Real AMD performance..

[qbazdz]

So, hope I won't stirr it too much but had been looking at recent benchmarks for AMD cards and I keep seeing that AMD marketing game is trying to show it in much better light than things actually are..

Is this actually true? I bet we have enough people here who had swapped GPU or did a comparison that may be a bit more independent than all reviews on either side of fence..

 

[Iron Man]

Read the reviews rather than the marketing blurb.

 

[FredFlint]

If the marketing says something = 10 then it = 10 in a very specific situation and the best one they can find. All company’s do this(NVidia--> its 4x faster!!!!!), wait for user benchmarks for more legit info.

 

[qbazdz]

Sorry, maybe I mistyped somewhere.. Where I meant marketing it was the amd fan base reviews plastered everywhere, using very specific games or known benches where they'd be on top.

Guys at UserBenchmark seem to be very allergic to AMD approach

 

[ElderKarr]

Sorry, maybe I mistyped somewhere.. Where I meant marketing it was the amd fan base reviews plastered everywhere, using very specific games or known benches where they'd be on top.

Guys at UserBenchmark seem to be very allergic to AMD approach

UserBenchMark is very anti-amd for unknown reasons, don't use their site even though they seems to be pretty good on the SEO side to stay relevant on google search

 

[Joxeon]

AMDs marketing said 1.7X faster reality is more like 1.35-1.4

Nvidia said 2-4x but that's only with fake frames.

 

[Twinz]

AMDs marketing said 1.7X faster reality is more like 1.35-1.4

Nvidia said 2-4x but that's only with fake frames.

This.

The XTX is as fast or faster than a 4080 at raw raster. (But not by as much as AMD implied it would be)

RT has caught up to Ampere, but still lags Ada.

Power efficiency is improved but not as efficient as Ada. (And not as good as AMD implied)

1% lows look good.

 

[Dalandius]

Marketing is lies they can get away with.

 

[Grim5]

RDNA3 launch announcement: "up to 1.7x performance compared to the 6950xt"

Hardware Unboxed: The 7900xtx is on average 35% faster than the 6950xt

Now who is in the wrong here; cause AMD's claim isn't wrong, if you cherry pick a RT game or you try F1 22 because it's a huge outlier then yeah you can see the up to 1.7x number.

So is it the customers fault for believing leakers and hype?

 

[Gerard]

RDNA3 launch announcement: "up to 1.7x performance compared to the 6950xt"

Hardware Unboxed: The 7900xtx is on average 35% faster than the 6950xt

Now who is in the wrong here; cause AMD's claim isn't wrong, if you cherry pick a RT game or you try F1 22 because it's a huge outlier then yeah you can see the up to 1.7x number.

So is it the customers fault for believing leakers and hype?

People don't seem to understand what "up to" means, if they can find a game and a scenario where they can achieve that metric then they'll use it in the marketing. They never said 1.7x faster across the board, yet that's what people seemed to take away from the presentation despite the slide clearly saying up to. According to the slides it was achieved in cyberpunk:

 

[Grim5]

People don't seem to understand what "up to" means, if they can find a game and a scenario where they can achieve that metric then they'll use it in the marketing. They never said 1.7x faster across the board, yet that's what people seemed to take away from the presentation despite the slide clearly saying up to. According to the slides it was achieved in cyberpunk:

Yep that's the slide that's coming up today as deceiving. I can kind of see how this occurred

Going back to rdna1 and rdna2, AMD showed performance across a wider range of games including ones where they were not so fast, so people knew what the expected range was and what the average would be.

Then for rdna3 we got that slide you posted; so using AMD's history as a reference point, I can see how people believed that this chart represents both average 4k gaming performance and the range of minimum and high performance - so people totally expected that the 7900xtx would be 50-70% faster in average and that a lot games would be at least 50% faster

AMD needs to go back to showing a wide variety of games including good and bad ones when they announce products.

And Nvidia got roasted for something similar this year, but at least with Nvidia people already knew they were being misleading because they've done it before, so when Nvidia said "up to 4x performance" people knew it's BS, but AMD usually doesn't mislead and so I can see how people took AMD's numbers at their word