Tuesday, November 29th 2016

AMD Socket AM4 "Bristol Ridge" APU De-lidded

Here are some of the first pictures of an AMD socket AM4 APU being de-lidded. De-lidding is the process of removing the IHS (integrated heatspreader), the metal plate covering the CPU die. Some PC enthusiasts remove the IHS to improve heat-transfer between the CPU and extreme cooling solutions, such as LN2/dry-ice evaporators. Overclocker Nam Dae Won, with access to a couple of socket AM4 chips (most likely 7th generation A-series "Bristol Ridge" APUs), de-lidded the chips, revealing a large rectangular die. AMD is using high-quality TIM between the die and the IHS, which could either be solder or liquid metal. There's also a clear picture of the underside pin-grid of the AM4 chip, which has a central cutout that lacks any SMT components. Socket AM4 has 1,331 pins.
Sources: Nam Dae Won (Facebook page), PCGH
Add your own comment

24 Comments on AMD Socket AM4 "Bristol Ridge" APU De-lidded

#1
RejZoR
That certainly looks like liquid metal compound.
Posted on Reply
#2
ZoneDymo
See what I dont get is, if GPU's today all have an IHS-less design, why not CPU's as well?
Like why dont they standard come with something like this instead:
https://www.pcper.com/files/imagecache/article_max_width/review/2014-06-16/26-die-guard-top.jpg
Where the CPU is exposed for the best contact with the cooler.

Or atleast have some "extreme" edition be like that, it would make sense.

Also why do these still have actual pins pins and not those ermm dots that Intel uses for a while now, seems much more prone to damage with actual pins.
Posted on Reply
#3
INSTG8R
I'm just a little surprised the CPU is using pins still. I would have thought it would be more like Intel with the pins in the socket but I guess both ways work.
Posted on Reply
#4
TheLostSwede
At least you can straighten out a bent CPU pin, have you tried to do that with Intel's current sockets? I can tell you that it doesn't work, as I was given a board with a few bent pins and it was no fun at all.
Posted on Reply
#5
wolar
TheLostSwede said:
At least you can straighten out a bent CPU pin, have you tried to do that with Intel's current sockets? I can tell you that it doesn't work, as I was given a board with a few bent pins and it was no fun at all.
Depends really , my brother still has a duo quad , motherboard misses 3 pins(broken) for more than 3 years and it still works perfectly.
Posted on Reply
#6
Fouquin
ZoneDymo said:
See what I dont get is, if GPU's today all have an IHS-less design, why not CPU's as well?
Because AMD and Intel both found that by adding the IHS they reduced the number of dies shattered and chipped during installation by the consumer. The added benefit of having a chunk of metal to take some thermal load off the CPU in the event of a poorly attached heatsink, allowing the chip to not just burn up, was enough to make it an industry standard. An IHS also allows for LGA locking mechanisms to be much more robust, giving a strong metal plate to balance the mounting pressure across.

Only recently has the IHS been any kind of hindrance, what with Intel skimping on the internal TIM.
Posted on Reply
#7
RejZoR
ZoneDymo said:
See what I dont get is, if GPU's today all have an IHS-less design, why not CPU's as well?
Like why dont they standard come with something like this instead:
https://www.pcper.com/files/imagecache/article_max_width/review/2014-06-16/26-die-guard-top.jpg
Where the CPU is exposed for the best contact with the cooler.

Or atleast have some "extreme" edition be like that, it would make sense.

Also why do these still have actual pins pins and not those ermm dots that Intel uses for a while now, seems much more prone to damage with actual pins.
It's because ALL graphic cards come with heatsink pre-installed and with specifically tailored one for the particular card model. Basically, when you slam a cooler on a PCB, it'll have more fixing points than just around GPU. And even there, the interposer usually has edges raised so you can't really press too hard on GPU just from one side and potentially crack it.

