Why is everyone thinking battery storage needs to be one big giant battery?

Imagine 10 000 residential single family houses. Then imagine:

- Each has a 20 kWh battery installed (200 MWh)
- For every fiftieth household, a 2 MWh buffer is installed (400 MWh)
- For every 1000th household a 20 MWh town buffer is installed (200 MWh)
- For every 10 000th household a 200 MWh battery is installed

So 1 GWh storage in total for 10 000 households for a total cost of perhaps $25k per house. Unrealistic?

As a consumer would you rather buy the $20k Toyota Corolla Electric or the $25k Toyota Corolla Hybrid? Are we there yet, no, Toyota Corolla Electric costs $35k at the moment (actually, Corolla Electric does not exist, but the equivalent BYD Dolphin does). We will be there before 2027 though. Perhaps even 2025.

So, let's say you go with the Hybrid. Many will not, you might. At ~15% fleet penetration, some gas stations will start to get dismantled. How far away can a petrol station be, before it is too inconvenient to refuel there? 10 miles? 20? 50? 100?

Those two factors more or less guarantee the demise of ICE engines. You simply cannot find any fuel for them by 2035.

Keep laughing, reality will hit you like a truck soon enough. :)

https://www.cer-rec.gc.ca/en/d...

No. Plain old capitalism is going to "outlaw" CO2 sources.

I don't understand why so few people understand this, but here it is again;

1. Power grids based on Wind, Solar and Batteries are an order of magnitude cheaper to build, maintain and generate electricity with over more traditional fuels such as coal, gas, oil, water and nuclear. We will decarbonize the energy grid before 2040 because doing anything else will cost us 10x as much. In countries like Australia, coal and gas is already no longer an option to burn for electricity in plants that are built and operational.
2. Transports will electrify since they are an order of magnitude cheaper to run on electric over gasoline and diesel. There are already electric trucks out there that outperform traditional ICE trucks in pretty much all relevant metrics - and their TCOs are as cheap, or even cheaper, than subsidised gas and diesel.
3. Traditional agriculture will be outcompeted by hydroponics (growing vertically with 1/10th the water and land area), meat production will be outcompeted by lab grown meat. Why keep animals alive when you can just grow the proteins in a factory? Same thing with dairy, artificial cow milk will soon be cheaper to produce than real cow milk.

So, yeah... Food, water and energy are soon going to be ubiqutous for everyone in the entire world. By 2050, 90% of human CO2 emissions will be gone.

Well, in practice this is only relevant if you are on a roadtrip.

Manwhile have fun driving 20 minutes to the pump and then the same back to refill your gas car. ðY(TM)

New BEVs in Sweden 2023 is roughly 125k and 40% of the market. That is 2.5% fleet replacement and this is expected to double by 2027.

However as more buy electric fewer will want to stop at a gas station -> mass dismantling of pumps -> more buying BEVs. So by 2030 we might have a couple of years with 500k new BEVs each year.

To be fair, writing kernel code in C++, where you want to keep things as close to the metal as humanly possible, ideally writing assembler if the damn thing wasn't so hard to trace... Is not the best of ideas.

In kernel space you can't use a ton of the features C++ brings to the table without incurring significant performance problems. Often times memory is constrained, and just invoking an iterator can allocate four bytes of memory extra. "Bruh who cares about 4 bytes don't you have two gigs spare?" and again, no you do not. Modern CPUs have L1, L2, L3 cache and keeping within the limits of these caches massively reduce memory paging and thus massively reduce the time spent waiting for, say, L3 calls.

When you are writing code that should fit within 16 kb of L1 cache, all of a sudden memory usage becomes very important. This is not to say C is the proper tool everywhere within the kernel though.

No, I am advocating replacement because the piece of shit junker hardware you are pulling is the equivalent of a petrol car that draws 3 miles per gallon.

Seriously, a 7900 build would almost guaranteed be an upgrade from what you are sitting on, would cost you $1500, and would save you $50 a month* in power bills alone. 2½ years later and the new one has paid for itself, and it will still be twice as fast as what you are pulling today.

This is before we take Spectre and Meltdown mitigations into account. If your Xeons are too old for that, then a full $1000 7600 build would run circles around your Xeons.

