Sapphire Radeon RX 9070 XT Nitro+

[tfdsaf]

Personally if I can find one for $600 I'd be buying it. It just depends if there are msrp cards available.

For 600 I think it's a great product. Essentially almost a 4080, but way cheaper.

 

[TSiAhmat]

Is it really inefficient?

Spoiler

Idle also half of an 5070 ti, this version was probably pushed to hard. (if you want performance per watt)

 

[Rover4444]

Giggles? Why not? It's as good a starting point as any.

Oh, so an arbitrary number not based on anything. I ask because the 9070 XT here doesn't do that number while gaming. That's pretty bad faith of you.

Yeah, average load. As a STARTING point. Here a hint for ya: Not going to happen in this gen on this performance tier.

I used the V-Sync 60 hz data and you're right, there's nothing more efficient at a higher performance tier - although that's at 80w, which is below your baseline. At higher wattages it's not as performant as the competition

Actually no, why am I doing this? Just read the review of the Pulse card's power consumption when it comes out.

Then buy yourself a 7800XT and quit whining about it here..

I apologize, but you're going to have to acommodate me. Whining is how I have fun.

 

[Krit]

36% performance increase for 20% more money over the 7800 XT with no VRAM upgrade is extremely bad.

On average it's 46% not 36% and in some cases even 50% look at Tech Yes City review! There are no comparison with blackwell at all even RTX 5090 gain is significantly weaker.

 

[lexluthermiester]

Is it really inefficient?

Spoiler

View attachment 388123

Idle also half of an 5070 ti, this version was probably pushed to hard. (if you want performance per watt)

And that is 31W of extra power. Not going to make a difference. But if you look at the graphs for RT or Full load, it's nearly 100W in favor of the 5070ti, but again, that's not going make a big difference..

I ask because the 9070 XT here doesn't do that number while gaming. That's pretty bad faith of you.

I does, just depends on what you're doing. Again, I'm just talking about a generic average, as an example. Seems you missed that point.

Just read the review of the Pulse card's power consumption when it comes out.

No thanks, don't really care.

Whining is how I have fun.

Hmm... This explains a lot..

 

[Rover4444]

On average it's 46% not 36% and in some cases even 50% look at Tech Yes City review!

For which resolution? I'm using the TPU chart which says it's 4k. At those numbers the value uplift is decent, but I wouldn't say enough to make people replace their midrange/high end RDNA 2/3 cards unless they like raytracing.

 

[TSiAhmat]

An that is 31W of extra power. Not going to make a difference. But if you look at the graphs for RT or Full load, it's nearly 100W in favor of the 5070ti, but again, that's not going make a big difference..

yep. thats the point!

 

[JustBenching]

Is it really inefficient?

Spoiler

View attachment 388123

Idle also half of an 5070 ti, this version was probably pushed to hard. (if you want performance per watt)

Yes, it is very inefficient. What game is that tested, cause getting 60 vsynced at 80w means it's a game like CS2 or something.

 

[Krit]

For which resolution? I'm using the TPU chart which says it's 4k. At those numbers the value uplift is decent, but I wouldn't say enough to make people replace their midrange/high end RDNA 2/3 cards unless they like raytracing.

TechPowerUp

RTX 5070 (Average FPS) @4k is 25% faster than RTX 4070 and RT @4k is about 16% faster than RTX 4070 so combined result is about + ~20.5%
RX 9070 XT (Average FPS) @4k is 46% faster than RX 7800 XT and RT @4k is about 85% faster than RX 7800 XT so combined result is about + ~65.5%

 

[Hecate91]

This is the issue you've compared it to extremely gouged prices. That doesn't make these cards good value it makes them the least bad option.

Compared to Nvidia's fake MSRP and purposely limited supply, these are are a much better value.
I think AMD could've undercut harder with the XT at $549, but the market needs something in stock at MSRP.

 

[Krit]

Can anybody remember those people saying that RX 9070 XT will be on the same level as RX 7900 GRE ? They probably were looking at blackwells absurd gains and except the final conclusions from that.

 

[TSiAhmat]

Yes, it is very inefficient. What game is that tested, cause getting 60 vsynced at 80w means it's a game like CS2 or something.

maybe not sure, sadly not mentioned on the graph. Might have to ask the W1zzard himself. nvm found it

 

[Rover4444]

TechPowerUp

RTX 5070 (Average FPS) @4k is 25% faster than RTX 4070 and RT @4k is about 16% faster than RTX 4070 so combined result is about + ~20.5%
RX 9070 XT (Average FPS) @4k is 46% faster than RX 7800 XT and RT @4k is about 85% faster than RX 7800 XT so combined result is about + ~65.5%

Is the 36% on the relative performance for the GPU database wrong, then? And yes RT performance is pretty compelling, but by how much remains to be seen. Great value if you care about RT.

Can anybody remember those people saying that RX 9070 XT will be on the same level as RX 7900 GRE ? They probably were looking at blackwells absurd gains and except the final conclusions from that.

I remember, others were saying it would be 7900 XT level, too. I also remember a bunch of people saying $700 or higher was a decent price.

 

[Krit]

Is the 36% on the relative performance for the GPU database wrong, then?

I'm looking at average fps for all games RX 7800 XT vs RX 9070 XT = +46% performance boost.

 

[Rover4444]

I'm looking at average fps for all games RX 7800 XT vs RX 9070 XT = +46% performance boost.

