FTA:
“In March 2025, Intel appointed Lip-Bu Tan as its new CEO,” Cotton wrote in the letter. “Mr. Tan reportedly controls dozens of Chinese companies and has a stake in hundreds of Chinese advanced-manufacturing and chip firms. At least eight of these companies reportedly have ties to the Chinese People’s Liberation Army.”
I don’t know about his investments, but one fact is clear: he was CEO of Cadence Design Systems, which has just pleaded guilty to federal charges for exporting technology to China. That alone should make him ineligible to lead a company with major government contracts.
If he resigns (and he will), the board should go with him.
Can you imagine the look you'd get if it was 1998 and you told me that AMD would have over twice the market capitalization of Intel in the next few decades? In 1998 Intel was 50x larger by market cap than AMD.
It is a company that has been catastrophically mismanaged.
To paraphrase, Intel has to go of the notion that for Intel to win AMD and TSMC have to lose. The strategy that follows from that might involve some painful choices.
It played out so smoothly over 3 months it seemed scripted from the start
That's not going to inspire people to step up.
Everything is legal for sure, lawyers were paid to agree. But it feels 100% like an orchestrated short of Intel. A GameStop situation.
I am fucking tired of 60+ year olds setting everyone else up for failure. They've been in charge as the problems came about. They're the problem then.
Boomers and GenX need to get out of the way or get stomped by youth.
Indeed. And his first action was to diss their own AI efforts. Because AI is just some niche area that they can ignore.
Just as Battlemage GPUs were getting decent reviews and sold above the MSRP.
Intel could totally try to capture the LLM market 'bottom up' if they wanted to.
As an underdog in the GPU market, all they need to do is start by making cheap boards with lots of VRAM. I'm talking 32GB boards under 1k.
They don't have to be fast. They just have to take these bigger models into VRAM and be fast enough that it's better than dealing with normal CPU+RAM.
That gets them into the market, and then they can follow up with more expensive and 'enterprisey' silicon that is faster for the data centers.
Alas, that's probably too alien a thought for Intel, as they prefer that thick margin...
1) How might Nvidia etc respond? They've made one-off SKUs for crypto, they could certainly respond quickly with a part that matched on memory but had much better software (meaning, more compatible with tools and better performance. AMD doesn't have the software, but their hardware is find and they could similarly up on-board memory. So Intel would really have to compete on price.
2) Ok, now we've found some 2nd or 3rd place success in a business built on logic fabbed at TSMC and DRAM from Samsung or Micron. If this is the future, why have fabs or any of the associated R&D?
I don't know what the right answers are but maintaining Intel at anything resembling its current size seems like a pretty tough puzzle.
https://newsroom.intel.com/corporate/lip-bu-tan-steps-in-the...
Why don’t you already understand this??? You are misleading everyone.
So AMD going from that to 2-3x intel's market cap is just... not quite as impressive as Apple's turnaround, but certainly in that direction.
Effective at what? Intel’s done, it’s on the Boeing trajectory, it’s just about grabbing what you can as the ship goes down.
Right — so that extends the time window for pillaging, and means that you can raid future taxpayer money, not just current assets. But there’s still no “turn around,” it’s just different flavors of getting paid massively to go down with the ship.
And not only did the Board not do their proper due diligence on Tan, but just let him toss Gelsinger's plan out? Shareholders should sue every one of them.
Frank Yeary, managing member Darwin capital advisors
James Goetz, partner Sequoia Capital
Andrea Goldsmith, dean of engineering Princeton University
Alyssa Henry, former square CEO
Eric Meurice, former CEO ASML
Barbara Novick, cofounder BlackRock
Steve Sanghi, CEO microchip
Gregory Smith, former CFO Boeing company
Stacy Smith, chair of Autodesk
Dion Weisler, former CEO HP
There are a lot of people that should not be on the board of a semiconductor company.
Investors have experience "managing" money to themselves. What is controversial about this? It's like saying pilots have experience flying planes. These investors have a lot experience financializing everything about a company.
We know where that leads, and it's not to success ... well, except for the investors with inside information that get out at the right time.
Autodesk turned into a rent seeking company over a decade ago.
