|
Interesting. That's probably why the 140W CPUs overclock so well despite the higher TDP?
|
# ? Dec 20, 2016 19:29 |
|
|
# ? Apr 28, 2024 13:17 |
|
EdEddnEddy posted:What the main bummer is, is why is Intel's TIM working as if they are using some cheap bubble gum crap instead of the high quality TIM similar to even the stuff they have already pasted on their box coolers? That stuff on first application was on par with Arctic Silver or other high quality TIM so to loose 30C of efficiency just be delidding, just leads me to think it was a bad part (early run with a bit of a gap between the die/heatspreader, or some other weird defect (Too much/thick glue for the heatspreader base or something).
|
# ? Dec 20, 2016 19:33 |
|
EdEddnEddy posted:Also of note, the -E Chips all remained soldered as the dies were bigger (not sure about the Quad Core parts, but the 6+ cores were) It's always been down to the gap between IHS and die and not the TIM itself as far as I remember. Combat Pretzel posted:Interesting. That's probably why the 140W CPUs overclock so well despite the higher TDP? This is just a wild rear end guess on my part because I don't know the physics behind it, but they might be somewhat easier to cool thanks to the heat generation being more spread out than smaller die chips? I dunno!
|
# ? Dec 20, 2016 19:42 |
|
GRINDCORE MEGGIDO posted:Any posters in here delidded their Skylake chips? No but a friend of mine has. He asked me to 3D print the tool for him. Looks very but he said it worked well. https://www.youtube.com/watch?v=4o-rDGHZH4s I can ask him if you have any specific questions.
|
# ? Dec 20, 2016 19:55 |
|
Product segmentation is actually a good thing for customers. If a product ranges from $200 to $800 in price across all the different SKUs and the company wants to get $400 on average for each product, they can charge everyone $400 for the same product with all features enabled, or they can charge a range based on included features. If the company can create a SKU with certain features excluded and sell it for less than $400, customers who don't care about or wouldn't use the excluded features are better off, because from their perspective they've paid less for what is to them an identical product. Customers who want all the premium features would rather get it for $400 than $800, but there aren't as many of them. Here's an analogy: You're invited to dinner at a restaurant with 9 friends, and you order a $3 lemonade and a $8 salad. One or two people at the table order an appetizer, steak, two cocktails and a dessert, $60 on the menu. The total bill for the table is $400, and the restaurant has a policy that if multiple cards are being used, they can only split the bill evenly among the cards (these are all people who don't carry cash) How does paying $40 sound to the people who ordered $11 worth of food? What about for people who ordered $60 worth of food? Next week, the three people who ordered the cheapest stuff (in the $11-15 range, and thus overpaid the most) decline the invitation, and now 7 diners are splitting a $350 bill. Now the bill split evenly is $50 each, and the average that everyone is paying has gone up as well as reducing the restaurant's business. What used to be a good deal for the people ordering $60 worth of stuff has become less of a good deal, but the deal is still bad for people who ordered less than $50 on the menu. In this analogy, the restaurant and most of it's customers get a bad deal because of the restaurants stupid policy, and they're losing customers. The customers who leave are unhappy because the experience of the food and companionship was worth more to them than $11, but less than $40. The $150 Pentiums are the low margin, happy hour specials priced as street fighters against AMD chips in the same range. They would sell dramatically fewer chips to those customers if they sold only a single SKU at $400. If they sold that mono-SKU $400 chip to customers accustomed to spending $800 on those parts, they're not going to sell much more volume to those customers in order to make up the difference for the lost Pentium customers. They're pretty much saturated in volume, and are going to laugh all the way to the bank. (I'm going to ignore differences in manufacturing costs here too, because they're actually significant. Server product dies take up way more space on a wafer than desktop/mobile products. Also, if you have two particle defects on a wafer that would kill the die, it's a much bigger yield loss if you've killed two of the few dozen server dice on the wafer compared to killing two dice on a wafer with hundreds of dice) Segmentation is actually a really good thing.
