News New Direct-Die Ryzen 7000 Water Block Reduces Temps by 25 Degrees

[Admin]

Der8auer is creating a new water block with direct-die cooling capabilities that achieves significantly lower CPU temperatures compared to standard AIO liquid coolers.

New Direct-Die Ryzen 7000 Water Block Reduces Temps by 25 Degrees : Read more

 

[-Fran-]

AMD, please go ask der8auer how not to suck at making an IHS, please? Ok, thanks, bye.

Yeah, yeah, I know. But come on... That IHS is horrendous. Like... I don't know what else to even say. I've been whining about it* since day 1 reviews came out and der8auer himself demonstrated how friggen bad it was for the CPU. If they created that IHS just to have vertical "buffer" for whatever stacking they wanna come up with, I have to say it was really toxic to make Zen4 early adopters pay for it in such a crappy way.

Seriously, AMD, re-consider that pig of an IHS as a viable solution and just make it better. I don't mind paying extra for an adapter on existing coolers to make them compatibles and I'm sure a lot of people would not mind.

Regards.

 

[gg83]

You gotta watch his video! It's really good. It's weird how a custom heatspreader can perform better than the stock IHS. The rgb looks really tight too. He also makes a nonrgb version

 

[gg83]

AMD, please go ask der8auer how not to suck at making an IHS, please? Ok, thanks, bye.

Yeah, yeah, I know. But come on... That IHS is horrendous. Like... I don't know what else to even say. I've been whining about it* since day 1 reviews came out and der8auer himself demonstrated how friggen bad it was for the CPU. If they created that IHS just to have vertical "buffer" for whatever stacking they wanna come up with, I have to say it was really toxic to make Zen4 early adopters pay for it in such a crappy way.

Seriously, AMD, re-consider that pig of an IHS as a viable solution and just make it better. I don't mind paying extra for an adapter on existing coolers to make them compatibles and I'm sure a lot of people would not mind.

Regards.

Exactly what I took away from his video.

 

[helper800]

AMD, please go ask der8auer how not to suck at making an IHS, please? Ok, thanks, bye.

Yeah, yeah, I know. But come on... That IHS is horrendous. Like... I don't know what else to even say. I've been whining about it* since day 1 reviews came out and der8auer himself demonstrated how friggen bad it was for the CPU. If they created that IHS just to have vertical "buffer" for whatever stacking they wanna come up with, I have to say it was really toxic to make Zen4 early adopters pay for it in such a crappy way.

Seriously, AMD, re-consider that pig of an IHS as a viable solution and just make it better. I don't mind paying extra for an adapter on existing coolers to make them compatibles and I'm sure a lot of people would not mind.

Regards.

I wonder if you cant just sand off 1-2mm of the useless IHS on an AM5 CPU and add spacers to the cooler mounts?

 

[-Fran-]

I wonder if you cant just sand off 1-2mm of the useless IHS on an AM5 CPU and add spacers to the cooler mounts?

You can and der8auer asked Jay2Cents to give it a try when he sent him a "testing" kit for what he wanted to sell for "sanding" the IHS of Ryzen CPUs.

Regards.

 

[bit_user]

I've been waiting for someone to do a direct-die water block for AM5!

The big question I have is: how much did it benefit performance???

 

[helper800]

I've been waiting for someone to do a direct-die water block for AM5!

The big question I have is: how much did it benefit performance???

Probably 25-50mghz per 5c or so.

 

[bit_user]

You gotta watch his video! It's really good.

The graphene thermal pads are neat. Where do you get those? Is that another one of his products?

I wonder if it makes sense to use those with a lapped heatspreader + lapped heatsink.

 

[Eximo]

Yes, he is selling those soon as well. I thought it an interesting compromise to improve the cooling to such a point that you don't need paste/liquid metal is good for a delidded solution. I've delidded one CPU, but was never brave enough for liquid metal, just used kryonaut which is still working some five years later.

Getting a CPU designed to peg 95C to run at 65C is just peace of mind to me. Never did like AMD's idea of that. Even just the heatspreader mod is useful.

Certainly look into it for myself for Zen 5 / Zen 5 3d. Depending on what Intel manages, quite curious.

 

[Makaveli]

really should put required delidding the cpu which will void the warranty. And the majority of users don't have the skills to do a proper delid.

 

[bit_user]

really should put required delidding the cpu which will void the warranty. And the majority of users don't have the skills to do a proper delid.

In the past, he's sold delidding kits which make it a lot safer and easier to do. Presumably, he'll sell a delidding kit for AM5, also.

 

[PEnns]

Very impressive and elegant solution.

I wish AMD (and even Intel) would watch this video and learn something new about cooling their CPUs!

 

[bit_user]

I wish AMD (and even Intel) would watch this video and learn something new about cooling their CPUs!

Did you see the part where his diamond-milled heatspreader costs 100 Euros? Also, its flatness presumes the heatsink mounted atop it is also perfectly flat. Making it slightly convex is the safer route.

