AMD Radeon RDNA2 Navi (RX 6500, 6600, 6700, 6800, 6900 XT)

[CarstenS]

Your proof that resolution scaling is IC-related, is…?


Navi 33's target is 1080p. It's by far the most power-efficient GPU for 1080p. It has 28 % better power efficiency than GA106:
https://www.computerbase.de/2021-10...3/#abschnitt_energieeffizienz_in_fps_pro_watt

Has to be 40 % over Ampere to be convincing or what?

I wonder why, if it's by far the most efficient GPU for 1080p, Navi23 has only an index rating of 93%. What's 100%?

A click on "+9 Einträge" reveals, Navi21 LE aka RX 6800 is at 100%. Incidentally the card i was trying to buy for months before giving up.

 

[CarstenS]

Another fun fact, as I've been looking at availabity of RX 6600 and numbers. German e-tailer mindfactory seemingly had a large contingent of sapphire pulse cards. So large, that an hour after sales started, more than 350 were sold already at 499 € and they claimed more than 5 in stock. Refreshing that particular page over a couple of hours until now revealed: still more than 350 sold, but not more than, say 400? And still cards in stock. Makes me wonder how reliable those numbers are.

edit: next morning, the counter seems to work again, maybe just a hiccup yesterday. 550 sold at 499 Euro.

edit2: roughly another 12 hours later, card has sold out at 670 pieces of stock moved.

 

[Subtlesnake]

I am starting to see the flaws of RDNA2's cache startegy, while it works great for top dog GPUs, it scales down miserably to low end GPUs, making them only fit for specific resolutions. The RX 6600 is suitable only for 1080p gaming, going 1440p makes it fall behind it's RDNA1/Turing/Ampere competitors, Ray Tracing is absolutely trash with the 6600 and 4K is a no go. These GPUs will probably be a disaster in regards to future proofing your purchase, as they only work within a narrow band of performance. It's also why I think consoles didn't go that route, and opted for regular old methods instead.

Also they are not terribly power efficient considering their 7nm origin and narrow memory bus, compared to their Ampere competitors.

Given its 128-bit bus and cutback CUs vs the 6600 XT, I think it makes more sense to think of the 6600 as a 5600 XT replacement – that's priced so high because we live in a world where the 3060 is retailing for $800+ on Newegg. Realistically, you shouldn't be buying this class of card to "future proof", since requirements are only going to go up over time.

 

[trinibwoy]

It’s amazing to see graphics cards are still being bid up to 2x MSRP on eBay auctions. I’m all for free market capitalism when it comes to luxury products like GPUs and it seems enough buyers are still valuing these cards at much higher than MSRP. As long as this continues it means the scalper price is the right price so the 6600’s $329 launch price is academic at best.

The only way prices come down is if people stop buying and let GPUs marinate on store shelves for a while.

 

[Deleted member 13524]

The only way prices come down is if people stop buying and let GPUs marinate on store shelves for a while.

Gamers might let GPUs marinate, but miners will never not buy graphics cards that start pumping out profits just 3 to 4 months after being bought.

 

[trinibwoy]

Gamers might let GPUs marinate, but miners will never not buy graphics cards that start pumping out profits just 3 to 4 months after being bought.

Yeah that's the problem. We don't have separate miner and gamer markets. So if miners say a 6600 is worth $700 then that's what it's worth. I don't know how we fix that.

 

[Deleted member 2197]

AMD has effectively recaptured the power efficiency crown from Nvidia, at least at this price point. It’s worth noting that achievement because AMD has struggled with power efficiency against Nvidia for at least the past eight years. If what you care about is power efficiency, the Radeon RX 6600 deserves some accolades. But as the spec sheet shows, it achieves some of these improvements by offering less RAM and a smaller memory bus. The onboard Infinity Cache is intended to compensate for the smaller memory bus, but while this works quite well for the 6700 XT and above, the 6600 XT proved to be less well balanced. Performance reports indicate the RX 6600 has problems as well:

https://www.extremetech.com/extreme/328188-the-radeon-rx-6600-is-amds-weakest-rdna2-gpu-yet

 

[Deleted member 13524]

Yeah that's the problem. We don't have separate miner and gamer markets. So if miners say a 6600 is worth $700 then that's what it's worth. I don't know how we fix that.