|
# ? Dec 20, 2016 20:14 |
|
canyoneer posted:Product segmentation is actually a good thing for customers. If a product ranges from $200 to $800 in price across all the different SKUs and the company wants to get $400 on average for each product, they can charge everyone $400 for the same product with all features enabled, or they can charge a range based on included features. This is highly debatable economically. Without going into a long and boring discussion of surplus value, it's easy to see that the value of product segmentation depends where you sit on the spectrum of how much you're willing to pay. Product segmentation results in the top end paying more and the low end paying less. It is good if you're on the low end, bad if you're on the high end. For people in the middle, it can often be a bad thing because otherwise you'd get more features at the same price. It is a fairly natural outgrowth though of extremely high capital costs + very low individual unit production marginal costs (which describes both software and chip-making like Intel does).
|
# ? Dec 20, 2016 20:23 |
|
eames posted:No but a friend of mine has. He asked me to 3D print the tool for him. Looks very but he said it worked well. That's cool! is your friend going to reuse the heatspreader, or cool it direct die? I'd like to direct die cool mine using aircooling (and a shim), but not seeing that done much.
|
# ? Dec 20, 2016 20:40 |
|
evilweasel posted:This is highly debatable economically. It's not quite as simple as "the high end pays more, the low end pays less, so the high end kinda gets screwed for the low end's benefit" though. Due to the costs of manufacturing different physical chips and whatnot, there's a very real chance that the added costs from running multiple production lines would eat up most--if not all--of the theoretical price difference allowed by market segmentation in the first place. And different production lines would be effectively required, because there'd be no way Intel would be able to plunk down a i7 and say "ok, this is the one chip that everyone will now use" as, even with averaged prices, they would be far too expensive to achieve the market penetration into the low-end and mobile sectors where Intel wants to push things. The thing that drives more features at a price point (at any level) is competition. Which Intel basically doesn't have outside the low end. So in that sense segmentation is probably a net gain for almost everyone, since it allows for lower prices for the bulk of the product lineup, and they can charge basically arbitrary amounts for the upper-end products anyhow.
|
# ? Dec 20, 2016 20:42 |
|
GRINDCORE MEGGIDO posted:That's cool! is your friend going to reuse the heatspreader, or cool it direct die? If you're going through that much effort, you might as well water cool it. A mid-grade AIO can be had (especially on sale) for not much more than a solid air cooler these days.
|
# ? Dec 20, 2016 20:44 |
|
I don't like that product segmentation thrives off of marketing extra features to people who don't know whether or not said features will actually be useful to them in practice or everyday use. Especially in products as complicated as microprocessors vs. average Joe consumers with no IT or compsci background. I don't like marketing
|
# ? Dec 20, 2016 21:14 |
|
Sidesaddle Cavalry posted:I don't like that product segmentation thrives off of marketing extra features to people who don't know whether or not said features will actually be useful to them in practice or everyday use. Especially in products as complicated as microprocessors vs. average Joe consumers with no IT or compsci background. Yeah but if we were just marketing CPUs based off of "what does the average person know about a processor" what would that even be? The raw clock speed maybe.
|
# ? Dec 20, 2016 21:20 |
|
DrDork posted:In both cases you are artificially limited in what your physical object can do by people who want you to pay more money to enjoy the full experience. Is softSKU still a thing? I don't understand why folks are utterly convinced that every single i5 has perfectly functional $feature, and it's just binned off out of spite. There's probably a good chunk of them that have a broken $feature and that's precisely why it's being sold as an i5 in the first place. Is the core of the complaint "CPU marketing is far to simple and I want every 2^(chicken bit) SKU's in every line"?
|
# ? Dec 20, 2016 21:21 |
|
JawnV6 posted:Is the core of the complaint "CPU marketing is far to simple and I want every 2^(chicken bit) SKU's in every line"? I don't think anyone is honestly arguing that allowing for chip harvests into lower lines is bad, since it lowers prices for literally everyone. There are absolutely people annoyed that they can't get a CPU that hits some specific subset of desires not currently available (usually cheap + medium power + ECC/vt-d for servers that can't justify a full Xeon), but I'm pretty sure they're in the minority. I think no one would be bitching about segmentation at all if Intel had been putting out anything resembling compelling new desktop products since the 2500k.