AMD knows perfectly well how to make a heatspreader that isn't horrible. Their stated reason for not doing so was AM4 cooler compatibility, but I think there are also some unstated reasons. I have a fantasy that their plan is to use a vapor chamber, in Zen 5 CPUs.

 

[Deleted member 2838871]

Getting a CPU designed to peg 95C to run at 65C is just peace of mind to me. Never did like AMD's idea of that. Even just the heatspreader mod is useful.

I'm not finding the temps all that bad TBH... I've twice done a 30 min all core burn test and in both cases the temps never got higher than 88C... and that was with an air cooler.

really should put required delidding the cpu which will void the warranty. And the majority of users don't have the skills to do a proper delid.

I once delid an i7 7700k... paid like $30 for the tool which makes it idiot proof. I did the whole liquid metal thing and everything and it worked out really well... dropping temps like 20C.

I just don't see a need for the new 7950x3D.

 

[PEnns]

Did you see the part where his diamond-milled heatspreader costs 100 Euros? Also, its flatness presumes the heatsink mounted atop it is also perfectly flat. Making it slightly convex is the safer route.

AMD knows perfectly well how to make a heatspreader that isn't horrible. Their stated reason for not doing so was AM4 cooler compatibility, but I think there are also some unstated reasons. I have a fantasy that their plan is to use a vapor chamber, in Zen 5 CPUs.

Yes, I saw that it cost 100 euros....for a single item.

But companies buy such items by the millions, so you can imagine the discount they get.. And yes, a vapor chamber is definitely a great idea.

 

[TerryLaze]

Yes, I saw that it cost 100 euros....for a single item.

But companies buy such items by the millions, so you can imagine the discount they get.. And yes, a vapor chamber is definitely a great idea.

There is no company that could make millions of perfectly flat heatspreaders....at all, let alone for cheap. No company would agree to that, they always have a certain amount of deviance.

 

[Deleted member 2838871]

Watched a video last night where a guy killed his CPU by apparently tightening incorrectly putting too much pressure and cracked the die.

 

[bit_user]

There is no company that could make millions of perfectly flat heatspreaders....at all, let alone for cheap. No company would agree to that, they always have a certain amount of deviance.

AMD's headspreaders aren't so bad because they couldn't make them better. They had (seemingly bad) reasons for making them like that. It wouldn't take such an overkill solution as this to meaningfully improve AMD's ability to keep their dies cool.

Watched a video last night where a guy killed his CPU by apparently tightening incorrectly putting too much pressure and cracked the die.

Must've been a bad tolerance somewhere, because there's supposed to be a gap between the heatspreader and the die. That's why you need the liquid metal... or was he maybe using graphene?

 

[Deleted member 2838871]

Must've been a bad tolerance somewhere, because there's supposed to be a gap between the heatspreader and the die. That's why you need the liquid metal... or was he maybe using graphene?

View: https://youtu.be/WTF9ydHPucE

 

[Eximo]

I'm not finding the temps all that bad TBH... I've twice done a 30 min all core burn test and in both cases the temps never got higher than 88C... and that was with an air cooler.



I once delid an i7 7700k... paid like $30 for the tool which makes it idiot proof. I did the whole liquid metal thing and everything and it worked out really well... dropping temps like 20C.

I just don't see a need for the new 7950x3D.

7950X3D has some additional limits. The regular chips certainly hit 95C without high end AIO.

I just used a razor blade on my 7700k. That was back in the thermal paste era rather than solder, so not much effort needed other than cutting away the goop. I had one really hot core, and that was my target. Dropped the distance and used regular thermal paste again. I want to say it was like a 9C drop on the problem core and 3-5C on the others. But I was running at 5Ghz with a ludicrous voltage.

 

[Deleted member 2838871]

7950X3D has some additional limits. The regular chips certainly hit 95C without high end AIO.

I just used a razor blade on my 7700k. That was back in the thermal paste era rather than solder, so not much effort needed other than cutting away the goop. I had one really hot core, and that was my target. Dropped the distance and used regular thermal paste again. I want to say it was like a 9C drop on the problem core and 3-5C on the others. But I was running at 5Ghz with a ludicrous voltage.

Yeah I don't miss those Intel thermals. In addition to AM5 being new the thermals were the other reason I dropped Intel after 20 years. I'm getting comparable performance to the 13900k with much better power draw and temps.

 

[gg83]

The graphene thermal pads are neat. Where do you get those? Is that another one of his products?

I wonder if it makes sense to use those with a lapped heatspreader + lapped heatsink.

I think he mentioned thermal grizzly. I'm pretty sure he works closely with them. I feel like as long as ther surface is really flat than the graphene should be the best?

 

[bit_user]

I feel like as long as ther surface is really flat than the graphene should be the best?

He said it's about 15 degrees worse than liquid metal, if I recall correctly? But, he conceded that it's worth such a deficit, for not having to deal with the various issues that liquid metal brings.

I wonder how "good" that graphene really is. He made the point that the lattice is oriented in the Z direction (perpendicular to the sheet), in order to provide maximum thermal conductivity. However, maybe there are enough imperfections that it doesn't represent the theoretical performance graphene should be able to provide.