Sorry, but there is a point when "Good Enough" is just not cutting it anymore. Treat yourself to something nice and modern. You've earned it. Your money, your time and your parts though :)

* Assuming electricity price of $0.15 per kWh.

Because you can double, triple or even quadruple the performance for half the money you paid for your current system?

It does make a difference and if you can't see why, you have clearly not run a modern PC for a while.

Ah yes, Xeons... Because running 400W CPUs when you could be running 50W CPUs for a modest upgrade is a really great way to save on electricity bills :D

Seriously, do have a look at NUCs here. Same performance for $300-$500 and saves you maybe 400W-600W, that boils down to ~50-100 kWh saved. And if you are using this as a server, not a desktop, why the heck have you not migrated to Linux yet, Docker allows you to run Windows servers in a VM nowadays you know...

Here is why. In 2017, a high end $1800 build looked like this:

CPU($350) - 7700k
Cooler($150) - 240 rad AIO
Memory($95) - 4x2GB sticks
Motherboard($175) - Z270 Motherboard
Storage($90+$70) - 256GB NVMe + 2TB HDD
Case($130) - Whatever fits ATX and decent airflow
PSU($150) - ATX 850W
GPU($590) - GeForce 1080
Total: $1800

In fall of 2023, a low-mid gaming build looks like this:

CPU($210) - 13400
Cooler($30) - 120mm Tower Air Cooler
Memory($50) - 2x16GB sticks
Motherboard($100) - Z690 Motherboard
Storage($80) - 2TB NVMe with DRAM cache and TLC
Case($100) - Whatever fits ATX and decent airflow
PSU($120) - ATX 850W
GPU($310) - Radeon 6700 XT
Total: $1000

So what do you get? Basically a doubling in everything except GPU and storage, for ~half the money. For a true doubling, get an RTX 4070 for $520 or a Radeon 7800 XT, and add another $80 for another 2TB NVMe disk, so that is like, $300 extra. For roughly 70% of the money you have doubled all stats of a high end gaming machine from 2017. Oh, and did I mention this one draws half the same power envelope?

If all you need is a surf PC though a simple NUC for $300-$400 should suffice.

Sooooo... I hope you are aware an Intel core i3 13100 CPU is more powerful than both anything on AM3, AM3+ and the 7900k? This means a $130 CPU beats whatever you are rocking right now.

You can get Mobo + RAM + CPU for $300-$400. A 5700X build could also be interesting, though 13th gen is a much better deal than AM4 builds.

You are aware NT was a basis from Windows XP and onwards right? Last DOS version was NE.

"No one should find manufacturing inspiration from the builder of the worst quality cars in the industry."

According to Sandy Munro they are now the best quality in the industry, but yes they did start with pretty bad quality cars.

Here is the thing... Ford needs 12-18 months to improve a specific manufacturing process. Tesla require 12-18 hours to improve something. Ford improves a lot of things and release a new revision after 18 months. Tesla release new revisions on a weekly basis, without ever needing to slow down.

Do not believe the oversell on this though, this method of working is not without drawbacks, it's just that Tesla can tap into a large pool of undervalued college educated Millenials and Gen Zs that makes this way of working viable. The reason Tesla can move as fast as they do is because everyone is allowed to try and change the production cells in a factory, and if that change turns out to be more efficient, it gets copied to the rest of the cells in that factory.

Problem with that is, not every change is great and the more you optimize the less likely it is that a cell can be optimized further. Still, if only 1/10 changes work better, then that is still 10 more ways of making things more efficient every week, if 100 cells were altered across all Gigafactories.

It is a great way to work in a rapidly evolving market but eventually the EV industry will hit a brick wall where new innovations will be very hard to come by, just like ICE vehicles did. Problem for Ford et. al. is, we will reach that point in 2035 sometime and most ICE vehicles will stop being acceptable on the market by 2028.

Really, with 4 TB SSDs costing less than $150 now, any HDD less than 8TB is a downgrade these days.

HDDs need to come out with 30 TB consumer drives now and 60 TB drives by 2026, or see themselves outcompeted by cheap $200-$300 32 TB SSDs...