Tech YouTuber Highlights ASRock X870 Motherboard's "Killing" of His Ryzen 9 9950X CPU

[Chrispy_]

Hell NO.

visual Cancer omfg.

4Minutes 34 Seconds. That video guy really thinks that HWINFO, a junk windows Software is measurement.

Get a proper oscilloscope and probes. Let it calibrate properly. Think about the measurement setup and than come again.

Really any teletubbie can make a videochannel these days ans post junk for the masses. "Fake news".

4 Minute 58 SEconds. Windows operating system and hwinfo is not a measurement. I'll not explain why. Basics - look it up when interested. This was discussed too often and proven also.

--

Especially in this case the root cause determination for 8D-Report needs proper measurement. when someone does not know the difference in measurement it's even worse.


The 600.000 youtube subscribers and the usual contents he makes speaks volume. He should have sticked to his usual contents. Air bubble videos about hardware without deep dive. Note: I only checked youtube video preview - not the videos. // Such videos are kinda easy. Read a few reviews and than present it like Hardware unboxed, Linus tech tips and a few other low quality youtube channel do.

I dislike his channel, but HWINFO clearly shows a discrepancy between Asrock boards and everything else he tried. That's better than blind guesses and speculation at this point, and it needs others to confirm his findings before we can have any faith that SoC voltages are the true cause of melting AM5 chips.

Take from it what you will, but I don't personally think the AM5 socket is flexing on those Asrock boards. For a start, those backplates are pretty robust, even if they look cheap compared to the LOTES ones. I can't say for sure, but I wouldn't be surprised to learn than the Asrock black backplate was stiffer and stronger than the LOTES flat steel plate. Those ridges and rolled edges make a massive difference to how much you can bend something.

 

[jnv11]

Hell NO.

visual Cancer omfg.

4Minutes 34 Seconds. That video guy really thinks that HWINFO, a junk windows Software is measurement.

Get a proper oscilloscope and probes. Let it calibrate properly. Think about the measurement setup and than come again.

Really any teletubbie can make a videochannel these days ans post junk for the masses. "Fake news".

4 Minute 58 SEconds. Windows operating system and hwinfo is not a measurement. I'll not explain why. Basics - look it up when interested. This was discussed too often and proven also.

--

Especially in this case the root cause determination for 8D-Report needs proper measurement. when someone does not know the difference in measurement it's even worse.


The 600.000 youtube subscribers and the usual contents he makes speaks volume. He should have sticked to his usual contents. Air bubble videos about hardware without deep dive. Note: I only checked youtube video preview - not the videos. // Such videos are kinda easy. Read a few reviews and than present it like Hardware unboxed, Linus tech tips and a few other low quality youtube channel do.

Sure, the Tech YES City channel’s owner allegedly has some serious problems, but even flawed testing that can’t stand up to serious scrutiny or cross-examination in court could give us a potential clue as to what kind of more serious testing that could stand up to formal scrutiny to either confirm or debunk the flawed test is needed.

The video showed that HWINFO records that the CPU VDDR_SOC voltage when running a 9800X3D CPU ranging from 1.185 V to 1.265 V when the SOC/Uncore OC Mode is left in the default “Auto” setting on an ASRock X870E Taichi Lite that is running Fortnite. The SOC/Uncore OC Mode setting forces the SOC or uncore to run at maximum frequency at all times, potentially boosting performance at the cost of more heat and using more power. Other boards held the same voltage nearly constant in their default settings. When the SOC/Uncore OC Mode is set to “Enabled” on the same X870E Taichi Lite as before, that motherboard held the same voltage nearly constant with a recorded range of 1.185 V to 1.187 V. HWINFO recorded the Asus ROG X870E Crosshair Hero holding the same voltage at a constant 1.250 V at default settings. Other brands’ X870E boards like Gigabyte and MSI had lower CPU VDDR_SOC voltages reported by HWINFO, and those voltages are held constant or to a range of a few millivolts. The ASRock B850 Steel Legend also has similar behavior to the ASRock X870E Taichi Lite. For comparison, the MSI MAG X870E Tomahawk Wi-Fi reported a maximum CPU VDDR_SOC voltage of 1.200 V, the Gigabyte Aorus X870E Master reported a CPU VDDR_SOC voltage of 1.191 V and the Asus X870E might be pushing close to the edge with a CPU VDDR_SOC voltage of 1.250 V.

