Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Microsoft and OpenAI end their exclusive and revenue-sharing deal

bloomberg.com

Post reply on HN
Print view View original

891–900 of 915 posts

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#891

Post by kllrnohj »

bigyabai wrote: post #692
mschuster91 wrote: post #675

Earlier quoted context omitted.

> Apple hertzes and apple bytes are much better than PC hertzes and bytes because they are made by virgin elves during a full moon. The thing that Apple has always been excellent at is efficiency - even during the Intel era, MacBooks outclassed their Windows peers. Same CPU, same RAM, same disks, so it definitely wasn't the hardware, it was the software, that allowed Apple to pull much more real-world performance out…

> NV didn't want to give Apple the specs any more to write drivers. I think that's a valid demand, considering Nvidia's budding commitment to CUDA and other GPGPU paradigms. Apple, backing OpenCL, would have every reason to break Nvidia's code and ship half-baked drivers. They did it with AMD's GPUs later down the line, pretending like Vulkan couldn't be implemented so they could promote Metal. Apple wouldn't have ma…

> They did it with AMD's GPUs later down the line, pretending like Vulkan couldn't be implemented so they could promote Metal.

It was even worse than that, they just stopped updating OpenGL for years before either Vulkan or Metal existed at all. Taking a Macbook and using bootcamp would instantly raise the GPU feature level by several generations just because Apple's GPU drivers were so fucking old & outdated.

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#892

Post by PhunkyPhil »

wg0 wrote: post #619
PhunkyPhil wrote: post #583

Earlier quoted context omitted.

Right, but this electron box led to one of the largest (if not the largest) media revolution that has transformed the course of humanity in a frightening way we're still trying to grapple with. Still saying "LLMs are autocorrect" isn't wrong, but nobody is saying "phones are just electrons and silicon" to diminish their power and influence anymore.

Electron box was reliable. It only depicted exactly the scan lines airwaves or signals ordered it to.

The people controlling what went on the screens were unreliable and nondeterministic. The algorithm on facebook/instagram is nondeterministic and I hope I don't have to convince you of the impact these algorithms have.

As far as I'm concerned, the nondeterminism argument is fruitless

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#893

Post by rishabhaiover »

teiferer wrote: post #742
rishabhaiover wrote: post #634

Earlier quoted context omitted.

You do realize that you need a human, a "SWE", to do the task that I just described? A computer can't do it.

You had a human to prompt the LLM to do the RCA, didn't you?

Your argument is not meant to tackle my core claim, it is to poke pedantic holes. What a waste of my time.

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#894

Post by wolvoleo »

gchamonlive wrote: post #853
wolvoleo wrote: post #809

Earlier quoted context omitted.

I understand how this came to pass (I didn't know it before so thanks for the insight!) But as a customer I absolutely hate working with AWS tech. Their stuff is a mess and I feel like I shouldn't have to get my head around their idiosyncracies. I prefer Azure even though Microsoft is a terrible company to work with. I find the AWS people and attitude a lot nicer but their services are a mess. If I do something new I…

I'm right there with you, don't get me wrong. Choosing a cloud provider is like choosing the lesser of many evils. We are, however, coming to a point where k8s is viable for most workloads, so it's less complex today to spin up a project with cloud mobility in mind than it was 10 years ago if you plan it right.

Oh really? I didn't know, thanks! I'll have a look at it.

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#895

Post by everforward »

Aerroon wrote: post #874
timacles wrote: post #852

Earlier quoted context omitted.

People are discounting the fact the almost no one wants a contraption on their head for an extended period of time. It is a universal preference Until VR is done via glasses or some wire you stick in your neck matrix style, it will never take off

People use helmets all the time for extended periods of time. I'm not sure that's such a deal breaker. I don't want to wear glasses for extended periods of time either, but I've had to get used to them.

The risk/reward tradeoff of helmets is potentially dying, and we still had to make laws requiring people to wear them.

Glasses also have a pretty compelling reason to wear them.

I don't think VR has as compelling of a reason, at least so far. Even if it does, you need people to get far enough in to see that reason which is a hurdle when it involves a device they don't want to wear.

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#896

Post by zozbot234 »

bigyabai wrote: post #886
zozbot234 wrote: post #791

Earlier quoted context omitted.

