AMD 6000 reviews thread

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

BFG10K

Lifer
Aug 14, 2000
22,709
2,995
126


Wow, even the number 3 card has 16GB VRAM and is faster than the 2080TI. And the $1000 6900XT matches the $1500 3090 in performance.

The 3000 parts don't look so hot now.

Post reviews edit:
It's astonishing what AMD have managed to achieve with both the Ryzen 5000 and the Radeon 6000, especially given the absolutely minuscule R&D budget and resources compared to nVidia/Intel. Lisa Su is definitely the "Steve Jobs" of AMD with such a remarkable turnaround.

6900XT:
(It's absolutely amazing to see AMD compete with the 3090)


 
Last edited:

dzoni2k2

Member
Sep 30, 2009
153
198
116
Looks like we go the 1st official paper launch from PowerCooler


Literally.
 
  • Haha
Reactions: lightmanek

thilanliyan

Lifer
Jun 21, 2005
11,944
2,175
126
I'm in a quandary trying to decide(I'm waiting until prices drop on both the Radeon RX 6800,and the RTX 3070),which company to go for.

My current card is a Nvidia GTX 1080 8 GB, hooked up to a 32-inch 1440p Samsung LCD monitor.

The last AMD video card I had was a Radeon X1900XT...IIRC.
Wait for reviews then decide.
 

tviceman

Diamond Member
Mar 25, 2008
6,734
514
126
www.facebook.com
I'm happily shocked at the jumps AMD is claiming. Waiting for final reviews, but everything looks great. RT and their answer to DLSS is still up in the air, but efficiency looks fantastic.

A 6800 XT just may be in my future!
 
  • Like
Reactions: lightmanek

ModEl4

Member
Oct 14, 2019
71
33
61
TPU do not test Bordelands with Badass mode and they use the same API for both cards. AMD said they used the best API for each card.

Cross comparing FPS numbers for different setups is a folly, don't do it. Far too many variables to account for.



They also showed the 5700XT as being faster than the 2070 and then when the real benchmarks came out the 5700XT was ahead by a few % more than AMD claimed in their event.

On top of that they avoided games like Death Stranding, F1 2020, Horizon:Zero Dawn where the 5700XT is faster than the 2070S @4k so they could have cherry picked a lot harder if they had wanted to.
That's why I didn't mention a specific fps result for Borderlands 3 in contrast with Gears 5. Still think the Borderlands result is around 2-3% off and the Gears result is even more off.
in any case I didn't take into account these differences, I think that the results will be at least -2.5% in relation with what AMD's slides are suggesting based on a more fair testing with more games. I just mentioned it because it's not out of the realm of possibility 6900XT to be just 1% better than 3080 in 4K in a 10900K without Rage which will be a huge fail at $999...
Based on Navi 22 rumors (2.5GHz, 64RBEs, 4 primitives, 40CU, 192bit bus, infinite cache) probably will be just 30% better than a 5700XT at most (meaning 2070S will be just -17% & 2080S just -4% vs 6700XT) Guys here said it will be around +5% vs 2080Ti, then it was at 2080Ti level, a few days after it was 2080Ti -5%, now we are at 2080S +10%, lol when does it stop?
 

Hitman928

Diamond Member
Apr 15, 2012
6,123
10,532
136
That's why I didn't mention a specific fps result for Borderlands 3 in contrast with Gears 5. Still think the Borderlands result is around 2-3% off and the Gears result is even more off.
in any case I didn't take into account these differences, I think that the results will be at least -2.5% in relation with what AMD's slides are suggesting based on a more fair testing with more games. I just mentioned it because it's not out of the realm of possibility 6900XT to be just 1% better than 3080 in 4K in a 10900K without Rage which will be a huge fail at $999...
Based on Navi 22 rumors (2.5GHz, 64RBEs, 4 primitives, 40CU, 192bit bus, infinite cache) probably will be just 30% better than a 5700XT at most (meaning 2070S will be just -17% & 2080S just -4% vs 6700XT) Guys here said it will be around +5% vs 2080Ti, then it was at 2080Ti level, a few days after it was 2080Ti -5%, now we are at 2080S +10%, lol when does it stop?

AMD's 3080 numbers in Borderlands 3 are right in line with what Hexus got when testing Badass quality.

b042e288-dce3-44e8-b054-5124bc8c8058.png


Their Gears 5 numbers line up too:

eeee5fa3-af3c-4c64-bc51-8fbebebd32b6.png


2020-10-28_11-13-38-jpg.32498


Obviously different setups may swing fps a little bit one way or another and we'll get the full picture when reviews come out, but at least in the two games you mentioned, it doesn't look like AMD was doing anything to decrease the 3080's performance in their comparison and several of the titles they chose weren't exactly AMD favorable titles, at least not for RDNA1 cards. I'd be pretty shocked if the 6900XT (without rage or SAM) is only 1% better than the 3080 at 4K, that would mean there would be no appreciable difference between a 6800XT and 6900XT.
 

alexruiz

Platinum Member
Sep 21, 2001
2,836
556
126
Looks like we go the 1st official paper launch from PowerCooler


I am looking forward to that one!
In my eyes, the Red Devil was the top RX 5700 XT.
 
