Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Trekkin]

Why do high-end (especially top-of-the-line) workstation GPUs come with obscene amounts of VRAM? Like, the Nvidia RTX 2080 Ti Super comes with 11 GB of VRAM (far more than you'd likely need for gaming, frankly), and the Quadro RTX 8000, its (highest-end) workstation counterpart, comes with 48 GB of VRAM. What are its users doing that would warrant needing that much VRAM?

Deep learning, primarily.

[wierd]

Also very high end CAD workstations, simulating an entire aircraft assembled from virtual individual components down to nuts, bolts, and rivets for stress simulation.

Things of that nature.

[Iduno]

Why do high-end (especially top-of-the-line) workstation GPUs come with obscene amounts of VRAM? Like, the Nvidia RTX 2080 Ti Super comes with 11 GB of VRAM (far more than you'd likely need for gaming, frankly), and the Quadro RTX 8000, its (highest-end) workstation counterpart, comes with 48 GB of VRAM. What are its users doing that would warrant needing that much VRAM?

Bitcoin mining, which is where most of the video cards end up.

[methylatedspirit]

Bitcoin mining, which is where most of the video cards end up.

But the Quadro RTX 8000 costs a whopping $5500. Entire computers are more expensive than that. You could buy 4 RTX 2080 Ti's and SLI two of them together, give away the rest, and you'd still have enough for an i9-9900K. The RTX 2080 Ti costs $1199, which is downright reasonable in comparison. They both run on the same graphics chip anyway, so you'd likely see a tiny performance increase, if any.

Bitcoin mining's already difficult to get a decent return on investment on, so I'm not sure why anyone would go for such obscenely-priced hardware. Plus, you don't need that much VRAM to do mining; a cursory search shows that you'd need 6-8 GB max.

In other words, I doubt it.

[Egan_BW]

Bitcoin mining's already difficult to get a decent return on investment on, so I'm not sure why anyone would go for such obscenely-priced hardware.

Being rich and dumb?

[methylatedspirit]

I forgot that idiots existed. The same rich idiot would also buy a Xeon processor for a gaming computer on the basis of "it's more expensive, it's gotta be better!".

[Naturegirl1999]

There was an experiment where the same cake had only the prices different, and customers reported the more expensive cake tasting better even though the cakes were the same save for prices. I think humans have been conditioned to think more expensive=better

[methylatedspirit]

Does the assumption of "expensive = good" hold up if the more expensive thing is slightly worse than the less expensive thing? I'm asking because Xeons are server and workstation processors; reliability is the main focus, not speed. As such, a Xeon is worse for gaming than an equivalent Core.

Of course, I'm in a space where you can objectively measure the difference between the things by running tests. You can run a bunch of benchmarks on two processors that are identical save for, say, clock speed, and you can conclude that one is better than the other in some areas. Cake can't really be compared objectively; you can't, like, increase the amount of sugar used by 30% and always expect a proportional increase in the number of people that like it. Some people are repulsed by very sugary food, so the number might start to decrease after some point.

[dragdeler]

-

[methylatedspirit]

And therein lies the problem. Not enough people research the things that they're gonna buy for companies to change their prices. Really, I think the only way to stop it is to research alternative products that still meet your requirements. But that's time- and brainpower-intensive. It's simply much easier to buy the first thing that comes up. So most people do that.

Random thought: I think people only start researching rigorously once they're cash-strapped, since at that point, every dollar matters. Every dollar saved is money that can be put elsewhere. So I suppose it's best to always pretend that you're broke to save money?

[Reelya]

There are definitely some weird psychological issues around pricing. For example, there are experiments where you put an additional option on a list that nobody ever takes, but the mere addition of the option causes people to choose differently between the other existing options, compared to when that extra option wasn't there.

Also for ranges of things, they have a really expensive one that they actually expect nobody to get, but because it makes the mid-range option look better. Generally, people like there being a choice, but the vast majority of people want a mid-range one, not cheap and nasty, not needlessly expensive. However, removing the cheap/nasty and overly expensive options change consumer behavior, even if the people making the decision never actually pick those.

EDIT: to give you an idea how this works, say you have "cheap option 1" that's $20, and "expensive option 2" that costs $100, and you get a voucher for a free cup of coffee. Now, a certain % of people will pick option 1, and a certain % of people will pick option 2. However, if you add "option 3" that's identical to option 2, but you don't get a free coffee, everyone agrees that option 3 is a shit option: it's just option 2 minus the coffee. So people pick option 2 instead of option 3. The weird part is that more people also start picking option 2 instead of option 1, even though nothing actually changed between those two options, and nobody ever picks the "extra" option 3. So, by adding superfluous options, you can actually steer people to choose what you want them to choose: add slightly inferior versions of the choice that you're trying to steer them towards, so that the desired option looks more appealing. In other words, to make a certain design more appealing, add some butt-ugly optional designs to pick from, which cost the same. If you have too many good choices, people will get choice-paralysis and not be able to decide. So you only want a couple of decent ones and some godawful ones you hope nobody picks.

This is good knowledge for game designers actually: Say you have a regular sword, and that's 20 gold, and you have a +5 sword, and that's 100 gold, and you want more people to buy the +5 sword. Well, add a +4 sword for 95 gold. Now, anyone would be a fool to buy the +4 sword, since the +5 sword is clearly better value for money. But ... now people will see the extra +1 of the +5 sword as a "bargain" so they'll be more inclined to shell out for the better sword. Even if nobody ever buys the +4 sword. Strange but true, and verified in experiments.

[itisnotlogical]

Has anybody experimented with putting on a play or other live stage performance in VR? For obvious reasons, searching "VR play" and "VR performance" gives unrelated hits.

I've seen a bit of buzz about musical concerts taking place in VR due to covid19, but I'd be very surprised if nobody's at least thought about it before, if nobody's actually tried it.

[Frumple]

Dunno about VR explicitly, but I'm pretty sure there's been stuff at least close to what you're talking about in relation to vocaloid junk?

[MaxTheFox]

Any recommendations for DF/Rimworld-like games?

[hector13]

Rinworld and Dwarf Fortress /lolsofunnyandclever