Intel Slips, and a High-Profile Supercomputer Is Delayed (nytimes.com) 77
The chip maker was selected for an Energy Department project meant to show American tech independence. But problems at Intel have thrown a wrench into the effort. From a report: When it selected Intel to help build a $500 million supercomputer last year, the Energy Department bet that computer chips made in the United States could help counter a technology challenge from China. Officials at the department's Argonne National Laboratory predicted that the machine, called Aurora and scheduled to be installed at facilities near Chicago in 2021, would be the first U.S. system to reach a technical pinnacle known as exascale computing. Intel pledged to supply three kinds of chips for the system from its factories in Oregon, Arizona and New Mexico. But a technology delay by the Silicon Valley giant has thrown a wrench into that plan, the latest sign of headwinds facing government and industry efforts to reverse America's dependence on foreign-made semiconductors. It was also an indication of the challenges ahead for U.S. hopes to regain a lead in critical semiconductor manufacturing technology.
Intel, which supplies electronic brains for most personal computers and web services, has long driven miniaturization advances that make electronic devices smaller, faster and cheaper. But Robert Swan, its chief executive, warned last month that the next production advance would be 12 months late and suggested that some chips for Aurora might be made outside Intel factories. Intel's problems make it close to impossible that Aurora will be installed on schedule, researchers and analysts said. And shifting a key component to foreign factories would undermine company and government hopes of an all-American design. "That is part of the story they were trying to sell," said Jack Dongarra, a computer scientist at the University of Tennessee who tracks supercomputer installations around the world. "Now they stumbled."
Intel, which supplies electronic brains for most personal computers and web services, has long driven miniaturization advances that make electronic devices smaller, faster and cheaper. But Robert Swan, its chief executive, warned last month that the next production advance would be 12 months late and suggested that some chips for Aurora might be made outside Intel factories. Intel's problems make it close to impossible that Aurora will be installed on schedule, researchers and analysts said. And shifting a key component to foreign factories would undermine company and government hopes of an all-American design. "That is part of the story they were trying to sell," said Jack Dongarra, a computer scientist at the University of Tennessee who tracks supercomputer installations around the world. "Now they stumbled."
Maybe... (Score:5, Funny)
Maybe they should have used AMD?
Re: (Score:1)
Naa, it has to be the best product! Obviously that must be an American one! Because America does everything better, larger, faster! (Like getting people killed or maimed with Covid-19, for example...).
Re: Maybe... (Score:2)
Well, this is indeed the American government. So can't they just have Intel manufacture AMD CPUs, and if necessary, upgrade the fab in the go? Come on... at thosr budgets!
Re:Maybe... (Score:4, Insightful)
Maybe they should have used AMD?
. . . or POWER . . . it seemed that the US Department of Energy had good success with that:
https://newsroom.ibm.com/US-De... [ibm.com]
. . . or maybe that is just marketing Gilhooley . . .
Re: (Score:2)
What is Gilhooley?
Re: (Score:1)
a person with questionable instincts; a crowd follower; an adherent to conventional wisdom.
Re: (Score:2)
. . . or POWER . . . it seemed that the US Department of Energy had good success with that:
I thought there are POWER outages all over the country due to lack of regulation and oversight...
Re: (Score:2)
Aren't AMD's high-end processors all fabbed in Taiwan? That doesn't seem like the best demonstration of "all-American" technology.
Re: (Score:2)
Neither does this.
Re: (Score:2)
So what. All of the Intel Xeon chips I use say "made in Costa Rica". Last time I checked that wasn't part of the USA.
Aside from prototypes, does Intel make any serious CPUs in the USA?
Re:Maybe... (Score:4, Insightful)
Re: (Score:2)
I didn't realize that was done with CPUs.
TIL
Re: (Score:2)
"which supplies electronic brains", come on.. (Score:2)
Re: (Score:1)
Is it really that big a deal? This is the part of the story that ticks you off?
Re: (Score:2)
Re: (Score:3)
I'm with him. I read "electronic brains" and I literally winced. Computers are mainstream enough at this point that there's absolutely no reason to be so ridiculously imprecise. They supply processors. If somebody really doesn't know what a processor is, they likely would have zero interest in a story about Intel delaying a super computer.
Know your audience. Or, better yet, hire a technical editor before posting an article about a technical subject.
Re: (Score:2)
Re: (Score:2)
oh im sorry, are journalist allowed to be completely ignorant on the subjects that they are reporting? This is not the "remmington - rand" days anymore
Re: (Score:1)
Re: Be kind (Score:2)
You take pedantry to brave new levels.
