Radeon 7900 Reviews

[Stuka87]

Will update this list as more come along.

ArsTechnica:
(Ryzen 5800X3D, Asus ROG Crosshair VIII Dark Hero, 64GB DDR4-3200, Windows ???)
https://arstechnica.com/gadgets/202...0-gpus-are-great-4k-gaming-gpus-with-caveats/

Gamers Nexus:
https://www.youtube.com/watch?v=We71eXwKODw

Guru3D:
(Ryzen 5950X, ASUS X570 Crosshair VIII HERO, 32 GB (4x 8GB) DDR4 3600 MHz, Windows 10)
https://www.guru3d.com/articles-pages/amd-radeon-rx-7900-xtx-review,1.html

Hardware Canucks
(Ryzen 7700X, Asus X670E ROG Crosshair hero, 32GB DDR5-6000, Windows 11)
https://www.youtube.com/watch?v=t3XPNr506Dc

Hardware Unboxed:
(Ryzen 5800X3D, MSI MPG X570S Carbon Max WiFi, 32GB DDR4-3200, Windows 11)
https://www.youtube.com/watch?v=4UFiG7CwpHk

Igor's Lab:
(Ryzen 7950X, MSI MEG X670E Ace,32GB DDR5 6000)
https://www.igorslab.de/en/amd-rade...giant-step-ahead-and-a-smaller-step-sideways/

Jay's Two Cents:
https://www.youtube.com/watch?v=Yq6Yp2Zxnkk

KitGuruTech:
(Intel 12900K, MSI MAG Z690 Unified, 32GB DDR5)
https://www.youtube.com/watch?v=qThrADqleD0

Linus Tech Tips:
https://www.youtube.com/watch?v=TBJ-vo6Ri9c

Paul's Hardware:
(Ryzen 7950X, Asus X670E ROG Crosshair Hero, 32GB DDR5-6000, Windows 11)
https://www.youtube.com/watch?v=q10pefkW2qg

PC Mag:
(Intel 12900K, Asus ROG Maximus Z690 Hero, 32GB 5600MHz, Windows 11)
https://www.pcmag.com/reviews/amd-radeon-rx-7900-xtx

Tech Power Up:
(Intel 13900K, ASUS Z790 Maximus Hero, 2x 16 GB DDR5-6000 MHz, Windows 10)
AMD: https://www.techpowerup.com/review/amd-radeon-rx-7900-xtx/
ASUS: https://www.techpowerup.com/review/asus-radeon-rx-7900-xtx-tuf-oc/
XFX: https://www.techpowerup.com/review/xfx-radeon-rx-7900-xtx-merc-310-oc/

Tech Spot:
(Ryzen 5800X3D, MSI MPG X570S, 32GB of dual-rank, dual-channel DDR4-3200 CL14, Windows ???)
https://www.techspot.com/review/2588-amd-radeon-7900-xtx/

TechTesters:
(Intel 13900K, ASUS ROG Maximus Z790 HERO, 32GB DDR5-6000, Windows 11)
https://www.youtube.com/watch?v=3uQh4GkPopQ

 

[Stuka87]

Thank you, appreciate it!
Do you think it would be possible to add the test setup of each review?

Written ones are done, I will have to do the video ones in a bit.

 

[DAPUNISHER]

Kitguru review is excellent as usual. His game suite had a gen on gen uplift of 39%. Outside of a couple of games I have seen tested, AMD did not deliver the claimed performance uplift. Dom used a 12900K for testing. He also experienced coil whine, same as GN. Looks like a partner card will be the way to go so far.

 

[Timorous]

XTX is about the same as the 4080 but has some issues - the expected lower RT performance, but also the infinity cache means it behaves like a card with too little memory (performance and 1% mins fall off in a similar way to cards that are starting to run out of memory). If they'd doubled the infinity cache so it wasn't struggling at 4k it would have been much stronger, which being as it's on chiplets should have been easy!

Still there are no bad cards, just bad prices, if they drop the price a bit then it's not such a bad buy. XT needs to drop more.

I might be feeling down on the card but it does not struggle at 4K at all. If anything it does relatively better when increasing resolution than the 4080 does.

 

[Unreal123]

So basically AMD increased clock speed of RX 7900 XTX to keep up with RTX 4080, which made 20% more power hungry than RTX 4080.

 

[Saylick]

I think this plot from TPU is very enlightening about what's going on with the silicon. It's power starved depending on the workload, likely as a result of them capping it to 355W TBP, but that itself normally should not be an issue if the VF curve wasn't all jacked up. It really does need a respin. At least they report a few instances of the clocks getting close to 3 GHz.

