Radeon RX 9070 Gaming Benchmark at CES Analysis

[@jayargeronimo]

Such a shame the outlets that are supposed to report on this accurately don't even know what Daniel mentioned about the settings not applying if you change it and don't restart the game. Such a shame IGN.

[@A3r0XxLol]

it's IGN, what did you expect?

[@fracti9443]

IGN employ literal chimps

[@jordanmackay6746]

That might not be true though. I just tested and it DOES apply the graphics settings in warzone without a restart. I doubt the benchmark is any different but i could be wrong.

[@kaseyboles30]

The IGN reporter was probably concerned about getting caught and a bit rushed. So I can see how they could forget to reset the game or even not know, reporter != reviewer.

[@westii7287]

there is a popup that says you need to restart the game for the settings changes to apply lol

[@zodwraith5745]

I can't say I'm even slightly surprised that an IGN writer doesn't know how to benchmark properly.

[@cire420siuol]

I'd be inclined to agree, but I think the bigger issue is that Activision/Treyarch/IW/whatever don't know how to implement settings properly... You shouldn't need to restart the game to change settings, and if for some reason you do need to, the game should tell you that...

[@angeltzepesh1]

​@@cire420siuolalmost every developer has this problem, settings need a restart

[@be0wulfmarshallz]

@@angeltzepesh1 Almost every developer? Nah, depends on their engine. Most games dont need this or tell you if you need a restart. IGN's problem here is that they should have looked at their low/med/high benchmarks to see if extreme changed or not...

[@MrBeetsGaming]

They weren't actually "benchmarking the GPU" they literally just ran the in game benchmark because that happened to be the game they had running at their booth for CES. We have no idea if they even changed any settings or if they did whether they restarted the game. There are plenty of reasons to criticize IGN but this isn't one of them, at least not without more information.

[@TheBoredT-Squered]

@@cire420siuol are you a developer? How do you know that it should not be necessary?

[@daMillenialTrucker]

I truly want to see AMD succeed..

[@UncleButterworth]

They are - with CPU's.. Only AMD can make their GPU's a compelling value. We'll see.

[@janisir4529]

Most people want AMD to "succeed" to be able to buy Nvidia GPUs cheaper.

[@mahshshsrklingfa7031]

They already given up.. cant compete with Nvidia at high end and they already have good mid range gpus..cpu is making good money with intel sudoku

[@GSP-76]

@@UncleButterworth They will beat Nvidia too...they were neck and neck with Nvidia with the RX6000 series so they're definitely capable of competing. With the RX7000 series, they pivoted to MCM because they realized what we're seeing in general now - silicon is rapidly coming close to EOL. Nvidia's new cards don't offer more performance with the 50 series. They simply added more power to get higher performance. Their entire presentation was focused on software... Nvidia is being affected by die fab processes way more than AMD right now because they've relied heavily on a monolithic die. The future for GPUs will require an MCM approach.

[@polla2256]

​@@GSP-76the future of all silicone processors is to do more with less so software is the only way forward until we have some other form hardware.

[@Antares-dw9iv]

5:05 It's absolutely mindboggling how these companies with marketing budgets in the high millions, end up with terrible slides like this that would have failed you a middle-school presentation. I genuinely think every single person I've ever met would have been able to create something with a clearer message and better information density.

[@twisted_ps2743]

the purpose of this was to stay relevant in the wake of nvidia's 50 series announcement.. they got the message across without spending millions .. they don't really seem to have a line up to compete with the 5090.. you'll get your high end slide when they fire back..

[@Juice_VI]

It’s on purpose

[@goldenhate6649]

@@twisted_ps2743 The 5090 is only a 3% uplift per watt compared to the 4090. Most of Nvidias uplift is in fake frames...but you need at least 60-90 fps natively for fake frames to even feel okay. I suspect AMD just wasn't ready. The 50 series has basically no increase in technology other than fooling the customer into thinking it is natively better than the 40 series until you start to realize you mouse buttons are pressing when you press them. Hence why its cheaper than the 40 series in regards to msrp.

[@Scytian1]

@@goldenhate6649 AMD was ready, it looks like cards are ready to launch right now, we have "leaks" saying that RDNA4 may launch in around 2 weeks, and these are not random leaks from random people, it's XFX posting about cards coming on 24th January and some shops saying that pre-orders start on 23rd. There must be other reasons to not announce these on CES, I don't know what these reasons are but if I had to say it on spot I would say that GPU division is just way too small for them to "waste" time on big stage, they'd rather give some stage time to Dell and make billions.

[@WayStedYou]

@@twisted_ps2743 the 5070 that only matches a 4070ti without DLSS 4.0? which is below what the 9070 is going to do

[@adeptuspotatocus6451]

CoD games are known to favor GPUs with a lot of cache. RDNA3 Infinity Cache is off-die on the MCDs while RDNA4 is moving all of that back into a monolithic design. This will give latency benefits and may give exaggerated benefits in cache-sensitive games.

[@broose5240]

you mean preforms better. Not using fake frames?

