4 ms·

Vote me down to the 9th circle of hell, but I find the article is simply off with the fairies. What do boards, directors, companies, capital, investors or any o

Vote me down to the 9th circle of hell, but I find the article is simply off with the fairies. What do boards, directors, companies, capital, investors or any of the self-important suit-wearing dog and pony show have to do with what's happening to Intel? Intel is dying because it simply cannot bring itself to make a product that people want instead of the things it thinks people want [0].That's its fundamental life-preserving responsibility to itself. Sure, semiconductor making is very cut-off from the reality of what it makes tech for. But mindless progress for the sake of progress? It's a headless chicken running about the farmyard. The basics of business still apply. Make stuff people ask you for. [0] https://techrights.org/n/2024/12/04/Technology_rights_or_responsibilities_Part_IX.shtml https://techrights.org/n/2024/12/04/Technology_rights_or_res...

9cb14c1ec0 2y ago
> Intel is dying because it simply cannot bring itself to make a product that people want instead of the things it thinks people want [0]. Not sure that the link adequately makes its case. The number of people that actually care about Intel Management Engine is extremely small. Intel's real problem is that their products have struggled to be competitive in the marketplace on a price and performance level.
- hakfoo 2y ago
  I could see a case for "we'll sell a few narrow SKUs of non-ME products for an audience that demands it" -- assuming there are some people who will vote with their pockets for it. It's probably up there with some of the other corner-cases in their product matrix. But I agree, that's noise. There are much bigger problems with their product line: * Adding and removing features (especially AVX, but also the whole market-segmentation nonsense with ECC memory) * The marginal and in some cases negative generation-over-generation performance boosts. * The unlocked wattage situation which screams "we're pushing an end-of-the-rope product that we can only get to decent clocks by feeding it an entire mains circuit". (See: FX-9590, Pentium 4 Extreme Edition) Intel might be able to say "Today's Glysophate Lake CPU is clearly better than the Haswell you bought 10 years ago", but there will be a lot of mumbling if you ask "how much better is it than last year's Paraquat Lake". That doesn't inspire confidence, especially when they have traditionally been the premium brand.
- amluto 2y ago
  > The number of people that actually care about Intel Management Engine is extremely small. I think this sentence, in isolation, actually kind of nails the problem. I realize this isn’t what you and the article you’re replying to have in mind, but: no one cares enough about ME to buy it. After years of development, Intel has a solution to server management (and client management?) that is arguably technically superior to the crappy, power hungry chips from ASPEED and the management stacks from the likes of Dell and HPE that, until very recently, were unbelievably bad. But Intel completely failed to execute! ME should have been a headline feature, made friendly to the point where people would buy Intel systems because of ME. Anyone considering AMD or Ampere or POWER would be annoyed because their organization heavily used ME and Intel’s competitors didn’t have it! But instead, Intel hid the good parts (iAMT) behind annoying SKUs, obnoxious and secretive software and weird proprietary lock-ins. And the result is: no one cares about ME! (Intel has messed up most of its recent features in almost exactly the same way.)
  - BobbyTables2 2y ago
    Fully agree — always thought the same way. Saw AMT once about 15 years ago — available for free on an Intel banded desktop motherboard. Now I can’t even pay money to use it on my modern PCs…
  - nonrandomstring 2y ago
    This is a good point that I cannot disagree with. One more round in the magazine of the Intel footgun. Lack of openness and engagement with customers cuts both ways. The ME tech is great for a market segment of industrial datacentres. If those buyers chose to prioritise energy and management over security, then hurrah to whatever works for them. But as you say, Intel hid it away. Like a guilty secret. They somehow forgot to tell even their most Evil™ customers about the Evil Inside™. Which is jolly suspicious. And they've hidden away other "features" that trash security. It may be the number of people who "actually care about Intel Management Engine is extremely small", but those who do matter and they make decisions. Yes performance is important but security factors are going to be gaining more and more influence now. So Intel were told time and time again that people want legible, well documented products with no surprises, and they ignored that.
