HN user
Just a guy named Burak.
burakbiyikli.com
I think there are a lot of ECEs that work in software. A lot of my classmates from my EE undergrad went to software despite getting a more hardware centric education because software was offering more money.
I imagine there are many people people that, if they pay was higher, would move back to more 'pure' EE/ECE work.
My understanding is that it is ARM's parent company Softbank that stands to benefit rather than ARM looking to reinvest the money to fuel growth.
ARM would benefit only slightly, as its employees have RSUs that the company said it would honor in the event that no sale of ARM occurs.
Softbank on the other hand has 10's of billions of losses since 2017 from its vision fund and may be looking to rebalance its books. I can see why ARM would be a good asset to sell from Softbank's perspective. ARM is unlikely to see explosive growth in a new segment, and the market as a whole, and the tech sector in particular has fairly 'healthy' PE ratios[1]. This means that Softbank could realize more cash from a sale now than a hypothetical future time when there is more pessimism in the market.
[1]P/E ratio is the ratio between the price of a stock and its earnings. Some notable outliers are Nvidia @ 232 & AMD @ 483 compared to Intel @ 15.36 or Texas Instruments @ 20.01; PE is higher for companies that are expected to grow and lower for companies that are mature or declining.
I read it more as, clients have a mix of questions. Clients don’t go through the system everyday and likely have some general worries and procedural questions as a result. Clients also have a case and need to coordinate with their lawyer. AI could enable more of the latter to happen within the limited time the lawyer has for each client.
The language on their website implies that they act collectively in the event of a veto “A vetoed bill can become law if two-thirds of the members of each house vote to override the Governor's veto.”[0].
Practically speaking, there are usually committees (may be area specific, or general such as scheduling/introductory) that make the first choice as to how the process will continue. For new york this seems to be the standing committee which decides what will be put to a vote, “ Members of Standing Committees evaluate bills and decide whether to "report" them (send them) to the Senate floor for a final decision by the full membership.”[0].
My understanding is that it is more a consequence of politics than the system... Sure you have a veto proof majority on this bill, but what about what comes next?
The governor is able to stop "tyranny of masses" by their veto power. Almost no bill in our partisan age is able to pass with a veto proof majority. So the governor has the ability to prevent good outcomes for specific members of their party in the legislature in the future. Sure a budget that makes the party and governor look good will pass, but members who "step out of line", will find that they can't "bring home the bacon" in terms of what is allocated to their district.
So could the bill be pushed through unmodified despite an governor's veto? Of course. The system expressly allows this given a 2/3rds vote. But the legislature needs to weigh their options:
1.) Have all members push the bill forward anyways 2.) Have just enough members push the bill forward 3.) Accept the governor's revisions and move the bill forward 4.) Drop the bill
As a representative it is hard to tell which of 1/2 you are choosing in the moment. The issue is that 2 may paint a target on your back. The members of the party opposing the governor are less incentivized to care about the veto. So do you want the governor to hold a grudge with "1 of X" representatives of their own party who sided against them? Granted the legislature is about 2/3rds democrat, so for a break with the governor to happen maybe half the democratic legislature would have to break with the governor.
3 Is appealing over 4, since you are able to claim victory for now. Something was passed after all, and the name of the bill alone is usually enough to make a good ad come re-election. and if the issues really are so severe, this is just another victory for the future you.
I think that you can at least choose for photos you take to not have location data at all in camera settings.
I found out because the feature of mapping of where your pictures where your pictures were taken did not work for me, because I stopped the camera from having location data.
True. That is part of the reason why only 3 of maybe 30-60 layers are that small.
Analog gets some signal integrity benefits from larger transistors, and often we have fairly large fets for low rdson and high voltage tolerance.
TI's fabs are mainly focused on analog, so the requirements are a bit different. IIRC we have 3 130-65nm layers for digital in our mixed signal designs which is roughly on par with a the process technology Intel used in 2005.
It looks like the scale rating the energy efficiency of the fridges aged out of being useful. From the article: "Eventually, more than half of the products on the market were labelled A++ and A+++".