  • Like
Reactions: lightmanek

Timorous

Golden Member
Oct 27, 2008
1,748
3,239
136
AMD's 3080 numbers in Borderlands 3 are right in line with what Hexus got when testing Badass quality.

b042e288-dce3-44e8-b054-5124bc8c8058.png


Their Gears 5 numbers line up too:

eeee5fa3-af3c-4c64-bc51-8fbebebd32b6.png


2020-10-28_11-13-38-jpg.32498


Obviously different setups may swing fps a little bit one way or another and we'll get the full picture when reviews come out, but at least in the two games you mentioned, it doesn't look like AMD was doing anything to decrease the 3080's performance in their comparison and several of the titles they chose weren't exactly AMD favorable titles, at least not for RDNA1 cards. I'd be pretty shocked if the 6900XT (without rage or SAM) is only 1% better than the 3080 at 4K, that would mean there would be no appreciable difference between a 6800XT and 6900XT.

I did some maths on 6800XT 'stock' vs an estimated 6900XT 'stock' and the 6900XT was about 8% faster on average. If we say stock 6800XT = 3080 (FE?) and use the computerbase 3090 FE review we get the 3090 FE being 11% faster than the 3080 FE on average so I expect the 3090 FE would be around 3% faster than the 6900XT without rage and SAM turned on. It seems like Rage should probably be turned on all the time so that will go down to a virtual tie and then if you have Ryzen 5k you get SAM to push you ahead.

Apparently a SAM like feature has been available in Linux for a long time now and can be used on most platforms with many GPUs so I would not be surprised to see it supported with other CPUs in the future but for now using it to push more Ryzen 5k systems is a smart business play.
 
  • Like
Reactions: lightmanek

Geranium

Member
Apr 22, 2020
83
101
61
Looks like we go the 1st official paper launch from PowerCooler

How come a teaser a paper launch??? Are you comparing apples to oranges.
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
I have some qualms with the pricing to an extent. That being said, the 6800XT and 6900XT look like great values. It appears these cards will equal (with Rage and SAM) their competitor. The 6900XT particularly is looking like a killer for the 3090. The smart buy seems to be the 6800XT if it really does roughly parallel the 3080 in 4K (and 1440p) performance, as it's only $70 more than the 6800 (a touch over 10%), yet offers a 30% gain in 4K performance.

Unless the 6800 drops to $479 or so I can't see it being viable against the 6800XT or the 3070. I think that at $479 you can justify that it represents a massive gain in performance (34%) at 4K over 5700XT with only a 17% increase in cost, which is what we all want - performance gains that exceed price increase.

CardPriceVRAMTFLOPSEst. 4K
RX 5700$3498GB7.959%
RX 5700XT$3998GB9.7566%
2080 Ti$999-$119911GB13.499%
3070$4998GB20.4100%
RX 6800$57916GB13.9~113% (SAM)
3080$69910GB39.8131%
RX 6800XT$64916GB18.6~131%
3090$149924GB35.7144%
RX 6900XT$99916GB20.6~144% (RAGE, SAM)
 
Last edited:
  • Like
Reactions: psolord

Z15CAM

Platinum Member
Nov 20, 2010
2,184
64
91
www.flickr.com
Man, after the RTX 3070 and RX 6800 introduction, I really bleed for those that bought a $1200 RTX 2080 Ti resonantly. Do not believe they could sell their used one for over $400 now.

On the other hand, I doubt anyone who can afford a $1200 GPU much cares !

It's unbelievable a GPU worth $1200 a month ago is worth less then $450 now !

That $70 charge difference between the RTX 3070 Vs the RX 6800 is just away AMD is rubbing nVidia's nose in the mud after all the years nVidia price gouging - LOL

I'm tossed between the RTX 3070 and RX 6800 or spend $100 more for an RX 6800XT that bests the RTX 3090 for less money; wishing I could afford $1000 to beat the $1200 RTX 3950 ;o(
 
Last edited:
  • Like
Reactions: lightmanek

ModEl4

Member
Oct 14, 2019
71
33
61
AMD's 3080 numbers in Borderlands 3 are right in line with what Hexus got when testing Badass quality.

