Radeon RX Vega 64 vs. GeForce RTX 2080, Adrenalin 2019 Edition Driver Update Benchmark Test

james.miller · 20 Jan 2019 at 23:11

16-235 is a throwback from analogue terrestrial transmission. 16 and 235 were set as the black and white points respectively in order to give some buffer to avoid clipping in the analogue transmission - Everything either side of those numbers is discarded. It's stuck for tv and film ever since. With the display set correctly (ie, either for full or limited) the resulting output should be near enough identical. Problems come when you calibrate a panel for one and then send it the other. Don't do that.

4K8KW10 · 20 Jan 2019 at 23:31

james.miller said:
16-235 is a throwback from analogue terrestrial transmission. 16 and 235 were set as the black and white points respectively in order to give some buffer to avoid clipping in the analogue transmission - Everything either side of those numbers is discarded. It's stuck for tv and film ever since. With the display set correctly (ie, either for full or limited) the resulting output should be near enough identical. Problems come when you calibrate a panel for one and then send it the other. Don't do that.

Yeah, the root of the problem is the missing word "simultaneous". R, G and B at 16 results in something like black but not quite black, and R, G and B at 235 results in something like white but not quite white.
But the spectrum also has all the colours which come from the different combinations of the values, including the ones missing from 0 and around 255.

james.miller · 21 Jan 2019 at 01:43

4K8KW10 said:
Yeah, the root of the problem is the missing word "simultaneous". R, G and B at 16 results in something like black but not quite black, and R, G and B at 235 results in something like white but not quite white

* When viewed on a screen calibrated for RGB full.

Gregster · 21 Jan 2019 at 03:06

Terrible testing methodology! Better when people know what they are doing show what is what, as opposed to some random fella.

shankly1985 · 21 Jan 2019 at 06:35

Gregster said:
Terrible testing methodology! Better when people know what they are doing show what is what, as opposed to some random fella.

Explain?

Gregster · 21 Jan 2019 at 08:38

shankly1985 said:
Explain?

Messed up the colours, clocked a 9900K down to 4Ghz. Overclocked and undervolted the Vega 64.

Yer, pretty shoddy really and surprised you used this.

Brazo · 21 Jan 2019 at 08:43

Gregster said:
Messed up the colours, clocked a 9900K down to 4Ghz. Overclocked and undervolted the Vega 64.

Yer, pretty shoddy really and surprised you used this.

And used lower graphics settings to (perhaps) introduce a cpu bottleneck at high FPS. After all most games still only use four cores so a 9900k isn’t as impressive as it sounds here.

Any professional test shows very different results.https://www.techpowerup.com/reviews/Zotac/GeForce_RTX_2060_AMP/32.html

shankly1985 · 21 Jan 2019 at 12:13

Gregster said:
Messed up the colours, clocked a 9900K down to 4Ghz. Overclocked and undervolted the Vega 64.

Yer, pretty shoddy really and surprised you used this.

Only thing I agree here is the colours. Cpu was at 4ghz on both GPUs.
Tweaking Vega is fair considering the 2080 was also tweaked vs a stock 2080

He clearly explains why he done this.

shankly1985 · 21 Jan 2019 at 12:17

Brazo said:
And used lower graphics settings to (perhaps) introduce a cpu bottleneck at high FPS. After all most games still only use four cores so a 9900k isn’t as impressive as it sounds here.

Any professional test shows very different results.https://www.techpowerup.com/reviews/Zotac/GeForce_RTX_2060_AMP/32.html

Bottleneck a 9900k? Really? Lol check the video and show me one game he tested only using 4 cores?
If anything the benchmark actually shows the games fully using the CPU cores.

Pure salt going around here.

D.P. · 21 Jan 2019 at 12:54

shankly1985 said:
Only thing I agree here is the colours. Cpu was at 4ghz on both GPUs.
Tweaking Vega is fair considering the 2080 was also tweaked vs a stock 2080

He clearly explains why he done this.

What was tweaked with with the 2080?

shankly1985 · 21 Jan 2019 at 12:56

D.P. said:
What was tweaked with with the 2080?

Read OP

Gregster · 21 Jan 2019 at 13:00

shankly1985 said:
Only thing I agree here is the colours. Cpu was at 4ghz on both GPUs.
Tweaking Vega is fair considering the 2080 was also tweaked vs a stock 2080

He clearly explains why he done this.

Even if it was only the colours messed up and nothing else, it is still complete rubbish. Read this thread to see what it has produced. People claiming the IQ of AMD is better and as someone who still owns a 290X, I have to say what tosh and because people are seeing this vid, they actually believe that AMD has better IQ. This is why I was surprised you used this. I remember having my NCP setting performance over quality (something I never set) and did a comparison vid and got ripped to bits for it but this is somehow acceptable?

D.P. · 21 Jan 2019 at 13:01

shankly1985 said:
Read OP

SO absolutely nothing was touched on the 2080 and it is showing out of the box performance with guarantees.