The change means a fridge that would be an A in the old system is now a D in the new one. They are hoping that the scale will provide a more accurate idea of the performance of the fridge
The text of the comunity post is:
'YouTube has decided our Arc A380 review is "18+ only" and "not suitable for ads." This hurts views and revenue. Please share the video around manually since YouTube is restricting who sees it! https://youtu.be/La-dcK4h4ZU There's not much that's more discouraging in this space than working almost 24 hours straight on a review, producing very high quality testing and video, and then getting told "you don't deserve to make money on this because reasons we won't explain instituted by people we won't put you in contact with." Despite having 1.6M+ subs, we no longer have a creator contact at YouTube because it's a revolving door position with a new employee every few months. Hoping this makes it back to them somehow and they fix the ad qualifications of the Arc review.'
Gasoline is maybe 5 dollars a gallon and contains some 35kwh of energy. When burning gas in an engine we don't get 35kwh [1], but that is what is there.
Lets imagine we are building a gallon of gas with perfect efficiency. 35kwh is already about $3.15-10.50 depending on the market [2]. Lets assume however, we are willing to take the capital risk and invest in our own solar plant which gives us a price of $2.10-2.80. We can take this even further.
Lets say we buy our own panels, and don't even connect it to the grid. These panels are only used for this, and we only make gas during the day. Here we pay $1.2/watt of capacity [0]. Solar panels have a life of about 10 years. 8x265x10 -> 29.2Kwh/dollar if we use the energy from the panels entirely efficiently. This is about $1.2 on the absolute low end. Note that panels require some maintenance, and don't just die after 10 years.
Now that we have the energy, lets get our raw materials. Gasoline has a chemical formula of C8H18, a ratio by weight of 96:18 or 16:3. A Gallon of gas weighs about 6 pounds, meaning about 5 pounds of carbon to 1 pound of hydrogen.
Hydrogen is not cheap [4], Electrolysis costs some 2.4K USD per ton, 1.2 dollars a pound. Electrolysis is more than energy in, hydrogen out. It uses up electrodes for instance. If we are allowed to use fossil fuel derived hydrogen the price comes down to 0.2 dollars a pound [5].
Carbon is $35/ton, but this increases to $75/ton if we want beverage grade [6]. You could argue that we will need to purify the CO2 ourselves either way so we can use the cheaper carbon. Carbon to oxygen by weight is 12 : 30 or 2 : 5. So for 1 pound of carbon we need 3.5 pounds of CO2. This brings our cost to $122.5/ton or about 0.12 cents a pound.
Thus our material cost, including no synthesis is already in the ball park of $2 -> $12.9 per gallon.
We should also consider that gasoline has costs that our hypothetical fuel would also need. Additives make up 30-70 cents per gallon. Distribution another 30-60 cents a gallon. Taxes another 40-60 cents a gallon [7].
Adding these costs, our hypothetical fuel now costs $3 -> $14.8 per gallon. But we still need to actually add the costs for turning the ingredients into fuel. We also don't have any labor costs, or costs of capital.
[0] https://a1solarstore.com/solar-panels/panasonic-solar-panels... [1] - https://physics.stackexchange.com/questions/98966/maximum-th.... [2] - https://www.electricrate.com/electricity-rates-by-state/#:~:.... [3] - https://homeguide.com/costs/solar-panel-cost#:~:text=Residen.... [4] - https://oilprice.com/Energy/Energy-General/The-Green-Hydroge.... [5] - https://www.chemanalyst.com/Pricing-data/hydrogen-1165 [6] - http://www.dotyenergy.com/Economics/Econ_Physical_CO2_Market.... [7] https://voltaoil.com/what-makes-up-retail-price-for-gasoline....
Its been a bit since I watched, but he raises a few issues:
1. Many NFTs are fundamentally lacking a copyright. Only humans, not animals or even programs can create copyrightable works. So NFTs that are generated may have one copyright collectively or maybe even no legal protection. This single copyright can only be held by one person. So here clearly NFT != ownership.