Re: (Score:1)
Re: (Score:2)
Prett sure I haven't seen the term "electronic brain" to refer to a computer or any part of one since I read The How And Why Wonder Book Of Robots And Electronic Brains [biblio.com].
Re:"which supplies electronic brains", come on.. (Score:4, Informative)
Re: (Score:1)
Electric brain actually.
Re: (Score:2)
They set themselves up for it years ago. Everything since skylake has been a rehash of skylake. In theory they have a 10nm process but its only in low volume mobile chips. It is incapable of scaling to desktop thermal and power budgets.
But Intel is a big company with a lot of talented engineers and a huge RnD budget right? How could they fall behind?
I'm not so sure. I have a feeling, although I can't confirm it, that they were neglecting RnD for quite some time. ANd they let themselves fall into this situat
Re: (Score:2)
But Intel is a big company with a lot of talented engineers and a huge RnD budget right? How could they fall behind?
I'm not so sure. I have a feeling, although I can't confirm it, that they were neglecting RnD for quite some time. ANd they let themselves fall into this situation. nonexistent competition for the last decade let them get complacent and lazy and with Ryzen being a decent product it came back to bite them. They've tried to reverse things but it isn't that simple. It takes years to undo years of neglect and once you lose a lead its hard to catch up again.
My take also. Remember AMD embarrassed them before. Multiple times, in fact. There is a reason we are using the AMD64 Instruction set today (yes, Intel chips as well). There are reasons AMD had the memory controller in the CPU years ahead of Intel. Also, the supposed "performance disaster" the Intel PR people managed to manufacture for the AMD FX CPUs was anything but. But it allowed Intel to appear to be ahead for a few more years that they apparently were incapable of using that time to good effect.
Re: (Score:2)
Also, the supposed "performance disaster" the Intel PR people managed to manufacture for the AMD FX CPUs was anything but. But it allowed Intel to appear to be ahead for a few more years that they apparently were incapable of using that time to good effect.
This one always amused me. I'm still running eight core FX CPUs around here. There's no compelling reason to replace them. They really are wonderful multithread platforms.
Re: Since when do we call "massively screwing up". (Score:1)
Sure, I've only got four floating point units but that's becoming less and less relevant with 'GPU Compute' becoming mainstream...
Re: (Score:1)
I've got mine hitting 5GHz on air, with FSB to RAM ratio of 1:1 (300MHz) and the DDR3 2400 @CAS 10.
Sure, I've only got four floating point units but that's becoming less and less relevant with 'GPU Compute' becoming mainstream...
Cue Austin Powers: Ya baby, squeeze the power from your gorgeous plump over-clockable...
nvm.
Re: (Score:1)
This one always amused me. I'm still running eight core FX CPUs around here. There's no compelling reason to replace them. They really are wonderful multithread platforms.
Same here. I was thinking about upgrading but at this point these 8-cores are still doing fine. The main point of lag in my system is the HD because I'm still running old SATA drives instead of SSDs.
Re: (Score:2)
Indeed. And recently I found that even in gaming, they are a lot better than many people thought. The thing is that Intel CPUs get choppy and jitter a lot on low frame rates, which is really unpleasant to watch. While on AMD performance stays smooth. So in one game, the people on Intel were only happy with >40 FPS, while 20FPS on an FX8350 was perfectly playable and non-problematic. It seems Intel sacrificed a lot more than just security to get their performance.
Re: (Score:2)
I generally agree with your examples but I'm going to pick some nits.
There is a reason we are using the AMD64 Instruction set today (yes, Intel chips as well).
That wasn't a case of bad engineering on Intel's part. It was a case of bad business decisions. There is no reason why they couldn't have made something like AMD64 but management wanted 1: a clean break with x86 which actually makes sense to some degree and 2: they wanted it to be all new and controlled by them so they could lock people in. So Intel pulled an IBM PS/2 and played themselves. The market will always go for the path of least r
Re: (Score:2)
PR people managed to manufacture for the AMD FX CPUs was anything but.
Eh, sorta. The FX were and are inferior to contemporary Intel chips. That is a fact. What muddies the waters is that for many use cases old chips are still perfectly viable today. I know people still using 2600K's.
Actually no. In some benchmarks, yes, but for example game performance on the FX CPUs in multi-threaded games is far smoother than on Intel and you get a good experience on far lower frame rates. One example I have seen is 20 FPS being perfectly fine on an FX8350, while on Intel people wanted 40 FPS or more for a good experience because things got choppy and jittery. Workloads that do not include floating-point (server) were also not really affected. And if you compare whole systems in the same price-class,
Re: (Score:2)
This is an often remarked myth about AMD performance. Gamers Nexus has videos both on the topic of "Smoothness" And retrospective processor performance with empirical tests. Its all BS invented by fanboys who want to justify their purchase 10 years later. And I'm saying that as someone who has and has had several AMD based systems.