- throwawaythekey 2y ago
  > Intel's real problem is that their products have struggled to be competitive in the marketplace on a price and performance level. That's not super true. AMD has had good outlier performance in gaming with the X3Ds for a while, but for regular use something like a 13400f has been as good of a deal as any AMD chip. Arrow lake seems to have definitely made the situation worse for intel though.
  - HelloNurse 2y ago
    For most people. "regular use" (and "gaming") is the expectation that the CPU doesn't self-destruct. Regaining the faith of users will take years if the best Intel has to say is "this SKU is unaffected by this flaw"; less years if they openly switch to more trusted TSMC processes and/or ARM designs.
- BlueTemplar 2y ago
  If it was only IME... But AMD has their own version with the PSP. So we're stuck on Bulldozer hoping for Risc-V or something to mature enough before Bulldozer is completely obsolete. (With Apple obviously never having been an option to start with, and Microsoft keeping to push for the likes of TPM and Pluton, thankfully still optional for now.)
LarsDu88 2y ago
You do realize that it takes at least 5 years for new products to come to fruition in semi right? And Intel brought GPUs to market under Gelsinger's tenure. This isn't some software startup that just pivots from widget A to widget B or some ad campaign Even Tesla took like 5 years from announcement (after several years of developing a prototype BEFORE the announcement) to bringing the Cybertruck to market, and unlike cars, you need to mostly build the entire assembly line from scratch for new nodes every few years. It will be at least 3 years before we even see the outcome of decisions Gelsinger made at the beginning of his tenure.
- nonrandomstring 2y ago
  Big companies are not agile. On the other hand I studied VLSI at school and we designed a 4 bit microprocessor (team of 10 students) in a couple of months and had it fabbed and back on the test bench before the end of semester. And we had to listen to what the instructor told us, which is the failing I'm chewing Intel out for. Being obstreperous and uncooperative with the market, not slow.
- sounds 2y ago
  Grandparent said: > it simply cannot bring itself to make a product that people want To which you responded: > You do realize that it takes at least 5 years Intel could lower the price on their HEDT chips tomorrow. Intel could enable AVX-512 tomorrow. Intel could more fully open up their fabs to other companies tomorrow. Intel could publicly commit to producing silicon with no Management Engine tomorrow. What would you do, if you were Intel CEO?
  - spoaceman7777 2y ago
    > Intel could enable AVX-512 tomorrow. They can't just enable AVX-512 on their chips that include both performance and efficiency cores, because it's only present on the perf cores. > Intel could more fully open up their fabs to other companies tomorrow. Their fabs are open to other companies. they have multiple customers already signed on to make their chips on 18A/1.8nm process that enters mass production next year, including Amazon and the DoD. > Intel could publicly commit to producing silicon with no Management Engine tomorrow. Do real people actually care about the presence or absence of the IME to such a degree that it would be worth redesigning a bunch of their products and greatly increasing the number of individual SKUs? ... and, your first point "Intel could lower the price on their HEDT chips tomorrow". Well.. they could I suppose? They are pretty irrelevant when it comes to HEDT chips anyway though considering how crazy Threadrippers are. Would they really sell more of them if they knocked a couple hundred off their 2000/3000 lines? probably not. they're all under 5GHz anyway.
AtlasBarfed 2y ago
Because if the rot is this bad at the top, imagine what it's like below it. The board smacks of people that spend their careers making and defending their career positions, rather than actually doing something or achieving something. These are the top dogs of middle management Machiavellis. Underneath them is just a bunch of lackeys built on loyalty, not competence. The talent has been chased out. The talent that remains in engineering is likely just counting days to retirement or a buyout package.