In comparison, here's the 4080:

 

[Hitman928]

I see this release as basically a , "meh". Not as good as their PR made it seem it would be, not nearly as bad as some past releases. Raster performance is a little less than I expected but RT performance is slightly better. Still needs some driver refinements it seems. In the competitive landscape, it doesn't really move the needle one way or the other.

 

[DAPUNISHER]

Paul tests with a Ryzen 7950X and gets right at a 40% gen over gen uplift in raster.

I expect the XTX will sellout immediately. Our crew gripes, but everything about this $1000 card is a solid buy. Paul includes RTX 30 series prices as of filming. The lowest I have seen a 3090 go for at any point was $950 NIB. This is a great buy in comparison.

 

[linkgoron]

7900xtx is weaker than what AMD let us believe. It's just faster than a 4080 by a hair.
IMO It's quite clear that something is wrong with the 7900xtx and it's either a design issue, or a bug and a respin might help it out.

Having said that, the 7900xtx is priced well vs Nvidia IMO, and should hopefully sell well.

 

[Stuka87]

My expectations may have been way off others, but I fully expected the 7900XTX to trade blows with the 4080. Even their benchmarks basically said this (though in a very indirect way as it compared to the 6900XT).

So to me at least, not underwhelming. Its right where I thought it would be, and its priced lower than the competition. Though I wonder how much better it may have been if it was not for the rumored hardware issue that caused higher power consumption.

 

[insertcarehere]

My expectations may have been way off others, but I fully expected the 7900XTX to trade blows with the 4080. Even their benchmarks basically said this (though in a very indirect way as it compared to the 6900XT).

Their benchmarks were showing 50-60+% increases in raster vs a 6950XT...

Next time people say AMD doesn't mislead or cherry pick I will refer back to this launch...

 

[Stuka87]

Their benchmarks were showing 50-60+% increases in raster vs a 6950XT...

Next time people say AMD doesn't mislead or cherry pick I will refer back to this launch...

But those FPS are within a few of what most reviewers show. So how is AMD's chart (that you posted) way off and misleading it if matches what independent reviewers are seeing?

Also note, as I recall AMD tested using a Ryzen 7000 series CPU. Reviewers are using Ryzen 5000 or Intels.

 

[Timorous]

Their benchmarks were showing 50-60+% increases in raster vs a 6950XT...

Next time people say AMD doesn't mislead or cherry pick I will refer back to this launch...

As you should. AMD have blown years of trust building with their marketing material only to blow it in a 20 minute presentation.

 

[alexruiz]

I think the cards are exactly where AMD wanted them to be:

7900 XTX faster in raster than RTX 4080, and faster than RTX 3090Ti in RTX while priced almost as a RTX 3090.
Overall win

7900 XT faster in raster than RTX 3090 Ti, same as RTX 3090 in RT while priced like a 3080 Ti.
Overall win

Let's not forget also the state of the drivers, with team red having a not so spectacular track record with launch day product maturity.
Expect them to get better and better as time goes on.

While many of us wanted to see the 7900 XTX closer to the 4090 and the 7900 XT beat the 4080, the price was a giveaway of where they would land.
AMD priced them right.
The disruption factor might come in the lower tiers GPUs.

 

[gdansk]

But those FPS are within a few of what most reviewers show. So how is AMD's chart (that you posted) way off and misleading it if matches what independent reviewers are seeing?

Also note, as I recall AMD tested using a Ryzen 7000 series CPU. Reviewers are using Ryzen 5000 or Intels.

It's misleading because they chose above average examples.

 

[Timorous]

But those FPS are within a few of what most reviewers show. So how is AMD's chart (that you posted) way off and misleading it if matches what independent reviewers are seeing?

Also note, as I recall AMD tested using a Ryzen 7000 series CPU. Reviewers are using Ryzen 5000 or Intels.

Just to use Pauls Hardware results as they did use Ryzen 7000 they have the 7900XTX as 52% faster than the 6950XT in CP2077 which is miles off of the 67% AMDs slides show.

This is repeated across multiple reviews with both Intel and AMD configs.

 

[PJVol]

So to me at least, not underwhelming. Its right where I thought it would be, and its priced lower than the competition.

Yep, the performance is here, though I expected at least +50% to the prev.gen, but power effieciency is way off this time.
The latter could potentially be improved in N32

 

[Stuka87]

It's misleading because they chose above average examples.

Then why don't the people that are complaining up and down about AMD's presentation not doing the same with nVidia? Or Intel? This literally happens every time a new CPU/GPU comes out. The manufacturer chooses the best benchmarks to display.

 

[Timorous]

It's misleading because they chose above average examples.

Not even that. Nowhere gets close to seeing a 67% uplift over the 6950XT in CP2077. Some struggle to see that vs the 6900XT. It is utter garbage marketing by AMD that has undone years of hard work.

