Thursday, January 30th 2025

AMD Radeon 9070 XT Rumored to Outpace RTX 5070 Ti by Almost 15%

It would be fair to say that the GeForce RTX 5080 has been quite disappointing, being roughly 16% faster in gaming than the RTX 4080 Super. Unsurprisingly, this gives AMD a lot of opportunity to offer excellent price-to-performance with its upcoming RDNA 4 GPUs, considering that the RTX 5070 and RTX 5070 Ti aren't really expected to pull off any miracles. According to a recent tidbit shared by the renowned leaker Moore's Law is Dead, the Radeon RX 9070 XT is expected to be around 3% faster than the RTX 4080, if AMD's internal performance goals are anything to go by. MLID also notes that RDNA 4's performance is improving by roughly around 1% each month, which makes it quite likely that the RDNA 4 cards will exceed the targets.

If it does turn out that way, the Radeon RX 9070 XT, according to MLID, should be roughly around 15% faster than its competitor from the Green Camp, the RTX 5070 Ti, and roughly match the RTX 4080 Super in gaming performance. The Radeon RX 9070, on the other hand, is expected to be around 12% faster than the RTX 5070. Of course, these performance improvements are limited to rasterization performance, and when ray tracing is brought to the scene, the performance improvements are expected to be substantially more modest, as per tradition. Citing our data for Cyberpunk 4K with RT, MLID stated that his sources indicate that the RX 9070 XT falls somewhere between the RTX 4070 Ti Super and RTX 3090 Ti, whereas the RX 9070 should likely trade blows with the RTX 4070 Super. Considering AMD's track record with ray tracing, this sure does sound quite enticing.

Of course, it will all boil down to pricing once the RDNA 4 cards hit the scene. If AMD does manage to undercut its competitors from NVIDIA by a reasonable margin, there is no doubt that RDNA 4 will be the better choice for most people. However, with NVIDIA's undeniable lead in ray tracing, paired with DLSS 4, will presumably make things more complicated than ever before. It is unclear what AMD has up its sleeve with FSR 4. Recent rumors do point at pretty good compatibility, but as with all rumors, be sure to accept any pre-release whispers with a grain of salt.
Source: MLID via YouTube
Add your own comment

305 Comments on AMD Radeon 9070 XT Rumored to Outpace RTX 5070 Ti by Almost 15%

#226
3valatzy
AMD718For those who said AMD wasn't competing at the high end because they didn't have the technology to compete at the high end / would need years to match a 4090 and realistically were just incapable of producing anything that could come close to a 5090 (because Nvidia is years ahead of everyone else or something), it should be clear now that AMD could have produced a 5090 competitor *this* generation, but chose not to for pure business reasons. The high-end will be up for grabs with UDNA vs. 60 series. I just hope AMD doesn't scrap high-end with UDNA. That would be a damn shame.
I'm looking at how bad the RTX 5000 series of Nvidia cards is, and think that AMD is in a big trouble. Now was the time to compete with the nothing-impressive RTX 5080 and under it.
I'm afraid that UDNA won't compete with RTX 6000 series, and that time AMD's top part will compete against RTX 6070.

It is a strategic mistake by AMD to give time to Nvidia.
ThomasKYou'll have to wait until march for that.
That's unknown. AMD could very well postpone it even further by other three months, or cancel the launches altogether.
Posted on Reply
#227
ThomasK
3valatzyThat's unknown. AMD could very well postpone it even further by other three months, or cancel the launches altogether.
Official info is that it launches in march, so that's what am choosing to believe in.
Posted on Reply
#228
3valatzy
ThomasKOfficial info is that it launches in march, so that's what am choosing to believe in.
Source?*

*Plans are subject to change.

There was no reason to delay the official launch from January to March, to begin with.
Posted on Reply
#229
ThomasK
3valatzySource?*

*Plans are subject to change.

There was no reason to delay the official launch from January to March, to begin with.
Amd themselves said in its presentation at CES, and if you quickly google it or just check a wikipedia page, you'll find out it says Q1, which ends in march.