|
# ? Dec 20, 2016 21:36 |
|
evilweasel posted:This is highly debatable economically. Without going into a long and boring discussion of surplus value, it's easy to see that the value of product segmentation depends where you sit on the spectrum of how much you're willing to pay. Product segmentation results in the top end paying more and the low end paying less. It is good if you're on the low end, bad if you're on the high end. For people in the middle, it can often be a bad thing because otherwise you'd get more features at the same price. Yeah. It works great in microprocessors because customers who feel like they're paying too much don't have very many other meaningful external substitutions. The low-end stuff is priced to be a good value against competitive offerings, and on the higher end the good stuff is just so much better than the competitor's stuff.
|
# ? Dec 20, 2016 21:45 |
|
DrDork posted:I don't think anyone is honestly arguing that allowing for chip harvests into lower lines is bad, since it lowers prices for literally everyone. There are absolutely people annoyed that they can't get a CPU that hits some specific subset of desires not currently available (usually cheap + medium power + ECC/vt-d for servers that can't justify a full Xeon), but I'm pretty sure they're in the minority. Enthusiasts complain about it in literally every market, including pricing strategies through unbundling. That's because they anchor their pricing expectations on the low end sku, but their desired features on a more deluxe one. Truck enthusiasts complain that they can't get the tow package on the lowest trim, and that they have to pay for a factory premium audio system that they're going to rip out and replace anyway. Phone enthusiasts complain that the cheapest phone in a lineup (starting at $499!) has a worthless 16gb of storage, and the manufacturer charges an extra $100 to spend an incremental $8 on flash to upgrade to 32gb. Guitar enthusiasts complain that having the factory upgrade the pickups in the instrument costs twice as much as buying them separate aftermarket and doing it themselves.
|
# ? Dec 20, 2016 22:00 |
|
GRINDCORE MEGGIDO posted:That's cool! is your friend going to reuse the heatspreader, or cool it direct die? If you end up doing it and buy/3d-print a delid tool, let me know if you want to sell the tool when you are finished. Maybe we can get a delid tool mail chain going.
|
# ? Dec 21, 2016 01:54 |
|
Would a 3D printed delidder be strong enough to withstand multiple deliddings? The typical filament used is pretty weak structurally, isn't it?
|
# ? Dec 21, 2016 02:06 |
|
DrDork posted:Tesla is really doubling down on this, with the S60 having a 75kWh battery, but the "stock" version being limited to only 60kWh unless you pay to unlock it. Same with the autopilot features: all the sensors are built in, but disabled until you pay to unlock them. Literally an over-the-air upgrade for them. Even if you never upgrade, you still benefit from having the larger battery installed because it delays range degradation.
