News Ryzen 9000 delid leads to first Zen 5 death — shared photos expose cracked dies and leftover solder

[Admin]

One of the world's first Ryzen 9000 delids went utterly wrong for extreme overclocker Tony Yu. The failed delid resulted in the I/O die of his Ryzen 9000 chip being cracked, making the chip completely unusable.

Ryzen 9000 delid leads to first Zen 5 death — shared photos expose cracked dies and leftover solder : Read more

 

[bit_user]

As this isn't the first CPU with a soldered IHS, I wonder whether there are somewhat reliable techniques for de-lidding them?

It seems to me that you'd basically have to heat the CPU to the melting point of the solder, while removing the IHS. That sounds rather tricky & more than a little dangerous.

I guess the other thing you might do is just grind off the IHS.

 

[hotaru251]

, I wonder whether there are somewhat reliable techniques for de-lidding them?

heating up the IHS would help as you dont "need" it to be melting point but anywhere hot enough to weaken its "grip" as you apply some force to ihs would help over trying to force it off cold.

 

[Notton]

I suppose the more painstaking method of sawing through the 8 dots of epoxy, and then heating up the IHS with a heatgun would give better results?

 

[drtweak]

Oh man. Miss the days of the P3 era Socket 370 CPU where the die was exposed. I remember taking my 600Mhz Celeron to 1 Ghz on a nice zalman all copper cooler with some Artic Silver. Ran better than the P3 in my main machine depending on the work load.

 

[SkyBill40]

I've never understood the point of this, which often results in situations like this. All for a very minimal performance tweak. For the average consumer, this is a bad idea and not worth any potential for gains as the risks far outweigh the reward.

 

[USAFRet]

I've never understood the point of this, which often results in situations like this. All for a very minimal performance tweak. For the average consumer, this is a bad idea and not worth any potential for gains as the risks far outweigh the reward.

For most people, yes, this is useless.

For those out on the edge, that do this as a hobby (or even making some money)...this is what they do.
Sometimes to the point of destruction, like here.

 

[bit_user]

I've never understood the point of this, which often results in situations like this. All for a very minimal performance tweak.

I think you're referring to when people remove the IHS, replace the TIM, and put it back on (usually without glue). That's the low risk/reward version and doesn't pay off like it did back in the Skylake era.

What you can't deny about removing the IHS is the benefits of doing direct-die waterblock cooling. It's a lot of trouble, but you certainly couldn't say it provides negligible benefits!

 

[SkyBill40]

I think you're referring to when people remove the IHS, replace the TIM, and put it back on (usually without glue). That's the low risk/reward version and doesn't pay off like it did back in the Skylake era.

What you can't deny about removing the IHS is the benefits of doing direct-die waterblock cooling. It's a lot of trouble, but you certainly couldn't say it provides negligible benefits!

Nah, I was referring to removing the IHS and had no consideration for those who choose to reapply TIM and reassemble. The opportunity for failure still remains and while the DD cooling you mention may have some significant benefits, is that still worth the risk for the average hobbyist that could end up destroying their CPU? I don't personally feel it is, but I'm not interested in trying it anyway. To each their own.

 

[USAFRet]

Nah, I was referring to removing the IHS and had no consideration for those who choose to reapply TIM and reassemble. The opportunity for failure still remains and while the DD cooling you mention may have some significant benefits, is that still worth the risk for the average hobbyist that could end up destroying their CPU? I don't personally feel it is, but I'm not interested in trying it anyway. To each their own.

Delidding your CPU is only slightly less extreme than adding nitrous oxide is in the car world.

Only for those out on the edge, and both can result in the death of your component.

 

[funguseater]

I miss the old days of whacking the CPU with a hammer, I havn't had to use my bench vise in years

 

[YSCCC]

For extreme overclockers, that's fine, you have to face risks when you are going to challenge the last mile

 

[TheHerald]

Nah, I was referring to removing the IHS and had no consideration for those who choose to reapply TIM and reassemble. The opportunity for failure still remains and while the DD cooling you mention may have some significant benefits, is that still worth the risk for the average hobbyist that could end up destroying their CPU? I don't personally feel it is, but I'm not interested in trying it anyway. To each their own.

It used to be worth it back when intel chips weren't soldered. Depending on how lucky / unlucky you were, you could see 20+ C drop by deliding. My particular 8700k had a horrible TIM job and I was constantly at thermal throttle temperatures pre delid. After the delid it was sitting at 70-75.

Nowadays it's just not worth it. CPUs are soldered and dies are huge, cooling modern cpus is kinda easy unless you are trying to push 300 watts for some reason.

 

[Vanderlindemedia]

I've never understood the point of this, which often results in situations like this. All for a very minimal performance tweak. For the average consumer, this is a bad idea and not worth any potential for gains as the risks far outweigh the reward.

AMD IHS's are quite large in height, resulting in higher temperatures then really needed. Ofcourse, it works, and no it's not a design flaw, just to keep coolers of AM4 compatible along with it and creating less e-waste.

However, IHS removal should be done properly. If cores start to crack the tin was not molten properly enough, or too much force was applied. The gains can be as high as 20 degree doing direct die cooling with liquid metal, and thus more headroom for higher clocks.

I've delid quite some CPU's in the past, but the ones with solder are kind of a waste of effort. You only do this now because of the IHS. But normally you could see quite large gains since a simple compound paste was used in between the IHS and core.

 

[eichwana]

I miss the old days of whacking the CPU with a hammer, I havn't had to use my bench vise in years

That’s how I delidded my 3770k, I used a bit of wood against the cpu and hammered that with the ish in the vice. So glad I didn’t drop it when it came off!

I used coolabatory Liquid Metal and got 10C drop in temps under all circumstances so it wasn’t a waste

 

[artk2219]

It seems like it was either too much heat, too fast, and there was a bit of vapor in the die causing a steam explosion. Or more likely, the opposite, too little heat, and too much force, breaking the die through physical stress. The result is the same either way, that is unfortunate, hopefully they finished all of their testing beforehand.

 

[hotaru251]

is that still worth the risk for the average hobbyist

your average hobbyist isnt delidding their cpu.

 

[adamXpeter]

Aww, a botched CPU tuning on the front page - nothing happens, everyone but the unlucky news editor is on summer holiday.

 

[SkyBill40]

your average hobbyist isnt delidding their cpu.

Fair enough.

 

[bit_user]

It seems like it was either too much heat, too fast, and there was a bit of vapor in the die causing a steam explosion.

Not sure how that would get there. Packaging should happen in a clean environment with humidity controls. Moreover, I'd expect the die will have gotten quite hot enough to evaporate any moisture, when the solder is initially applied.

 

[artk2219]

Not sure how that would get there. Packaging should happen in a clean environment with humidity controls. Moreover, I'd expect the die will have gotten quite hot enough to evaporate any moisture, when the solder is initially applied.

I agree, it's most likely physical stress without enough heat, it's just those were the two most likely scenarios.

 

[tumbbi]

This guy used Thermal Grizzly's delidding tool completely wrong, which caused the break. He moved an IHS once in the tool, then pulled the CPU out and used a thermal paste spatula to force the IHS out. You must move the IHS back and forth until it is loosened due to metal fatigue (no need to heat anything).

Please check Der8auer’s video:

View: https://youtu.be/jJzSlXe_aDA
.