 

[Timorous]

Then why don't the people that are complaining up and down about AMD's presentation not doing the same with nVidia? Or Intel? This literally happens every time a new CPU/GPU comes out. The manufacturer chooses the best benchmarks to display.

NV didn't this time. The 3 games they showed raster performance gains in ended up averaging within 5% of the final performance delta for the 4090 and 4080.

Also it is something AMD have avoided doing for years. There will always be mild discrepancies within a few % but a 15% miss is an entirely different category.

 

[GodisanAtheist]

Big OOOOOOOOOfffffffff on those review numbers.

If it takes AMD 530mm2 of silicon to match NV's 376mm2 of silicon in raster and outright lose in RT then we've basically gone back to the Vega gen (which is, needless to say, bad). Hopefully something is just broken with the N31 GCD and this isn't an indictment of he chiplet strategy on gaming GPUs, we'll see.

4090 is like the 1080Ti, floating above the rif-raff down below.

4080/7900XTX are fighting neck and neck for 30% faster than the prior gen, and then you don't even really get any substantial $$$/performance improvement gen on gen from there.

I figured a good amount of the "counter hype" we'd been seeing over the last few days was AMD trying to reign in the hypetrain it had created, but I didn't know it was going to go this poorly.

 

[gdansk]

Then why don't the people that are complaining up and down about AMD's presentation not doing the same with nVidia? Or Intel? This literally happens every time a new CPU/GPU comes out. The manufacturer chooses the best benchmarks to display.

Because Nvidia almost totally avoids providing any useful numbers in their slides. It's all mixed with useless DLSS numbers. It's pretty bad but not like RTG cherry picking to show the 7900 XTX as 50% faster than the 6950 XT in rasterization when it's not even close to that.

RTG marketing group is consistently the worst within AMD. But this time they had an impossible task: make an overpriced dud look appealing. So we got AMD's most misleading presentation in years.

 

[gdansk]

Big OOOOOOOOOfffffffff on those review numbers.

If it takes AMD 530mm2 of silicon to match NV's 376mm2 of silicon in raster and outright lose in RT then we've basically gone back to the Vega gen (which is, needless to say, bad). Hopefully something is just broken with the N31 GCD and this isn't an indictment of he chiplet strategy on gaming GPUs, we'll see.

4090 is like the 1080Ti, floating above the rif-raff down below.

4080/7900XTX are fighting neck and neck for 30% faster than the prior gen, and then you don't even really get any substantial $$$/performance improvement gen on gen from there.

I figured a good amount of the "counter hype" we'd been seeing over the last few days was AMD trying to reign in the hypetrain it had created, but I didn't know it was going to go this poorly.

For AMD's Radeon group I hope it is just Navi 31's problem and not all of RDNA3 that's messed up. Or they'll slip below 10% market share in no time.

 

[coercitiv]

Then why don't the people that are complaining up and down about AMD's presentation not doing the same with nVidia?

I don't think you want us to come down on RDNA3 as hard as we did with Ada or Arc.

 

[Kaluan]

I think I enjoyed Igor's review so far the best, one of the few that actually put some effort in it.

They have averages (with loads of percentile data) for raster, RT/hybrid AND generalized averages (which are 6:4 raster:RT, which I think is spot on if you value RT) versus 4080 baseline.

Edit: Here's a sample from thr review (raster+RT avg)


Hardware Unboxed's is decent too. But parts of the conclusions made me cringe pretty hard. Especially the bit about buying better features with nVidia, "better encoding" in partucular... which they didn't even test (WTF?). And going by some initial skimming on encoding/streaming, OptimusTech's showed that at least under AV1, RDNA3 shows potential to be on par with Lovelace in streaming quality.
I expect HUB to needlessly ramble on sometimes, but commenting on stuff their review didn't even cover (and especially so early on) is an epic derp on their part.

 

[GodisanAtheist]

Because Nvidia almost totally avoids providing any useful numbers in their slides. It's all mixed with useless DLSS numbers. It's pretty bad but not like RTG cherry picking to show the 7900 XTX as 50% faster than the 6950 XT in rasterization when it's not even close to that.

RTG marketing group is consistently the worst within AMD. But this time they had an impossible task: make an overpriced dud look appealing. So we got AMD's most misleading presentation in years.

- Yeah it was wild you could almost see the exact point where the "spin cycle" started going in reverse over the last few days with discussion of bugs in N31 etc.

I feel like we get messages in a bottle from AMD's marketing from time to time, where their official stuff has to display the product in the best light while they have to also, simultaneously, keep everyone's expectations in check.

I do not envy anyone in that position.