They are in a deep hole, and it is difficult to see a future where they can restore their former glory in the foreseeable future.
ARM isn't doing any such thing. Apple & Qualcomm are, though. ARM itself if anything looks weak. Their mobile cores have stagnated, their laptop attempts complete failures, and while there's hints of life in data center it also seems mediocre overall.
ARM aren’t trying to force Qualcomm to use ARMs cores. They’re trying to force them to update the licensing royalties to make up for the (as ARM sees it) licensing term violations of Nuvia who had a design license for specific use cases.
The part you reference (using ARM designs) is the fallback if Qualcomm lose their design license.
The destruction of the chips is standard practice to request in cases of license and IP infringement .
I'm not enough of a lawyer to figure out who is right.
That is the entire crux of the issue. ARM gave Nuvia a specific license, and then Nuvia was acquired which transferred the IP to a license that ARM did not extend it to.
And the fact that Qualcomm got just about everyone to endorse the acquisition ahead of announcing it but didn’t even tell Arm is a bit of a tell.
But even if they meant ARM-the-ISA, that'd still put it in a fragile position when the 2 clear market leaders in the ARM-the-ISA space have no particular allegiance to ARM-the-ISA (Apple having changed ISAs several times already, and QCOM both being active in RISC-V and also being sued by ARM-the-company)
It’s a really bad sign when a customer decides it can out innovate you by internally copying your entire business and production line.
Not necessarily, at scale, especially Apple's scale, vertical integration can make a ton of sense - for control, innovation, price, risk hedging.
It is the fact they can build a much better chip in almost any metric so far ahead of Intel is the red flag.
And historically, wasn't this juts an extension of their fight with Samsung in the mobile space more than a rejection of Intel?
I think you’re correct that this was initially because Apple didn’t want to be at the behest of Samsung. But as Apple iterated on their mobile and tablet chips, they had the choice to bring them to their desktop and laptop lines or not to. They took that gamble and here we are today with non Intel Macs.
It's pretty much two decades at this point.
I'm seeing the same thing now with Intel QAT / IAA / DSA. Only niche software support. Only AWS seems to have it and those "bare metal" machines don't even have local NVMe.
About 10 years ago Intel Research was publishing a lot of great research but no software for the users.
Contrast it with Nvidia and their amazing software stack and support for their hardware.
Linus seems to disagree https://m.youtube.com/watch?v=tQIdxbWhHSM
The video is 12 years old. A lot changed in the meantime.
AMD has open source drivers and crashes often. NVidia has (or more precisely had) closed source drivers that nearly always work.
Kernel Module Driver which is most likely less than 5% of drivers.
(I can't watch videos)
Every aspect of that document was just dripping in corporate dinosaur / MBA practices.
For example, they include 4 cores of these accelerators in most of their Xeons, but soft fuse them off unless you buy a license.
Nobody is going to buy that license. Okay, maybe one or two hyperscalers, but nobody else for certain.
It's ultra-important with a feature like this to make it available to everybody, so that software is written to utilise it. This includes the starving university student contributing to Postgres, not just some big-enterprise customer that merely buys their software!
They're doing the same stupid "gating" with AVX-512 as well, where it's physically included in desktop chips, but it is fused off so that server parts can be "differentiated".
Meanwhile AMD just makes one compute tile that has a uniform programming API across both desktop and server chips. This means that geeks tuning their software to run on their own PCs are inadvertently optimising them for AMD's server chips as well!
PS: Microsoft figured this out a while ago and they fixed some of their products like SQL Server. It now enables practically all features in all SKUs. Previously when only Enterprise Edition has certain programmability features nobody would use them because software vendors couldn't write software that customers couldn't install because they only had Standard Edition!
Intel started doing this kind of right recently with their Xe cores (ideally same programming model between their integrated GPUs and their datacenter HPC GPUs), but we’ll see how different the Xe-LPG and Xe-HPC end up being when Falcon Shores really comes out (I’m very worried about this description of it as a XPU, which seems like real confusion about what ML people want).
And this destroys 99% (maybe 99.99%) of the actual economic value for Intel! What Intel needs is for people to integrate these accelerators into software the way that AVX is integrated into software. Then software and libraries can advertise things like “decompression is 7x faster on Intel CPUs that support such-and-such”, and then customers will ask for Intel CPUs. And customers will ask their providers to please make sure they support such-and-such feature, etc.
But instead we have utterly inscrutable feature matrices, weird licensing conditions, and a near-complete lack of these features on development machines, and it’s no wonder that no one uses the features.
Which means that any fool that did utilise this feature has tied their cart to a horse that's going to be put down.
Smarter people can see this coming a mile off and not bother investing any effort into support.
It's so predictable that it's just depressing.
For example, could "Intel A" continue to own the foundries, split off "Intel B" as the owner of the product lines, and then do some rebranding shenanigans so that the CPUs are still branded as "Intel"?