No, the miner market needs to die. Proof of work is a greed-driven cancer that needs to be stopped, for the sake of humanity and the environment.
The governments of the U.N., E.U., U.S.A. etc. are all using tax payer money to promote green energy at a ridiculous pace, pressuring industries to change faster than feasibly possible in many cases, while at the same time letting massive mining operations to proceed tax-free that consume as much power as entire countries.

One year we're worshipping Greta for talking down to actual industry and environmental experts, but the next year we're all happy about creating mining operations run out of fossil fuel in the U.S. because we bought 0.00001 BTC and we're "profiting" from it.

 

[Rootax]

Who are "we" ? Governments don't care about your well being, or ecology... It's all about making profits. If "green" sell, then push green, come here little girl, talk about stuff you don't know about (but a lot of ppl believe because it make them look righteous). Oh now crypto is a danger to banks ? Look, we make CBDC soon !

Eth mining will die soon anyway, and btc use mosty asics for a while now, so it's not impacting gpu much. Some will still mine eth classic&co, but I doubt ALL the gpu miners will follow.

 

[Davros]

Techpowerup Doom scores seem a little confusing
3060 12gb scores double that of 3070 8gb which makes you think the 8gb isnt enough but then the 3060ti 8gb which should have roughly 3070 performance is about 25% faster than the 3060

 

[Malo]

Techpowerup Doom scores seem a little confusing
3060 12gb scores double that of 3070 8gb which makes you think the 8gb isnt enough but then the 3060ti 8gb which should have roughly 3070 performance is about 25% faster than the 3060

Maybe you should link to the actual benchmark page you're referring to.

 

[DegustatoR]

https://www.techpowerup.com/review/powercolor-radeon-rx-6600-fighter/16.html

https://www.techpowerup.com/review/gainward-geforce-rtx-3080-ti-phantom-gs/15.html

Seems fine to me.

 

[Pete]

Maybe you should link to the actual benchmark page you're referring to.

Davros is talking about the Doom Eternal 1440p RT benchmark about ten posts up:

Spoiler

3070 8GB 42fps
6600 8GB 65fps
3060 12GB 97fps.

Different drivers or game versions, maybe?

 

[Deleted member 13524]

At first glance it looks like Nvidia cards with 8GB or less running out of VRAM, but then there's that 3060Ti that should never outperform a 3070.
Perhaps it is some driver mishap indeed.

 

[iroboto]

Davros is talking about the Doom Eternal 1440p RT benchmark about ten posts up:

Spoiler

3070 8GB 42fps
6600 8GB 65fps
3060 12GB 97fps.

Different drivers or game versions, maybe?

The texture pool size setting can really leap up fast once you go beyond ultra level setting. I think on Nightmare, the 3070 will usually be at it's break point. But yea, not understanding why a 3060TI with 8GB is outperforming a 3070 here.

 

[Davros]

Thanks Pete
(you missed out the 3060ti)
3070 8GB 42fps
6600 8GB 65fps
3060 12GB 97fps.
3060ti 8GB 124fps (nearly 3x the performance of the 3070)

 

[Silent_Buddha]

Thanks Pete
(you missed out the 3060ti)
3070 8GB 42fps
6600 8GB 65fps
3060 12GB 97fps.
3060ti 8GB 124fps (nearly 3x the performance of the 3070)

That almost makes me wonder if the NV driver is doing somethign differently on the 3060's and not rendering the same things as the 3070. Or the 3070 driver is bugged since in no way should it ever be slower than the 6600. Or a combination of both.

Regards,
SB

 

[PSman1700]

3060Ti and 6700XT are really where the line starts of the better performers there. 5700XT/2070 seem aged already, tech's still movin' fast lol.

 

[Malo]

Davros is talking about the Doom Eternal 1440p RT benchmark about ten posts up:

Ah ok, thanks. Couldn't see posts due to ignored user.

That makes zero sense for the 3070. Are we sure their data isn't just wrong, that they made a mistake inputting their numbers for the 3070? Are those results mirrored anywhere else?

 

[Davros]

In the other games benchmarked by Techpowerup the 3060ti/3070 are were they should be, the 3070 slightly faster than the 3060ti
Deathloop
3060ti 91.9fps - 3070 103.6fps

F1 2021
3060ti 107.3fps - 3070 123.1fps

Metro Exodus
3060ti 92.5fps - 3070 108.1fps