Nvidia Is Now Worth More Than Meta (pcgamer.com) 60
Nvidia is now a larger company than social media giant Meta. PC Gamer reports: In a meteoric turn of events, Nvidia has surged to become the 7th largest company in the US, despite being nowhere close only a few years ago, and helped along by Meta's recent share price collapse. Meta's fall from stock market grace this past week saw 30% of its share value wiped out, leaving it with a total value in shares, or market cap, of just $615.70B (at time of writing). That's clearly still a lot of money, but it's notably less money than it was worth at the beginning of last week -- around $260B less.
Compare that to Nvidia's market cap of $657.06B, and the green team is out on top. Perhaps not for long, but we'll see. That's still a little shy of Berkshire Hathaway in 6th place at over $720B, but it's markedly higher up than Nvidia was only a few years ago, when its share value was a small fraction of what it is today. [...] Nvidia officially terminated its attempt to buy Arm, the UK-based chip designer, for $40B, and that did see some value wiped off its share price in the following days. Though clearly that dark cloud hasn't stuck around Nvidia's Santa Clara HQ, as it's now back up to around $260. That's over 40% up on its lowest point this year, and just under 22% down on its all-time high of $334.
Compare that to Nvidia's market cap of $657.06B, and the green team is out on top. Perhaps not for long, but we'll see. That's still a little shy of Berkshire Hathaway in 6th place at over $720B, but it's markedly higher up than Nvidia was only a few years ago, when its share value was a small fraction of what it is today. [...] Nvidia officially terminated its attempt to buy Arm, the UK-based chip designer, for $40B, and that did see some value wiped off its share price in the following days. Though clearly that dark cloud hasn't stuck around Nvidia's Santa Clara HQ, as it's now back up to around $260. That's over 40% up on its lowest point this year, and just under 22% down on its all-time high of $334.
write drivers? (Score:2)
And yet, despite all that money, they are incapable of writing drivers that don't suck.
Re: (Score:2)
Well I have 2 Windows machines with NVIDIA cards that have never had a problem with drivers. I do however also have a Linux box and I have had issues with those drivers. So I can say Linux drivers suck I guess.
Re: (Score:2)
No Wayland support, but that doesn't really bother me, as X is still well-supported for now.
Re: (Score:3)
No Wayland support, but that doesn't really bother me, as X is still well-supported for now.
It's taken 13 years to fail to displace a system that wayland's developers declared to be really rather bad.
The trouble is that it's just not a very well thought out system. It pushes almost everything on to the compositors and toolkits, so the devs can tell us that's "out of scope" and what's in scope works. But the result feels somehow much more balkanised than under X. You can have a compositor that does remoting.
Re: (Score:2)
It performs better for a reason. That's all it was really designed to do.
Now, I'm one of the rare people who do use X forwarding every single day, so I'm not eager for the transition to Wayland, but it's happening. It's the default display server on every distribution I can think of now.
You can still make X forwarding work via XWayland and funky environment vars to tell the TKs to render to the X server/bridge instead of their local window, but it's a little janky.
Ult
Re: (Score:2)
It performs better for a reason. That's all it was really designed to do.
Does it in practice though? There are definitely a few cases where it undoubtedly performs better but in an irrelevant way: it certainly cuts down on the number of context switches to process a keypress compared to X with a compositor. That maybe halves the latency, but that's taking it from microseconds to microseconds, so it's not something that really bothers me.
For straight line graphics performance, X has some pretty decent paths
Re: (Score:2)
Does it in practice though? There are definitely a few cases where it undoubtedly performs better but in an irrelevant way: it certainly cuts down on the number of context switches to process a keypress compared to X with a compositor. That maybe halves the latency, but that's taking it from microseconds to microseconds, so it's not something that really bothers me.
Not just keypresses- all input, and associated actions, like moving a window.
Also, the direct rendering pathway is much leaner. You're talking directly to the compositor instead of the compositor via the X server. Generally, in benchmarks, there's around a +10% uptick, which who cares about honestly, really it's about the 10% less work done under normal circumstances meaning better power utilization for the increasing number of things that use the GPU for mundane tasks.
I haven't seen a very coherent argument to that effect. I've seen some pretty disingenuous ones from people who I know know better.
It took 12 years to get middle click paste ironed out and forwarding is still janky 13 years later! On the balance of evidence I'd say both have been tried and while TK development sounds better in practice it has not proven to be.