|
# ? Dec 21, 2016 02:30 |
If someone is rich enough to afford the high end market segmentation, they generally are rich enough to where they don't give a poo poo. Now, if its something that costs little to nothing to add (money and time) then yeah its economically good but morally bad to segment it, but alot of things aren't. Hell, some times the high end stuff is subsidizing the low end as again, they are generally rich enough to not give a poo poo about an extra $#. I know for me, 20-30 is something I don't even blink at, 100 is minor hesitation, 1000 is serious thought, but for someone making minimum wage 20-30 is literally being able to get to work for the next week or the food for an entire week! Basically, you can think of it as relativity: What's another 50 when its already 800 vs 80? But seriously, I can guarantee that the raw materials going into each wafer are only on the order of like a grand, and with probably 350 good kaby lake dies per wafer you can see how cheap it is that way. However, given the ArF and KrF litho tools are loving expensive (1-2B alone for a fab) and that's not even mentioning the rest, $3-400 for a good enough die on a piece of precision made fiberglass is an ok deal. As the E/X stuff is 2.5 times larger, that increases the costs a fair amount as it takes just as much time, money, and effort to move a lot of 140mm2 dies as it does 350mm2 ones, and thats not even taking into account the fact that an increased die size increases the odds of a random die being defective (like what, 6 times as probable? or am I messing up my math?). Intel is kinda hurting too, so they are looking for ways of filling the coffers and funding the exponentially expensive nodes. Honestly, given the brick wall they are desperately speeding towards (which given their anemic foundry poo poo means it'll hurt worse than tsmc or samsung or whoever) Can you really blame them for trying to milk things for as long as they can? For the Tesla thing: yeah thats sketchy as poo, but remember they are basically a start-up and as they've learned with the suv, the more separate process lines they have the harder and more expensive things are. It would probably cost them more to change things up enough to put larger batteries and/or the sensors on only some of the cars they make (as thats multiplying the number of lines by 4 effectively). They probably could just have the one option, but I wouldn't be surprised if the cost to build the car is based around the higher price. Having software upgrades is a compromise as it allows them to sell to a larger population, and they have the ability to still get the full return on the vehicle. Watermelon Daiquiri fucked around with this message at 02:39 on Dec 21, 2016 |
|
# ? Dec 21, 2016 02:31 |
|
PerrineClostermann posted:Would a 3D printed delidder be strong enough to withstand multiple deliddings? The typical filament used is pretty weak structurally, isn't it? 3D printing has gotten a lot more "real" in the things that it can produce with a consumer/hobbyist printer that costs under a thousand dollars. Low end printers only do PLA, which is a biodegradable plastic made from starches and often used in recyclable plastic containers. Beefier printers can work with ABS filament, which is what most extruded plastic commercial goods are made out of -- including Lego bricks, and think of how solid those are when you step on them. You also need a better printer to work with ABS, since it requires hotter temperatures and a heated bed usually covered in mildly adhesive tape in order to not warp. Printed PLA is more brittle than ABS, and not usually something you want to make actual working tools out of. Printed ABS can be really strong, though, and survives a fair bit of mechanical stress, but it's not food-safe without sealing. While both PLA and ABS should be printed in a well-ventilated area, this goes doubly for ABS since it releases fumes that are potentially toxic if inhaled too much.
|
# ? Dec 21, 2016 02:37 |
|
Lowen SoDium posted:If you end up doing it and buy/3d-print a delid tool, let me know if you want to sell the tool when you are finished. Sounds good. I gotta do some research to find out if i can reuse my Noctua nhd14 with direct die or not. Also not sure if i can retain the stock board retention mechanism.
|
# ? Dec 21, 2016 02:56 |
|
Watermelon Daiquiri posted:Now, if its something that costs little to nothing to add (money and time) then yeah its economically good but morally bad to segment it, but alot of things aren't. Morality doesn't really factor into economics basically ever. The only companies who can/do worry about morality are either in such a dominant position that they don't have to care about the economic repercussions, or they're small shops who aren't exactly optimizing their economic output in the first place. Watermelon Daiquiri posted:For the Tesla thing: yeah thats sketchy as poo,.... My point actually wasn't that it was sketchy, but simply that this sort of thing does actually occur in other physical goods. As you point out, there are some pretty solid reasons a company might decide to go down that route. Cheaper in bulk and all that.