2. Some NFTs that could have copyrights are not from their rightful owners. As a buyer it is hard to tell. While the origin of the NFT is easy to verify by design, any non web asset associated with it is hard.
3. Selling a copyright as a NFT is legally dubious. The first sale could legally work, but its not clear how you would reliably bind all subsequent buyers/sellers to the same contract.
4. Even if the work in the NFT is copyrightable, author provided, and backed by a legally sound process, there is no guarantee the “rights” granted have any real meaning. The example given was sport clips where your “ownership” while strong and clear, was limited to basically viewing it on the site/app for non-commercial purposes.
The issue kinda becomes, its hard to tie an NFT to any kind of transferable ownership in a legally sound way. Imo it sounds possible, like how homes with HOAs force buyers to agree to and perpetuate the terms of the HOA. Its just hard. Hard to setup, and hard to verify.
I should have mentioned explicitly that:
average power = capacity factor * max power
Thus it is clear why nuclear is .9, it doesn’t vary much in its generation. The max is only a bit more than the average. Capacity factor is in some sense, but not exactly, the variability.
I believe that what is happening is that the rated capacity of a power plant is its maximum output times its capacity factor. Say we have a solar plant that produces a maximum of 1 GW in optimal conditions with a .15 capacity factor. This plant is not a 1GW plant but a 150MW one.
In general fossil fuels have capacity factors around .55, nuclear about .9, hydro about .4, and solar is between .1-.3
So you can imagine a scenario where we have 200MW of natural gas and 200MW of solar, and a grid load of 300MW. The natural gas could produce 400MW, powering the grid alone. Similarly, we could have up to 1 GW of solar, more than enough. But the averages are what is important, and what we use in ratings.
These stories are mostly “feel good” and without meaning imo. We will hit 100% energy from renewables for x minutes, long before we have enough power for the grid. If we were trying to make all the renewables come online at once, you don’t actually need that much capacity to hit 100%. The capacity factor says we only need 1/5 of the grid as solar to hit “100%” renewables. The reason we haven’t is intentional, we want more consistent power from these typically inconsistent renewable sources.
My fellowship email said 34k/yr plus 10k in tuition. This was for UT.
In terms of expenses? Maybe about 2k/month. Depending on what standard of living you want 600-1400 for rent.
Separately, I was also a TA. This ıs ın the ballpark of 25/hour wıth a 80% tuıtıon rebate. Beıng an undergrad TA only paıd 10/hr wıth no rebate though.
[1] “ monthly fellowship stipend ($2,222) paid to you for nine months during fall and spring ($20,000 value) · full tuition paid for classes required for your degree for the fall and spring semesters ($10,554 value) · two payments to you in the amount equal to the premium for student health insurance ($4,000 current value) · plus, an additional $9,000 engineering fellowship, renewable for three additional years ($36,000 total value)”
I could be reading this wrong but it looks like the star is likely to come close, and the uncertainty is such that they may even collide.
“ An exceptional event will take place in early May 2028, when alpha Cen A will come within 0.015 +/- 0.135 arcseconds of the mK = 7.8 star 2MASS 14392160-6049528 (hereafter S5). In terms of impact parameter and contrast, this is the most favorable stellar conjunction of alpha Cen within at least the next three decades. With an angular diameter of LD = 0.47 +/- 0.05 mas, it is likely that S5 is a red giant or supergiant located at several kiloparsecs.”
I took it to mean double digit percent, but now I am also thinking it is confusing wording.
Fortunately, waiting is easy :)
I think that this is a really interesting move by Intel. The conspiracy theorist in me thinks this is an interesting way for Intel to get at TSMC. A few years back there was a cross licensing agreement between TSMC and GlobalFoundries. The agreement is broad, and covers all patents made by TSMC for the next 9 years [1]. It is easy to see how many problems this could create for TSMC. Either the deal is void, in which case TSMC is and has been likely infringing on Global Foundries' (with an acquisition Intel's) patents... Or Intel has a broad cross licensing agreement with TSMC... The deal was good for TSMC at the time, as Global Foundries had given up on more advanced nodes as the investment was just too much for their shareholders, but with new ownership I see it as an interesting weakness.