Anyone Ex Boeing is sus, but maybe I'm wrong.
Ex HP CEO is also sus because HP has been slowly circling the drain for the last 20 years.
i share rhat to say, i think it's got positive connotations atm
"Has a stake in hundreds" is true of anyone who owns a global index fund.
In Soviet Russia you follow their rules and do everything/anything to win.
In US you follow the rules there and do everything/anything to win.
In China/India/country you follow the rules there and do everything/anything to win.
Through law, politics, advertising.
The ultimate goal is to win globally right?
A true capitalist leader can operate with complete lack of attachment for the sake of the corporation.
I don't debate his history at Cadence Design is concerning from a national security point of view, but the approach the administration took really shows how we're in a different era of politics.
Please let’s not sanewash what is happening right now.
https://www.politico.com/story/2009/03/gm-ceo-resigns-at-oba...
Sen. Warren:
https://www.pbs.org/newshour/economy/embroiled-scandal-wells...
That was my original "norm" I stated. What has gotten more specific about that?
And that happened as part of the government bailing GM out.
The president commands an enormous amount of power, and has an army of people who will do his bidding and simply adopt his opinions on any number of subjects. Shouting out to millions of his followers to state that the CEO of a private company is "CONFLICTED" and must resign is, by any definition, propaganda. Propaganda that changes the minds of the citizens of the country, riles up the base, and does nothing productive except to stoke anger and fear.
Working privately with this CEO, having a professional discussion with him, investigating the facts, determining that the best course of action for national security would be for him to step down, and maybe even putting some political pressure on that person to do so, and then publicly announcing the facts of what happened, is responsible governance.
It's genuinely an enormous difference.
It seems like youve had an agenda since before you asked the question
If the government has more info that’s even more reason to make this a matter of governance, and not twitter, IMO.
Genuine question, what do you think was the purpose of trump making that a public grievance instead of working on it directly with Tan?
The public pressure puts Tan on defense which gives Trump leverage in negotiating with him. Not sure what Tan/Intel will need to give up to address any potential conflicts but remains to be seen what happens.
If you genuinely don't see the problem here, I don't know what to say.
But you’re right, the Obama administration didn’t publicly bully him, just privately did so. I don’t see one method as better than the other, there’s a use for either but they both have the same goals.
Fortunate for Tan, a lot of people don’t like Trump and he’ll probably gain more public support and potentially stay on as CEO. This is certainly where this backfires on Trump as his method of publicly shooting from the hip doesn’t always work. Tan becomes the target of the day, forgotten about a week later. This is where Trump diplomacy could work better through private methods, but we also don’t know if that’s been tried. Quite honestly, I think he just had somebody whispering in his ear and he just decided to tweet it.
I'd be more concerned about non-public dealings that Trump might have learned from Roy Cohn, but these are probably off limits for discussion here. In general, what is on Truth Social does not matter.
The Wells Fargo CEO presided over a major scandal involving customers being signed up for services they never agreed to.
What has the Intel CEO presided over during his short tenure that measures up to those?
In that Constitution story, a government website that has the Constitution's text was updated in a peculiar way. It could be interpreted as having been related to habeas corpus rights, as that was in the middle of the removal. It could also be interpreted as unintentional, as the deletion started in the middle of Article I Section 8. You'd think a targeted deletion wouldn't include so much unrelated text. Then again, you could say that it's just an incompetently done targeted deletion. It's debatable! Maybe it was intentional and maybe the order came from the top. Or maybe it was just a run of the mill tech SNAFU.
In this situation, Trump, on Trump's social media platform, posted that he wants this CEO to resign. That's not debatable, it's verifiable fact. It happened. We know the man at the top is saying this.
So yeah, stop with the false equivalencies and pay attention to what's actually happening.
It's a good story, but what I'm remembering and is relevant here is this:
> At some point, I realized that if I wrote a wiki page and documented the things that we were willing to support, I could wait about six months and then it would be like it had always been there.
Authoritarians and fascists recognize this potential to create new "truths". If you say it enough, it's the truth. If you change things and say it's always been that way, then it was. If you're willing to drag through the mud, fire, prosecute, imprison, harm, or kill those who push back, fewer people push back. Even if everyone "knows" it's false, it no longer matters - most operate as if it is true.