Where you seeing that? On the 7800 XT page it says 136% for the 9070 XT.

 

[tfdsaf]

How did you figure that? The 9070xt consumes more power than a 5080, a 5070ti, a 4070ti super, a 4080, a 4080 super while being as fast or slower than all of these. Almost every nvidia card of the last 2 generations (bar the 4060) is more efficient

https://tpucdn.com/review/sapphire-radeon-rx-9070-xt-nitro/images/energy-efficiency.png

It's a custom model with 40w more at baseline and depending on max clock even more. The difference between 304w and 350w is 4% performance. It's up to you to decide which model to get. There are baseline models at msrp or higher power oc models for 650 or more money.

 

[Broken Processor]

Compared to Nvidia's fake MSRP and purposely limited supply, these are are a much better value.
I think AMD could've undercut harder with the XT at $549, but the market needs something in stock at MSRP.

We shouldn't be comparing to Nvidia we all know how bad there release has been. Instead compare it to 7700 and 7700xt and see how it stacks up. At the end of they day I'm not saying they are bad cards just the the price increase on the new generation is no good.

 

[Krit]

Where you seeing that? On the 7800 XT page it says 136% for the 9070 XT.

54.4 fps + 46% = 79.5 fps

 

[Broken Processor]

All I'm seeing here is "blah blah blah"...

You've added zero to the thread. Instead of saying anything intelligent you're just attacking.
The quote sum's you up nicely.

 

[medi01]

Is it really inefficient?

Spoiler

View attachment 388123

Idle also half of an 5070 ti, this version was probably pushed to hard. (if you want performance per watt)

Efficiency figures are all over the place.

 

[hardwarefan]

Nice card, at msrp at least.
It still trails Nvidia in RT. But they closed the gap quite a bit.
Given that, the fact most 9070 (xt) use the regular 8 pin connectors instead of the still suspect 12v-2x6, the missing rops on some Nvidia cards, the loss of support for 32 bit physx games, and the utter stagnation of 50 series Nvidia cards in raster and RT, AMD should regain some market share.
Hopefully the increased competition from AMD would push Nvidia out of stagnation and the 60 series becomes a real improvement over the 40/50 series.

 

[Hecate91]

We shouldn't be comparing to Nvidia we all know how bad there release has been. Instead compare it to 7700 and 7700xt and see how it stacks up. At the end of they day I'm not saying they are bad cards just the the price increase on the new generation is no good.

Why not? It's competition to Nvidia so I'm not just going to ignore the 50 series no matter how bad it is. The 9070XT is 30-40% faster than the 7700XT depending on the game, and much better RT and upscaling performance. A 30% gain is much better than what the 50 series has provided comparing it to previous gen cards.

 

[Spunjjatron]

Yes, it is very inefficient. What game is that tested, cause getting 60 vsynced at 80w means it's a game like CS2 or something.

So it's "very inefficient" even though it's measurably surprisingly efficient in a like-for-like comparison of equal work done? You sure? You really want to pin your credibility to the arbitrary "no watts wasted over a hypothetical competitor at any price" standard you've apparently adopted for this thread?

Come on.

Let's not be silly. It looks pretty clear that AMD delayed the launch for a few reasons:
1) Getting software ready
2) Building stock
3) Tweaking final clock speeds

After the abortion of the 5080 and 5070 Ti launches, AMD likely realised that rather than their face-saver chip only duking it out with Nvidia's last gen high-end as they'd expected (and telegraphed with those old slides), they had a chance to goose these things and make them competitive in a price/performance sense with the current high-end, especially with Nvidia having dropped a tremendous wet fart with stock.

Take a look at the power draw for a stock 9070. Where are those extra 134 watts for an overclocked 9070 XT going? They're going on putting it within single digits of a 5070 Ti. For the people who want that, they're going to get it. That's pretty great!

For those of us who have always given a shit about finding a magic triad of price/power/performance - no matter which card it means buying - the undervolting potential might well be very interesting indeed. We're too early in the review cycle to have good data, but the indications just based on what a defective die can do in the stock 9070 are good. This has already been mentioned up-thread. It's not quite rekindling the magic of the venerable GTX 970 in that regard, but it's pretty damned good for 2025.

Me, well - I was going to buy a 5080 on launch. After three generations of woe, I'd decided to throw in the towel on value-for-money for a change and get something really fast, try out DLSS and MFG, really spoil myself.
Then I couldn't get one on launch, despite sitting and refreshing the page at 2pm as fast as my browser would let me.
Then the prices shot up.
Then used prices of the 4080 shot up, too.
But ah, the 5070 Ti, similar performance for far less money! That'll do!
Nope, same fucking deal.

So, I'll be online at 2pm, taking a punt on a card that's going to sacrifice a little of the performance and power efficiency I wanted, but - and this is crucial - I literally don't care anymore if it means I can actually buy one. Because some of us compare more than one metric at once.

 

[AusWolf]

This is the issue you've compared it to extremely gouged prices. That doesn't make these cards good value it makes them the least bad option.

What gouged prices? The MSRP on the 5080 is $1000. You don't have to gouge that to make it sound bad.

 

[TSiAhmat]

What gouged prices? The MSRP on the 5080 is $1000. You don't have to gouge that to make it sound bad.

and 1,169 € in Europe (if you live there, also VAT included)