bgnn

38 ms·

571.
by bgnn 3y ago
10GBASE-T is rated up to 300m cable. I don't think these numbers include 300m cable latency. It's just the PHY latency itself.
572.
by bgnn 3y ago
10G-baseT ethernet uses 128-DSQ modulation (LDPC coded). It's 128 symbols sent over 2 cycles. These complex modulations have a huge drawback though: latency. You need to pack, ramdomize, encode and modulate at the transceiver, and undo
573.
by bgnn 3y ago
This isn't fully correct. Drawn 7nm is like 20nm, but effective gate length is shorter due to short channel effects. So, this numbers are almost 1-to-1 reflection of effective gate length. People mix up the transistor pitch with gate l
574.
by bgnn 3y ago
In principle 7nm has 57, 64 or 76nm gate pitch. Routing is quite inefficient in 57nm gate pitch, so 64nm is used most often. As far as I can see in the design rule manual there's no possibility for 60nm pitch. 40nm metal pitch for colo
575.
by bgnn 3y ago
This. Intel 10nm and TSMC N7 aren't comparable in their development. Intel didn't simply scale down their 14nm but took a different parth. TSMC N7 is basically a scaling of their N16. These two processes are 90% comparable. Only b
576.
by bgnn 3y ago
This. Basically it's the price of not having proper social security against illness. I live in the Netherlands and opioids are unheard-of except cancer patients etc.. A friend of mine broke his collar bone, which was bulging out a bit,
577.
by bgnn 3y ago
coronavirus
578.
by bgnn 3y ago
Ultrasound is routinely used to break kidney stones. With more precision like this, why not? I don't know if there would be risks though.
579.
by bgnn 3y ago
yeah. yet people believe AI can do new things without training.
580.
by bgnn 3y ago
Depends on the EU country. Phd students earn more than 40k during their studies in NL, and a starter salary for a phd is 80k or so. Add the social security contributions on top, it's closer to 100k Euros for the employer. 40k would be
581.
by bgnn 3y ago
True. Luckily the competition is catching up fast. Though Marvell isn't so much different.
582.
by bgnn 3y ago
because trying to do that will get you arrested. it's not so different than Russia.
583.
by bgnn 3y ago
it's because of the CEO Hock Tan. Broadcom became a sort of investment fund rather than an engineering company. Hock even wanted to buy Qualcomm but got blocked by Trump. This was because it was very likely Hock would just divide it in
584.
by bgnn 3y ago
It's like 100 usd/vial in some countries in Europe, more expensive in some others. Still ridiculously expensive.
585.
by bgnn 3y ago
There are certainly cases in big concert halls etc to run like 100m. 900m should be more a stadiums.
586.
by bgnn 3y ago
Yeah but that's the same fir any car manufacturer aynwhere. VW is extremely subsidized by Germany, US bailed out their bankrupt manufacturers after 2008, same with France, Japan, Korea. UK didn't do that, or stopped doing that, an
587.
by bgnn 3y ago
nor fully correct. lumped circuit theory doesn't only depend on the frequency but also the cable length. EM waves travel at around 0.6C in solids (copper). wavelength = V/f where V=0.6C, and frequency is say 20kHz, which is 9km. I
588.
by bgnn 3y ago
They get raw wafers from other suppliers AFAIK. It comes pre-doped even.
589.
by bgnn 3y ago
They don't even consider the energy consumed to produce the chips AI/ML is running on at all. The biggest chip producer TSMC ysed 22000 gigawatt hour last year. This doesn't include oder foundries, memory etc.. AI is now more
590.
by bgnn 3y ago
I see your point. There are more factors than talent being different in the Bay Area I think. I work for a Bay Area start-up in the Netherlands. They opened their office here for a specific talent. This talent can be found in the Bay Area t
591.
by bgnn 3y ago
5nm is good yield at the moment. It's even fully automotive qualified, which is a testament to its yield. But performance advantage vs cost doesn't justify to move from 7nm for a lot of designs. So tech adoption is getting stickie
592.
by bgnn 3y ago
yeah lose. I don't know why I keep typing this wrong..
593.
by bgnn 3y ago
This is so true. US foreign policy thinking is like a bad parody of Hollywood movies.
594.
by bgnn 3y ago
There is literally no good reason other than it being the standard I think. Analog design engineers are very averse to change. Everone learned the quirks of Virtuoso by now and nobody dares to try something else because they are not so much
595.
by bgnn 3y ago
Chip design software GUIs are known to be unintuitive and unfriendly. I'm using them daily and it's in the way all the time. So I don't know what changed after these but I'm not seeing it as a user. Oh, also, when will w
596.
by bgnn 3y ago
output doesn't capture accumulated know-how. If you fire researchers you loose know-how, hence you loose potential.
597.
by bgnn 3y ago
One can always use clock repraters or a reclocking with a new PLL in the middle if the cable length gets too long and caused ISI (inter-symbol-interference). I don't think 30 ft requires this. At 60ft you might need it though it depend
598.
by bgnn 3y ago
I think this would be amazing to ttack movements but the combined images would be extremely noisy wouldn't it?
599.
by bgnn 3y ago
Yeah GPS is a good tool to sychronize over very long distances. Locally locked clocks might still be better for an array not far away from each other if done properly. GPS might be easier to implement for amateurs though.
600.
by bgnn 3y ago
Short answer is: no. The long answer is: Speed of light on glass is not so different than electromagnetic signal speed in copper, approximately 0.66C. Considering clock sources output electrical signals, using a coax cable instead of fibre

More ›