|
# ? Dec 21, 2016 02:56 |
|
Watermelon Daiquiri posted:But seriously, I can guarantee that the raw materials going into each wafer are only on the order of like a grand, and with probably 350 good kaby lake dies per wafer you can see how cheap it is that way. However, given the ArF and KrF litho tools are loving expensive (1-2B alone for a fab) and that's not even mentioning the rest, $3-400 for a good enough die on a piece of precision made fiberglass is an ok deal. As the E/X stuff is 2.5 times larger, that increases the costs a fair amount as it takes just as much time, money, and effort to move a lot of 140mm2 dies as it does 350mm2 ones, and thats not even taking into account the fact that an increased die size increases the odds of a random die being defective (like what, 6 times as probable? or am I messing up my math?). No way, material cost per wafer has to be O($100). I happen to know that plain 100mm electronics grade silicon wafers in small volumes are ~$20 each (this is actually incredibly amazing by the way--electronics grade silicon may be the purest material known to man, and yet it is so cheap). Intel, although they are buying much bigger wafers, probably are able to get a much better price/area than I could. The other materials on the wafer used to make the transistors and the wires are used in such small quantities, being very thin films on the wafer, that I'd be shocked if they greatly increased the cost over the wafer cost, even though now they are often oddball rare elements. The package material cost is probably greater than the die material cost. You are right in pointing out that the cost of manufacturing the wires, transistors, etc. is probably way more than the material cost. This article estimates that the loaded production cost (I'm not sure exactly what that means, I assume it has the capital cost of all of the lithography and the other manufacturing equipment baked into it) is $5k per wafer produced for the 14nm node. Watermelon Daiquiri posted:Intel is kinda hurting too, so they are looking for ways of filling the coffers and funding the exponentially expensive nodes. Honestly, given the brick wall they are desperately speeding towards (which given their anemic foundry poo poo means it'll hurt worse than tsmc or samsung or whoever) Can you really blame them for trying to milk things for as long as they can? Let's not get too silly here, Intel is reaping the full benefits of having a monopoly on server and PC chips. PC sales are dropping and development cost of new chip technology is greatly increasing, but it doesn't mean that you have to feel bad for the company.
|
# ? Dec 21, 2016 03:36 |
|
Why is everyone focusing on wafer cost and not fab plant cost + engineer salaries?
|
# ? Dec 21, 2016 04:04 |
|
The alternative to segmentation isn’t “everyone gets a Xeon for an i5 price”; it’s “everyone pays i5 prices for an i3 equivalent”. Those big margins on the high end drive development, which benefits everyone in the long term.
|
# ? Dec 21, 2016 04:10 |
|
thebigcow posted:Why is everyone focusing on wafer cost and not fab plant cost + engineer salaries? Once the fab plant is set up, it doesn't cost that much to operate (engineer salaries included). The vast majority of the cost is capital costs, developing the process node and setting up the fab. Then it's cheap to run it to whatever designed capacity. This is obviously not the world's largest chip - but since you get more chips per wafer you get better yields and it costs less to throw away a die. A circa-2009 Intel Atom processor that measures 22mm^2 costs about $6-8 per chip to produce and sells for up to $135 per chip. That's roughly normal. Capital costs are huge, production costs are tiny. Your average HEDT processor probably costs $200-250 to produce I'd think. Crappy chips get sold near cost as 6800Ks, then the server market and the "gamers with too much money" segment pay out the rear end for Xeons and 6950Xs. Paul MaudDib fucked around with this message at 04:17 on Dec 21, 2016 |
# ? Dec 21, 2016 04:14 |
I did include all that, although obliquely.silence_kit posted:No way, material cost per wafer has to be O($100). I happen to know that plain 100mm electronics grade silicon wafers in small volumes are ~$20 each (this is actually incredibly amazing by the way--electronics grade silicon may be the purest material known to man, and yet it is so cheap). Intel, although they are buying much bigger wafers, probably are able to get a much better price/area than I could. Yeah, the films and all are thin, but you have to consider the costs of things like delivery vehicles (as in silane, arsine, or N2 or noble atmo rather than truck), disposal costs, engineering wafers (every little process changes needs at least a cursory look at how it affected the device parms), cmp slurry, other cleaning stuff, wet etch chemicals, PR, not to mention the fact that a LOT of material gets wasted in some of the processes, spare parts (gotta replace those beam slits once they get worn down enough from all those plasmonic ions hitting it, among other things), plasma targets in particular can be $$$$, and a whole boatload of stuff I'm sure I'm forgetting. Granted, I am rounding up a bit (and when I say on the order of, I mean just around a grand. Not 1-10k), but consumables are one of the things that affects my bonus lol. And oh yeah, I'm well aware of Intels intentions, I was just trying to give a bit of their perspective. If they truly are going to ditch silicon for <10nm stuff, hooo boy will xeons cost a leg. You think those cheap silicon wafers are expensive... Watermelon Daiquiri fucked around with this message at 05:12 on Dec 21, 2016 |
|
# ? Dec 21, 2016 05:09 |
|
Watermelon Daiquiri posted:If they truly are going to ditch silicon for <10nm stuff, hooo boy will xeons cost a leg. You think those cheap silicon wafers are expensive... If they do it, they probably aren't going to switch away from the silicon wafer. They'll deposit/grow the new transistor channel materials on top of the silicon wafer. silence_kit fucked around with this message at 05:43 on Dec 21, 2016 |
# ? Dec 21, 2016 05:33 |
silence_kit posted:If they do it, they probably aren't going to switch away from the silicon wafer. They'll deposit/grow the new transistor channel materials on top of the silicon wafer. Oh duh. Yeah, they already use sige in there.