X struck what turned out to be a very good balance (there was no requirement that libinput or mode setting were ever in the server, and on old unix workstations that often wasn't the case with the equivalents). Moving those things out had no effect on the X protocol, or the programs using X11, it was purely an implementation detail of the server.
Alright, that's fair.
Ultimately thou
Re: (Score:2)
Not just keypresses- all input, and associated actions, like moving a window.
I'm skeptical of the claim. Once the X server or compositor has the window contents, it just redraws the surface elsewhere.
Also, the direct rendering pathway is much leaner.
X has direct rendering.
I'd also argue that the 12 year metric isn't quite fair. Wayland integration into live desktops didn't really begin in earnest until recently, and it's advancing very quickly.
They've supported wayland for ages, but they haven't supported
Re: (Score:2)
I'm skeptical of the claim. Once the X server or compositor has the window contents, it just redraws the surface elsewhere.
Every movement of the mouse is a pass from the kernel, to the X server, to the compositor, to the client.
X has direct rendering.
Sure does. However, the communications via the DRM is through the X server, and the X server has to handle contention between its clients (including the compositor)
We're on DRI3, now, which almost doesn't suck (we can finally allocate our own buffers, now!)
Of course, that comes with its own layer of cruft- specifically PRIME, which sucks butthole.
They've supported wayland for ages, but they haven't supported wayland well for ages. We also heard a lot about how X sucks over and over and over. So far it's still taken 13 years and about the best users say with a wayland setup is it works about as well as X and there aren't too many bugs. I think the claims about X sucking just aren't bourne out by the reality of the transition.
OK, that's just nonsense. Who is this mystical "they"?
G
Re: (Score:2)
Every movement of the mouse is a pass from the kernel, to the X server, to the compositor, to the client.
Oh ok I misunderstood. But I still don't think that's significant.
OK, that's just nonsense. Who is this mystical "they"?
GNOME wasn't fully ported to Wayland until 2015.
Well quite. It took quote a long time to do the porting.
KDE, work started last year.
According to this URL:
https://community.kde.org/KWin... [kde.org]
people were giving presentations about it in 2014, which means work took 8 years.
No, absolutely not.
Re: (Score:1)
Re: (Score:2)
The bar is indeed very low but nVidia is still worlds away from clearing it. The AMD driver takes 269MB out of a 1.2GB kernel source with tens thousands of other drivers, doesn't work when built not as a module -- but at least I can't blame it for a single crash, not being able to run a -rc or even released kernel, artifacts on the screen, etc.
Nvidia is next (Score:5, Insightful)
I think it's sitting on a giant bubble, if cyrpto comes down, Nvidia is partially going down with it.
Re: (Score:2)
I think it's sitting on a giant bubble, if cyrpto comes down, Nvidia is partially going down with it.
Nvidia hardware has uses besides crpto mining, you know. All a crypto-currency bubble bursting will do is bring video card prices back down to a sane level.
Re: (Score:2)
Nvidia and AMD vendors should start making cards that are just the graphics engine with no video output so that the f***ing miners can use them, and then the rest of us can buy graphics cards for a sane price.
Re:Nvidia is next (Score:4, Informative)
Like this: https://www.nvidia.com/en-us/d... [nvidia.com] ?
Re: (Score:2)
I think they tried lockouts, the miners figured them out and rendered them useless.
Re: (Score:2)
AMD tried. the RX 6500XT is the result. Terrible doesn't even com close...
Driver lockouts are the Green Goblin's domain. Doesn't work either.
Re: (Score:2)
"Videogames are a bigger industry than movies and North American sports combined, thanks to the pandemic"
graph [marketwatch.com]
Re: (Score:2)
That's probably why they want to use their bubble money to buy stuff. Diversify as fast as possible.
Re: (Score:1)
In another way I might say Nvidia was always more valuable than Meta but the market has just caught up with the fact.
Re: (Score:2)
Shows you don't know NV's revenue streams very well. There ARE threats to their future, but that isn't a major one.
Re: (Score:2)
Yeah, no one wants to see computers make pictures.
Re: (Score:2)
You can't freaking buy a card to make it make pictures, they are being sold to turn electricity into heat so people can win a lottery.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
There they offer a lot of proprietary solutions that artists (for video games, advertising, and many other fields in modern media) have become dependent on over the last decade.
