Hacker News Reader

Comment by dyauspitr | Hacker News Reader

Im still bullish on them at this price though. The terafab (already under construction) collab with Elon and 14A being the most powerful chip when they get it right has a lot of potential.

View on HN

I'm still salty at them for fumbling Optane, Gaudi, Altera and the Xeon Phi. the company seems like it's full of talented engineers working under terrible leadership.

It's a little odd they are completely missing the ddr5 gold rush when they are sitting on fabs.

Fabricating memory requires different techniques than logic, and it would take Intel several years to develop a commercially useful DRAM process.

Memory fabs are quite different from logic chip fabs.

The list of shit that Intel was within the ballpark of the right tech or 5 years mistimed from having a huge hit is ridiculous. Fuckers don't have good management, but also must have terrible luck to keep _just_ missing the mark?

No. I worked on p1274 and p1276. They didnt take years to hit profitability because of leadership. There was a decade of brain drain and we just could not overcome the technical challenges. But that is heresy to ears of techies. The fact that something is just too hard for the talent at hand is just never offered up. Intel engineers, to put to bluntly: suck. (Me included. I was a senior principal.) There just isn’t enough talent there because they don’t pay enough. I took voluntary retirement and went to work for a smaller instrument maker. I’m way happier.

> They didnt take years to hit profitability because of leadership.

Oh, it wasn't leadership?

> There just isn’t enough talent there because they don’t pay enough.

Ah, so it was leadership.

ex-Intel engineer here.

> Intel engineers, to put to bluntly: suck.

It hurts, but it is true.

> I took voluntary retirement and went to work for a smaller instrument maker. I’m way happier.

I was not so lucky. I would have loved the voluntary separation package. They did get around to doing straight large layoffs one year. My manager got laid off, I was not so lucky. I wanted the layoff package. I told him I'd happily trade places with him. He tried, but was denied. I ended up leaving anyway. Now, like you, working for a small company and much much happier.

me and a quarter of my company are looking up the barrel of a voluntary separation package. but I just moved to a team full of brilliant people working on a project that's a zen garden of algorithm design, so I won't go gently into that dark night.

I replied to another Intel engineer saying something similar: after seeing Panther Lake I just can’t believe that Intel has an engineering problem.

Intel engineers absolutely do not suck.

Even at small companies, there can be huge variance in capabilities across teams.

"Employee experience" varies, too. I've even heard from people who are very happy in some corner of Intel.

I hear you about the heresy, but are you sure you aren't being too hard on yourself? mismanagement can turn brilliant, passionate people into aimless zombies through demotivation and throwing you at problems that don't fit you.

also, my old boss thought Larabee was cool as hell when he got to play with it at MSR.

The "Larrabee New Instructions" were an instruction set much better designed than any other instruction set designed at Intel, presumably because for this they had brought in some people from outside Intel, with great experience in writing low-level optimized programs.

The hardware of the first Larrabee implementations was bad, being designed by some third-tier team in older technologies, without the resources used by the A-team that designed the Sandy Bridge CPU and its successors.

Nonetheless, after a couple of rebrandings the "Larrabee New Instructions" have become "AVX-512", and more recently "AVX10", and they are now the best part of the Intel-AMD instruction set.

Intel has tried to reserve the better instruction set for big spenders, but fortunately AMD has forced them to unify again their ISA, so starting next year, when Intel should launch the "Nova Lake" CPUs, Intel will have for the first time really competitive CPUs for personal computers, since the launch of AMD Zen 2 (mid 2019, so about 8 years before the launch of Intel Nova Lake; Zen 2 was when AMD switched to TSMC and they also doubled the computational throughput of their desktop CPUs in comparison with the best Intel desktop CPUs).

Maybe this is a layperson’s misunderstanding but after seeing Panther Lake hit the consumer scene I have a hard time believing Intel has a talent problem.

I’d even say what Intel is doing with Arc graphics is incredibly impressive. Coming out of nowhere to legitimately compete with a decent set of AMD/Nvidia consumer and workstation graphics cards is no small feat.

Panther Lake is very nice, but they did not sell much of it, in part because of the huge increase in memory price, but also because Panther Lake is much more expensive than the Arrow Lake H that it replaces, even twice more expensive for the variants with good GPU.

The great increase in price of Panther Lake is obfuscated by the fact that some part numbers that look like replacements for the corresponding part numbers of Arrow Lake H, and which do not have much higher prices, in reality are not replacements because they have much worse specs, e.g. a much smaller GPU, so for a real upgrade you must go to a higher part number, with a much steeper price increase.

I am among those who wanted to buy a Panther Lake based mini-PC, but I dropped my plans after analyzing the new prices, even those of the barebone systems, which have not been affected by the memory crysis.

The GPU of Panther Lake continues to be made by TSMC, in about the same technology as for the previous Intel CPUs. Because the greatest increases in Panther Lake prices are in the models with big GPUs, I assume that TSMC has increased the prices that it demands from Intel, due to the shortage of production capacity caused by the demands for AI datacenters, so it is likely that Panther Lake is also an indirect victim of the AI companies.

I really hope that Intel will continue to design ARC GPUs with good FP64 performance, because they are the only GPU vendor with good FP64 throughput, which has even increased in the Battlemage generation, while all the other GPU vendors attempt to reduce more and more all the "professional" features of their "consumer" GPUs, in order to reserve them for overpriced "datacenter" GPUs, which are not affordable for small businesses or for individuals.

I don’t see how it’s a problem that TSMC fabs the GPU, it’s not like Qualcomm, ANd, or Apple have their own fabs. Apple product pricing has increased recently, too. The whole industry is struggling with consumer sales, not just Intel.

> The fact that something is just too hard for the talent at hand is just never offered up.

Those of us who've worked at less illustrious companies are all too familiar with that. It's definitely unpopular to mention, but it drives a lot of consulting sales.

Panther Lake shows they’ve still “got it.”

Intel ARC is not only impressive just for the mere feat of “almost catching up” this quickly, but Intel absolutely can be an AI GPU supplier of note.

I don’t claim to fully understand the business especially on the enterprise side but I don’t think it’s the sort of company to count out.

Intel bring the knowledge but doesn't operate and it's not the only partners in talk. They are not tied into the production at any step beyond the blueprints.

Knowing them they'll have all this stuff online just as the AI bubble pops. Basically the same thing they did when they belatedly tried to get into mobile chips

How are they doing fab wise ? I was a serious investor at around 20 dollars before Trump came in and messed timeline. Foundry is an incredibly hard business and idiots bought in just because of Trump, China and AI. I’m wondering if foundry made tangible progress in the last 1.5 years?

Reply on news.ycombinator.com