|
|
# ? Dec 21, 2016 06:19 |
|
Watermelon Daiquiri posted:Oh duh. Yeah, they already use sige in there. SiGe is in the source and drain of at least the p-type transistors in the latest integrated circuit manufacturing processes. It isn't the channel material. It is also in the base of bipolar transistors in certain, more specialty manufacturing processes, but that isn't really a technology for computer chips. Even if you aren't using the silicon in the transistor, they already know how to make the silicon wafers very flat, very pure, and at low cost. Silicon also has a much better thermal conductivity than other low cost non-silicon substrates like sapphire or quartz. I would be shocked if they actually changed the transistor channel material, but I'd be even more shocked if they changed the substrate material.
|
# ? Dec 21, 2016 18:37 |
silence_kit posted:SiGe is in the source and drain of at least the p-type transistors in the latest integrated circuit manufacturing processes. It isn't the channel material. Yeah, exactly.
|
|
# ? Dec 21, 2016 18:51 |
|
DrDork posted:If you're going through that much effort, you might as well water cool it. A mid-grade AIO can be had (especially on sale) for not much more than a solid air cooler these days. That's a great option, but just popping the IHS and fitting a shim isn't much bother, except I'm not sure if it's as easy as that with this NHD14 cooler. Would a Noctua NHD14 work on a delidded, shimmed, Skylake goons? If not, replacing the TIM and reusing the IHS is what I'll do. GRINDCORE MEGGIDO fucked around with this message at 00:27 on Dec 22, 2016 |
# ? Dec 22, 2016 00:15 |
|
I don't think there's any conceivable gain going from a a chip with the TIM replaced with liquid metal pro to direct die cooling. Especially not when you consider how necessary something in the form of the heatspreader is to even lock the chip in the socket correctly/position the heatsink at the correct height. There's pretty much only fraught dangers of loving your chip up vs just replacing the TIM.
|
# ? Dec 22, 2016 02:36 |
Real nerds use Mercury
|
|
# ? Dec 22, 2016 02:48 |
|
Well, he did say liquid metal, so...
|
# ? Dec 22, 2016 02:52 |
|
Watermelon Daiquiri posted:Real nerds use Mercury Check out my clock speeds now! This is insane!
|
# ? Dec 22, 2016 03:29 |
Well, it does have a thermal conductivity an order of magnitude higher
|
|
# ? Dec 22, 2016 03:58 |
|
Double Punctuation posted:Check out my clock speeds now! This is insane! "Mad Hatter Thermal Paste: The only TIM that requires you to sign a legal waiver before we ship it!"
|
# ? Dec 22, 2016 04:08 |
|
Intel's new Minamata Bay architecture.
|
# ? Dec 22, 2016 04:42 |
|
|
# ? Apr 28, 2024 13:17 |
|
PerrineClostermann posted:Would a 3D printed delidder be strong enough to withstand multiple deliddings? The typical filament used is pretty weak structurally, isn't it? One of the related vids is the same tool made with poo poo materials failing. They don't show the aftermath but the plastic of the tool gives before the lid pops.
|
# ? Dec 22, 2016 23:55 |