So should I believe that, which is the info coming from the manufacturer of the product, or some random guy on the comment section conjecturing things?
Posted on Reply
#230
AusWolf
3valatzyI'm looking at how bad the RTX 5000 series of Nvidia cards is, and think that AMD is in a big trouble. Now was the time to compete with the nothing-impressive RTX 5080 and under it.
I'm afraid that UDNA won't compete with RTX 6000 series, and that time AMD's top part will compete against RTX 6070.

It is a strategic mistake by AMD to give time to Nvidia.
That's exactly why AMD isn't in trouble. The 5000 series isn't just bad, but also available in extremely low quantities, which buys AMD some time. I just hope they won't overprice the 9070 XT to follow Nvidia.
Posted on Reply
#231
Dahita
AusWolfThat's exactly why AMD isn't in trouble. The 5000 series isn't just bad, but also available in extremely low quantities, which buys AMD some time. I just hope they won't overprice the 9070 XT to follow Nvidia.
I agree 100%. The way I see it, I'd rather wait and for AMD to release a good card with an FSR that can compete with DLSS, rather than them releasing earlier to fill the gap at the cost of credibility. Let's be patient and hopeful!
Posted on Reply
#232
Unregistered
ThomasKAmd themselves said in its presentation at CES, and if you quickly google it or just check a wikipedia page, you'll find out it says Q1, which ends in march.



So should I believe that, which is the info coming from the manufacturer of the product, or some random guy on the comment section conjecturing things?
you can believe on AMD's own ads saying it was out for purchase weeks ago. Or that retail already had the gpus weeks ago. Or you can just use your brain and realise Q1 2025 started a month ago. Or not i guess and say it was march all along, because you are the random guy making the conjecturing thing even when faced with the facts

but hey it's the internet never let facts stop you
#233
mkppo
AusWolfAda $1,000 vs Blackwell $1,000. That's pretty much what gen on gen is.
This, 100%.

Baffling that anyone is even remotely trying to justify the 5080. I remember people trying to justify why the 4060Ti is a good card when it was barely any faster than 3060Ti. This is no different, just without a node advantage.

I just don't see why people do it. Like if something is shite, call it out. Don't buy it, don't praise it, defend it, nothing. It's that simple but nooo, go ahead and enable nvidia's anemic progress and give them your $. Smh
AusWolfThere's no need to keep bringing the 7900 XT up. Everybody knows it was a crappy offer, just like the 7700 XT.

Still, where's the 40% you mentioned?
I'll tell you what the lines of defence are and honestly this needs to stop now, it's universally established that the 5080 is terrible progress gen on gen. No amount of defense will sway the ones who think that, and the ones who are defending are essentially wasting their time but whatever.

1) 4080 was $1200, this is 12% faster for $200 less which gives their 38% or whatever perf/$ uplift. Ignore the fact that nvidia lowered the price because it was that terrible and released the 4080S for $1000. No no, must compare it to the 4080 which was so badly priced that it was rotting in shelves. Hell, at least 5 different shops tried to sell me a 4080 when I went to buy USB drives or something random. It was a strange time. Anyway, here's an example that should hopefully put a stop to this nonsense.

Nvidia released the GTX 280 for $650 because, you know, greed and they thought they had it and left AMD in the dust. Fast forward one month and AMD launched the 4870 and 4850 which came close to it's performance with half the die size and cost. What did nvidia do? They were forced to lower the price from $650 to $500 in a month.