Re: (Score:2)
Not sure why gweihir has been modded a troll for stating the bleedin' obvious. Intel has missed so many important dates that "slipping" seems to be the rule rather than the exception.
Re: (Score:2)
Don't want the CEO up my butt, in my code (Score:2)
I gotta disagree. The CEO may not have done their job well, but designing CPUs or foundry equipment isn't their job.
With over 100,000 employees, Intel has a lot lot of people whose job it is to design CPUs. The CEO's job isn't too design CPUs, but do things like:
Hire good VPs, who will effectively hire and manage the best leaders for CPU-design teams
Decide whether or not they should try to buy Nvidia or some other company, at what price
Set a productive company culture where the CPU designers can work effec
Re:Don't want the CEO up my butt, in my code (Score:4)
No on is suggesting the CEO should be designing chips. We are suggesting, they should understand the process of designing and manufacturing chips in order to lead well. One of these two CEOs obviously does understand enough to lead well, while one is obviously of the "An MBA gives me all the knowledge I need to lead any company in any field, specific domain knowledge is for lesser mortals" school of thought. One has succeeded, the other has failed, proving you need to at least have a basic understanding of the business you plan to lead in order to succeed.
There is more to business leadership than just saying "Go make a product and sell it for me."
Re:Don't want the CEO up my butt, in my code (Score:5, Insightful)
The point is MBAs should not be in leadership positions. If they must be present at all, stick them in an advisor position where their damage is throttled down by people with actual knowledge of the company's core business. Putting an MBA in the top position of a company is a certain recipe for disaster as their singular focus on short-term profit over all else inevitably leads to a failure to plan ahead. Which leads to what we see here.
Re: (Score:2)
The MBA is a leadership degree. Does that mean that everyone who gets that degree is automatically the best person to be CEO of a huge company? Of course not. No more than everyone who has a degree in electrical engineering is world-class chip designer. But to say that people in top leadership positions shouldn't study leadership - well that's just as silly as saying chip designers shouldn't study electrical engineering.
Re: (Score:2)
The point is MBAs should not be in leadership positions.
MBAs are literally the business degree studied by people with other knowledge. It's not uncommon for MBAs to also have degrees in engineering or science. Your blanket statement just shows how little you know about the degree or the people involved in running companies.
Re: (Score:3)
Intel were doing pretty good when Andy Grove was in charge. A leader and an engineer who was with Intel from the beginning.
Grove is credited with having transformed Intel from a manufacturer of memory chips into the world's dominant producer of microprocessors for PC, servers, and general-purpose computing. During his tenure as CEO, Grove oversaw a 4,500% increase in Intel's market capitalization from $4 billion to $197 billion, making it the world's 7th largest company, with 64,000 employees. Most of the company's profits were reinvested in research and development, along with building new facilities, in order to produce improved and faster microprocessors.
https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:3)
It is not hard to understand the problems at Intel.
The MBA garbage is an american problem, not an Intel problem.
Production reliability problems (Score:5, Informative)
Can't turn something that big on a dime (Score:2)
After decades of growing dependence on foreign semiconductor fabs, it was both vain and foolish to expect that any American company, or even a group of them, could bring it all back home that quickly.
The fact that it originated in the States doesn't mean that all the folks you outsourced it to (in the name of extra profits) didn't learn a lot and improve upon it. It also doesn't mean that they won't turn the dinner you're expecting into their breakfast when you try to compete with them. You've grown fat, la
Intel need to get their manufacturing fixed (Score:5, Interesting)
This is a big problem for a manufacturing company, yes that is what they are. Not a technology company. The fact that CEO Robert Swan is thinking of having their silicon made in other fabs is very telling of a much bigger problem.
They're new fab issues have been going on for a number of years. Maybe some of their strategies have finally caught up with them. Such as:
-Giving early "retirement" or separation packages to older employees, and replace them with "recent college graduates". You know, some of them old peeps you got rid of have turned on a fab or 2 and have really great knowledge. RCGs are cheap have no experience. By the way, can't tell you how many engineers/developers just ship from Intel straight to Amazon. Weird Amazon doesn't know about silicon and server development.. except when they hire the whole team away....