b042e288-dce3-44e8-b054-5124bc8c8058.png


Their Gears 5 numbers line up too:

eeee5fa3-af3c-4c64-bc51-8fbebebd32b6.png


2020-10-28_11-13-38-jpg.32498


Obviously different setups may swing fps a little bit one way or another and we'll get the full picture when reviews come out, but at least in the two games you mentioned, it doesn't look like AMD was doing anything to decrease the 3080's performance in their comparison and several of the titles they chose weren't exactly AMD favorable titles, at least not for RDNA1 cards. I'd be pretty shocked if the 6900XT (without rage or SAM) is only 1% better than the 3080 at 4K, that would mean there would be no appreciable difference between a 6800XT and 6900XT.
Let's hope you're right, my assumption was that Borderlands was 2-3% off and Gears 3-4% off, averaging 2.5% in the low end 3.5% in the upper end. I'm not going to nit pick reviews trying to prove my case (I don't care), trying to find 2.5%, it doesn't natter so much, the difference is very small anyway.
77fps on 3950X in Nexus, 77fps on 5900X at 4K, maybe 5900X vs 3950X doesn't gain even 1fps at 4K in Gears 5.
We should wait for reviews, in any case AMD had a good showing, I'm not even worrying too much about the absence of ray tracing results, it will be a little bit slower but it will be just fine, every game eventually will be optimized for consoles, no big deal, what I want to see (when is ready) is the quality of SuperRes vs DLSS, this would be interesting.
 

Timorous

Golden Member
Oct 27, 2008
1,748
3,239
136
I have some qualms with the pricing to an extent. That being said, the 6800XT and 6900XT look like great values. It appears these cards will equal (with Rage and SAM) their competitor. The 6900XT particularly is looking like a killer for the 3090. The smart buy seems to be the 6800XT if it really does roughly parallel the 3080 in 4K (and 1440p) performance, as it's only $70 more than the 6800 (a touch over 10%), yet offers a 30% gain in 4K performance.

Unless the 6800 drops to $479 or so I can't see it being viable against the 6800XT or the 3070. I think that at $479 you can justify that it represents a massive gain in performance (34%) at 4K over 5700XT with only a 17% increase in cost, which is what we all want - performance gains that exceed price increase.

CardPriceVRAMTFLOPSEst. 4K
RX 5700$3498GB7.959%
RX 5700XT$3998GB9.7566%
2080 Ti$999-$119911GB13.499%
3070$4998GB20.4100%
RX 6800$57916GB13.9~100%
3080$69910GB39.8131%
RX 6800XT$64916GB18.6~131%
3090$149924GB35.7144%
RX 6900XT$99916GB20.6~144%

The 6800 is more like ~115% in your chart. No idea where you got 100% from.
 
  • Like
Reactions: Tlh97

antihelten

Golden Member
Feb 2, 2012
1,764
274
126
AMD's 3080 numbers in Borderlands 3 are right in line with what Hexus got when testing Badass quality.

As can be seen from this roundup from 3Dcenter.org, Hexus isn't the only site that got numbers like this.

In fact it's pretty clear that Techpowerup is the outlier here (since they didn't use Badass settings).

The same thing can also be seen for Gears 5, where Techpowerup again got the highest results. Looking into this a bit more, it looks like TPU runs their Gears 5 bench without motion blur enabled, which apparently can make a difference of about 5%. So with motion blur enabled TPU would have gotten somewhere around 80 FPS

With all that being said though, there is definitely something weird going on with AMDs Gear results. During the presentation they reported that the 6800XT gets 78 FPS, however if you go to their website, they report that it gets 84 FPS. One possible explanation might be that the latter result is from enabling Rage and SAM.

All in all, none of this really matters though. Either the 6800 XT is equal to the 3080 (78 versus ~77 FPS going with the low results) or it is still equal to the 3080 (84 versus 84,4 FPS going with the high results).
 
  • Like
Reactions: lightmanek

Hitman928

Diamond Member
Apr 15, 2012
6,123
10,532
136
With all that being said though, there is definitely something weird going on with AMDs Gear results. During the presentation they reported that the 6800XT gets 78 FPS, however if you go to their website, they report that it gets 84 FPS. One possible explanation might be that the latter result is from enabling Rage and SAM.

If you look at the footnotes on the website results, SAM and rage mode are on versus off in the presentation.
 
  • Like
Reactions: lightmanek

Z15CAM

Platinum Member
Nov 20, 2010
2,184
64
91
www.flickr.com
One of my favorite platforms was the Intel Z68 Chipset with the I7 2700K Sandy Bridge running at 5G's with CrossFire 290X's (Under H2O) back in the Day - Still have 2 or three of them rigs Rigs with 2300 Mhz Samsung Green DDR3 and love them.