9cb14c1ec0 2y ago

> Intel is dying because it simply cannot bring itself to make a product that people want instead of the things it thinks people want [0]. Not sure that the link adequately makes its case. The number of people that actually care about Intel Management Engine is extremely small. Intel's real problem is that their products have struggled to be competitive in the marketplace on a price and performance level.

hakfoo 2y ago
I could see a case for "we'll sell a few narrow SKUs of non-ME products for an audience that demands it" -- assuming there are some people who will vote with their pockets for it. It's probably up there with some of the other corner-cases in their product matrix. But I agree, that's noise. There are much bigger problems with their product line: * Adding and removing features (especially AVX, but also the whole market-segmentation nonsense with ECC memory) * The marginal and in some cases negative generation-over-generation performance boosts. * The unlocked wattage situation which screams "we're pushing an end-of-the-rope product that we can only get to decent clocks by feeding it an entire mains circuit". (See: FX-9590, Pentium 4 Extreme Edition) Intel might be able to say "Today's Glysophate Lake CPU is clearly better than the Haswell you bought 10 years ago", but there will be a lot of mumbling if you ask "how much better is it than last year's Paraquat Lake". That doesn't inspire confidence, especially when they have traditionally been the premium brand.
amluto 2y ago
> The number of people that actually care about Intel Management Engine is extremely small. I think this sentence, in isolation, actually kind of nails the problem. I realize this isn’t what you and the article you’re replying to have in mind, but: no one cares enough about ME to buy it. After years of development, Intel has a solution to server management (and client management?) that is arguably technically superior to the crappy, power hungry chips from ASPEED and the management stacks from the likes of Dell and HPE that, until very recently, were unbelievably bad. But Intel completely failed to execute! ME should have been a headline feature, made friendly to the point where people would buy Intel systems because of ME. Anyone considering AMD or Ampere or POWER would be annoyed because their organization heavily used ME and Intel’s competitors didn’t have it! But instead, Intel hid the good parts (iAMT) behind annoying SKUs, obnoxious and secretive software and weird proprietary lock-ins. And the result is: no one cares about ME! (Intel has messed up most of its recent features in almost exactly the same way.)
- BobbyTables2 2y ago
  Fully agree — always thought the same way. Saw AMT once about 15 years ago — available for free on an Intel banded desktop motherboard. Now I can’t even pay money to use it on my modern PCs…
- nonrandomstring 2y ago
  This is a good point that I cannot disagree with. One more round in the magazine of the Intel footgun. Lack of openness and engagement with customers cuts both ways. The ME tech is great for a market segment of industrial datacentres. If those buyers chose to prioritise energy and management over security, then hurrah to whatever works for them. But as you say, Intel hid it away. Like a guilty secret. They somehow forgot to tell even their most Evil™ customers about the Evil Inside™. Which is jolly suspicious. And they've hidden away other "features" that trash security. It may be the number of people who "actually care about Intel Management Engine is extremely small", but those who do matter and they make decisions. Yes performance is important but security factors are going to be gaining more and more influence now. So Intel were told time and time again that people want legible, well documented products with no surprises, and they ignored that.
throwawaythekey 2y ago
> Intel's real problem is that their products have struggled to be competitive in the marketplace on a price and performance level. That's not super true. AMD has had good outlier performance in gaming with the X3Ds for a while, but for regular use something like a 13400f has been as good of a deal as any AMD chip. Arrow lake seems to have definitely made the situation worse for intel though.
- HelloNurse 2y ago
  For most people. "regular use" (and "gaming") is the expectation that the CPU doesn't self-destruct. Regaining the faith of users will take years if the best Intel has to say is "this SKU is unaffected by this flaw"; less years if they openly switch to more trusted TSMC processes and/or ARM designs.
BlueTemplar 2y ago
If it was only IME... But AMD has their own version with the PSP. So we're stuck on Bulldozer hoping for Risc-V or something to mature enough before Bulldozer is completely obsolete. (With Apple obviously never having been an option to start with, and Microsoft keeping to push for the likes of TPM and Pluton, thankfully still optional for now.)

LarsDu88 2y ago

You do realize that it takes at least 5 years for new products to come to fruition in semi right? And Intel brought GPUs to market under Gelsinger's tenure. This isn't some software startup that just pivots from widget A to widget B or some ad campaign Even Tesla took like 5 years from announcement (after several years of developing a prototype BEFORE the announcement) to bringing the Cybertruck to market, and unlike cars, you need to mostly build the entire assembly line from scratch for new nodes every few years. It will be at least 3 years before we even see the outcome of decisions Gelsinger made at the beginning of his tenure.

nonrandomstring 2y ago
Big companies are not agile. On the other hand I studied VLSI at school and we designed a 4 bit microprocessor (team of 10 students) in a couple of months and had it fabbed and back on the test bench before the end of semester. And we had to listen to what the instructor told us, which is the failing I'm chewing Intel out for. Being obstreperous and uncooperative with the market, not slow.
sounds 2y ago
Grandparent said: > it simply cannot bring itself to make a product that people want To which you responded: > You do realize that it takes at least 5 years Intel could lower the price on their HEDT chips tomorrow. Intel could enable AVX-512 tomorrow. Intel could more fully open up their fabs to other companies tomorrow. Intel could publicly commit to producing silicon with no Management Engine tomorrow. What would you do, if you were Intel CEO?
- spoaceman7777 2y ago
  > Intel could enable AVX-512 tomorrow. They can't just enable AVX-512 on their chips that include both performance and efficiency cores, because it's only present on the perf cores. > Intel could more fully open up their fabs to other companies tomorrow. Their fabs are open to other companies. they have multiple customers already signed on to make their chips on 18A/1.8nm process that enters mass production next year, including Amazon and the DoD. > Intel could publicly commit to producing silicon with no Management Engine tomorrow. Do real people actually care about the presence or absence of the IME to such a degree that it would be worth redesigning a bunch of their products and greatly increasing the number of individual SKUs? ... and, your first point "Intel could lower the price on their HEDT chips tomorrow". Well.. they could I suppose? They are pretty irrelevant when it comes to HEDT chips anyway though considering how crazy Threadrippers are. Would they really sell more of them if they knocked a couple hundred off their 2000/3000 lines? probably not. they're all under 5GHz anyway.

AtlasBarfed 2y ago

Because if the rot is this bad at the top, imagine what it's like below it. The board smacks of people that spend their careers making and defending their career positions, rather than actually doing something or achieving something. These are the top dogs of middle management Machiavellis. Underneath them is just a bunch of lackeys built on loyalty, not competence. The talent has been chased out. The talent that remains in engineering is likely just counting days to retirement or a buyout package.