I don’t know if it’s legally possible, but HP shows the branding bit can kinda work.
Splitting Intel is necessary but probably infeasible at this point in the game. Simple fact is that Intel Foundry Services has nothing to offer against the likes of TSMC and Samsung - perhaps only cheaper prices and even then it's unproven to fab any non-Intel chips. So the only way to keep it afloat is by continuing to fab Intel's own designs, until 18A node becomes viable/ready.
That means either he knew and allowed it to happen, which is bad, or he didn't know and allowed GPU division to squander the resources, which is even worse. Either way, it was an adventure Intel couldn't afford.
There is a similar story in other areas.
Intel deserves a lot of blame but they also got hit by some really shit circumstances outside of their control.
This only reinforces my previous point. He had good ideas, but couldn't execute.
On a side note getting people in russia write your drivers sounds a bit insane. Yea lower cost and probably ok quality, but the risks...
a chief executive officer, the highest-ranking person in a company or other institution, ultimately responsible for making managerial decisions.
Maybe you mean COO?
They chose to outsource the development of their core products to a country like Russia to save costs. How was that outside of their control? It's not like it was the most stable or reliable country to do business in even before 2022...
e.g. There are plenty of talented engineers in China as well but it would be severely idiotic for any western company to move their core R&D there. Same applied to Russia.
War in Afghanistan (2001–2021)
US intervention in Yemen (2002–present)
Iraq War (2003–2011)
US intervention in the War in North-West Pakistan (2004–2018)
Second US Intervention in the Somali Civil War (2007–present)
Operation Ocean Shield (2009–2016)
Intervention in Libya (2011)
Operation Observant Compass (2011–2017)
US military intervention in Niger (2013–2024)
US-led intervention in Iraq (2014–2021)
US intervention in the Syrian civil war (2014–present)
US intervention in Libya (2015–2019)
Operation Prosperity Guardian (2023–present)
Wars involving Russia in the 21st century: Second Chechen War (1999–2009)
Russo-Georgian War (2008)
Russo-Ukrainian War (2014–present)
Russian military intervention in the Syrian Civil War (2015–present)
Central African Republic Civil War (2018–present)
Mali War (2021–present)
Jihadist insurgency in Burkina Faso (2024–present)Also, I am not American and not an I conditional supporter of their foreign policy. And considering the trajectory of American politics it is obvious that any foreign multinational developing in the US should have contingency plans.
Putin's concentration of power has been alarming, but only since around 2012, to be honest. It was relatively stable between 2000 and 2012 in general (minus isolated cases of mysterious deaths and imprisonments). Russia was business-friendly back then, open to foreign investors, and most of Putin's authoritarian laws were yet to be issued. Most of the conflicts Russia was involved in were viewed as local conflicts in border areas (Chechen separatism, disputed Georgian territories, frozen East Ukrainian conflict, etc.). Only in 2022 did the Ukraine war escalate to its current scale, and few people really saw it coming (see: thousands of European/American businesses operating in Russia by 2022 without any issue)
So I kind of see why Intel didn't do much about it until 2022. In fact, they even built a second R&D center in 2020... (20 years after the first one).
i.e. if you are an American/European company and you are doing business in Russia you must account for the potential risks of suddenly. The sanctions after 2014 were a clear signal and Intel had years to take that into account.
> So I kind of see why Intel didn't do much about it until 2022.
I'm pretty sure that the consensus (based on pretty objective evidence) is that Intel was run by incompetent hacks prior to 2021 (and probably still is).
> thousands of European/American businesses operating in Russia by 2022
Selling your products there or having local manufacturing is not quite the same as outsourcing your R&D there due to obvious reasons...
Then again Yandex kind of pulled it off.
I will never understand this line of reasoning. Why would anyone expect an initial offering to match or best similar offerings from the industry leader? Isn't it understood that leadership requires several revisions to get right?
Intel had money and decades of integrated GPU experience. Any new entrant to the market must justify the value to the buyer. Intel didn't. He could sell them cheap to try to make a position in the market, though I think that would be a poor strategy (didn't have financials to make it work).
Honestly, even with their iGPU experience, Arc was a pretty impressive first dGPU since the i740. The pace of their driver improvement and their linux support have both been impressive. They've offered some niche features like https://en.wikipedia.org/wiki/Intel_Graphics_Technology#Grap... which Nvidia limits to their professional series.
I don't care if they have to do the development at a loss for half a dozen cycles, having a quality GPU is a requirement for any top-tier chip supplier these days. They should bite the bullet, attempt to recoup what they can in sales, but keep iterating toward larger wins.
I'm still upset with them for cancelling the larrabee uarch, as I think it would be ideal for many ML workloads. Who needs CUDA when it's just a few thousand x86 threads? I'm sure it looked unfavorable on some balance sheet, but it enabled unique workloads.
