Why so many people are switching from NVIDIA to AMD?

[C1REX]

I've switched to AMD as well, but I wonder why recently so many people recommend AMD over NVIDIA. Don't people care about Ray Tracing and DLSS? Is it some kind of pushback against NVIDIA's pricing? I thought my decision was rare and not that popular.


EDIT: I understand that NVIDIA dominates with over 80% of the GPU market. I simply wonder why some AMD GPUs even appear in top selling GPUs on NewEgg, mindfactory.de or overclockers.co.uk. Why people on reddit recommend AMD so often. Why AMD did so well in recent Hardware Unboxed survey (40% AMD).

[KeegsterMann]

引用自 Lord Flashheart：

引用自 C1REX：

I've switched to AMD as well, but I wonder why recently so many people recommend AMD over NVIDIA. Don't people care about Ray Tracing and DLSS? Is it some kind of pushback against NVIDIA's pricing? I thought my decision was rare and not that popular.


EDIT: I understand that NVIDIA dominates with over 80% of the GPU market. I simply wonder why some AMD GPUs even appear in top selling GPUs on NewEgg, mindfactory.de or overclockers.co.uk. Why people on reddit recommend AMD so often. Why AMD did so well in recent Hardware Unboxed survey (40% AMD).

It seems gamerrs are no longer a priority for nvidia.
Typically the cards lack VRAM, more expensive, with locked down features.
The upscaling is the obvious one.
Also still to display port 2.1. Combined with lack of VRAm is an issue for longevity.

I'm going to agree with you on that fact. Nvidia is overpriced, lacks enough VRAM, they lock their features and they are too reliant on AI.

[Shadow]

As much I like to use my GPU for Ai Training.

Nvidia seems to be the most supported. I would switch, AMD is lacking Ai Support.

However, I do agree I wish they strictly made GPU's for Gaming only.

At the moment, if they keep pushing AI with GPU products and if this is the future.

Gaming console might be more of an option just strictly for gaming, which I hope never does.

[Karumati]

引用自 Shadow：

As much I like to use my GPU for Ai Training.

Nvidia seems to be the most supported. I would switch, AMD is lacking Ai Support.

However, I do agree I wish they strictly made GPU's for Gaming only.

At the moment, if they keep pushing AI with GPU products and if this is the future.

Gaming console might be more of an option just strictly for gaming, which I hope never does.

unfortunately, consoles are quite limited in what they can offer.
mediocre fps, lack of ultrawide support are few examples.

[Shadow]

引用自 Karumati：

引用自 Shadow：

As much I like to use my GPU for Ai Training.

Nvidia seems to be the most supported. I would switch, AMD is lacking Ai Support.

However, I do agree I wish they strictly made GPU's for Gaming only.

At the moment, if they keep pushing AI with GPU products and if this is the future.

Gaming console might be more of an option just strictly for gaming, which I hope never does.

unfortunately, consoles are quite limited in what they can offer.
mediocre fps, lack of ultrawide support are few examples.

Exactly, also things like Mod support and having to pay for online multiplayer is ridiculous.

[Skkooomer Lord]

Hardware / intellect sigmas like me exist who usually don't tout it on our sleeves, but when Nvidia users out number me 100:1 it's quite an experience.

For example when the remaster for Horizon released, I am one of few with no issues whatsoever and I made a detailed post, not a single response from anyone.


https://steamcommunity.com/app/2561580/discussions/0/4631483020295014455/

[MancSoulja]

AMD aren’t even offering a high end GPU option next gen, they’re only bothering to compete with low end and mid range. That’s how well they’re doing 💀

[r.linder]

That's pretty much what happens when you have to crank the power to 11 just to keep up with the competition.

RDNA2 was pretty efficient, more-so on the lower end, but RDNA3 saw a big uptick in power demands almost across the board, but most extreme in the high end because they had to push hard just to match the 4080.

But considering that the 5080 is going to be roughly 4090 performance, I can see why AMD is dropping out temporarily, they have no choice but to work on a more efficient design that they can really push to catch up with nVidia.

IMO, RDNA isn't working anymore and they need a complete redesign, not something that's based on GCN (which mostly failed them) and improved incrementally per generation. They have to do better if they're ever going to expect to steal market share from nVidia.

[Karumati]

引用自 r.linder：

That's pretty much what happens when you have to crank the power to 11 just to keep up with the competition.

RDNA2 was pretty efficient, more-so on the lower end, but RDNA3 saw a big uptick in power demands almost across the board, but most extreme in the high end because they had to push hard just to match the 4080.

But considering that the 5080 is going to be roughly 4090 performance, I can see why AMD is dropping out temporarily, they have no choice but to work on a more efficient design that they can really push to catch up with nVidia.

IMO, RDNA isn't working anymore and they need a complete redesign, not something that's based on GCN (which mostly failed them) and improved incrementally per generation.

looking at 40 and 30 series 4070 is roughly 3090 performance, so if nvidia will do decent work - we should expect 5070 with 4090 performance

[r.linder]

引用自 Karumati：

引用自 r.linder：

That's pretty much what happens when you have to crank the power to 11 just to keep up with the competition.

RDNA2 was pretty efficient, more-so on the lower end, but RDNA3 saw a big uptick in power demands almost across the board, but most extreme in the high end because they had to push hard just to match the 4080.

But considering that the 5080 is going to be roughly 4090 performance, I can see why AMD is dropping out temporarily, they have no choice but to work on a more efficient design that they can really push to catch up with nVidia.

IMO, RDNA isn't working anymore and they need a complete redesign, not something that's based on GCN (which mostly failed them) and improved incrementally per generation.

looking at 40 and 30 series 4070 is roughly 3090 performance, so if nvidia will do decent work - we should expect 5070 with 4090 performance

4070 SUPER is 3090 performance, regular 4070 is only 3080 performance

Leaks already suggested that the 5080 would only be roughly equivalent to a 4090D (the cut down version for China)

[_I_]

and nvidia doesnt really have a low-mid gpu anymore
x60 is mid tier, but no x50 or lower cards, unless you go back to the 950/940/930/920

[r.linder]

引用自 _I_：

and nvidia doesnt really have a low-mid gpu anymore
x60 is mid tier, but no x50 or lower cards, unless you go back to the 950/940/930/920

They have two variants of the 3050 though? Also the mobile 2050, and the 1050 and 1650 cards, but yeah below that they definitely stopped developing basic display adapters after the GT 10xx series

I would also argue that x60 is low tier now and x70 is mid tier, x80 being high end, with x50 being entry level

[Guydodge]

i bought a amd card for the first time for the wife was removed and sold within 3 months
the most unstable no frills card i have ever had the displeasure of owning.i want to game
not play with drivers and have the foot print of there lousy software interfering with said goal. but if your on a budget i guess its worth the trade off but not IMO.

and as far as dlss and raytracing go if you aint using it you have no idea the
benefits of it. especially the new dlss 3.7

[_I_]

for the longest time x50 was entry level
any gt/gs/le was for office/desktop tasks

x80-x90 were always high end for non gaming, more rendering or dev tasks
but they kinda got moved toward gaming

they could number them from x30-x90 and not use so many with super or ti, to help divide them bit
but that would probably force them to make a large price gap between them too

amd, nvidia and intel have always had trouble naming their products
but atleast nvidia has kinda been sticking to the same theme for a long time