[@deathtoinfidelsdeusvult2184]

​@@broose5240yes, but its like nvidia benchmarking with raytracing to favor them.

[@adeptuspotatocus6451]

@@broose5240 Yes. It should just run some games faster. Not all, but some.

[@DonPatro92]

Its a stretch but entirely possible idk.

[@Ouss-2077]

Good point

[@baronvonslambert]

Someone else mentioned it, but why is it wrong to use CoD as a benchmark because it's AMD optimized when Cyberpunk is an incredibly common benchmark for reviewers and card comparisons while not just being Nvidia optimized it's basically an Nvidia coproject? It would have been better to just say one data point isn't enough to draw a conclusion and maybe to be a bit conservative regarding performance speculation due to the CoD being optimized for the architecture, rather than saying the benchmark isn't the best because AMD typically performs better on said benchmark.

[@Lowlocity]

Because it's not biased unless you turn on raytracing.........for obvious reason.

[@Yurikane1211]

Cyberpunk is really bad example. It runs really good on rastering on AMD cards

[@ProtectusCZ]

@@Yurikane1211 it doesn't even support the latest FSR, meanwhile it supports immediately every new nvidia bs

[@tguit-fiddler5692]

@@ProtectusCZ Doesn't mean its bad at rasterized performance. I get 120FPS at 1440p with my 7900xt without FSR.

[@roki977]

CP2077runs great on AMD as well, without RT ofc..

[@Cinnabuns2009]

If you recall also, the 7900XTX was faster than the 4090 in another COD game when launched so it may just be the high outlier. (you do mention this ~@3:25) I think you're exactly correct. They changed the setting to ultra and ran the bench and didn't restart because they didn't want to draw attention. 9070 is supposed to have 3-4x the RT perf though so that may be factoring in with the settings?

[@KM_47]

The game doesn't have any form of RT

[@WayStedYou]

black ops 6 doesnt have ray tracing

[@junyaiwase]

@@WayStedYou doesnt need it either its already an unoptimized pos anywho

[@inkredebilchina9699]

I'm gonna wait on third party benchmarks when the cards are out before pulling the trigger. I'm still rocking 5600G as my main GPU. although technically it's an APU but yeah. that's my everyday PC and it's rock solid. if this turns out to be true, and IF (big if) the price isn't insanity like on team green I just might go with the non XT. but, that remains to be seen and verified.

[@EhNothing]

"I gotta go teach first period math" lmao. I love it.

[@Venlorz]

yup I didn't know he teaches in school. No wonder I understand him clearly (as a person who's not that smart and a bit slow)

[@SuperShado101]

gl to his students, math and first period don't mix

[@joewills7582]

Awesome spot here Daniel. Your theory seems very likely, I'm kind of shocked IGN didn't consider this given how baffling their bench result was.

[@Dras7]

Lmao they couldn’t even bother to get a comparison before posting the article, of course they have no idea what they are doing.

[@helldiego]

B- bullshit, tipical IGN

[@Spr1ggan87]

IGN are console people like most review outlets, expecting them to know anything beyond plug and play is not realistic.

[@andersjjensen]

IGN will run with certifiable click-bait without as much as blinking. Revenue is revenue, so screw it if they stuff up the launch for AMD by setting expectations that leads to disappointment. They ain't the ones losing the money, right?

[@PURENT]

His theory might be wrong. Ancient Gameplays benchmark of this game got 113 FPS and 98 FPS for 7900 XTX/XT. Which 9070 being 99 FPS is about where AMD said this card would sit. Apparently this game loses like 40% of it's performance when played on Steam vs Battlenet which would explain Daniel's really poor benchmark.

[@nisem0no]

The memory usage is also a giveaway. Why would the 9070 use 4gb less VRAM at the same settings? That's a huge disparity. It'd be great if it was even partially true, but it's unlikely. *kicks pebble*

[@broose5240]

3dv cash?

[@zachb1706]

@@broose5240that makes zero sense

[@RuanLFC]

Its likely the vram scale settings in Bo6 it has a scale you can choose i think its 60% as 60% of 16 is about 9 and 60% of 20 is 12 i think 60% is the default settings this can be increased but i think thats about stock test later on my rig

[@RuanLFC]

Doesn't explain frames rate likely not extreme settings as i doubt its more powerful than an xtx but it would explain vram usage at least :)

[@Dionyzos]

Daniel should've also tested a 16GB card

[@FlamingJune-ol3ry]

your news are 100% faster than 5090

[@kleddo3126]

Frame Gen you forgot in the end

[@AthamAldecua]

​@kleddo3126 no, cus Daniel doesn't make news up as he goes. He delivers only truth, not fake crap 😂

[@zpr3d4t0r6]

@kleddo3126amd has frame gen too, they likely used it on that benchmark

[@FcoEnriquePerez]

Is IGN, I wouldn't doubt those bunch of incompetents did f*** it up somehow.

[@TheRealHypervizor]

@@AthamAldecua 5090 is plenty fast w/o framegen tho..