Let's not forget that years later, Trump felt the need to inflict revenge on NOAA employees connected to sharpiegate: https://www.nbcnews.com/science/science-news/two-high-rankin...
That's why this shit makes people nervous, why people are on edge about information changing on government websites for no apparent reason. Trump has repeatedly shown a willingness to inflict his view of reality on others, with force, and without regard for facts.
You trivialize it at your own peril.
But we're going around in circles and should probably just let this go.
>Eh. Without getting anywhere near the merits of this particular fracas, the federal government has gotten deeply involved in critiquing the management of companies like Lockheed and Boeing, both for national security reasons and because of the importance of those companies to the economy. Easy to see Intel fitting into that mold in 2025.
Which was you saying "this Intel thing is NBD and basically the same as a bunch of past things presidents have done."
That's the original false equivalency I called you out on. And which you now disagree with apparently, because now you think the Intel thing is potential malfeasance? I don't think you really know what your position is anymore.
You keep saying it's the same old shit but when people point out the differences you change the subject. You're basically a troll.
Personally I'd say it's 99.99% in favor of run of the mill SNAFU. The 0.01% is mostly an allowance for the tendency of authoritarian systems to occasionally act in incredibly clumsy and incompetent ways. I feel this may be a factor of personal loyalty and ideological alignment receiving more consideration than competence in hiring and advancement decisions within authoritarian systems.
Asking a company that wants a nearly $17 billion bailout to make some changes is not at all unreasonable or unusual.
As far as I know Intel is not asking the government for anything special, so having the government commenting on their internal organization is unusual.
If it doesn't affect them directly, or they can't perceive how it will affect them directly, they simply do not care.
https://leger360.com/wp-content/uploads/2025/08/U.S-Politics...
Page 28
Here's a fun experiment. Take someone you know who's super passionate about a political figure, doesn't matter which one.
Then tell them "everything you think you know about this person has been told to you by someone else".
Not a sitting president and the NSA doesn't need a warrant for foreign targets.
That is correct. IIRC, FISA made that the law of the land since like the 1970s. However, Congress felt the need to provide retroactive immunity to the telcos who assisted in the FISA-violating wiretaps that the NSA demanded of them around the turn of the century. See Title II on printed page 32 of this [0] for more information, and check out newspaper coverage about the "FISA Amendments Act of 2008" around July, 2008.
This grant of retroactive immunity was particularly outrageous because it mooted in-progress civil suits against those telcos, which is not something that's supposed to be done at scale... especially for civil liberties violations.
That's a really odd thing to do if no law was violated, don't you think?
[0] <https://web.archive.org/web/20101207052813/http://frwebgate....>, found via following the chain of [1] -> [2] (because THOMAS is down today) -> [3]
[1] the July 9th, 2008 entry here: <https://www.eff.org/nsa-spying/timeline>
[2] <https://web.archive.org/web/20101209001911/http://thomas.loc...>
[3] The PDF here of version 4 of the bill, because archive.org doesn't have the text version archived. <https://web.archive.org/web/20101207012221/http://thomas.loc...>
https://apnews.com/article/business-china-asia-beijing-race-...
Trump in particular is essentially trying to make sure Intel lives despite market forces. It is effectively a quasi-nationalized entity akin to major military-industrial complex entities.
Given that, we are not talking about a random private entity. A US President making such statement about Intel is entirely justified.
This truly never happened before. Oh wait:
https://www.politico.com/story/2009/03/gm-ceo-resigns-at-oba...
GenX and Millennials watched too many "Sprite; you do you" commercials and grew up believing in their autonomy and freedom.
But American government was always at the wheel, via regulation, law, and drinks in DC off the clock.
It's all PR to say our system free market and the like. Intel wouldn't exist without falling in line over the decades.
This is why Tim Apple presented Dear Leader with a 24K gold award and so was rewarded by a tariff exclusion.
