Earlier quoted context omitted.
You might be surprised to discover that roughly half of their population is women.
Not women who'd work in a tech company.
We just lost a good (female, Australian) grad student we were recruiting, to UT Austin.
Intel Honesty
361–370 of 426 posts
Re: Intel Honesty
#362Earlier quoted context omitted.
I'm not sure we're really disagreeing. The thing is that I don't really want a budget Chromebook (which certainly exist). I want something for travel mostly. Ended up getting a new iPad Air recently which isn't all that light but probably fits my needs best as it now has a pretty functional keyboard and can easily be used without the keyboard as an entertainment/consumption device.
Context upthread was for a development box, though. Airs are pretty limited for that, but the bigger Chromebooks do great.
Fair enough. I lost that part of the thread. I don't do development when I travel which is what I would be potentially interested in a Chromebook for.
Re: Intel Honesty
#363Earlier quoted context omitted.
I thought historically companies always wanted "second sources" for things like this. Surely all the fabless semis can't be blithely running off a cliff with TSMC as their only parts source? That would be a catastrophic risk to take down the track if there's only a single vendor for advanced chips and they've decided to hike prices and collect all of the benefits alone from their advanced processes and share none of…
Buyers will obviously always want more sellers. That is not a sufficient condition for sellers to exist, however. Selling what TSMC sells might not be easily reproducible.
As it currently stands, it is reproducible. In 5 years, it might not be, if Samsung/Intel throw in the towel.
I don't really see that happening yet, but at some point if TSMC maintains its lead and the alternative options don't get the sales to continue investing, TSMC could well become completely unassailable at the leading edge.
Re: Intel Honesty
#364Earlier quoted context omitted.
Part of the problem is what is available. If you're a Japanese or a Taiwanese, that detached single family home with a massive garden around it is simply not an option, so you don't feel worst off than anyone else by living in a small house or in a high rise. Other advantage is that keeping cities very dense and preventing urban sprawl means everyone has access to nature quickly, through public transports.
Oh, those kind of houses are definitely available in Japan. Houses in Japan are worth negative money; if you want to sell one the buyer will want you to knock it down first. The reason they're so cheap is that nobody wants to live in those areas because cities are better and have better jobs.
There's also a culture in Japan that houses are essentially disposable. Whereas on the east coast at least of the US, 100-200 year old houses aren't rare. The default may be to do some renovations if you buy one but not to knock it down. And a lot of those houses may not be in cities but they're often within an hour or so and many of the jobs are outside the cities anyway.
Re: Intel Honesty
#365Earlier quoted context omitted.
I think lack of focus on US domestic chip capability is semi-intentional and is an eventuality. Semiconductor industry, and many other manufacturing industries too, seem to flourish at very outer edges of the free world. It could be simple as the free world not wanting manufacturing capability at home. It could be wanting only grain farms and money printers.
Manufacturing in the west is expensive compared to cheaper developing countries. Also, manufacturing has a massive impact on the environment, this includes semiconductor manufacturing. It has politically been impopular for a while now.
Because of the resource curse - everything that isn't the best cost/benefit ratio to a country tends to become unworkably expensive. The USA's cursed resource is money - literally - it prints the world reserve currency.
Re: Intel Honesty
#366> today AMD has both better designs and, thanks to the fact they fab their chips at TSMC, better processes On paper AMD is better, but in the scientific community, it seems that Intel has much better performance for things like NumPy and SciPy. The reason seems to be "Intel® MKL" ( https://en.wikipedia.org/wiki/Math_Kernel_Library ). I'm using a lot of weasel words like "seems to" because I haven't rigorously proved…
> But anecdotally, my company's AI pipelines which are NumPy/SciPy heavy run an order of magnitude faster (2 seconds vs. 20 seconds) on my laptop's Intel i7 than the do on my Ryzen 7, despite the Ryzen 7 being a newer gen than the i7. I very very seriously doubt this is a like-for-like comparison. I suspect one case is multithreaded and the other not, or that one is using generic non-SIMD fallbacks or something like…
All I know is, when I run `python3 ai_script.py`, it's much slower if I invoke that command on my Ryzen 7 desktop PC vs. my i7 laptop. I briefly profiled it once, and it seemed like `skimage.measure.label` ran much slower on AMD.
Re: Intel Honesty
#367Earlier quoted context omitted.
Buyers will obviously always want more sellers. That is not a sufficient condition for sellers to exist, however. Selling what TSMC sells might not be easily reproducible.
As it currently stands, it is reproducible. In 5 years, it might not be, if Samsung/Intel throw in the towel. I don't really see that happening yet, but at some point if TSMC maintains its lead and the alternative options don't get the sales to continue investing, TSMC could well become completely unassailable at the leading edge.
Intel seems like it’s at least a few years (but I would bet many years) behind TSMC on obtaining/training the necessary people with necessary expertise with the necessary management changes.
It took TSMC a long time and many expensive, risky bets to beat Intel. I imagine it will take Intel just as many resources and time to catch up to TSMC. Hence the divergence in their market caps.
Re: Intel Honesty
#368Earlier quoted context omitted.
How did Amazon (possibly no less "toxic", Idk) embarrassed combative disagreement and sublimated that natural emergent expression/desire during stressful situations into something that worked for the company, whereas Intel sounds like it failed to do that? All that worked (or seemed to work) for Intel for many years - 30+ years of profits and market dominance. I don't know the exact abuse level of Amazon, Microsoft,…
This is a good point. I get it probably gets ingrained, more so than a more open culture that would be open to bettering it self. My view was Intel had a good culture until external pressures resulted in turning toxic.
My gosh you can read the posts on this thread of myself and others talking about the toxicity of Intel's culture in the 90s.
If you're in doubt, I'd recommend David M. Gordon's Fat And Mean for a discussion of how toxic culture persists and intensifies in corporate America.
Edit: I think can credit the long period of success Intel had to shrewd leadership and leading in a massively expanding market. The thing about a chip company is that it is constantly gambling on extremely costly investments in machinery and people. Abusive tactics that get good people working really hard for average salaries are really useful and "only come due over time" and often when the company foundering anyway.
Re: Intel Honesty
#369Intel is trapped. Its debt repayments alone are massive. The poor performance in stock has encouraged 20years of wage stagnation to the point where you can literally earn over 50% more at amd or nvidia or even an startup for equivalent roles so they aren’t hiring the best. Their pay in the Bay Area is double tsmc in taiwan for equivalent roles in raw dollar terms but the ppp differences means your better off working…
Intel's debt is not bad at all. My reading has it at $53 billion. Before their revenue collapsed in the last few years that debt is only about 50% of revenue. High but manageable compared to att or warner discovery who actually have unmanageable debt. Intel's problem is that revenue is decreasing, and quickly.
Re: Intel Honesty
#370> today AMD has both better designs and, thanks to the fact they fab their chips at TSMC, better processes On paper AMD is better, but in the scientific community, it seems that Intel has much better performance for things like NumPy and SciPy. The reason seems to be "Intel® MKL" ( https://en.wikipedia.org/wiki/Math_Kernel_Library ). I'm using a lot of weasel words like "seems to" because I haven't rigorously proved…
MKL detects non-Intel processors and uses a slow code path.
Indeed, and ignores the CPU feature flags for things like AVX2.
There is various hacks, some to the binary (to edit out the if then slow) or even just ENV flags. This fixes things like matlab's use of MKL and shows a significant performance increase.