Intel has historically been willing to play dirty, like messing with compilers to benefit Intel over Amd[2]. Or like when they gave kickbacks to OEMS [2]. Or when they paid vendors to not support AMD fully in software[2]. Or just plain old false advertising [2]. Actions like these is what led to a 1.25 Billion dollar settlement with AMD [3].
Outside of playing dirty, it is a good strategic move for their new foundry business. You get the tools, know how, processes, and most importantly customers that can make the venture a success. Intel has had a foundry business before [4], but they closed it. "ecosystem is everything with the foundry business and that takes time, money, and technical intimacy," [4]. Intel has money, and they are buying the years in the foundry business they missed along with technical intimacy to more standard tools. So, I think it makes sense from a business perspective, as it gives Intel a recurring revenue for its older processes and in turn greater advantage from its vertical integration.
[1] https://www.anandtech.com/show/15038/globalfoundries-and-tsm... [2]https://www.anandtech.com/show/3839/intel-settles-with-the-f... [3] https://www.cnet.com/news/intel-to-pay-amd-1-25-billion-in-a... [4] https://semiwiki.com/semiconductor-manufacturers/intel/7912-...
I have some done some tests on 2900mAh Samsung cells from 2008. I have tested about 50 cells so far that were pretty abused, meaning they were discharged to less than 2.2V for several months. This is very bad as their normal operating range is 4.2-3.2V if you are really pushing them.
About half of the cells are effectively dead, internal shorts or electrolyte leaks. Included in this half are cells with mA discharge currents which is very out of spec. Not suitable for use.
The other half lived with an average capacity of 2050mAh with a std deviation of 150mAh. This is only a degradation of about 30% which was pretty surprising to me.
I guess the benefit of google’s speed test is that if ISPs try to artificially speed it up, they run a good chance of accidentally increasing the bandwidth of youtube.
I took ıt to mean that theır completive advantaged was mıtıgated because of the pandemıc. Normally wıth thıs lead TSMC would make much more than Intel, SMIC, ect., but because of the demand by the pandemıc all companıes are makıng more and very busy.
Lıke ın a normal year TSMC would make lots and competıtors fabs would be more empty
Generally old chips are good, depends on the purpose. Some things like the f35 would like to use TSMC’s fabs. But not every chip is for only processing and larger process have their advantages for certain applications like power, analog, and em hardening
I think what you say is correct, its so much easier to fire a missile than it is to stop one. Honestly, reliability and cost are bigger drivers than raw performance. The scenario is a bit bad I can admit, and did admit.
But if we were in desperate times, I think the limits would matter and fab tech would be an important strategic asset for what it enables. The nature of it is that its easier to attack, but we hopefully have something to defend.
Intrinsically larger process nodes are better for analog and have uses even as new fabs come along. In 7nm your analog signals are not treated very nicely, but these are important for sensors. Also power electronics are not really on the newest nodes either. The microchips are old because they get the job done and I don’t think we feel so threatened to push the limits.
I would argue the opposite, that it is a huge deal for defense.
Imagine if the situation is missile defense, both the missile and the defense systems are computerized and not only the milliseconds but also the microseconds are the difference between a hit and a stopped missile. Whoever had the newest fab node would have a notable advantage in being able to push their system harder. It could be better ML in missile to lock on and dodge, vs the defense system predicting paths as fast as possible with as low latency as possible to defend a ship or similar asset.
Its asymmetric in the example of course, but both sides would be doing both shooting and defending.
Mainly more points of failure, and additional weight to my knowledge.
The electric motor allows superior city driving due to the nice fact that you have regenerative breaking and much more efficient low speed operation where an ICE would not be able to run at its most efficient RPMs. While the gas engine gives a lot of convivence that you might otherwise lose.
The downside is now you have two drivetrains that need to integrate into a very limited space, with limited weight as the whole goal was to make an more efficient car. Its also non-trivial to make a good transmission let alone one that has two inputs, or even an integrated electric motor.