CPU's on other hand are handle by butterfingered users who install it by themselves, there are bunch of different coolers with different contact plates (direct or distributed, flat or slightly curved etc) and different weights. Anyone who has worked with naked cores back in the days knows how fragile thy are and how quickly you can chip edges of the core away. When you're attaching 1kg of metal to it, it can such hard if you crack a 400€ CPU and make it useless. IHS prevents chipping the core, distributes the applied force and is in general a good all around thing for the masses.
Posted on Reply
#8
syncrod
TheLostSwede said:
At least you can straighten out a bent CPU pin, have you tried to do that with Intel's current sockets? I can tell you that it doesn't work, as I was given a board with a few bent pins and it was no fun at all.
its not fun but it is possible I've straightened 3 bent pins on my z87 after remove my 4770k for delidding.
Posted on Reply
#9
P4-630
The Way It's Meant to be Played
ZoneDymo said:
See what I dont get is, if GPU's today all have an IHS-less design, why not CPU's as well?
If they did, the CPU manufacturers wouldn't give any warranty since the die is very easy to damage.
Posted on Reply
#10
Sempron Guy
INSTG8R said:
I'm just a little surprised the CPU is using pins still. I would have thought it would be more like Intel with the pins in the socket but I guess both ways work.
Kudos to AMD. I'd imagine it be easier to straighten bent pins on the processor than on the cpu socket.
Posted on Reply
#11
Arrakis+9
RejZoR said:
That certainly looks like liquid metal compound.
That is indium solder just like Intel HEDT chips. Uses the same technique it looks like as well. der8auer did a pretty nice write up on this type of chip to IHS soldering technique, It's woth a read if your interested.
Posted on Reply
#12
RejZoR
Can you delid a soldered IHS? Only way I can think of is heating up the IHS to melt the solder, hoping you don't cook the CPU. Or was that epoxy compound which can't be heated up to release IHS...
Posted on Reply
#13
eidairaman1
The Exiled Airman
ZoneDymo said:
See what I dont get is, if GPU's today all have an IHS-less design, why not CPU's as well?
Like why dont they standard come with something like this instead:
https://www.pcper.com/files/imagecache/article_max_width/review/2014-06-16/26-die-guard-top.jpg
Where the CPU is exposed for the best contact with the cooler.

Or atleast have some "extreme" edition be like that, it would make sense.

Also why do these still have actual pins pins and not those ermm dots that Intel uses for a while now, seems much more prone to damage with actual pins.
They did during P3 and Athlon XP CPUs, the issue was End users and even oem installers were cracking the dies.
Posted on Reply
#14
RejZoR
I had my AXP 2400+ all chipped on the edges and I was remounting coolers carefully. Luckily it didn't affect it.
Posted on Reply
#15
eidairaman1
The Exiled Airman
RejZoR said:
I had my AXP 2400+ all chipped on the edges and I was remounting coolers carefully. Luckily it didn't affect it.
Yup. Hence why shims came out, yes i used them after killing a 3200 by accident...
Posted on Reply
#17
Arrakis+9
RejZoR said:
Can you delid a soldered IHS? Only way I can think of is heating up the IHS to melt the solder, hoping you don't cook the CPU. Or was that epoxy compound which can't be heated up to release IHS...
m1dg3t said:
Yep https://m.youtube.com/results?q=removing+soldered+IHS+from+cpu&sm=3
Its actually easier to use the vice method for this rather than applying heat. You have to apply 156c to the heat spreader to liqify indium metal vs just shearing it off with the vice method. Indium is actually a soft metal and a weaker joint than the bga/epoxy keeping the silicon on the pcb. From the pictures shown you can actually see the sheared indium on the IHS and the die.

Watch this video if your still unsure


I'm really excited to see what zen can actually do, it gets more and more promising with each snippet of info.
Posted on Reply
#18
INSTG8R
Arrakis+9 said:
That is indium solder just like Intel HEDT chips. Uses the same technique it looks like as well. der8auer did a pretty nice write up on this type of chip to IHS soldering technique, It's woth a read if your interested.
True enough I have a "wonky" P67 Sabertooth in a box because I just couldn't get that one "pin" straightened out right in the socket...
Posted on Reply
#19
eidairaman1
The Exiled Airman
INSTG8R said:
True enough I have a "wonky" P67 Sabertooth in a box because I just couldn't get that one "pin" straightened out right in the socket...
Easier to use a pencil on a PGA lol.
Posted on Reply
#20
m1dg3t
Arrakis+9 said:
Its actually easier to use the vice method for this rather than applying heat. You have to apply 156c to the heat spreader to liqify indium metal vs just shearing it off with the vice method. Indium is actually a soft metal and a weaker joint than the bga/epoxy keeping the silicon on the pcb. From the pictures shown you can actually see the sheared indium on the IHS and the die.