I'm now playing with an ASUS X470 CH7, 32GB CL14 @ 3200Mhz (Can do 16Cl at 3600Mhz' with a little more voltage) and the CPU R7 3700X CPU at 4400Mhz's, NMVe PCIe3 drives - Some what of a Mid range PC by today's standards but would not mind either a RTX 3070 or RX 6800 to replace my Crossfire 290X's (Some what better than a GTX 1080). Monitors are a Korean beauties 27" 96 Hz IPS 1440P QNIX display or a Japanese 165 Hz IPS Pixio Prime 7 27" display. Funny thing is I'm not into gaming - If I was it would be the 165hz 1440P 2K IPS Pixio with either a nVidia RTX 3070 or AMD RX 3800 @ 1440P on this platform.

Really I'm looking at around $800 CDN for a Video card just to upgrade a bit ;o(

Doubt I would get anything for my Crossfire Sapphire XSPC H20 290X's = The R9 295X2. Capable of running a solid 1175 X 1600 Mhz's at under 75C under full load with a spare card - Wattage is horrendous at least 800 Watts on the Intel Z68.

Buildzoid would certiantly be interested. The issue is the cost of shipping 10lbs of cards from Canada to the UK !

To think of it I will notify him ;o)
 
Last edited:

antihelten

Golden Member
Feb 2, 2012
1,764
274
126
If you look at the footnotes on the website results, SAM and rage mode are on versus off in the presentation.

Aah, you're right, I missed that. Thanks

Judging from AMD presentation (and the message that AMD is trying to pass - that it has a card that can compete with 3090) as far I can tell we would have in 4K:

So AMD just released the full numbers used in the presentation (here).

Someone on reddit collated all of the numbers and calculated the averages:

6900XT is 3.1% faster than 3090 at 4K.
6900XT is 7.3% faster than 3090 at 1440P.
6800XT is 4.8% faster than 3080 at 4K.
6800XT is 9.3% faster than 3080 at 1440P.
6800 is 15.3% faster than 2080 Ti at 4K.
6800 is 19.3% faster than 2080 Ti at 1440P.

Now all of these numbers are with SAM and possibly also Rage on (the page only mentions SAM), so without those the numbers would probably look something like this (assuming 5% average boost from SAM and Rage):

6900XT is 1.9% slower than 3090 at 4K.
6900XT is 2.2% faster than 3090 at 1440P.
6800XT is 0.2% slower than 3080 at 4K.
6800XT is 4.1% faster than 3080 at 1440P.
6800 is 9.8% faster than 2080 Ti at 4K.
6800 is 13.6% faster than 2080 Ti at 1440P.

If we furthermore apply the suggested 2.5% to the Nvidia results to account for cherrypicked benches, then we get the following:

6900XT is 4.2% slower than 3090 at 4K.
6900XT is 0.3% slower than 3090 at 1440P.
6800XT is 2.7% slower than 3080 at 4K.
6800XT is 1.6% faster than 3080 at 1440P.
6800 is 7.1% faster than 2080 Ti at 4K.
6800 is 10.8% faster than 2080 Ti at 1440P.
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
Aah, you're right, I missed that. Thanks



So AMD just released the full numbers used in the presentation (here).

Someone on reddit collated all of the numbers and calculated the averages:

6900XT is 3.1% faster than 3090 at 4K.
6900XT is 7.3% faster than 3090 at 1440P.
6800XT is 4.8% faster than 3080 at 4K.
6800XT is 9.3% faster than 3080 at 1440P.
6800 is 15.3% faster than 2080 Ti at 4K.
6800 is 19.3% faster than 2080 Ti at 1440P.

Now all of these numbers are with SAM and possibly also Rage on (the page only mentions SAM), so without those the numbers would probably look something like this (assuming 5% average boost from SAM and Rage):

6900XT is 1.9% slower than 3090 at 4K.
6900XT is 2.2% faster than 3090 at 1440P.
6800XT is 0.2% slower than 3080 at 4K.
6800XT is 4.1% faster than 3080 at 1440P.
6800 is 9.8% faster than 2080 Ti at 4K.
6800 is 13.6% faster than 2080 Ti at 1440P.

If we furthermore apply the suggested 2.5% to the Nvidia results to account for cherrypicked benches, then we get the following:

6900XT is 4.2% slower than 3090 at 4K.
6900XT is 0.3% slower than 3090 at 1440P.
6800XT is 2.7% slower than 3080 at 4K.
6800XT is 1.6% faster than 3080 at 1440P.
6800 is 7.1% faster than 2080 Ti at 4K.
6800 is 10.8% faster than 2080 Ti at 1440P.
If you can run your 6900XT with SAM and RAGE on and it doesn't void the warranty, and it's just a setting to flip, why should we exclude those results from testing?

I wonder what testing policies will be. I know many review sites like TPU test overclocking, for 3090 it gains 4.8%, but the overclock voids the warranty.

Odd situation to be in.