Intel was (and arguably still is) too large relative to its current technical capabilities. Yet even in this current “bad chips” era, Intel is only, at worst, about 10% behind in gaming performance (largely due to cache disparity) and is on par or better in most other workloads. From the K10 era until Zen 3, AMD processors were objectively worse (sometimes comically so) and AMD still managed to survive.
Intel’s mobile CPUs remain extremely competitive. Their integrated GPUs are the fastest in the x86 space. And their SoC+platform features: video decode/encode, NPUs, power management, wifi, and so on are the best in class for x86 CPUs; they are usually a solid second place or better regardless of architecture.
Subjectively, the most interesting “mainstream” laptops on the market are still, and historically have been, Intel-based. I understand that in an era where the M4 Max, Snapdragon 8 Elite, and Strix Halo each serve as best-in-class in different segments, “mainstream appeal” no longer equates to market dominance. And that is bad news for an Intel that historically just make a few CPUs (the rest being market segmented down versions of those chips), but still, to suggest they will disappear overnight seems... odd.
Intel can’t rely on the same. They haven’t been directly impacted by another larger company, they rely too much on a single technology that’s slowly fading from the spotlight, and they can’t compete against AMD on price.
Maybe if they ended up in a small and lean desperation position they could pivot and survive, but their current business model is a losing eventuality.
As I said, AMD survived by going into a lean pivot out of desperation. Intel has that opportunity as well, but the deck is stacked against them due to their size and over-reliance on specific IPs.
2) Ok, so there is expensive workstation available. It is a step forward I guess.
3) Call me when it is available and I can buy it in any normal computer shop.
Look, I hate the x86 architectur with a passion, having grown up with MS-DOS and the horrors of real mode. But the truth is that if I want to buy a computer right now, today, it is either a x86 PC or an Apple, and I have zero interest in Apple's closed ecosystem, so a PC it is.
3) PM me your number.
The fact is just that x86-CPU gets less and less important in the overall picture. It's not shrinking in total volume but in relative volume and fabs get exponentially more expensive. If Intel can make it in this generation they can't finance the next generation making only their own CPUs. They missed the big growth markets. They lost Apple as a customer. They have no GPU and no mobile chip that matters. They invested billions and billions into technology that went nowhere. Now they have to ask TSMC already to manufacture competitive CPUs. The only way this could change if Apple and/or NVidia massively start buying future capacity at intel. But why would they?
The current “bad chips that are only 10% behind” are fabbed by TSMC, not Intel.
Gaming is irrelevant.
For AMD, gaming (both console and PC combined) is less revenue than embedded-- things like those routers you can get off of aliexpress and Synology NASes.
Enterprise, cloud, and AI are the only things that matter, and even enterprise is falling off.
Back in 2020 with the second wave of AMD EPYC Rome, after I had gotten a couple of R7525s in hand and put them through their paces I started saying that you are professionally negligent if you, as a technology professional, recommend an Intel solution unless you have some very specific use cases (AVX512/Optane-optimized options). In 2022 everyone started agreeing with me.
Now you are professionally negligent if you recommend Intel at all.
Enterprise cares about speed, cloud cares about clients per socket, and AI cares about bandwidth. Intel is not competitive in any of those.
Even in the consumer space, for running bullshit workloads like Copilot on a laptop the difference is negligible. Intel is ahead, by about 10%-- at ONE HUNDRED AND SIXTY WATTS (if the OEM even allows it) while you trade that 10% for 75W on AMD.
No human being on earth cares that the scan to identify if there's a cute dog in the photo they just saved to disk takes .255 or .277 seconds. They do care about battery life.
And gaming isn't just irrelevant due to revenue, once you look at margins you start realizing that AMD could never again spend a single cent on marketing X3D chips to gamers and instead redirect that money to target other sectors and they would probably be better off for it.
Look at Nvidia. Gaming went from their cash crop to burdensome baggage in just a couple of years. Gaming went to less than 9% of revenue from like, 80%. They don't care about people buying an RTX card and having to deal with OEMs and distributors and retailers and marketing and RMAs and driver patches at whatever piss-poor margin it is due to everyone taking their cut when enterprise clients are putting in POs directly to them for tens of thousands of Data Center cards at a time at high margins-- and they didn't have to spend barely anything on marketing.
