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

Nvidia buys $5B in Intel

tomshardware.com

Post reply on HN
Print view View original

411–420 of 638 posts

Re: Nvidia buys $5B in Intel

#411

Post by KronisLV »

kmacdough wrote: post #395
xp84 wrote: post #386

Earlier quoted context omitted.

Agree. Not only would there be no money in it to try to replace Iris graphics or whatever they call them now -- it would be ultra pointless because the only people buying integrated graphics are those where gaming, on-device AI, and cryptocurrency aren't even part of the equation. Now, that is like 80%+ of the PC market, but it's perfectly well served already. I saw this move more as setting up a worthy competitor to…

Calling BS on "gaming not part of the equation". Several of my friends and I have exclusively games on integrated graphics. Sure we don't play the most abusively unoptimized AAA games like RDR2. But we're here and we're gaming.

> Sure we don't play the most abusively unoptimized AAA games like RDR2.

Wait, RDR2 is badly optimized? When I played it on my Intel Arc B580 and Ryzen 7 5800X, it seemed to work pretty well! Way better than almost any UE5 title, like The Forever Winter (really cool concept, but couldn't get past 20-30 FPS, even dropping down to 10% render scale on a 1080p monitor). Or with the Borderlands 4 controversy, I thought there'd be way bigger fish to fry.

Re: Nvidia buys $5B in Intel

#412

Post by adabyron »

sfilmeyer wrote: post #311
techsystems wrote: post #267

The reason why Nvidia is buying now does not have to do anything with Arc or GPU competition. There are mainly two reasons. 1) This year, Intel, TSMC, and Samsung announced their latest factories' yields. Intel was the earliest, with 18A, while Samsung was the most recent. TSMC yieled above 60%, Intel below 60%, and Samsung around 50% (but Samsung's tech is basically a generation ahead and technically more precise),…

>2) It's just generally good to diversify into your competitors. Every company does this, especially when the price is cheap. This definitely isn't a thing that every company does (or even close to every company).

Not every company, but the largest ones do.

Microsoft once owned a decent amount of Apple & Facebook for example.

Re: Nvidia buys $5B in Intel

#413

Post by cherioo »

tiffanyh wrote: post #363
cherioo wrote: post #275

Earlier quoted context omitted.

I am curious where you get your information about Samsung being more “precise”. I was recently looking into 2nm myself, and based on wikipedia article on 2nm, TSMC 2nm is about 50% more dense than the samsung and intel equivalent. They aren’t remotely the same thing. Samsung 2nm and Intel 18A are about as dense as TSMC 3nm, that’s been in production for years.

This information is a bit dated but ... Since "nm" is meaningless these days, the transistor count/mm2 is below. As reference: TSMC 3nm is ~290 million transistors/mm2 (MTr/mm2). IBM TSMC Intel Samsung 22nm 16.50 16nm/14nm 28.88 44.67 33.32 10nm 52.51 100.76 51.82 7nm 91.20 237.18 95.08 5nm 171.30 3nm 292.21 2nm 333.33 https://news.ycombinator.com/item?id=27063034 https://www.techradar.com/news/ibm-unveils-worlds-fir…

I think the intel 7nm is unrealistic. If true intel wouldn’t be “behind”

According to Wikipedia intel 7nm density is ~62 MTr/mm2. I cannot find the source wikichip page mentioned in your reference post.

FWIW, I am not in the semi industry and all my info are from Wikipedia https://en.m.wikipedia.org/wiki/7_nm_process https://en.m.wikipedia.org/wiki/2_nm_process

Re: Nvidia buys $5B in Intel

#414

Post by Sohcahtoa82 »

mysteria wrote: post #379
Sohcahtoa82 wrote: post #352

Earlier quoted context omitted.

FWIW, my slowness is because of quantizing. I've been using Mistral 7B, and I can get 45 tokens/sec, which is PLENTY fast, but to save VRAM so I can game while doing inference (I run an IRC bot that allows people to talk to Mistral), I quantize to 8 bits, which then brings my inference speed down to ~8 tokens/sec. For gaming, I absolutely love this card. I can play Cyberpunk 2077 with all the graphics settings set to…

Hold on, you're only getting 45 tokens/sec with Mistral 7B on a 5090 of all things? That gets ~240 tokens/sec with Llama 7B quantized to 4 bits on llama.cpp [1] and those models should be pretty similar architecturally. I don't know exactly how the scaling works here but considering how LLM inference is memory bandwidth limited you should go beyond 100 tokens/sec with the same model and a 8 bit quantization. 1. https…

My understanding is that quantizing lowers memory usage but increases compute usage because it still needs to convert the weights to fp16 on the fly at inference time.

Clearly I'm doing something wrong if it's a net loss in performance for me. I might have to look more into this.

Re: Nvidia buys $5B in Intel

#415

Post by mosura »

jerf wrote: post #362
bee_rider wrote: post #319

Earlier quoted context omitted.

The US government isn’t (or at least shouldn’t be) profit-motivated anyway, so it isn’t obvious what their incentives are WRT Intel’s stock.

They want a source of chips for the wars they want to conduct that is not either controlled by the party they want to go war with, or way way closer to the party they want to go to war with than they are. Buying a chunk of Intel is a way of making sure they do the things the government wants that will lead to that outcome. Or at least so the theory goes; I've got my own cynicism on this matter and wouldn't dream of t…

> Right now if the US wants to go to war with China

The US is desperate to not have that war, because they spent so long in denial about how sophisticated China has become that it would be a total humiliation. What you see as the US wanting war is them simply playing catch up.