Then when the GTX 480 released it was pretty terrible and got bashed pretty hard in reviews when it launched at $500. Now do you think anyone defended it because hey, the GTX280 launched at $650 so this must be a good deal at $500? No, there were no grounds because the GTX280 was lowered to $500 which effectively became it's MSRP. When there's a MSRP correction, that becomes the baseline. Nvidia corrected it to $1000, so that's what this will be compared against. Bringing up $1200 4080 price to make the 5080 look good is gasping at straws.|

edit: 5080S is reserved for comparisons to the 4080S? Nope, 5080S is to be compared to 5080 to see if there's a meaningful upgrade. If comparing gen on gen, both 5080S and 5080 will be compared to 4080S. That's the best nvidia had in that class in the prior gen. At the end of the day you can make all the comparisons you want, but the fact remains that this is a terrible gen on gen uplift. No two ways about it really
Posted on Reply
#234
GodisanAtheist
ContraYou are spreading misinformation. None of your statements are true.
Before posting false information, please read and compare the RDNA2 and RDNA3 specifications and read the architecture comparison. Thank you.
- Can you just tell me A) what I am wrong about and B) why I am wrong using the information in those links?

Cause i'm not going to read random links posted by a stranger on the internet on a Saturday afternoon
Posted on Reply
#235
ThomasK
Bomby569you can believe on AMD's own ads saying it was out for purchase weeks ago. Or that retail already had the gpus weeks ago. Or you can just use your brain and realise Q1 2025 started a month ago. Or not i guess and say it was march all along, because you are the random guy making the conjecturing thing even when faced with the facts

but hey it's the internet never let facts stop you
The ads were pulled, much like when nvidia had to pull its ads when they tried to sell you a 4080 with less vram, but ended up having to call it a 4070, because of the backlash.

You're grasping at straws here, just give up with your baseless arguments and stop talking about price/performance, when the MSRP doesn't mean anything anymore.

Bye.
Posted on Reply
#236
mkppo
GodisanAtheist- Can you just tell me A) what I am wrong about and B) why I am wrong using the information in those links?

Cause i'm not going to read random links posted by a stranger on the internet on a Saturday afternoon
Yeah i'm not sure why he decided to not counter your points but link to the specs and chips and cheese article. That one is a good read though (the C&C one). Anyway these were your points:
GodisanAtheist- Beefed up their CUs to have higher IPC. RDNA3's CU's essentially performed exactly like RDNA2 CUs, you only got more performance when you increased the number of CUs.
- Increased the boost clock (one of RDNA3's biggest missteps was missing clock targets)
- Tweaked or rebalanced infinity cache/L2 Cache to make-up for the loss of bandwidth.

Ultimately, the thing is Nvidia is able to get 4080/5080 performance out of a 380mm^2 256-bit piece of silicon on the N4 process, so (again, theoretically) there isn't a reason AMD cannot do the same (assuming N48 is in fact ~380mm^2).
So RDNA3 didn't necessarily perform the same as RDNA2 per CU, but the uplift was less than what they expected because it just didn't clock as high as they wanted. Plus you have to factor in RT improvements which were pretty substantial and doesn't come for free in terms of die space. Looking at the arch, there are expected hurdles to get dual-issue working first time around which I'm almost certain will be improved in RDNA4. Convert more PS to wave 64 to get more dual issue streams working without scheduling and not relying as much on the compiler? There are ways and i'm sure they know better. A couple of low hanging fruit in there too but being in it's third iteration, it's architecturally relatively polished other than dual issue being more miss than hit. It doesn't take up much die area though.

Second point is absolutely valid, clocks will go a long way.

Third point is also valid, there will be L2 changes in RDNA4 or so I believe. The bus was wide enough at 384 bits and I would argue they need to focus more on L2/registers and not Infinity cache, especially if they run higher speed GDDR6 modules. Plus, going monolithic will save them memory latency, remember that they sacrificed ultimate performance just to split GCD and MCD.

I think people forget that, had AMD made 7900XTX monolithic, it would've certainly been faster than 4080S (edit: before things go up in flames, I mean in raster). Had they made it monolithic and hit their desired clockspeed targets, it wouldn't be that far off a 4090. There's at least 80mm^2 of die area taken up just to get the fanout links and receive/send circuitry that wouldn't be needed in a monolithic die either. If I were to take a very rough guess, a monolithic 7900XTX on the same 5nm would be around 440mm2, so yeah not much larger than a 4080S considering it's 384bit . But they decided to do an experiment, one that didn't necessarily benefit the 79xx launch but rather the learnings they can take moving forward. Splitting the MCD/GCD is far, far more complex than people think.
Posted on Reply
#237
Contra
mkppoSecond point is absolutely valid, clocks will go a long way.

