Quote from: NikoB on August 26, 2023, 12:12:00why do people buy gaming video cards with a markup of 350% relative to 2008? Where do people get the money to overpay +350% for the same class as in 2008?

If I wanted a gaming card for 3D gaming, I would get the iGPU 680M and use modest settings. Apparently, some 3D gamers appreciate high settings so much that they want fast, new dGPUs and pay the price. Only they can answer why they pay several times as much as several years ago.

In 2008, software for my current dGPU use case (machine learning) did not exist and, if it existed then, the dGPUs then were several factors too slow for it. Around 2017 or 2018, software occurred and dGPUs started being just fast enough but both were not really good enough yet. Since 2019, the software was good enough, and since late 2020, the dGPU hardware was good enough but the mining / Corona crisis occurred with prices until early 2023 exaggerated by manifacturer / seller excess greed beyond any reason. Since May 2023, prices are high compared to 2008 but not short-term extra-high.

Would I want low prices? Of course. Will I delay upgrading by several years due to the high prices? Of course. Have I bought my "first" dGPU and its PC? Yes. Now you wonder why.

I do need it for my use case. Expensive - sure. Worth the price? Absolutely! In fact, for machine learning, Nvidia dGPUs and libraries are much faster than you would expect from all those work / pro tests by countless reviewers. They never test genuine machine learning. It is almost as fast on "gaming" dGPUs as you expect entry Hopper to be. I am delighted by the speed for my use case. It is several times as fast as I hoped. My graphics card model is also great. (Only the crapware sucks.)

Could I use 1000x speed? Yes. However, it would be way too loud and expensive. My card is almost silent and its price as a one-time investment manageable. Good enough and for the realistic speed class much faster than hoped. Expensive but worth the price for pro-like applications needing such speed.

Paper work or CPU computing. Brain or AI computing on the GPU / TPU. Countless brains or cluster AI computing. Applications can justify high prices or don't. The manufacturers charge for application use cases instead of hardware costs. Some applications are worth certain excesses, as long as monopolies / oligopolies exist.

Let me get this straight: you want more AMD gpus so Ngreedia can, hopefully, lower their exorbitant prices and you can buy a RTX gpu anyway? Gotcha!

*Plot twist: AMD is no angel releasing cards at $100 or even $200 more than they should be priced at launch.

The oligopoly from NVidia+AMD artfully manipulates prices, spreading suckers for money.

But the reality is that AMD's top 2008 cost $330 (pre-top 230). Just like Nvidia. And now everything is worth, in exactly 15 years, 5 times more expensive in nominal $ prices.

Has the dollar depreciated 5 times since 2008? It seems not. Only 2 times, i.e. real margin for nothing x2.5 times.

The question is - why do people buy gaming video cards with a markup of 350% relative to 2008? Where do people get the money to overpay +350% for the same class as in 2008?

I perfectly understand your future intention but could you please explain why you bought a 3090? Was this not exactly what you would avoid in the forseeable future? Had you waited several years and did you absolutely need such a fast card? At what price did you buy?

I'm not buying a new graphics card or new equipment in the near future at all. I have an RTX 3090 with an Intel i9 11900K and 32GB of RAM with one of the fastest NVMe SSDs on the market. Nvidia can wait with its overpriced RTX 40-series and upcoming 50-series. I'm sure someone will want it, not me.

I have 3060 12 GB, paid for it 303€ brand new a couple of months ago. This here is a $340 card, not a $1 more.

Entry market GPU pricing is still ludicrous.