Re: Nvidia buys $5B in Intel

#416

Post by clifflocked »

Sohcahtoa82 wrote: post #272
tw04 wrote: post #191

Earlier quoted context omitted.

>Consumer gpus are totally different products from the high end gpus now. Intel has failed on the gpu market and has effectively zero market share, so it is not actually clear there is an antitrust issue in that market. It would be nice if there was more competition but there are other players like AMD and a long tail of smaller ones I'm sorry that's just not correct. Intel is literally just getting started in the GP…

I don't know what anybody would do with such a weak card. My RTX 5090 is about 10x faster (measured by FP32 TFLOPS) and I still don't find it to be fast enough. I can't imagine using something so slow for AI/ML. Only 2.2 tokens/sec on an 8B parameter Llama model? That's slower than someone typing. I get that it's a budget card, but budget cards are supposed to at least win on a pure price/performance ratio, even with…

I feel as though you are measuring tokens/s wrong, or have a serious bottleneck somewhere. On my i5-10210u (no dedicated graphics, at standard clock speeds), I get ~6 tokens/s on phi4-mini, a 4b model. That means my laptop CPU with a power draw of 15 watts, that was released 6 years ago, is performing better than a 5090.

> The 5090 is 10x faster but only 6-8x the price

I don't buy into this argument. A B580 can be bought at MSRP for 250$. A RTX 5090 from my local Microcenter is around 3250$. That puts it at around 1/13th the price.

Power costs can also be a significant factor if you choose to self-host, and I wouldn't want to risk system integrity for 3x the power draw, 13x the price, a melting connector, and Nvidia's terrible driver support.

EDIT: You can get an RTX 5090 for around 2500$. I doubt it will ever reach MSRP though.

Re: Nvidia buys $5B in Intel

#417

Post by mmaunder »

Intel should never have existed in the first place. We should have gone the RISC route in the 80s and 90s and it took 30 years for the world to realize that with ARM. It’s like we’re continuously resuscitating a zombie to terrorize us, instead of just letting it die.

Re: Nvidia buys $5B in Intel

#418

Post by agentcoops »

evanjrowley wrote: post #211

Nvidia's stake in Intel could have terrible consequences. First, it is in Nvidia's interest to kill Intel's Arc graphics, and that would be very bad because it is the only thing brighing GPU prices down for consumers. Second, the death of Intel graphics / Arc would be extremely bad for Linux, because Intel's approach to GPU drivers is the best for compatibility, wheras Nvidia is actively hostile to drivers on Linux.…

The article hints at it, but my guess would be this investment is intended towards Intel foundry and getting it to a place where NVIDIA can eventually rely on them over TSMC — and the ownership largely to give them upside if/when Intel stock goes up on news of an NVIDIA contract etc. It isn’t that uncommon of an arrangement for enterprise deals of such a potential magnitude. Long-term, however, and without NVIDIA making the call that could definitely have the effect of leading to Intel divesting from directly competing in as many markets, ie Arc.

For context, I highly recommend the old Stratechery articles on the history of Intel foundry.

Re: Nvidia buys $5B in Intel

#419

Post by Sohcahtoa82 »

dragonwriter wrote: post #325
Sohcahtoa82 wrote: post #272

Earlier quoted context omitted.

I don't know what anybody would do with such a weak card. My RTX 5090 is about 10x faster (measured by FP32 TFLOPS) and I still don't find it to be fast enough. I can't imagine using something so slow for AI/ML. Only 2.2 tokens/sec on an 8B parameter Llama model? That's slower than someone typing. I get that it's a budget card, but budget cards are supposed to at least win on a pure price/performance ratio, even with…

> My RTX 5090 is about 10x faster (measured by FP32 TFLOPS) and I still don't find it to be fast enough. I can't imagine using something so slow for AI/ML. Only 2.2 tokens/sec on an 8B parameter Llama model? That's slower than someone typing. Its also orders of magnitudr slower than what I normally see cited by people using 5090s; heck, its even much slower than I see on my own 3080Ti laptop card for 8B models, thoug…

Yeah, I must be doing something wrong. Someone else pointed out that I should be getting much better performance. I'll be looking into it.

Re: Nvidia buys $5B in Intel

#420

Post by KeplerBoy »

ninetyninenine wrote: post #164
jonbiggums22 wrote: post #153

Earlier quoted context omitted.

Way back in the mid-late 2000s Intel CPUs could be used with third party chipsets not manufactured by Intel. This had been going on forever but the space was particularly wild with Nvidia being the most popular chipset manufacturer for AMD and also making in-roads for Intel CPUs. It was an important enough market than when ALi introduced AMD chipsets that were better than Nvidia's they promptly bought the company and…

nvidia does build SOCs already. The AGXs and other offerings. I'm curious why they want intel despite having that technical capability of building SOCs. I realize the AGX is more of a low power solution and it's possible that nvidia is still technically limited when building SOCs but this is just speculation. Does anybody know actual ground truth reasoning why Nvidia is buying Intel despite the fact that nvidia can m…

Why is Nvidia partnering with Mediatek for CPU cores in the dgx spark? Different question, probably the same answer.
Post reply on HN