Third point is also valid, there will be L2 changes in RDNA4 or so I believe. The bus was wide enough at 384 bits and I would argue they need to focus more on L2/registers and not Infinity cache, especially if they run higher speed GDDR6 modules. Plus, going monolithic will save them memory latency, remember that they sacrificed ultimate performance just to split GCD and MCD.
@GodisanAtheist
The average frequency of RDNA3 is 10-30% higher than RDNA2 depending on the boards being compared. Most AIB cards easily support 3+GHz.

As for the throughput or cache, you could tell from the C&C article that the throughput of the RDNA3 IC is over 3.6TB/s, which is almost 1.8 times higher than RDNA2, providing about the same latency. And all this contributes to the IPC improvement, given that it is a double-digit %.
focus more on L2/registers
This is ridiculous and stupid. rdna3 register files are 196k, while RDNA2 has 128kB, and rtx 30/40/50 has 64kB. The one that really needs improvement here is NV) That's why we see 0% IPC SM improvements for RTX40->50
As for L2, NV went to such a waste of crystal area for one (may be not only) function - SER, which reloads all register files via L2. But they were wrong about the scale of this feature. One game in two years is nothing.

Your claims about no changes are not true. The recommendations for configuring the last level cache and memory are also incompetent.
Posted on Reply
#238
medi01
GGforeveraccording to MLID
Oh boy.
The Norwegian Drone Pilotgreat features
The Norwegian Drone Pilotawesome features
The Norwegian Drone Pilotawesome feature set
So much awesomeness in just 3 sentences. Awesome!
The Norwegian Drone Pilotis one of the reasons why
Aha.

The 380mm^2 has been debunked. 9070XT is 350mm^2 at most.
3valatzyI'm afraid that UDNA won't compete with RTX 6000 series, and that time AMD's top part will compete against RTX 6070.

It is a strategic mistake by AMD to give time to Nvidia.
I prefer more R&D in middle range, than chasing elusive "high end" balloons, where stuff won't sell well because <yet to be released feature by NV that is absolutely totally crucial to have>, thank you very much.
Bomby569Blackwell sold out
Without knowing number of cards sold, this is a very meaningless metric.

All we've heard at this point was "very low volume".
Posted on Reply
#239
k0vasz
Super XPAMD abandon the dGPU market? Would be a death blow to the company. Not happening of course.
AMD is going through the transition from RDNA to UDNA, that's all. They placed most of its Radeon Group resources onto UDNA. And choose to stuck to the mid range market with its final RDNA4 run. That was a good decision from the company.
dGPUs are most probably just a "side product" for AMD from the lucrative professional (AI) segment and the console segment. they'd be fine without it, but if they can sell a few extra chips for gamers with minimal effort, than why wouldn't they do it?
Posted on Reply
#240
Krit
medi01The 380mm^2 has been debunked. 9070XT is 350mm^2 at most.
Where you get this info ?
Posted on Reply
#241
Sound_Card
3valatzyI'm afraid that UDNA won't compete with RTX 6000 series, and that time AMD's top part will compete against RTX 6070.
RTX6K is more like Q1 2027. UDNA on the other hand, might actually be out by summer 2026.
Posted on Reply
#242
medi01
KritWhere you get this info ?
380mm^2 was based on chip sizes on photos. The original measurements were off.
Posted on Reply
#243
RaphaelOne
Sorry for going off-topic a bit. @ For many years GPU manufacturers were loyal to the unwritten rule of decency, which set the upper limit of TDP of graphics cards at 300 W. Unfortunately, AMD and nVidia later abandoned these rules.
Today, the media glorifies graphics cards at the price of a car, and TDP exceeds 500 W. Seriously, seeing a home desktop computer that uses about 1 kW of power under load, is hot and noisy during 3D loading, I think this is the wrong way to go.
I love efficient and quiet machines, so recently I lowered the clock and voltage of the RX6800/16GB, which brought huge benefits.

