwah]

Maybe AMD will permit a en mass return if this cannot be fixed via BIOS. I suspect it can but will that effect and advertised speeds, be that RAM or CPU.

I’m watching this unfold and will not be powering my machine back on line until it’s known what the deal is.

 

[Harlequin]

Been reading that AMD tech docs were ignored and this is a motherboard vendor caused problem, primarily Asus (again, they actually have a history if this)

 

[Jamin280672]

ASUS definitely need to revise its boards. Shipping boards with non functional OCP is truly terrible.

Not non functional, as per GNs video, theyve just not fully utilised all the features, he said they were simply being lazy, im pretty sure this can be done with bios updates.

What a mess, and impressively so. Ironically said it was a mess earlier on and people were very quick to defend AMD - how do you feel about it now? xD

The fact that ASUS have rushed out so many BIOS versions too is making me pause on upgrading my own purely because I have to question how tested and reliable these updates are, especially given the lack of testing originally. At least it does for the most part still look to come down to SoC voltage, so with mine set manually at 1.2v I can't see any reason to upgrade for now, until the current/next BIOS have matured and been actually tested somewhat.

It doesnt need to be thoroughly tested now though, they know what the problems are, they just need to test those parts.

I’m so ****** off with this I would return this crap if I was still within 14 days return policy.
I don’t want to run soc at 1.3v as GM showed that actuall voltage can be higher by about 0.05v and who knows if not more.
Right now I’m running with manual soc voltage.
Started at 1.23v and had a reboot after few hours of use (it rebooted when idling)
then increased it to 1.24v and again had a reboot after a while.
I’m now at 1.25v.
Should have gone Intel after all.

Im surprised you are needing that much SOC voltage, im running 1.15v SOC, 1.25v MC voltage and 1.1v MISC, Ryzen 9 7900 and my ram specs are the same as yours, 6000mhz, but CL32, however im using buildzoids tightened ram timings, I ran memtest pro yesterday for 14 hours, no errors, 3000%.

 

[jigger]

Not non functional

Seems Asus OCP is non functioning to me. Not the first board to suffer from this type of issue either. Asus have been playing loose and fast with Intel and it’s Intel boards for years. Even pushing to the point of self destruction.

 

[z10m]

Not non functional, as per GNs video, theyve just not fully utilised all the features, he said they were simply being lazy, im pretty sure this can be done with bios updates.


It doesnt need to be thoroughly tested now though, they know what the problems are, they just need to test those parts.


Im surprised you are needing that much SOC voltage, im running 1.15v SOC, 1.25v MC voltage and 1.1v MISC, Ryzen 9 7900 and my ram specs are the same as yours, 6000mhz, but CL32, however im using buildzoids tightened ram timings, I ran memtest pro yesterday for 14 hours, no errors, 3000%.

I’m running cl30 XMP profile.

 

[Jamin280672]

Seems Asus OCP is non functioning to me. Not the first board to suffer from this type of issue either. Asus have been playing loose and fast with Intel and it’s Intel boards for years. Even pushing to the point of self destruction.

13.47 to 14.50 in the GN Video.

I’m running cl30 XMP profile.

CL30 isnt that much different to CL32, I wouldnt expect it to need that much more SOC voltage, and XMP would be way easier than buildzoids timings, for a start ive reduced TRAS from 96 to 28.



Going to flash the 1410 beta now, see what difference it makes with EXPO enabled to the default voltages.

 

[jigger]

13.47 to 14.50 in the GN Video.



CL30 isnt that much different to CL32, I wouldnt expect it to need that much more SOC voltage, and XMP would be way easier than buildzoids timings, for a start ive reduced TRFC from 96 to 28.

Seems non functional. Asus should revise its boards.

 

[delta0]

Seems non functional. Asus should revise its boards.

It’s not just ASUS and it’s being fixed with the bios updates.

 

[Gazanimal]

Updated my Asus ROG Strix X670E-A BIOS to the latest build (1303) before I put it all together.
Ran it initially without EXPO on the ram (Corsair 32GB DDR5 6000MHz C30) and saw not problem with SOC voltage. Enabled EXPO and seems the same.
Currently HWInfo is reporting CPU SOC as v1.288 max / CPU VDDCR_SOC Voltage as v1.244 max.
Only one I'm concerned with is CPU VDDIO / MC which is v1.447 max - but not sure if that's relevant to the issue although it is the integrated memory controller voltage.
Bit of a minefield that's for sure.

 

[jigger]

It’s not just ASUS and it’s being fixed with the bios updates.