Batteries in a hybrid are also pushed much harder in a way. Batteries are often rated over 20hr discharge cycles as you can generally get more power out of a battery by discharging it slowly. Asking 1A from a 100Ahr battery vs 1A from a 1A battery is very different in terms of effective capacity as well as the longevity of the cell. Given that batteries are so heavy and you need two drivetrains you can only add so much battery capacity. The end result is that its easy for the batteries to age faster in a hybrid.
A surmountable issue is heat, batteries don't play well outside a range of limited range of temperatures. Too cold and they will have less effective capacity, too hot and they can be damaged or degrade faster. This issue can be overcome, but is just another way there is more complexity than Tesla's 'the floor is a battery approach'.
Hybrids are cool, but they are hard to make right is the tldr.
The trick is mixing in some good ol analog, and having 2^4 voltages that can be loaded and stored in each transistor
Yes and no. TSMC's process node gives some hard upper limit to what can be built, but whether you can make your design that compact is an open question.
One consideration is that while we think of each transistor as individual and isolated, in reality they are simply wells of n and p type silicon with some oxides and metal thrown in. Because of this we can get unwanted parasitic components and latches. A BJT transistor for example is just a pattern of three silicon types, so it is somewhat inevitable that these will occur, whether they cause issues, and to what extent depends on the design.
Packing everything as tight as possible gives noise concerns, thermal concerns, signal integrity issues ect.
This is not to devalue the achievement of TSMC, but to state that design work is still a work as well.
Intel uses monolithic dies, meaning everything is on the same chunk of silicon. This has given them some improvements to latency[1] and power usage[2] in the past but hurt them on yields. AMD has a chiplet design, which improves yields[3] and may allow a more modular approach as mentioned.
[1] can be overcome via caching and other considerations, but purely from this aspect the impact is this. [2] Longer traces lead to higher capacitance, and the power estimation formula P=C(V^2)f*a shows that this one aspect will change power use. Everything on one die means less parasitic capacitance. [3] If the defect density is the same, and if you have 10 errors per wafer, then you will see different yields if you make 10 vs 100 vs 1000 chips on that wafer. Chiplets are smaller than monolithic designs, so we can put more of them on one wafer which improves yield independent of process
I have heard rumors that AMD had a closer relationship with TSMC than NVIDIA. I have found that a bit weird as NVIDIA seems to move some higher margin silicon despite vague claims that NVIDIA is hard to work with. These rumors show NVIDIA either moving or being pushed to Samsung 8nm [0].
This article brings up the interesting point that AMD is spread thin trying to manufacture CPUs, APUs, GPUs, and has products for xbox and playstation both last and current gen. It paints it like a problem, and I would agree that predicting demand is challenging in the best of times. But maybe the diversity of the lineup is the real drive of TSMC partnering more closely with AMD; TSMC is even willing to make special variations of its process for AMD as seen in [1].
I think its clear why AMD wants to partner with TSMC, but I was always wondered whats the incentive the other way around that makes AMD attractive to TSMC.
One possible point is also mind-share as AMD has been getting many headlines but as TSMC is B2B, so I find that unlikely as TSMC's main value add from AMD over other manufacturers.
Foundries are expensive business, and TSMC likely wants big commits to make sure its bets pay off. TSMC only cares about what they are paid and NVIDIA likely isn't passing on much of its margin, so AMD's money is just as good. Except its not just AMD's money, its also Sony's and Microsoft's money. I would be shocked if there weren't minimum volume clauses in the contracts for making those console chips, custom silicon is expensive and AMD probably wants to have a guarantee it will break even. The TLDR is AMD can feel safer in making bigger commitments to fab time from its console agreements and product lineup, and TSMC can feel safer that this capacity will actually be used.
[0] https://www.techpowerup.com/269347/nvidia-geforce-ampere-gpu... [1] - https://www.pcgamer.com/amd-zen-4-specific-5nm-enhanced-node...