And here is the problem. You are discussing a dream scenario with unlimited money. This thread is about how CEO of Intel has retired/was kicked out (far more likely) for business failures.
In real world, Intel was in a bad shape (see margins, stock price ect) and couldn't afford to squander resources. Intel couldn't commit and thus it should adjust strategy. It didn't. Money was wasted that Intel couldn't afford to waste.
I brought up Intel's insane chiplet [non-]strategy elsewhere in the thread as an example where it's clear to me that Intel screwed up. AMD made one chiplet and binned it across their entire product spectrum. Intel made dozens of chiplets, sometimes mirror images of otherwise identical chiplets, which provides none of the yield and binning benefits of AMD's strategy. Having a GPU in house is a no-brainer, whatever the cost. Many other decisions going on at Intel were not. I don't know of another chip manufacturer that makes as many unique dies as Intel, or has as many SKUs. A dGPU is only two or three of those and opens up worlds of possibility across the product line.
Weren't they pretty good (price/performance) after Intel fixed the drivers during the first year or so after release? The real failure was taking so long to ship the next gen..
Sure Intel GPUs are inferior to both Nvidia and AMD flagship offerings, but they're competitive at a price-to-performance ratio. I'd argue for a 1st gen product, it was quite successful at opening up the market and enabling for cross-selling opportunities with its CPUs.
That all said, I suspect the original intent was to fabricate the GPUs on IFS instead of TSMC in order to soak up idle capacity. But plans changed along the way (for likely performance reasons) and added to the IFS's poor perception.
So with that, they are outsourcing production of these chips to TSMC and using nearly cutting edge processes (battlemage is being announced tomorrow and will use either TSMC 5 or 4), and the dies are pretty large. That means they are paying for dies the size of 3080s and retaling them at prices of 3060s.
RTX 3070 Ti: 17,400 million transistors
A770: 21,700 million transistors
https://www.techpowerup.com/gpu-specs/geforce-rtx-3070-ti.c3...
https://www.techpowerup.com/gpu-specs/arc-a770.c3914
It has taken Nvidia decades to figure out how to use transistors as efficiently as it does. It was unlikely for Intel to come close with their first discrete GPU in decades.
That said, it is possible that better drivers would increase A770 performance, although I suspect that reaching parity with the RTX 3070 Ti would be a fantasy. The RTX 3070 Ti has both more compute and more memory bandwidth. The only advantage the A770 has on its side is triple the L2 cache.
To make matters worse for Intel, I am told that games tend to use vendor specific extensions to improve shader performance and those extensions are of course not going to be available to Intel GPUs running the same game. I am under the impression that this is one of the reasons why DXVK cannot outperform the Direct3D native stack on Nvidia GPUs. The situation is basically what Intel did to AMD with its compiler and the MKL in reverse.
In specific, information in these extensions is here:
https://gpuopen.com/amd-gpu-services-ags-library/ https://developer.nvidia.com/rtx/path-tracing/nvapi/get-star...
Also, I vaguely recall that Doom Eternal used some AMD extension that was later incorporated into vulkan 1.1, but unless ID Software updated it, only AMD GPUs will be using that. I remember seeing AMD advertising the extension years ago, but I cannot find a reference when I search for it now. I believe the DXVK developers would know what it is if asked, as they are the ones that told me about it (as per my recollection).
Anyway, Intel entered the market with the cards stacked against it because of these extensions. On the bright side, it is possible for Intel to level the playing field by implementing the Vulkan extensions that its competitors use to get an edge, but that will not help it in Direct3D performance. I am not sure if it is possible for Intel to implement those as they are tied much more closely with their competitors’ drivers. That said, this is far from my area of expertise.
Wouldn't that just pretty much guarantee that the foundry business would fail since Intel wouldn't have any incentives to shift most of their manufacturing to TSMC? The same thing happened with AMD/Global Foundries..
Anyway, whatever woes GloFo is facing you can’t blame them on AMD. They had an exclusivity deal for a decade which only got broken when it was no longer tenable and AMD still buys a ton of their wafers. I suppose AMD may have bought more wafers if their products didn’t suck for most of that time but it had nothing to do with shifting production to TSMC which only happened after GloFo gave up.
They will have to focus, that means getting out lines of business which may likely die.
That would be better than going bankrupt and your competitors picking the pieces
I, personally, found my life to improve when we decided that the cleaning lady could be someone from outside the marriage.
He should have cut 25% of the workforce to get started (and killed the dividend).
Also - the European expansion and Ohio projects, while good short-term strategies, were too early.
Cut the ship down to size and force existing sites to refocus or be closed. Get alarmist. Make sure you cut out all the bad apples. Prune the tree. Be ruthless and determined.
They should hire Reed Hastings now. He's the OG turnaround king.
BK will go down in history as the person who destroyed a once great engineering firm.
An alien from Vega looking at our constellation of tech companies and their leadership might point at an obvious answer…