The very last thing, after figuring out absolutely everything else, that Intel should care about is what their chips are benchmarking at in the latest video game.
I had a friend who legitimately could not understand why Nvidia didn’t care about their reputation in the gaming market souring even after I showed him the numbers on how much nvidia is selling to corporations now.
I don’t know if it was an inability to deal with the numbers or if it’s just culture shock at going from being a valued client to as you said “baggage”, but it was a surprising number of people in that camp
Huge gaming demand and easy retail availability of nvidia's cards was providing economies of scale. If a few professors were buying the GeForce 8800 to look at this new 'CUDA' thing that was mostly a marketing thing.
Around the same time there were also one or two Playstation 3 clusters - but a year or two later Sony removed support for that. HPC being inconsequential, and a distraction from their core business, presumably.
It's only in recent years the stuff that used to be marketing decoration has become reality.
As a parallel, imagine hearing that the IPhone 13 was the biggest selling device in history. Then suddenly the IPhone 14 is $4000 and mostly sold to enterprise. It doesn't make any logical sense without following the money. Even then it may not make much sense.
Sooner or later, someone else will fill the need. That may be AMD, it could be Intel if they just focus for more than a year, or it'll be some cheap Chinese GPU from a company you've never heard of. (Likely named by mashing the keyboard while capslock is on.)
It's like how the mainframe market is bigger than it has ever been, despite being an irrelevant rounding error in the minds of the "Wintel" server providers, cloud vendors, etc...
>Look at Nvidia. Gaming went from their cash crop to burdensome baggage in just a couple of years.
Yes, marking up your consumer hardware by 4-5x to appeal to crypto miners surely does have an effect on your market. Arguably, AI saved them from a crash due to their over investment on Crypto/NFTs. It's not like gaming demand diminished this decade.
Gaming isn't THE way out. But it's one avenue to consider. It does seem like companies c. 2025 prefer to fall into the AI bubble, though.
If the hype dies and they're back to selling 5090s to gamers, can they afford to pay those bills?
But we know that won't be enough for shareholders and their stock would tank regardless. Because 2020's speciation isn't about having a reasonable long term portfolio. It's just extremely abundant pumping until you need to dump and pump the next trend. It's not enough these days to be a good, sensible business.
To be fair, their marketing was all that gaming related BS you listed, plus developing and maintaining CUDA. Winning gaming got them the winning AI architecture and mindshare around it, and CUDA got them the winning developer interface to it and locked in that mindshare.
Steam [1] tells me gamers use Intel by 59:41
Price-performance scatter plots [2] say although Intel isn't battling AMD for the >$1000 threadripper territory, they have some competitive products in the sub-$500 price band.
And while Intel missed out on the smartphone market, I've heard people comparing their N100 CPUs favourable to the latest Raspberry Pi hardware.
Sure, Intel has had major troubles with their next process node. And one of the best performing laptops is ARM-based. But Intel are nowhere near defeated.
[1] https://store.steampowered.com/hwsurvey/ [2] https://www.cpubenchmark.net/cpu_value_available.html#xy_sca...
It really might be as bad a mistake as not having Intel Isreal's futher development of Pentium 3 would have been. (in other words, no Pentium M, no Core 2 Duo, no Nehalem...)
Intel has a competitive iGPU in the low-power mobile space. Their iGPUs in general are also pretty solid for general desktop use. But even in the x86 space, AMD has better-performing iGPU options than anything Intel has ever offered.
I think the M4 is a fanless marvel of a chip and noticeably more interesting than the M4 Max. A fanless 6+2+10 configuration M5 with 128gb of ram would the most interesting thing in the mobile space.
But since we are splitting hairs, how good is an iGPU if you can't play most games? x86 -> windows or proton. One can't even run Linux, let alone proton, on an M4 (Asahi support stops at the M2).
The margins on their desktop products are also way down, their current desktop product isn't popular due to performance regressions in a number of areas relative to the previous generation (and not being competitive with AMD in general), and their previous generation products continue to suffer reliability problems.
And all this, while they're lighting billions of dollars on fire investing in building a foundry that has yet to attract a single significant customer.
Intel's not in a good spot right now.