Their profits will shrink if crypto mining goes down, and I hope it crashes and burns in hell. They'll finally have to make decent graphics cards again, because people would not longer buy whatever crap they release, because of lack of choices. But they're nowhere near in trouble b
Come a long way (Score:5, Interesting)
I remember when NVidia was hot on the scene with the Riva TNT card and as someone who was using a Voodoo 1 card being a little jealous that you could have just one graphics card for both 2d and 3d. Just a few years later they were buying 3dfx no less.
Honestly well deserved, especially compared to Facebook. NVidia are kind of dicks but they have consistently been the ones pushing a lot the graphics and computation advances the past decade. Would like them to have more competetion though.
Re: (Score:2)
Guess you never got a Voodoo Banshee or Voodoo3?
Re: (Score:2)
Never had a Banshee, but the Rush was the first combined card anyway, not the Banshee.
Then a Voodoo5 5500. I was invested in 3dfx.
Then I learned that the GeForce supported Harware T&L, which made my Voodoo5 look like poopoo. It got wrecked in every benchmark. Unfortunately, 3dfx, though pioneers, were never able to keep up with NV.
It's no surprise NV ended up purchasing their skeletal remains.
Re: (Score:2)
I had dual Voodoo Banshees with some hacked up support in Xfree86 to make it go. They made the Xterm scrolling go brrrr.
Once the GeForce 256 came out, I bought it and never looked back.
The PC VGA industry has always been a meat grinder. You fall behind in a product's release time, its performance, or fail to meet the price of your competitors. And you have wasted a lot of capital and essentially need to start over and invest in brand new technologies. Only companies that either never make mistakes or have a
Re: (Score:2)
3dfx fucked themselves by deciding to make all of their chip customers their competition when they stopped being a chip designer / supplier and started retailing their own cards. Literally every one of their OEMs gave them the double-bird and selected Nvidia and S3 (remember them?) chips for their next generation and 3dfx revenue collapsed.
Nvidia only started doing the same after there was basically no competition that wasn't already doing the same (AMD). Now it's just expected that you can buy the refere
Re: (Score:2)
Voodoo Rush was so bad it's almost worth not mentioning. I had one. It was disappointing to say the least.
Also if you look at Voodoo Rush it was more of a two-card solution. They had a 2d card with a 3d daughtercard.
Re: (Score:2)
Voodoo Rush was so bad it's almost worth not mentioning. I had one. It was disappointing to say the least.
It wasn't bad at the time. It as 2D+ 3D, which was really cool.
Also if you look at Voodoo Rush it was more of a two-card solution. They had a 2d card with a 3d daughtercard.
Mine sure as hell didn't. Though the "3dfx" chips, and the 2D chips were on different halves of the card. If memory serves, you could get them with one of several different 2D chipset vendors, so I wouldn't be surprised if some kind of stacked card did exist.
Re: (Score:2)
Pretty sure mine was a stacked card. Had one in my k6-233 back in 1997.
Re: (Score:1)
NVidia are kind of dicks but they have consistently been the ones pushing a lot the graphics and computation advances the past decade.
summed up quite nicely.
also, their products have been "the good shit" basically since inception.
looks like AMD will be the first to launch MCM GPUs, and those should be very strong performers...
but given nvidia's history, im more excited about nvidia's theoretical response than AMD's actual product.
Facebook could be replaced in a week (Score:4, Insightful)
Nvidia, on the other hand, makes hard-to-design and hard-to-manufacture GPUs. If they disappeared tomorrow, other manufacturers could step up but it wouldn’t be easy or fast.
Nvidia should be worth WAY more than Facebook, but markets do extremely non-rational things.
Re: Facebook could be replaced in a week (Score:4, Insightful)
You could summarize your wise post this way:
N idea does something genuinely useful. FaceBook does not.
Re: (Score:2)
Exactly this. Nvidia actually makes a product that can do useful things. Facebook just collects product (your data) and sells it to literally anyone that can agree on a price. They are digital middle-men.
Re:Facebook could be replaced in a week (Score:4, Insightful)
Replacing Nvidia requires an end to the chip fabrication bottleneck/supply issue AND a company that can design better and/or cheaper cards. Agreed that it's harder than coming up with the next BaceFook; but they've got a nice target on their back now. US and Europe are rushing to build chip fabs because they finally woke up and realized that outsourcing it all to Asia might not have been such a bright idea. In 5 years there could be a turn-around. We could actually have a glut of fabs, falling chip prices, and a fantastic opportunity for the other players to step up and eat Nvidia's lunch.