GPU Clock - Reduced by 5% to 2000MHz
GPU Voltage - Reduced by 9% to 935mV
Card Power Limit - Reduced by 8% (lowest available)

As a result, during normal 3D load GPU power consumption ranges from 70W to 140W depending on the scenario, usually around ~110W in 1440p@165Hz mode.
The whole computer is quiet and efficient. It's common sense, a real pleasure and elegance in everyday work with the machine.

P.S. As for the silicon lottery, my RX6800 is slightly below average, so many RX6800s on the market are able to achieve even better results. I RECOMMEND
Posted on Reply
#244
OneSorcerer
out of curiosity, what are your guesses about pricing throughout the whole gen?

9070 ~ GRE for 499? and then amdiscounted to 449 later on...
9060XT 12GB ~ 7800XT for 399?
9060 12GB ~ 7700XT for 299?
Maybe a 9050 8GB ~ 7600 for 229?
Posted on Reply
#245
Sound_Card
´´´´´´
OneSorcererout of curiosity, what are your guesses about pricing throughout the whole gen?

9070 ~ GRE for 449?
9060XT 12GB ~ 7800XT for 379?
9060 12GB ~ 7700XT for 299?
Maybe a 9050 8GB ~ 7600 for 239?
9070xt ~ 7900xt for $600 or less
9070 ~ 7900gre for $500
9060xt ~ 7800xt for $400
Posted on Reply
#246
OneSorcerer
Sound_Card´´´´´´


9070xt ~ 7900xt for $600 or less
9070 ~ 7900gre for $500
9060xt ~ 7800xt for $400
My 6650XT can't sustain good frames at 1440p =P 9070 may be my next gpu as long as it goes a little cheaper than 500.
4070 S is on my radar as well...
RaphaelOneSorry for going off-topic a bit. @ For many years GPU manufacturers were loyal to the unwritten rule of decency, which set the upper limit of TDP of graphics cards at 300 W. Unfortunately, AMD and nVidia later abandoned these rules.
Today, the media glorifies graphics cards at the price of a car, and TDP exceeds 500 W. Seriously, seeing a home desktop computer that uses about 1 kW of power under load, is hot and noisy during 3D loading, I think this is the wrong way to go.
I love efficient and quiet machines, so recently I lowered the clock and voltage of the RX6800/16GB, which brought huge benefits.

GPU Clock - Reduced by 5% to 2000MHz
GPU Voltage - Reduced by 9% to 935mV
Card Power Limit - Reduced by 8% (lowest available)

As a result, during normal 3D load GPU power consumption ranges from 70W to 140W depending on the scenario, usually around ~110W in 1440p@165Hz mode.
The whole computer is quiet and efficient. It's common sense, a real pleasure and elegance in everyday work with the machine.

P.S. As for the silicon lottery, my RX6800 is slightly below average, so many RX6800s on the market are able to achieve even better results. I RECOMMEND
Did the same, undervolted my 6650XT by 40mV and temps are down 10c. I really appreciate the 40 series for its' low tdp, 4060 115w is beautiful. Let's hope for a decent 5060/Ti with GDDR7 fixing the bandwidth
Posted on Reply
#247
Krit
Sound_Card9070xt ~ 7900xt for $600 or less
9070 ~ 7900gre for $500
That's an awful improvement! Right now RX 7900 XT costs 635€ and RX 7900 GRE 546€. Similar improvement to blackwell.
Posted on Reply
#248
Rabit
Also 9070xt 32GB edition ~679
Posted on Reply
#249
Krit
RabitAlso 9070xt 32GB edition ~679
32GB for RX 9070 XT is the same as 16GB for RTX 4060 Ti (a waste of money)
Posted on Reply
#250
Rabit
if price right 9070xt 32 Gb will be mine personal pick for local LLM
Also we have similar situation in past RX 480 4Gb and 8GB not much difference when released by after some time we seen it when game star using above 4GB.
Posted on Reply
Add your own comment
Nov 17th, 2026 15:08 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts