Intel Engineer on: Die Thinning, CPU Thermals and Liquid Metal
Vložit
- čas přidán 4. 06. 2024
- More Information about Seasonic Vertex-GX PSUs:
seasonic.com/vertex-gx
---------------------------------------------------------
Support me on Patreon:
/ der8auer
---------------------------------------------------------
Music / Credits:
Outro:
Dylan Sitts feat. HDBeenDope - For The Record (Dylan Sitts Remix)
---------------------------------------------------------
Paid content in this video:
- Seasonic Ad
- The video and content itself is not paid but Intel paid for my flight to the US
Samples used in this video:
- /
---------------------------------------------------------
Timestamps:
0:00 Intro
1:05 Die Thinning Part 1
2:58 Seasonic (Advertisement)
3:38 Die Thinning Part 2
8:01 Insights into manufacturing
12:18 Research, development, data collection
15:58 CPU Temperatures
17:34 Too much solder?
19:11 Thinning at home & liquid metal
25:26 Summary / Conclusion
26:25 Outro - Věda a technologie
This is the type of content the internet needs, more engineers and less PR people. Keep the good content coming der8aur!!!!
Its not just what the internet needs, the whole worlds needs it 😅
@@tessierrr True! Very true!
seriously, nobody should ever talk to PR people, their job is to lie.
@@kucing1087 public relations, aka pretending we care about customers when we clearly dont 😅
YES!
It's been a surprising change of methodology with Intel lately. Plant / manufacturing tours, engineers giving interviews and explanations if not how-to's, and just information in general being shared in a way that they'd NEVER been keen on previously. I love it! This kind of relationship with your buying public is exactly how you gain their trust and support. You're doing it right Intel! Didn't always, lol... used to be a VERY locked down company, one that used to share nearly nothing, with everything being confidential. This though.... this is good!
And when you have quality assets in people as you clearly do, allowing them to speak, showing just how qualified and intelligent your employees are. That also goes a long way into gleaming trust in your company.
If you look closely he didn't really reveal any information that wasn't already public knowledge. Basically the only thing you got out of here was that he said that they could make the cpu's better, but won't because it's less profit.
@@ShadowManceri I also only ask questions where I know that I can get a good answer on. That's why I also picked the 9900K as example. It's kind of old tech and it's no problem for Intel to talk about it. A lot of people would try to get information about upcoming or classified tech which would just get boring as we won't get any good reply.
@@ShadowManceri Right, I'm sure you can dig a lot of this stuff up, maybe even more than what he goes into. But it's a good look, and I definitely got some insight that I didn't have, nor had I the attention span nor desire to do that research in the first place. I've come away with a much more in depth understanding of the fundamental stuff going on and I think that was the point.
@@ShadowManceri you walked away with that the reason they don’t do it is less profit? Firstly of course that is a metric, be real if it was you in that position it wouldn’t be any different. But that wasn’t the only metric. Intel and any company that is trying to make a reliable product needs to understand the risks prior to implementing a change.
As he stated if they don’t get good data on a change they push it back to gather more data.
@@jagdeeprana7504 The more complex process you need to have, more it will cost you in machining time, labor and materials. It will ALL make it less profitable. And then you have the competition, you can't just charge ultra-engineered BOM from the customer without eating your profits to zero. It's a hard balance to make as good product as your competition while keeping it at price customers are willing to pay. You don't always afford to take those extra steps because it will eat your profits off. I'm not saying profit is bad. But let's not bullshit about it like it's some sort of forbidden word. 26:05 the guy says it himself. Cheers.
Easily one of my favorite industry interviews in recent memory.
Only one that I liked more was GN doing the EVGA wrap up.
I think it may have been the only one like it that i've ever seen. Like, that was just good engineering time. ANd it's not like they spilled any secrets that could give any of their competition a leg up. Just nice to know stuff for enthusiasts or people who are not "in the know" so to speak. As another commenter pointed out "there was nothing here you couldn't have learned online". Well sure but it's not an intel engineer sitting in an interview giving a masterclass in the very fundamentals of the process.
@newdeathscope the evga "not 4090" overclocking video w/ Vince, Biso Biso, and crew hit me hard in the feels which is also a top recent for me...
this guy was freaking awesome. great interview wish it was longer tbh
Video's like this and the one Steve from GN recently did with the cooler engineer of Nvidia really sets these channels apart. From the run of the mill "I do benchmarks so I'm a tech youtuber" channels that would never dare to do technical interviews like this. If you know your stuff you aren't afraid to talk with experts. Keep it up, and also kudo's to Intel for the behind the scenes access 👍
Agreed. I always enjoy the informative, behind-the-scenes dives into technicals. People often can be accurate in their speculations and explanations, but it's enlightening getting details from those involved with the design and manufacturing processes. PCWorld also does these types of interviews (some great ones with Dell, IceGiant, and Frore Systems, among others).
TechTechPotato (Dr. Ian Cutress) also does some good technical interviews, more on the architectural side.
I thought this engineer was cooler than the last...
der8auer should have a lot more subscribers, makes you wonder how many people who follow tech channels, are actually interested in tech. Some of the largest channels are the dodgiest.
I think these kinds of videos will also help attract more people to engineering careers. You get to do a lot of cool things that hard to imagine without exposure.
@@-opus That's the thing most ppls ain't really interested in tech from technical perspectives that's why linus is so big cuz ppls just want entertainment and watch someone do stupid shit that makes no sense , same thing why mythbusters were such huge hit show back in the days. Random stuff that costs tons of money and 99.99% ppls wouldn't try it anyway.
Der bauer should have a podcast where he invites engineers from intel, amd, nvidia and etc to talk a bit about what goes on behind the scenes. Would be really interesting.
true, but sadly, NDA 😔
As a materials science and engineering student, this interview was amazing. not only was it nice to finally understand the technical details (the electrical eng stuff is not my expertise lol), but Mike explained some maths-heavy concepts really well and his diagram and explanation of it were really easy to follow. You can tell you are both super interested in this topic!
Please do more interviews with engineers! Learning about the technology, manufacuring, limitations (cost, reliability) etc is so important for understanding why products are made they way they are and perform they way they do.
Very interesting. I like the idea that intel have a whole team for each tiny part of the CPU even down to soldering team. Amazing.
Everyone has when soldering optimization can make/save you millions every year.
Maybe there's too many teams and the lack of collaboration between the PCB Warpage Team, the Die Flatness Team, the IHS Underside Curvature Team and the Retention Socket Frame Team is why Rocketlake had that issue
thanks Mark!
great to see engineers chatting and not PR teams interfering. enough info for enuthist without giving away secrets. nice
Great content, you don't often get this in-depth discussion with actual engineers... and no PR/marketing team involved.
I never quite understood why the IHS made a return following the direct-die era of the late 90s/early 00s. I know it eases installation for OEM's and novices etc but it does seem to have been a consistent problem as far as cooling and as desktop chips seem to be reaching new record TDP's it seems like direct-die chip options could be something Intel/AMD could offer on certain product lines for enthusiasts or simply manufacturers that require the very best thermal performance.
Obviously they don't want to honour warranty replacements to users who damage chips during installation but that wouldn't be expected and was never offered back when stock direct-die chips were a thing either.
The main reason is the move to LGA socket. It requires quite high pressure to ensure contact. This pressure has to be evenly distributed and without some kind of metal cover this would be much more complicated without damaging the PCB. It's mainly this plus the fact that it protects the chip and is better for very cheap cooling solutions.
@@der8auer-en The IHS came to the late model Pentium 3s (tualatin), that was socket 370. Intel went through another two sockets (PGA 462 and 478) before switching to LGA. Also, at the time the die exposed chips (both AMD and Intel) were prone to the corners getting chipped off when the heatsink was installed and uneven pressure was applied.
@@aetch77 The Pentium 4 needed that IHS. It was so hot, a bit of a heatsink mistake was detrimental (but will not kill the CPU with the thermal protections)
The IHS helped spread the heat to a larger area - it was still 180nm or 90 nm (Prescott) so not that power dense. The Pentium IIIs were cooler running in comparison
@@ruikazane5123 An IHS is never needed outside what de8auer and aetch77 outlined. You are placing multiple interfaces between the die and the sinking mass, this is never thermally desirable in a passive solution. The area of the IHS is irrelevant, as heat is only conducted through the die upper surface and the heat sink/TIM would always fully interface with that conductive area.
I remember CPUs that had the heatink directly soldered to the socketable processor.
As someone who still plays on a 4790K (which has been heavily overclocked since day one), I greatly appreciate Intels research and testing team.
yeah u can overvolt overclock have 90c and those cpus dont die. Intel is very good in terms of quality of thier products. Sadly nvidia and amd gpus arent always that good long term
That was fascinating! More content like this please, it feeds that hunger for knowledge that so many of us engineering/physics overclockers simply can't satisfy any other way.
Excellent, Roman is experienced enough with the science and engineering to know what questions to ask and Mark is extremely intelligent and capable of simplifying the concepts to make it captivating. I loved this interview.
For people interested, what he mentions is the Arrhenius equation, and there are other models of Accelerated Failure Time. The whole field is interesting
Talking to engineers is always fun. Every conversation is not wasted time as you always learn new and amazing things.
After working for some years supporting a fab NXP/Freescale/Motorola I really enjoy content like this sitting down with engineers picking their brains.
This is pure gold, great interview!
Accelerated stability testing is super common in the pharmaceutical industry as well. When making a new drug, nobody knows the shelf life of either the active substance or the final product (e.g. tablets). So once they are reasonable confident the formulation development (e.g. tablet design + packaging) is over, they start accelerated stability testing of the batch intended for clinical trials. This has to be done to get estimate of what the drugs may look like 6 months or 1 year down the line, but you want to know within 3 months, so you can start the clinical trials as soon as possible.
These are done at higher temperature and humidity (usually 40 °C and 75% RH) with periodic sampling to confirm whether the product still conforms to specification, such as active substance content, impurities, appearance and many more. If the accelerated tests pass, their results can be carefully extrapolated to 2× the accelerated test duration. So if your accelerated test shows no degradation after 3 months, you can be reasonably certain the drug is stable for 6 months.
Of course, room temperature long-term stability testing is performed in parallel, because there's only so much the accelerated tests can do. So when you see shelf life indication on a drug box, its length was determined by long-term stability testing.
Amazing interview! I really like your content in the past 2 years or so. Especially visiting and interviewing experts (the electron microscope comes to mind) please do more of these if possible. I don't like videos often, as I use it as a playlist of stuff I want to show others or watch again. This has definitely earned a like :)
This was a great interview. It is nice to hear the prospective of the people actually building the product. Thank you for putting this together.
This was super interesting. Thanks for bringing us content like this and thanks to Mark for taking the time to sit down for the interview.
I can always appreciate direct answers from tech manufacturers to let us know their perspective on design characteristics. Great video!
Not only refreshing and educating, but also the conversation felt good throughout with Mark. Loved it! Hoping to see more in the future! :)
This type of more in depth engineer discussion is super interesting content. Props to intel for putting their engineers forward for this type of excellent content.
Please bring more content with engineers in the future if you can. It is really informative and gives a different but true perspective on many things which are otherwise only speculated and wrong in many instances!
This is just awesome! You learn so much with every episode. Thanks Roman!
I’m so glad this information is public, thank you for doing this!
A very interesting and friendly conversation between two big experts, what a pleasure to watch this video, thanks to both of you guys!
Absolutely incredible interview. Just superb. Thanks Mark!!!
Super cool interview and nice to hear it straight from the engineers!
I just bought a R16 with a i7 14700kf cpu and this put my mind at ease so much with temps. I've only ever seen mine get to the low 80s while gaming. Everywhere on the internet are people saying you need to be 60c max 70c for longevity and health. Ill listen to this engineer who went to school for this.
Thank you for bringing us this awesome technical interview
Love these interviews that give an inside look at the manufacturing process
Very nice interview. Would love to see more content featuring ppl who actually work in the field like this. Cheers!
Very insightful. I really appreciate the open discussion.
Great job! @der8auer EN Would be cool to get some more interviews with Mark in the future. Fun as well as educational to hear the insight and knowledge he has. Plenty more that people would love to hear from him if he's willing to do another interview in the future. Or just anyone else at all from Intel & AMD would be awesome. Longer interview would be even better, like an hour.
Absolutely loved this interview... More of this content or this type should I say... It's mind-blowing
Awesome content. Mark seems like such a great guy, and explained everything in a great manner. Very insightful!
I've been watching your videos for a long time, today i finally subscribed, please keep it up and hope for more videos like this "behind the scenes"
Really awesome detail. Love it when the engineers get a chance to speak directly -- nothing lost in translation.
Absolutely fascinating, and I'm not a overclocker. Just the design diagram and what is allowed on camera is very informative in my opinion and if the desire ever is needed and let's face it if chips keep getting hotter then new answers are going to be needed before they burn through the silicon and leave a hole. But yeah thanks very informative
So much more respect for the thought and work that goes into building a cpu
Wasn't really information I was interested in, but glad I watched this video. Really awesome info! Thanks!
Thanks a lot for this content, it's refreshing to hear something from so close to the source.
Awesome interview! Thank you for this one
That was actually the most informative thermal dynamics conversation I have ever heard. 👏🏾👏🏾👏🏾
It's crazy this dude came to the campus I work at! Super exciting to see people like Der8auer and Linus get videos of what happens inside the very labs I work daily in
It's so much more interesting when an engineer is answering questions, they actually answer! PR People are a whole load of "I'll get back to you" and "I don't know if I can disclose that" and tangents about stuff they want to market. If an engineer goes on a tangent it's still interesting because it's actually related to the question.
As someone who got into PC building and gaming over Covid, I didn't even know about the amount of interviews these channels made! This is so awesome!
Great interview. Thank you!
Really liked this video really interesting! Would love to see more like this thank you!
Really pleased with these engineering interviews with Intel. Good job, keep them coming!
Welcome to the US for your visit, as always a great video.
While I understand they can play around on the IHS to account the thickness variation of the die itself, so it fits on existing socket and cooling combo units, I wish they would offer a direct die product as they used to do back on the old days. As we've seen with custom kits, yours included, there are multiple benefits in using direct die cooling - with the drawback of less experienced users possibly cracking the die. Great discussion Roman.
Honestly with the way everything is going I'm suspecting this may become reality at some point. As density and clock speeds continue to increase, everything will get harder and harder to cool. Direct die options seem inevitable for enthusiasts but I don't see them offering it without other safeguards in place. They could potentially modify the mount and/or put something around the die to reduce the chance of damage.
At the end of the day, it'll just be another risk like damaging socket pins. If beginners know the warranty will likely not cover damage, they'll either pay to have it installed or buy a model with an IHS.
I wish they would sell direct die CPU’s along with ones with integrated ihs. I may direct die cool future Ryzen 7000 or whatever it is when I buy it, but I don’t like the idea of having to rip the ihs apart. I may just end up sanding it down to be thin.
There’s too much risk to buyers damaging the die. Just not worth the hassle for intel. Few people crack their die and post on forums and it gives intel bad press as a product you have to e super careful with. Not worth their reputation.
brilliant vid, thanks Roman & Mark
I love this stuff! Getting an inside view at the physics and what-not
Awesome interview, keep up the great work!
Awesome interview. 10:57 is so important--that's why I continually pick intel for my builds every 4-5 years. They just haven't had issues in my experience, even if they run hot. Love my 13900!
Loved this! And thanks Mark for sharing.
Awesome interview, more content like this would be great.
This was epic. Love these technical insights not dumbed down and filtered by PR types.
Excellent conversation and content. TY.
This is just amazing, two engineers talking in scientific terms.. LOVE IT!!
excellent conversation. So many people complain chips aren't progressing fast enough and forget armies of engineers work their butts off to make it happen.
Love these deepdives! So informative, thanks. And to Intel, this guy is gold, excellent explainer.
Incredible interview, thanks
It is so nice to see people that put passion into their work instead of talking heads focused on manipulating public opinion. This interview and from other channels the engineer from Nvidia and the interview with JonnyGuru are the recent ones that I've watched and remembered that actually technology used to be fun, but it's hard to see nowadays behind the clouds of PR, claims and official statements of personnel not involved with the actual engineering work.
Hope you enjoyed your visit to Oregon, Hillsboro off all places!
Awesome content and kudos to Intel for making it possible and to Roman for having the knowledge to ask interesting questions!
Thanks! It was an enjoyable conversation.
This is what I come to CZcams for! Love from Australia ❤
He did a great job explaining some of these concepts we heard about we didn't fully understand, especially, in my opinion, why CPU manufacturers don't just shave down their dies as much as possible on a technical level.
This was extremely interesting!
Thank you very much!
Great video, i'm not rely into direct-die cpu cooling, but it was filled with good info from somebody who actualy knows what he's doing.
👍👍
Amazing content, thanks for this video. I'm really interested on all of these ENG insides.
Such good content sir. Keep it coming, you're amazing.
This is another brilliant video, thanks very much Roman :)
❤ Beautiful interview!
Thanks Roman!
Great video, really really interesting.
Wonderful interview.
This was fantastic, I hope intel keeps doing things like this, I hope they appreciate the good will this generates. Great interview, super fascinating, thanks.
Such topics are much appreciated
Amazing interview!
Thought the thumbnail was Kingpin in a bad disguise.
Enjoyed that, had to watch twice but was worth it. Great insight.
Great interview, good job
All power to you Mark, thank you for your service!
Thanks; just helped me revise for my upcoming physics exam.
fantastic conversation. really enjoyed this
Very rich information
Thank you very much
Keep doing that 👍🏻
Great Great video, Thank you to bring this to us!!!!
Amazing video mate!
Great interview!
Fascinating, thanks for this
I am absolutely loving this!!
Good discussion. Thanks for the video, Roman. Greetings from Finland.
This is great
thank you for your work
Thanks guys, that was insightful.
This.. This right here. This is exactly what we need. I could listen to people like this talking all day long. Even if they mostly talked about the, now quite old Core i9 9900. I can understand why Intel doesn't want to talk too much in detail about the current line of products. And also, alot of these concepts are probably more or less the same even with todays parts.
Would love more content like this.
Did not know Weird Al Yankovic is also an intel engineer! Jokes aside this was very useful information. Great content. Keep it up man!
Interesting videos. Touched on some good subjects.