AMD Fine Wine Analyzed

I'd still prefer that performance out of the gate. I generally don't hold onto my GPU's long enough for this to become relevant.

 

Downvoted. I wonder how much time that person spent making it. Probably ages, because it must have been pinnacle of his life.
And that was damn good example on: "How not to deliver information."

 

So what is the conclusion here? AMD drivers getting better, NVidia drivers getting worse, a combination of both?

 

My opinion is that the graphics cards are catching up to the performance AMD could have had with more resources and time but they do a pretty good job although it can take a while although unless the game is completely broken the performance gain tends to be pretty low so it's not like the GPU's are launching in a completely disastrous state either performance wise.

NVIDIA sees performance gains and increases too though, lately from what I'm reading they've really caught up in DX12 and Vulkan support and performance plus their existing lead with DX11 and OGL.

 

Video uses relative performance between cards year after year. But games tested change from year to year.

This does not mean that AMD is gaining and nVidia is losing performance in particular game(s).
It means that AMD's relative performance is getting better with newer games.
(You can read it other way around: nVidia does optimize games less for older cards as time passes.)
Or that games are natively made in way which is starting to suit AMD better.

Use of averaged values and relative performance basically prevents you from drawing exact conclusion. Only conclusion is that FineWine [tm] does apply in relativistic way.
And that translates into:
- If you buy GPU every year, you are better off with nVidia's card which delivers higher performance at same price at start of its life cycle. (please excuse 20x0 series here)
- If you buy GPU every 3 or more years, then AMD is bit better in long run.

 

A proper testing method would be better for this. Having a test with a system that runs the same spec with different cards and testing with different drivers across a broad spectrum of games and synthetics to see the differences between drivers.

Doing it properly you could even test whether MS updates had any effect so going from Win 7 through Win 8.1 and 10 and using the relevant drivers etc...

Could be an interesting project...

 

If someone have 980Ti and FuryX or 280X & GTX680 or 290X & 780 then we can have good starting point.

One is known for everybody:
1. AMD/ATI Tweaking & giving a life into GCN GPUs even today
-> Thus many people have FPS uplift even for Thaiti etc.
2. nV is Money Oriented Corp. *
-> thus only new uArch have proper Tweaking/Driver support, for older Gen they just pull the plug Off (killing old gen)
so:
a. New GPUs are better in benchmarks
b. thanx to this, sell more newer GPUs

Note.
Yesterday me & my Bratan' sit down and check our e-tailer for FreeSync vs G-Sync sales (just for fun, to check Quality gaming for Polaris & Vega in a FreeSync angle)

>3900 Units of FreeSync sold and only ~300 of G-Sync
Cheapest FreeSync is at ~100€ and G-sync counterpart is at ~400€
Most sales have some LG IPS (8Bit) one -> so people are aware of Quality Colors of the IPS panel.

For 400€ you have 1080p 240Hz FreeSync TN and for similar nV you need to pay a lot more.
In short Radeons + FreeSync are selling like creazy in EU (Those 3 games for free w/AC: Odyssey)

 

This again...I guess for the attention, someone is always uncorking this rancid topic.

 

We need to remember that most of Gamers runs RX4xx/5xx GPUs not even cheaper Vega 56 + FreeSync so the budged is ~300-500€ for whole GFX system (GPU+LED)
For those who have deep pockets nv is the only way (right now), you get for ~4500€ ~60-70FPS across the board in 4k HDR -> ATI have no answer for this hudge chips nv made.

Also it's no mistery that for the long run Radeon is better in budged
This is no Toxic thread at all, just Editorial for ATI users.

Also as the nv user you can understand that RX580 w/FreeSync is a way better deal than 1060 or even the V56 vs 1070 no G-sync.

 

980TI was one of the best card nVidia did, it was and still better choice than Fury X, i'm glad i picked this card.

However, 7970GHZ and 290X are far better than any kepler, Kepler was a disaster in the long run, i had the 680 once.

 

Yup, those 980Ti or 1080 should age better IMO, but all in all this depends on nv Driver department guidelines.

I had Fiji with FreeSync so no complains on my side on this uArch (Those HBM Mods )
Also final price was ~500€ for GPU + LED -> Thanks to this i had Biger budget for other things.

 

Hehe, though credit for nvidia that the 680 outperformed the 7970 till AMD released the secret sauce in December 2012, the 7970 and 290X remains one of AMD's best cards.

Oh and the 2015 Kepler downgrade gate was REAL, it's amazing how everyone noticed that and forced nVidia to "correct it", mine 780 was a pathetic shower at the Witcher 3 till the magic driver came.

 

complete burning his whole house

 

I would not sell any of my HW to friend if I expected that he'll suffer damage as consequence. Well, not even to anyone else.

 

Sweet revenge?? hehe, well the stock 290X cooler left a LOT to be desired, the AIB were the saving grace for the card.

 

Yep, I had the 290x Tri-X. It was an awesome card. I regret selling it when I did.

 

A hard lesson to learn nonetheless , i would say that nVidia fine wine would be 980TI and Pascal.

 

Yes, but you need to stay at driver 3xx 4xx is for RTX nowdayZ

 

Did you see that 390X go in Hilbert's Tomb Raider article



https://www.guru3d.com/articles_pag..._graphics_performance_benchmark_review,1.html