-Stop pushing so hard to hire women/minorities. Some people might bristle at this, but I'm ok with it. Hire the best employee, not because they have a different color skin or are a different sex. I know a woman engineer at Intel that was pissed to think she was hired because she was a woman and not off of her merits.
-Focus on what you're supposed to be. Manufacturing. If you can't tell by now I worked for Intel for a number of years. I can't tell you how many stupid ideas they tried to come up with and fail, or better yet make something that is good and stop at 90% done. The article mentions Xeon Phi. This is the GPU used for compute. That hardware was originally developed years ago and before NVidia had a compute GPU. Then they just pulled the plug, waited a couple of years and tried again after others held the market. How much money did Intel waste there? Don't get me started on phones. Everyone in the entire company knew it would fail. I could go on and on.
So what I see is a company that was coasting on it's past, dumping alot of people which is knowledge and experience, hiring to the pressures of society and not getting the best. It take a while for the giant to fall, but if Intel doesn't change their ways soon...
Re: (Score:3, Insightful)
replace them with "recent college graduates"
An American problem, not an Intel problem.
ire the best employee, not because they have a different color skin or are a different sex.
An American problem, not an Intel problem. So far, most of Intel's issues are due to American culture. I blame MBAs and communists
Re: (Score:1)
I guarantee that TSMC don't have a diversity program. Their focus on making the world's best chips is why they are worth twice as much as Intel.
Apparently for every 1000 good engineers you hire, you would take on another 1000 welfare cases with the same pay and responsibilities. Apart from wasting vast amounts of money, you'd have teams bogged down with idiots making bad decisions.
Hire the best 1000 engineers period. That isn't sexist or racist.
Re: (Score:2)
Apparently for every 1000 good engineers you hire, you would take on another 1000 welfare cases with the same pay and responsibilities. Apart from wasting vast amounts of money, you'd have teams bogged down with idiots making bad decisions.
Even non-whites and non-males can be educated and qualified.
That isn't sexist or racist.
Correct. It is both sexist and racist. Congratulations.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
if Intel has money to burn on diversity
Ultimately your idea is good, but you propped up a straw man. Diversity programs do not lose money, but instead create a net gain for preventing astronomically expensive discrimination lawsuits and settlements.
Re: (Score:2)
Re: (Score:2)
Maybe this one [business-humanrights.org]. It's not that it was claimed that the whole workforce was hiring insufficient minorities, but that there were insufficient minorities in senior positions, i.e. part of their workforce was less than X% minority.
Re: (Score:2)
Can you cite a company that has had to pay astronomical settlements because their workforce was less than X% minority/women? I haven't heard of that and I work in tech.
I can think of a few in the financial industry, but these cases were pivotal, and changed attitudes across most other industries. In the late 1990s and early 2000s, Morgan Stanley paid $54 million, and Smith Barney and Merrill Lynch more than $100 million each, to settle sex discrimination claims. In 2007, Morgan faced a new class action which cost the company $46 million. In 2013, Bank of America Merrill Lynch settled a race discrimination suit for $160 million. Cases like these brought Merrill’s tot
Re: (Score:2)
The problem is that hiring unqualified people
Straw man. No one said diversity programs are intended to hire unqualified people.
Re: (Score:2)
They can do both! Hire the 1000 best engineers regardless of skin color or gender AND hire worthy minorities.
You're admitting right here that you're hiring a group of people that aren't the best aside from the ones you actually want.
Re: (Score:2)
Re: (Score:2)
so you want a company to hire the best talent, and some tokens for what reason? isn't that reverse racism "oh I have zero reason to actually hire you in this organization other than I need to check this box that we have exactly 12 black people in these 3 departments, you are just a token here, try not to get fired"
Re: (Score:2)
What you're implying here is that one of the reasons for the decline of Intel is that there are too many women and minorities working there, and that they are much worse at their jobs than white men.
It's an extraordinary claim that you fail to back up. I'm sure your friend was indeed pissed at the thought that she might not be best for the job, but presumably that's because she's very skilled. It may very well be that without whatever programs they have in place, they would have bypassed her expertise. Do n
Re: (Score:2)
Hire the best employee
Research has shown time and time again that if you identify the "best" employee in isolation you won't have the best overall functioning team or organisation. "Best" includes applying research which shows that diverse teams of like skills almost universally out-perform their non-diverse counterparts.
Who knew making chips could be complicated? (Score:2, Offtopic)
Re: (Score:3)
Or said properly: (Score:2)
well at least (Score:2)
i mean, $300m is what? ~1/10 of one year's profit? still, priorities.
Can't believe Jack Dongarra is still at UT. (Score:2)
He was there when I was in grad school in the early 90's.