Watch this video if your still unsure


I'm really excited to see what zen can actually do, it gets more and more promising with each snippet of info.
I guess if you got the dosh for a 6950 you can swing 1 of these. I am, however, tighter than a jew on christmas LoL

Zen must be doing something right? OCable dual core w HT, mainstream hexcore and 32/64 on the server side from Intel now. I can't wait :D
Posted on Reply
#21
BeepBeep2
RejZoR said:
That certainly looks like liquid metal compound.
This was speculated on other news sites by writers and forum members.

- The die is soldered on just like every CPU since Athlon 64 (though it looks to be applied in two parts in this case).

- He killed the CPU in the process if I read his facebook translation right.

Nam Dae Won is "namegt" around the forums and HWBOT.

To reply to the user that said a vice method is better, that is impossible on AMD CPUs due to the SMT components on the top of the PCB which are the same height or taller than the die. Hint: If you cut past the bonding rubber, they break off easily.
Posted on Reply
#22
Jism
ZoneDymo said:
See what I dont get is, if GPU's today all have an IHS-less design, why not CPU's as well?
Like why dont they standard come with something like this instead:
https://www.pcper.com/files/imagecache/article_max_width/review/2014-06-16/26-die-guard-top.jpg
Where the CPU is exposed for the best contact with the cooler.

Or atleast have some "extreme" edition be like that, it would make sense.

Also why do these still have actual pins pins and not those ermm dots that Intel uses for a while now, seems much more prone to damage with actual pins.
AMD's Socket A Athlon, Duron had a IHS-less design:



It did had advantages on cooling but the weakest part is that cores could be crushed upon bad installation of the cooler. Another issue was that cores where able to burn up when the cooler was'nt properly attached. You can actually see on above picture that this core has slight damage on corners of chip.

The IHS protects the CPU from being damaged by improper installation. A question in the past was often which thermal compound, solder or whatever was sitting in between the IHS & core. Intel for example used poorly thermal compound which weared out over time. With this, i'm pretty sure AMD and users are able to cool the chip as best as possible since there's almost no loss in cooling capacity.
Posted on Reply
#23
Xajel
ZoneDymo said:
See what I dont get is, if GPU's today all have an IHS-less design, why not CPU's as well?
Like why dont they standard come with something like this instead:
https://www.pcper.com/files/imagecache/article_max_width/review/2014-06-16/26-die-guard-top.jpg
Where the CPU is exposed for the best contact with the cooler.

Or atleast have some "extreme" edition be like that, it would make sense.

Also why do these still have actual pins pins and not those ermm dots that Intel uses for a while now, seems much more prone to damage with actual pins.
Other guys already answered, basically.. because GPU's comes already with the cooler installed.. CPU's are not.. most peoples doesn't have the necessary experience to handle a IHS-less CPU design, a high chance of CPU die damage will occur resulting in mass angry/frustration...

And regarding LGA and PGA, well all have negatives and positives... I do have an LGA 1155 motherboard with i5 3570K, unfortunately I don't know how and when I saw two pins in the CPU sockets are bent... and It's not cool at all... I knew that from over 1.5 years now and finally knew why I couldn't run the RAM in Dual Channel, and why the CPU is not stable at any rise in clock speed when I was thinking I had a faulty mobo...

These things are damn small and very hard to figure out how they're bent actually to see how you can fix them, you can't actually "see" the wrong thing you just see different reflection pattern, you need a macro lens just to be able to see clear... I already have a macro lens but I still don't have time to check, you know I have to remove the motherboard from the case.. it's not just a removable CPU anymore...
Posted on Reply
#24
eidairaman1
The Exiled Airman
Jism said:
AMD's Socket A Athlon, Duron had a IHS-less design:



It did had advantages on cooling but the weakest part is that cores could be crushed upon bad installation of the cooler. Another issue was that cores where able to burn up when the cooler was'nt properly attached. You can actually see on above picture that this core has slight damage on corners of chip.

The IHS protects the CPU from being damaged by improper installation. A question in the past was often which thermal compound, solder or whatever was sitting in between the IHS & core. Intel for example used poorly thermal compound which weared out over time. With this, i'm pretty sure AMD and users are able to cool the chip as best as possible since there's almost no loss in cooling capacity.
Yup, killed a barton 3200 that way (adjusted clip tension too far and crushed the die )
Posted on Reply
Add your own comment