SOTA models are reportedly MoE, not dense.

A 5T MoE model is still bottlenecked by streaming weights from SSD, in addition to compute bottlenecks during prefill and decode.

True but a cluster built on pipeline parallelism can naturally stream from multiple SSD's in parallel. That probably makes offload somewhat more effective. And you also have RAM caching available as a natural possibility.

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#897

Post by bigyabai »

zozbot234 wrote: post #896
bigyabai wrote: post #886

Earlier quoted context omitted.

A 5T MoE model is still bottlenecked by streaming weights from SSD, in addition to compute bottlenecks during prefill and decode.

True but a cluster built on pipeline parallelism can naturally stream from multiple SSD's in parallel. That probably makes offload somewhat more effective. And you also have RAM caching available as a natural possibility.

You won't be RAM caching much of anything with experts that are 220b parameters worth of layers.

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#898

Post by teiferer »

rishabhaiover wrote: post #893
teiferer wrote: post #742

Earlier quoted context omitted.

You had a human to prompt the LLM to do the RCA, didn't you?

Your argument is not meant to tackle my core claim, it is to poke pedantic holes. What a waste of my time.

The argument I and others here are making is that what you call "intelligent" is a property that also other tools exhibit which are rarely called "intelligent". You can certainly do that, but that does not prove us wrong (and also doesn't fit what most people would consider "intelligence", as fuzzy as that concept might be).

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#899

Post by etempleton »

FartyMcFarter wrote: post #813
etempleton wrote: post #538

Earlier quoted context omitted.

Naming your company off a product that doesn't really exist yet and then ultimately fails is a pretty crazy and stupid thing to do. A bit cart before horse.

I think they were trying to disassociate themselves from the PR disasters Facebook was facing back then (privacy related IIRC).

Oh, I know, but they could have named themselves anything. They believed in the metaverse, They thought it was the next big thing. The same way Zuckerburg thought voice assistants, and AI will be a thing.

Re: Microsoft and OpenAI end their exclusive and revenue-sharing deal

#900

Post by jorvi »

kllrnohj wrote: post #666
jorvi wrote: post #660

Earlier quoted context omitted.

On Geekbench 5, the M1 hits 483 FPS and the RTX 3090 hits 504 FPS. There are other workloads where the M1 actually beats the 3090. Apple does plenty of hyping but it's always cute when irrational haters like you put them down. The M1 was (well, is) a marvel and absolutely smokes a 3090 in perf per watt.

What geekbench 5 fps are you talking about? Geekbench only has OpenCL and Vulkan scores for the 3090 as far as I can tell, and the M1 Ultra is less than half the OpenCL score of the 3090. And the M1 Ultra was significantly more expensive. Find or link these workloads you think exist, please > The M1 was (well, is) a marvel and absolutely smokes a 3090 in perf per watt. The GTX 1660 also smokes the 3090 in perf per wa…

Here you go[0]. 'Aztek Ruins offscreen'. Although I misremembered the exact FPS, the 3090 is at 506 FPS.

Also note how the M1 Ultra is pushing 2/3 of the FPS of the 3090 despite 1/3 of the power budget and the game itself being poorly optimized for the M-series architecture.

And here[1] you have it smoking an Intel i9 12900K + RTX 3900. The difference doesn't look too impressive until you realize the power envelope for that build is 700-800W.

Also, the GTX 1660 (technically an RTX 2000 series, but whatever) is about 26% less efficient than an 3090[2].

> Being more efficient while being dramatically slower

That's my whole point and what you're refusing to see. The M1 is not dramatically slower than an i9 or 3090 despite having dramatically lower power use.

The proof for this will really start to come once Qualcomm and Mediatek have gotten a handle on their PC ARM chips and Valve decides they're good enough for a Steam Deck 2 or 3. You'll get to see 2-3x the battery life along a modest performance increase.

[0]https://techjourneyman.com/img/blog/m1-ultra-vs-rtx-3090-ben...

[1]https://techjourneyman.com/img/blog/m1-ultra-vs-intel-i9-129...

[2]https://bestvaluegpu.com/comparison/geforce-rtx-3090-vs-gefo...

Post reply on HN