Without any sort of way to calibrate the board’s voltage measurement, there is no way to validate the reported voltages. Also, I don’t know how frequently the voltage measurements are taken, so I am wondering if the reported maximum SOC/uncore voltages on the ASRock motherboards are too low due to possibly not being able to catch transient voltage overshoot when the voltage has to be boosted when the SOC/uncore needs more voltage when going to a higher performance mode. A proper test with a well-maintained and calibrated oscilloscope that can measure and capture voltages much faster and more accurately than the motherboard’s built in voltage monitor should be run on these boards to see if there is voltage overshoot that is not being caught by the motherboard’s hardware monitor.

I initially hypothesized that overshoot is involved because I initially miscalculated that the the Asus X870E Crosshair Hero had a 6 millivolt difference with the ASRock X870E Taichi Lite when it comes to maximum voltages reported by HWINFO, possibly due to misreading one of the voltages before my initial miscalculation. The actual difference is 15 millivolts. I can’t tell if the measured voltages are accurate on both boards or not. If the measured voltages are accurate, the ASRock boards either are not catching voltage overshoot that is frying the CPU due to not sampling frequently enough to catch the overshoot or the ASRock motherboards' hardware monitors are catching the overshoot and 1.265 V is either at the absolute edge for or is too much voltage for some 9800X3Ds. If it is voltage overshoot that is frying the CPU, then forcing the SOC/uncore to maximum performance at all times would eliminate the need to suddenly raise the needed voltage when the SOC/uncore transitions from low performance modes to higher performance modes, eliminating the probable cause of the voltage overshoot.

EDIT: Grammar fixes

EDIT 2: Caught and fixed a math error on my part

EDIT 3: Realized the math error could force a conclusion change.

EDIT 4: More significant conclusion change.

EDIT 5: Adding comparison motherboard SOC voltages.

 

[jnv11]

Tech YES City reports that ASRock has released new BIOSesfirmwares that supposedly fixes the fried CPU issues. He is reporting that ASRock is not giving him a straight answer about who is at fault for the fried CPUs:

 

[alexbirdie]

Last month I changed my son's and my PC from AM4-3950x to AM5-9950x3d on an ASUS crosshair x870e hero. I flashed the newest BIOS ( 1401), and because I read before about the temperature-problems, I set SOC-voltage to 1.2 V. both systems are not overclocked ( for me that would not make any sense at all), CO is -22 for ccd1 and -17 for ccd2, and both systems are running cool and flawlessly since the upgrade. Hopefully there will not rise anothe temperature-issue in the future ( or another BIOS-bug perhaps).

 

[Caring1]

Tech YES City reports that ASRock has released new BIOSesfirmwares that supposedly fixes the fried CPU issues. He is reporting that ASRock is not giving him a straight answer about who is at fault for the fried CPUs:

That's NOT what he said at all, listen to the video.

 

[AVATARAT]

Tech YES City reports that ASRock has released new BIOSesfirmwares that supposedly fixes the fried CPU issues. He is reporting that ASRock is not giving him a straight answer about who is at fault for the fried CPUs:

In this video everything is unclear.

If we accept Asrock's (Tech City Yes) theory that the PBO is the problem because it does "high spikes", that's very interesting, because then why doesn't the overclocked by a clock generator via FCLK processor with a frequency boost to 6GHz, which uses really more voltages, die?
And all this when the reported problems are usually after using Expo, which add to vSoC and RAM voltages and shouldn't touch PBO voltages. To me that is a "very strange" explanation.

 

[de.das.dude]

View attachment 398730
Hmmmmm....

Charring means arcing.
Arcing means high voltage.
There are no HV sources in or around a CPU. However, when a loaded circuit is severed, the voltage rises intensely, causing arcing (and charring).

Given the evidence presented and my own experiences w/ AM5's socket flexing...
This CPU 'popped' out of place in its socket while under load; or, pin shifted slightly. Probably, a combination of PCB warpage / cooler torque and thermal expansion.

IMHO, start including a socket reinforcing bracket with CPUs, coolers, and boards as a norm.
It's pennies worth of material to help prevent RMAs and User Complaints. Who cares if most people would end up with more than one?

high current can also lead to arcing right?

 

[jnv11]

Gamers Nexus released a video about this issue. Steve Burke corroborates Tech YES City's claims that ASRock released BIOSes that supposedly fixes a PBO issue that ASRock believes is the root cause. Much like Tech YES City, Steve Burke does not believe that he has a straight answer. Steve thinks that there is an issue unrelated to PBO. Mr. Burke speculated about a possible factory contamination issue being another possible root cause. However, Steve does admit that he has higher priority investigations going on at the same time.