They're minus signs. The AI is evolving.
This is why we can’t have nice things.
I'd bet 98% odds that Intel CEO and/or his comms team drafted this using an LLM trained on a competitor's chips.
Bearish. As if we even needed to know, at this point.
edit: Downvoters -- would you honestly take the other side of that bet?
Apple licenses that and develops their own chip, which is then manufactured by TSMC.
So I guess if Intel dies the US will still have a few good CPU design firms, but no manufacturing
Also note that Foxconn (China) assembles the iPhones
Eg https://www.businessinsider.com/apple-iphone-factory-foxconn...
What you’re talking about is an ARM IP license, which allow the company to build their own implementation of the standard. Only a few companies have those and, of those, even fewer actually use it. Apple is one of those that does.
Plus Britain and Japan are both somewhere between close allies and client states. Nobody cares if we license from them.
Apple is near unique only in that they’ve pretty much never used reference implementations (since the PA Semi acquisition, at least) from ARM and stick to their pure bespoke microarchitectures. But they’re not the only company that could.
But this is the country that the US wants (said as a born and bred US citizen) these are the results of it. Every CEO is kissing Trumps ass because that’s the only way you get ahead in the US now.
The media, the other two branches, colleges, tech companies etc have all bent a knee and bribed the President in one way or the other.
I’m sure it’s killing him as much as us.
I'm not. If someone were playing the pure numbers, I'll bet they'd have some exactly this. It's how you maximize profit under Trump.
Sad but not implausible
ASML manufactures the machines that TSMC uses to produce chips - they have an even more critical and irreplaceable role in chip production than Taiwan does. ASML is headquartered in Veldhoven, NL. That would absolutely affect chip production - no new nodes, no replacement parts. There are other critical technologies for semiconductor manufacturing made in USA as well.
Intel, even in its current weakened state, did nearly double the revenue of AMD last quarter.
I would have also unironically gotten a Core Ultra CPU if the pricing was actually... well, not insane for the value (or lack thereof) that they have. A 245K would still be an improvement over my current 5800X, though I have no idea what they were thinking of with that pricing - if their CPU prices were as competitive as their GPU prices (vendors ignoring MSRP be damned), that sidegrade release might have not been as horrible. They're still modern CPUs that work pretty well.
Though when I buy Intel, it's mostly so I don't give even more money to Nvidia and support competition in the market (otherwise my build would be all AMD).
What's the point of being Intel CEO if you give up?
He should resign.
Intel is just bending over for shareholders instead of doing actual engineering. A big reason the previous CEO got yeeted.
Andy Grove would turn in his grave.
A sort of corporate communications-whitewashed version of the My Cousin Vinny "Everything that guy just said is bullshit. Thank you."
> According to the U.S. Department of Justice, Cadence, in July, agreed to plead guilty to resolve charges that it violated export controls rules to sell hardware and software to China’s National University of Defense Technology, which is linked to the Chinese military. Tan was the CEO of Cadence when the company violated the rules between 2015 and 2021.
https://apnews.com/article/intel-trump-cotton-yeary-tan-2061...
Ashkenazi specifically are mixed at most, but still have middle eastern DNA. Other Jewish groups are nearly entirely middle eastern.
Pretty incredible how you can find a way to insert anti-Semitism into a post about the Intel CEO.
The only people who disagree are people engaging in a settler colonial political project, which is geopolitics, not fact.
1. They have excellent engineering resources - why didn't they just take their ARC cards and add more RAM to them? People are dropping $2K+ on the 5090 with 32GB and would surely pay $1200+ for an ARC with 32GB or even much more for one with 64GB or higher. Absolute performance wouldn't be the benchmark; being able to load larger models would make for excellent price/performance, for many lower-end uses.
2. We've been stuck at typical network speeds of 1GBit Ethernet for literally 20 years at this point. A first generation Opteron server like the Sun V20z (made by Newisys or Celestica, really) had dual-Gigabit interfaces; Intel should be pushing for 10Gb or higher as the bare minimum - and they make the 10Gb chips! More bandwidth capacity, even on the low end, will grow the computing market. And Intel has a big chunk of the market.