Who else?

 

[jigger]

Updated my Asus ROG Strix X670E-A BIOS to the latest build (1303) before I put it all together.
Ran it initially without EXPO on the ram (Corsair 32GB DDR5 6000MHz C30) and saw not problem with SOC voltage. Enabled EXPO and seems the same.
Currently HWInfo is reporting CPU SOC as v1.288 max / CPU VDDCR_SOC Voltage as v1.244 max.
Only one I'm concerned with is CPU VDDIO / MC which is v1.447 max - but not sure if that's relevant to the issue although it is the integrated memory controller voltage.
Bit of a minefield that's for sure.

Those are voltages reported by Asus and those figures can’t be correct for this level of failure happen. Something(s)* truly terrible has to be going on.

 

[Andromeda1983]

And now I am glad I decided to skip this generation ( watched GN video ) and also I think i'll give ASUS a wide birth for a while until they stop being lazy. I know heating homes in the UK is hard right now but I didn't think ASUS wanted to make it easier

 

[Rawr]

Slightly worrying findings for any of us who were using it with 1.35v due to EXPO. Guess I'm glad I only had it at that for a few days before I dropped it to 1.25v. I'm not too worried it probably degraded in that time on my MSI motherboard as Wendell's findings on a different MSI board lines up with what I saw on my board, with it effectively adding about 0.014v from what it said in the bios when under a all core stress test. Which is a lot closer than the 0.5v difference that ASUS seems to be off by lol...

 

[Andromeda1983]

Rapid Unscheduled Disassembly - Asus " For those who dare "

 

[Jamin280672]

Who else?

Gigabyte, ASRock, MSI, all of them, watch the GN video, Gigabyte are mentioned right at the beginning of the video.

Just flashed the 1410 beta bios, which I dont really like doing, but hey ho, with EXPO enabled, the previous SOC voltage was 1.35v, now its 1.3 ish volts, ive backed it off to 1.2v manually set and also backed off the MC voltage as it really doesnt need 1.36v, ive set that to 1.25v, there is also an option in the Advanced menu for PROCHOT which defaults to Auto, but auto could be disabled, so ive enabled that.




This is with EXPO enabled:


This is EXPO disabled, hard set everything:

 

[jigger]

Gigabyte, ASRock, MSI, all of them, watch the GN video, Gigabyte are mentioned right at the beginning of the video.

Just flashed the 1410 beta bios, which I dont really like doing, but hey ho, with EXPO enabled, the previous SOC voltage was 1.35v, now its 1.3 ish volts, ive backed it off to 1.2v manually set and also backed off the MC voltage as it really doesnt need 1.36v, ive set that to 1.25v, there is also an option in the Advanced menu for PROCHOT which defaults to Auto, but auto could be disabled, so ive enabled that.




This is with EXPO enabled:


This is EXPO disabled, hard set everything:

I can’t believe all manufacturers have non functional OCP. Sorry chap but I’m not buying it.

I’ve made an attempt to watch the video, but it’s just a random guy that reviews stuff on video at the end of the day.

 

[Troezar]

I can’t believe all manufacturers have non functional OCP. Sorry chap but I’m not buying it.

I’ve made an attempt to watch the video, but it’s just a random guy that reviews stuff on video at the end of the day.

To be fair they have a lot of industry contacts. A bit more professional than "a random guy".

 

[Jamin280672]

I can’t believe all manufacturers have non functional OCP. Sorry chap but I’m not buying it.

Thats up to you, but its true, just requires a bit of Googling.

Heres one of the ASRock ones, theres another ASRock one floating around which is a video of the chip swollen up, (cant find the link right now, its in this thread somewhere), https://twitter.com/hms1193/status/1650879743947079680?ref_src=twsrc^tfw|twcamp^tweetembed|twterm^1650879743947079680|twgr^092e6c481927e11ff2967777267d2251e4f865a5|twcon^s1_&ref_url=https://forums.overclockers.co.uk/threads/amd-ryzen-7-7800x3d-cpu-burns-up.18971154/page-7

This was a 7700X, the only way they swell up like that and burn the pins in the socket is when they are still being fed high current and overheating after already being killed, failure of the OCP.

Der8auer 7900X too did the same thing, and shows the other ASRock chip damaged at 1min 10 seconds in his video, also shows the MSI boards damaging chips, and shows the chip from a Gigabyte X670 board which go so hot it de-soldered itself: https://youtu.be/arDqhxM8Wog

Thats why all board manufacturers are releasing fixes via bios updates.