The problem i see is that I dont see anyone who is suitable to be a ceo of Nvidia. It’s a special blend of hardware and software chops that no other CEO has imo.
Very very few HW CEOs have even the slightest ability to appreciate the key strength of Nvidia which is ability to build hardware and then drive it with outstanding software.
Look at AMD. Great hardware but hamstrung by their software for 20 years.
Funny how that perspective completely changed because CUDA ended up being a better idea than initially anticipated, while the need for well-designed and -maintained linux graphics drivers.. somehow subsided? (my nvidia drivers still deadlock way too often)
He has poor judgement but I’ve yet to see any of the major consumer hardware companies that understand both hardware and software in the way he did. Or really the value of a fully vertically integrated stack.
Nvidia thrived because it paired great hardware with great software.
AMD and Intel think hardware is enough; it’s not. There’s a reason neither has ever been able to compete with Nvidia.
They have a giant ego, but I’d argue it’s probably well deserved.
Apple is/was similar. They also see that hardware without great software is useless. Spend as much; if not more on software to make your hardware shine.
Intel has some good compilers that take advantage of their hardware don't they? I think AMD now understands the software side of things, and is trying to fix it. Wonder how long it will take for them to get close to Nvidia's software side?
That was 15 years ago. They haven’t done it since.
Look at optane; amazing hardware but total flop.
Nvidia made their market, they had these GPUs which were great at linear algebra so they found new systems to drive more demand for more powerful chips, things like PhysX, CUDA, DLSS, AI inference, etc.
They induce demand where others chased it.
Doesn’t excuse his faults but you can’t ignore what the company did.
According to you, in comparison to the current competition, hardly Microsoft Google, Amazon, Nvidia, AMD, the Koreans, the Chinese, the Linux world, they don’t add up not when it comes to hardware and software and the ability to create an ecosystem, you’re actually dead wrong.
Another thing is wrong was thinking that Apple should run in and build some type of AI model that’s proving to be well wrong the only thing in Apple’s way right now is memory. certainly isn’t money, and they certainly have the talent and house to fix the memory problem. The question is whether or not they have the will to do so.
No one equals Apple at this point in time when it comes to seamlessly integrating multiple devices in a useful way. Be it a desktop computer, laptop a smartphone, tablet, smartwatch, Apple Vision, AirPods with each iteration, once again in comparison to their competition, they seem to have the best hand at this time.
Memory cost is the biggest problem Apple has right now and it’s not like they don’t lack the money and with the Apple Silicon design group they certainly have the capability to fix it, or I should say design around it.
Yes, it might take 2 to 5 years to do it, but it’s not impossible, the main question is, do they have the will? And they’ve certainly shown it when it comes to processors, GPUs, modems, and other support chips that they can execute over time.
They may not execute on what you want or what I want, but in comparison to their competition, who is out there right now that can combine both the software and the hardware at a high level across the board?
I have no idea what you're trying to say in relation to my comment, but you do try to set up and take down several wild straw man arguments which I never made, so you might as well respond to your own comment and keep it going, I don't even need to participate.
This is such an arrogant, presumptive and mean-spirited comment that I cannot imagine you actually want an answer or a healthy conversation about the topic.
You need to review Hacker News site guidelines and refrain from posting flamebait.
Yes, we’re gonna find out how important he was. If he is on the level as some of the other great CEOs Nvidia gonna have trouble, this little flare could be the beginning of the AI bubble bursting. Because I don’t see Intel or AMD rising to the same level as Nvidia. Microsoft seems to be having bad luck with their team-ups.
Yeah. Every major self driving model that I’m aware of is fully e2e at this point. Going from fused sensor output to control+debug vectors.
This is more generalised.
But also since there’s a huge volume of data it’s too expensive to just keep scaling compute up (per car overhead) so there are necessary tricks involved.
I do think having a large model that can do this means that a small specialised model could be distilled form it though. Which is probably the most feasible path to production IMO.
I know it sa toy, but if someone gets the bright idea to do this for real: please learn from npm and dont duplicate packages for every clone of a project. Share them between projects like every sane package manager.
Before uv there was poetry, and before poetry there was pipenv, before pipenv there was virtualenv. I wish people would stop portraying Python tooling completely wrong like that. uv is not the first tool to solve most problems in this space. It may be the best performing though and may be the best overall currently.
I think you’re underselling it. You’ve listed several projects with many thousand man hours in development.
It’s telling that in Python there have been multiple package managers over the years to fix the broken status quo. Typescript/Javascript notwithstanding, in other languages and environments you get the one tool that’s been there for over a decade without competitors.
Very few languages have suffered from the same package manager issues as JS/TS and Python.
You don't want to install all dependencies globally. You only want to cache them globally. Each project should resolve its own dependency tree independently. If the wanted version is already in cache, use that, otherwise download from the repository even if an older / newer version is present in cache (unless the user specifically says to use the cache only). As long as your dependencies are inert (they don't mess with the system just by being present), this will work perfectly fine.
For non-inert packages, such as applications installed via the language's dependency manager, that's a different use case and should be handled differently.
Virtual venv existed before uv. So you could still just generate a virtual venv and then use pip to install packages in the local venv. (2007: virtualenv was released as a third-party tool).
Very very few HW CEOs have even the slightest ability to appreciate the key strength of Nvidia which is ability to build hardware and then drive it with outstanding software.
Look at AMD. Great hardware but hamstrung by their software for 20 years.
Hardware is not enough.
reply