Teh Vega on the otherhand has been heavily tweaked and voltages adjusted without any guarantee of that being stable

Apples and oranges.

shankly1985 · 21 Jan 2019 at 13:29

Gregster said:
Even if it was only the colours messed up and nothing else, it is still complete rubbish. Read this thread to see what it has produced. People claiming the IQ of AMD is better and as someone who still owns a 290X, I have to say what tosh and because people are seeing this vid, they actually believe that AMD has better IQ. This is why I was surprised you used this. I remember having my NCP setting performance over quality (something I never set) and did a comparison vid and got ripped to bits for it but this is somehow acceptable?

And I didn't post this video for image quality comparison. Other users brought that uppon themselves.
All I care about is the performance /end

Gregster · 21 Jan 2019 at 13:35

shankly1985 said:
And I didn't post this video for image quality comparison. Other users brought that uppon themselves.
All I care about is the performance /end

Of course they did lol. It is what will happen when a flawed comparison vid is shown. Hence my reasoning behind why this is a **** video and surprised you used it.

melmac · 21 Jan 2019 at 13:37

Why are people so upset about the colours used in the Video? The colours don't matter, they don't have any relevance to what the video was about. The colours don't change the performance. He said he forgot to set the range to full in the Nvidia control panel. So people saying that PQ is better on one brand than the other based on this video are talking drivel. But, the same goes for people who are saying that the video is rubbish just because the colours aren't right. It has no impact on the subject of the video.

It's also funny that so many people are getting upset about the overclock on the Vega card. When Vega came out there were quite a few posts and threads about overclocked 980Ti's matching the Vega 56 and showing that the 980ti lasted well to be able to match the second best card in the AMD lineup. None of the people complaining in this thread about the Vega card been overclocked had any issues with the 980Ti been overclocked in comparison to a stock Vega card. But, now that there is a video showing a overclocked Vega getting close to a 2080, it's not an apples to oranges comparison. The Vega card is "heavily" overclocked compared to a lightly overclocked 2080.

The "heavily" overclocked Vega comments are making me laugh.

melmac · 21 Jan 2019 at 13:42

Kaapstad said:
The guy should not have given the Vega card an overclock as it is too big and totally trashed the comparison making the video pointless.

The 2080 card used in the video has an 1815mhz boost clock compared to a Founders Edition card with 1800mhz, less than 1% difference.

The founders editions cards come factory overclocked. So the 2080 card used in the video had a 115hz overclock.

The Vega overclock is too big? lol no it's not, it's barely tweaked at all.

Gregster · 21 Jan 2019 at 13:43

melmac said:
Why are people so upset about the colours used in the Video? The colours don't matter, they don't have any relevance to what the video was about. The colours don't change the performance. He said he forgot to set the range to full in the Nvidia control panel. So people saying that PQ is better on one brand than the other based on this video are talking drivel. But, the same goes for people who are saying that the video is rubbish just because the colours aren't right. It has no impact on the subject of the video.

It's also funny that so many people are getting upset about the overclock on the Vega card. When Vega came out there were quite a few posts and threads about overclocked 980Ti's matching the Vega 56 and showing that the 980ti lasted well to be able to match the second best card in the AMD lineup. None of the people complaining in this thread about the Vega card been overclocked had any issues with the 980Ti been overclocked in comparison to a stock Vega card. But, now that there is a video showing a overclocked Vega getting close to a 2080, it's not an apples to oranges comparison. The Vega card is "heavily" overclocked compared to a lightly overclocked 2080.

The "heavily" overclocked Vega comments are making me laugh.

I very much doubt anyone is upset over it but if someone can run something like that and not even notice til it is pointed out to him, how can he be respected as a Youtuber showing a fair comparison? Seriously, I am an amateur Youtuber but can do a far better job.

melmac · 21 Jan 2019 at 13:52

Gregster said:
I very much doubt anyone is upset over it but if someone can run something like that and not even notice til it is pointed out to him, how can he be respected as a Youtuber showing a fair comparison? Seriously, I am an amateur Youtuber but can do a far better job.

You didn't have any problems with the DLSS video that had terrible colour differences. It was a completely flawed video because the colour saturation did affect how it looked.

whereas in this video the colour doesn't matter a damn, it's about the performance of the cards. Everything else in his testing was fine. His reasoning was sound, a lightly overclocked Vega card against a lightly overclocked 2080 card.

Gregster · 21 Jan 2019 at 14:16

melmac said:
You didn't have any problems with the DLSS video that had terrible colour differences. It was a completely flawed video because the colour saturation did affect how it looked.

whereas in this video the colour doesn't matter a damn, it's about the performance of the cards. Everything else in his testing was fine. His reasoning was sound, a lightly overclocked Vega card against a lightly overclocked 2080 card.

I liked the DLSS vid that was FFXV and thought it looked good. No good for me thus far though, as I am 3440x1440 and not supported (that I am aware of). I feel it is a good way of bringing 4K to the masses without having to have such a beast of a card and I believe AMD are jumping on a similar thing and sure it will be better than DLSS. If people think having an undervolted and overclocked card running against a ref card and lowering a 9900K processor to 4Ghz and not putting full RGB on in the NCP is a decent way of testing, then happy days to those guys. Personally, I prefer the guys who do it for a living or as an enthusiast.