If you look at the details you can clearly see SDL3_GPU is wildly different from this proposal, such as:
- It's not exposing raw GPU addresses, SDL3_GPU has buffer objects instead. Also you're much more limited with how you use buffers in SDL3 (ex. no coherent buffers, you're forced to use a transfer buffer if you want to do a CPU -> GPU upload)
- More modern features such as mesh shading are not exposed in SDL3_GPU, and keeps the traditional rendering pipeline as the main way to draw stuff. Also, bindless is a first class citizen in Aaltonen's proposal (and the main reason for the simplification of the API), while SDL3_GPU doesn't support it at all and instead opts for a traditional descriptor binding system.
SDL3 is kind of the intersection of features found in DX12/Vulkan 1.0/Metal: if it's not easily supported in all of them, it's not in SDL3-- hence the lack of bindless support. That means you can run on nearly every device in the last 10-15 years.
This "no api" proposal requires hardware from the last 5-10 years :)
Yup you've actually pointed out the most important difference: SDL3 is designed to be compatible with the APIs and devices of the past (2010s), whereas this proposal is designed to be compatible with the newer 2020s batch of consumer devices.
Over a couple of decades VCs have invested in vanity startups that cost billions of dollars like it's nothing, countless times.
I think half a billion isn't that expensive for a program that searches for a potential "theory of everything" that can profoundly change our understanding of the universe (even if it brings no results!)
We're long past the point where people can claim string theory has contributed nothing. AdS/CFT correspondence helped us understand what happens to information in black holes, and brought us the holographic principle which now is looking like it might potentially be the next big conceptual revolution in physics. Holography is making meaningful predictions in nuclear and plasma physics right now
Holography desperately needs it's own Brian Greene style ambassador to share the good news. In terms of momentum and taking center stage, it's now in the place where String Theory used to be like 10-15 years ago as the frontier idea with all the excitement ever momentum behind it, and it has been borne from the fruit of string theory. It's quite amazing times we're living in but I think there's been no energy in the post covid world to take a breath and appreciate it.
Last i’d heard, ‘they provided an interesting alternative way of thinking of the problem’ but provided no unique insights or additional testable behaviors. The folks using the alternative theories ended up being able to formulate them more directly using other (‘normal’) physics later. Do you have a cite for anything contrary to that?
It doesn’t help that when something does finally seem experimentally provable (Craig Hogan noise for example), but then gets tested and seems disproven, then it gets ‘removed from Canon’ as it were and ‘is not string theory’.
I'd like to express my doubts about your ability to understand their theories more colourfully, but I'm afraid I'm also under close scrutiny around here.
This exact long running issue with string theory is surely my imagination, and I’m the only one who has commented on it. luckily it’s easy to prove me wrong.
We already have an abundance of natural resources (at least... for the developed countries) The problem is that they aren't evenly distributed enough, to the extent that a lower population cannot counter the increasing wealth inequality.
>to the extent that a lower population cannot counter the increasing wealth inequality.
Wealth inequality has nothing to do with it; some of the countries with the lowest wealth inequality like Northern Europe have the lowest birthrates. A hundred years ago wealth equality in most countries was much higher than now and people were much poorer, yet they were still having many more children than people today.
We already have an abundance of natural resources (at least... for the developed countries)
I'm not convinced this is true.
You can blame the billionaires who own the vast majority of the wealth but that's mostly due to the stock market giving them that value on paper. Physical resources stay finite.
Kinda interesting that they were able to resurrect the WebKit engine on Windows, might have been quite an engineering feat. In the past (2007-2012) Apple distributed a Windows version of Safari, but it's been more than a decade since it was deprecated and the engine only supported Apple's ecosystem.
It's a wild read, that not only explains how this idea is technically horrible - he also analyzes the social and psychological effects this would have on humanity, and it's harrowing:
- We will be unable to see blue skies anymore, it will simply become white. This will be a traumatic event for humanity, and would have consequences beyond our understanding. Also, sunsets will be much more bloody-red, and we wouldn't be able to see the stars as clearly as before.
- The nature of aerosol deployments is that once we start this to combat global warming, corporations and governments will just rely on this instead of also reducing CO2 emissions. So as time passes, an exponentially larger amount of aerosols would be needed to block out the sun at every year, which makes the side-effects of geo-engineering grow immensely over time.
- Once aerosols hit a limit, then there lies the termination shock - once we "give" up on emitting aerosols due to it becoming exponentially expensive to maintain over time, the Earth's temperature will suddenly shoot up at unprecedented rates, quick rendering the planet inhabitable for humanity. His analogies to Freud's theory of repression is apt - we constantly repress our traumatic thoughts with stopgap measures, until the fantasy becomes too expensive to maintain and enter into the fully destructive phase.
If you want to read how horrible and scary the status quo is, just look at any scientific projection of current trends in 10 or 50 years.
> corporations and governments will just rely on this instead of also reducing CO2 emissions
They already rely on people wanting cheap gasoline and cheap beef and high profits to the companies that give generous political donations. So, no change in actual behavior.
Geo-engineering is already happening and has been for decades. It's just been as a side effect of industry. Doing a little bit of intentional work to counteract that is a reasonable response. Hoping that tomorrow everyone wakes up and decides to do the right thing about emissions is a fantasy.
Not all jobs are fun, but they can be bearable if meaningful enough (whether that being useful for other people, or even just provide a living wage to support your family)
If they really cared about performance (which I think is the case since they're using C++ for this) - they should use something else than nholmann-json. It's simple to use, but has one of the most atrocious performance for a C++ JSON library. (The safest choice would probably be RapidJSON)
As a side note, Intel's discrete GPUs are also famous for high-quality video transcoding - it was quite popular for streamers who needed a second helper PC only for OBS streaming.
Probably they're using the Intel MKL library for their linear algebra (which is severely gimped on AMD - SIMD is disabled and only the scalar fallback code runs).
If they've wrote SIMD code themselves then the gap between the two shouldn't be big (AMD's are actually better for SIMD nowadays, since the recent models support the AVX-512 instruction set while Intel ended support for that due to the P/E core split fiasco.)
- It's not exposing raw GPU addresses, SDL3_GPU has buffer objects instead. Also you're much more limited with how you use buffers in SDL3 (ex. no coherent buffers, you're forced to use a transfer buffer if you want to do a CPU -> GPU upload)
- in SDL3_GPU synchronization is done automatically, without the user specifying barriers (helped by a technique called cycling: https://moonside.games/posts/sdl-gpu-concepts-cycling/),
- More modern features such as mesh shading are not exposed in SDL3_GPU, and keeps the traditional rendering pipeline as the main way to draw stuff. Also, bindless is a first class citizen in Aaltonen's proposal (and the main reason for the simplification of the API), while SDL3_GPU doesn't support it at all and instead opts for a traditional descriptor binding system.
reply