Re: (Score:2)
but they've got a nice target on their back now. US and Europe are rushing to build chip fabs because they finally woke up and realized that outsourcing it all to Asia might not have been such a bright idea.
Hardly. The chip shortage that is being addressed in EU and USA have nothing to do with the cutting edge fabrication of NVIDIA and the rest of the PC chip industry. Only Intel's fabs being built is going to compete with them. All other investments in this space are a completely different node size.
Re: (Score:2)
Nvidia is a fabless chip designer. More fabs is *good* for them. You don't think Ford is screwed as soon as people can buy STM32s again, do you?
Nvidia's really good at designing GPUs, and they only have one real competitor. But I think their real strength is the work they've put into CUDA. I'd like OpenCL to succeed, but it just doesn't seem to have the reach CUDA does.
Re: (Score:1)
Wait. (Score:5, Funny)
And it should be (Score:5, Insightful)
nVidia tech is actual technology that requires our brightest brains to develop and create, whereas Meta is like anyone with a GED can come up with ideas for it, like hey dudes let us make legless avatars and call it metaverse. CPU design on the other hand requires you to know actual academic level shit. You have to do actual STEM. At least for two or three years until the EDA software can entirely design the cpu and put together IP blocks according to the fab rules. Then it will be analog, semiconductor materials, and device physics that will need actual STEM skills.
Re: (Score:2)
While Facebook itself is a bucket of shit dreamt up by someone who stole someone else's bucket of shit ideas, dismissing all of Meta as the stupidity for the metaverse is just stupid.
They do pour huge amounts of R&D into datacentre design, network design, they laid some serious foundation work to big database design, they created and maintain React (framework used by very many applications). They are also providing the largest R&D for VR in general, not just the fucking stupid metaverse concept. Not
Re: (Score:2)
There are a lot of wild projects going on at Meta. Both Google and Facebook have long spent money on some pretty crazy stuff that is outside of their primary business. I think subconsciously they knew their business model is shit and are looking for "the next big thing" to pivot into if the grift is ever revealed.
Other companies like Nvidia are more focused on their hardware and software platforms. Pretty much all public info on what NV does in their R&D seems to be somehow related to GPUs. Gaming, comp
Re: (Score:2)
nVidia tech is actual technology that requires our brightest brains to develop and create, whereas Meta is like anyone with a GED can come up with ideas for it, like hey dudes let us make legless avatars and call it metaverse. CPU design on the other hand requires you to know actual academic level shit. You have to do actual STEM. At least for two or three years until the EDA software can entirely design the cpu and put together IP blocks according to the fab rules. Then it will be analog, semiconductor materials, and device physics that will need actual STEM skills.
It's always amusing to me when someone says something like this. It's so obvious, yet for some reason guys like you and I didn't manage to create a company hundreds of billions of dollars, with hundreds of millions of daily interactions, who manages to collect data on people from countless other sites who stick share icons and use Facebook for logon authentication, and made it's founder a multi-billionaire.
I don't know why you didn't do this yourself. What Nvidia does is amazing. What Facebook does, also
Nvidia actually makes things (Score:2)
As opposed to just being a website with an overblown sense of purpose. Nvidia should be worth more, and always was in our non-damaged, parallel world.
Re: (Score:1)
Re: (Score:2)
It's not a matter of its value to society or even to the individual. It's a matter of how much revenue could potentially be generated. How many GPUs does each person in the first world need? 1 or 2 maybe?
Now how many products can Facebook hawk to users, thousands a month to each user. How much personal data can they collect from users to feed premium revenue for targeted ads? Ultimately the scale of the business for Facebook is just bigger than what a graphics card company will see. Facebook's business mode
Meta = one half (Score:2)
Sit nomen omen
MySpace (Score:2)
Meta? (Score:2)
Thiel left Meta (Score:1)
myspace (Score:1)
In more than one sense (Score:2)
Congrats! (Score:2)
To their marketing dept.
They have created an immense group of drones that will blindly worship Nvidia no matter what crap they do to the industry or their own customers.
They are also in charge of sending the lovely loyalty checks to all the websites like TechSpot and Youtubers like LTT, so they continue posting crap about Nvidia non stop which ends up as..food for the nvdrones!
I mean, watch any tech video in youtube and even if the video itself is not about nvidia, they will make sure to mention them and of