3. Intel did the same with their only offering dual-memory channel and thus much lower bandwidth of CPU <-> RAM ; unless you are buying an expensive server, you only get 2 channels to RAM; Apple increased their RAM bandwidth significantly and as it turns out, customers liked that, and bought more Apple CPUs.
Intel has to become "hungry" once more and stop their sedate, sclerotic ways. Maybe caring about their customers would help, too.
The moonshot efforts are around better Wi-Fi, which is, of course, at best a "good enough" solution that keeps people from running proper wires. But even as someone eager for hard-line networks, I wouldn't have good advice for a typical consumer.
If you run copper in your walls, you're really only good up to 10Gb and perhaps not even that. But if you want an optical-centric solution, that's an entirely new ecosystem that's a lot more complex. It's not just "buy a box of cable at the Home Depot and a crimp tool" anymore-- your devices might need 10GbE cards and SFP modules, you'll probably need some switches that still expose copper ports.
I wonder if there's a market for optical versions of the early "LAN in a box" kits that came with a couple of cheap ISA bus cards and a spool of cable-- just selling to people something that's all-inclusive and eliminates high-frustration mismatched parts.
This one also drives me bonkers, but my guess is that it doesn't capture that margin they want.
Intel seems to be kinda bad at starting from an 'underdog' standpoint in a market.
> Intel did the same with their only offering dual-memory channel and thus much lower bandwidth of CPU <-> RAM ; unless you are buying an expensive server, you only get 2 channels to RAM; Apple increased their RAM bandwidth significantly and as it turns out, customers liked that, and bought more Apple CPUs.
Apple has much tighter integration; i.e. they don't have sockets for the CPU or memory. What you buy is what you get unless you're brave enough to solder.
AMD provides a sort of in-between with Threadripper and it's pro variant, however it seems that they have a bit of a limitation on bandwidth based on CCDs for the low core count Pro models [0].
> Intel has to become "hungry" once more and stop their sedate, sclerotic ways. Maybe caring about their customers would help, too.
TBH I think whatever body (JEDEC I'd guess) needs to either make DDR6 128bit, or at the very least (and if even possible) work with memory controller folks to figure out a way to have a 'one stick, two channels' standard that simplifies board routing and keeps OEM usage simple.
It's really curious to me that over the last decade I've only dealt with one machine that only supported single channel, and the ones that 'were' single channel absolutely could have been dual channel but the OE could save a few bucks by going single channel due to the cost of two DIMMS vs one.
[0] - https://old.reddit.com/r/LocalLLaMA/comments/1mcrx23/psa_the...
That would require twice the number of pins and traces on each DIMM with corresponding extra pins on the CPU package as well and associated interconnects which would add heat as well. You end up with similar problems when you increase the number of channels per DIMM. CPU packages are strained as it is; I don't think the tradeoffs are worth it to increase system memory bandwidth on consumer systems.
Fair, I'll admit I wrote this after a very long work day and forgot to refresh myself on what a DIMM's pinout was actually like.
That said, Maybe instead of a DIMM it would be better to switch to a PGA or even LGA?
> and traces on each DIMM
Well, Traces are a problem regardless, at the same time it seems that Apple has managed to solve the problem by going to surface mounted chips. That said maybe(?) something like a PGA or LGA would allow for traces to be more clustered in making it a little easier to handle trace length differences...
> I don't think the tradeoffs are worth it to increase system memory bandwidth on consumer systems.
Well, I'd assume that for consumer systems you'd just have one 'channel', since at that point it's 128 bit and the equivalent of dual channel DDR5.
> CPU packages are strained as it is
Apple has managed to provide some fairly wide options in the M series though, so what are they doing that the others are not (aside from that they are effectively using more but smaller channels)?
They aren't using sockets. They are soldering it directly to the board. I figure that consumers aren't going to a buy a motherboard with a CPU and RAM soldered onto it. I could be wrong about that, but I wouldn't buy it.
They are also much more expensive. It's not that it can't be done since it is done with servers already -- it is that it can't be done for the price consumers want.
24GB and 48GB cards are supposed to be coming.
rwmj•6mo ago
rco8786•6mo ago
cosmicgadget•6mo ago