2060 vs 3050 vs 6600 xt

[Maximilian17]

Hi,

I'm comparing between those three GPU. Approximate price in Australian dollars at a local PC shop near where I live.

Rtx 2060 $459-$599
Rtx 3050 non ti $499-$599
6600 xt $649-$779

This build is what I have in mind:
https://au.pcpartpicker.com/list/wd8F8r

If you have $400-$600, which GPU are you going to pick? According to reddit 2060 is better than 3050, and 6600 xt is better than 2060.

I think I'd pick rtx 2060, just because it is cheaper than 6600 xt.

I'm not a hardcore gamer anyway. All I want is a better system than my 9 years old desktop PC. Initially I wanted to buy 1050ti, but it is selling for $299. Not wanting to pay that much for an entry level GPU, although it is a huge upgrade from 650 ti boost.

Currently I can use my old 650 ti boost on a new system while observing the GPU price. (Probably 650 is going back to heaven anytime soon)

Thank you for your responses.

[PopinFRESH]

Origineel geplaatst door UserNotFound:

Origineel geplaatst door PopinFRESH:

Not sure I'd call FSR a "strong competitor" to DLSS. FSR 1.0, and RSR are noticeably worse than DLSS. Good video that goes through comparison between them.

FSR 2.0, being temporal instead of spatial, while not exactly on DLSS quality, I believe it'd be close enough. Just like the recent Unreal 5 The Matrix Awakens demo has shown, TSR (believed to be temporal) provides pretty good image scaling and is hardware agnostic.

And no, FSR 1.0 was a stop gap solution, not meant to compete with DLSS which is limited to only the RTX cards, FSR 1.0 can be used by both AMD and nVidia cards, so the GTX cards can reap the benefits that FSR 1.0 brings to the table. RSR should NOT be compared to DLSS either, instead, it should be compared to nVidia's NIS (similar to RSR, these are driver based, the GTX cards would then be able to reap the benefit of a software based upscaling tech). I find it odd that NIS is never compared to DLSS, and should never as they work differently, yet so many use RSR to compare with DLSS.

Not sure why you say TSR is "believed to be temporal" as it is in the name.

Origineel geplaatst door Epic Games:

Temporal Super Resolution (TSR)

Unreal Engine 5 [www.unrealengine.com]

Just because they work differently doesn't mean they aren't attempting to accomplish a similar result. Using the logic of they work differently thus shouldn't be compared we shouldn't compare FSR, TSR, XeSS, or DLSS.

The GN video I linked I believe does a sufficient job at describing the differences in approach between them so they can be compared. Regardless, this is getting well off topic from the OP's question so I'm going to leave this at what I said before

Origineel geplaatst door PopinFRESH:

Regardless, within the budget the OP is looking at the Radeon 6600 is probably the best option and given the OP is going to be gaming at 1080p (assuming 60Hz given they didn't mention a higher refresh rate) this will certainly deliver a solid 60fps in most games at high/ultra settings without FSR/RSR.

[Χάρης]

I'd never buy a gtx or rtx with a " 50 " on it. Either 60, 70 or 80. So just get the RTX 2060 (or 3060, if you can afford).

[UserNotFound]

While I won't argue an FSR 1.0 vs DLSS comparison, they are similar in that game developers must add supprot for them to work. RSR and NIS on the other hand, are driver based and works without such optimizations, hence one can expect worse off PQ compared to DLSS. That's why you don't hear peeps comparing DLSS to, say, NIS, because it is not a fair comparison. Likewise, RSR should NOT be compared to DLSS when its real rival is NIS, that is all I'm saying.

My bad about the Temporal part, I keep saying TSR, but I do sometimes forget that the 'T' stands for Temporal, and hence, self explanatory..

[Kissa]

Origineel geplaatst door Bad 💀 Motha:

Are you folks not keeping up with tech reviews? The RTX 3050 is a complete joke.

is it really? i'd imagine that it would improve performance significantly just because it has 8gb vram. running out of that is the source of performance issues with the older gpus that have 4gb or even less. i noticed that because i have an ancient gpu and everything would run smoothly until the monitoring program started showing that the memory was getting full

[r.linder]

Origineel geplaatst door Kissa:

Origineel geplaatst door Bad 💀 Motha:

Are you folks not keeping up with tech reviews? The RTX 3050 is a complete joke.

is it really? i'd imagine that it would improve performance significantly just because it has 8gb vram. running out of that is the source of performance issues with the older gpus that have 4gb or even less. i noticed that because i have an ancient gpu and everything would run smoothly until the monitoring program started showing that the memory was getting full

For newer GPUs it's just expected to have at least 6GB or 8GB (depending on the memory bus used) but 4GB is the dead minimum to be able to game at decent settings in newer titles.

VRAM above what you actually need for a game is worthless, it doesn't add performance. The 3060 Ti is easily 20~30% faster than the 3060 but it has 4GB less VRAM. It's the core that matters first and foremost, what happens with VRAM is just an afterthought.

The 3050 is overlooked because it's typically just bad value, it has pretty much the same price that the 2060 did last gen, but it performs 10~20% slower. The 3060 is easily 30% faster in many games, often getting 30+ FPS when the GPU isn't absolutely slammed. That's why it's a joke.

To put it in perspective, the 2060 was roughly on par with a GTX 1080 when it came out, and the 3060 is on par with the 1080 Ti. The 3050 is slower than a GTX 1080 in quite a few instances as a result of it being slower than a 2060, and effectively it's just a raytracing refresh of a 1660 SUPER or Ti that has no business existing. RT performance is already bad enough with a mid range card like the 3070.