[@Zappen88]

Seems like the differences in allocated ram doesn't matter. COD allocates roughly 70% of available VRAM in all three cases

[@jayruka1977]

I was going to mention this.. the available ram is 4gb less compared to the XT and 8 gb less than the XTX. What does a GRE get? Like 75 fps?

[@Comycboy06]

Is COD still alive..?

[@nightmarepotato5000]

​@@Comycboy06 Unfortunately so it seems

[@mrpicky1868]

other reviewers told that 9070 does not have finished driver and can't run full power yet

[@hochhaul]

Yeah, it was partly why they said the "leaks" about its performance were fake. None of their board partners have finished drivers.

[@jonredcorn862]

They are saying nearly a 20% uplift. This card may fucking cook yall, if they price it right I am snatching one up for sure. Sticking my middle finger up to Nvidia and its FAKE AI FRAME GEN FUTURE!!!

[@w1se_w0lf]

So nothing new, AMD proves once again their drivers are trash, so buy Nvidia if you don’t want to waste money

[@greateminemfan]

@@w1se_w0lf Well the GPU hasnt launched yet. I joined AMD GPUs after the legendary 1080ti and went with 6950 xt and 7900xr. Not faced any driver issues yet and i play many different games... Dont know maybe some games have it but the drivers are okay. Nvidia has issues aswell but maybe less often or less big issues. Still i dont regret to get more fps/$

[@9tailedyokai]

@@w1se_w0lf Legit Ngreedia has been having driver issues too, but people are zealots when it comes to them

[@kerotomas1]

Still, the 9070/XT probably going to have better generational uplift compared to the Rtx 50 series as the Radeon parts moved from 5nm to 4nm while the 50 series remained on the same 4nm as 40 series.

[@erickalvarez6486]

That and isn't rdna4 finally a chiplet? Gpu stacked over another gpu? I just hope it performs more similar to a 7900xt rather than 7900GRE and I'm buying it

[@blondegirl7240]

My 4070S shows its 5nm not 4nm

[@Gotoxico]

If a nanometer less in process node made a huge uplift, what matters is the chip design, the upscalling technologies, the framegen AI...

[@hb-hr1nh]

Dream on

[@dralord1307]

@@erickalvarez6486 No i dont think its stacked. Need AMD to release detailed info though. All current info says it might be a chiplet but its not stacked or connected processing chiples

[@giantmonkey101]

How the heck are you only getting 67 fps at 4k extreme with your 7900 XT?? My XFX 7900 XTX even at stock will hit 98 fps using the same settings with upscaling completely OFF. Im sorry Daniel Owen but an XTX is not 46% faster than a stock 7900 XT. Make sure you have Smart Access Memory on and check your computer for some kind of performance causing issue please lol

[@gaav87]

You prolly had variable shaders on my friend...

[@Speejays2]

Yeah Daniel's results are a lot lower than they should be. Ancient Gameplays did a 4K Extreme benchmark (no upscaling) and their 7900 XT got 98 FPS with no frame generation. kzbin.info/www/bejne/n5_GfpmrZ69onMk The 4080 Super got 96 FPS, but as stated in other reviews, CoD typically has better results with AMD GPUs.

[@joaogabriels.f.5143]

probably he was screen recording it with the gpu. Naive rookie mistake; (screen recording take gpu power away to encode the video)

[@LoneWolf-tk9em]

Hey bro, is there any video on yt about optimizing pc for maximum fps?

[@adriankoch964]

He uses a separate capture card, not the actual GPU to capture the content.​@@joaogabriels.f.5143

[@tomthomas3499]

It's on preview driver, most likely the artifacting IGN mention is different than that you get, imho better wait for official announcement and just treat this report as another leak.

[@MrBeetsGaming]

As much as I love piling on IGN to be fair this person just ran the benchmark at their booth at CES because they had Black Ops 6 running, it's not like they were actually "benchmarking" the GPU. We also have no idea if they changed any settings or whether the game was restarted if they did, so it's kind of silly to immediately criticize them.

[@MrScuffyMissesALot]

Sure, the 'benchmark' may have been rushed, but they should have spoken about these considerations in the article, which it seems was also rushed as so many are nowadays.

[@andersjjensen]

To not be fair to IGN, but rather speak to my experience with them over many years, this is exactly what I expect from them. AMD has just been on a "temper expectation spree" and the first thing IGN does is post BS that anyone with a brain would question right off the bat. This is borderline click-bait as usual.

[@DethronedCrown]

@@MrScuffyMissesALot Its very clear in the article what the situation was....AMD were demoing the 9070 in a booth with blops 6 so the guy was able run the benchmark tool.

[@Someone-lr8os]

This probably happened because a change in the shader quality setting is only applied after a restart which has a big impact on fps. You should test both the 7900xt and xtx at extreme settings but with shader quality on the lowest setting and then you should see performance that is similar to the one on this ign bench.

[@decliningship4208]

BO6 has different performance depending on what platform you are playing it on

[@jackmills7758]

they might have variable rate shading enabled in bo6's settings, which boosts performance significantly

[@UKGUN1T]

I always have that 💩 off make it blurry ,

[@jackmills7758]

@@UKGUN1TI turn it off to see how powerful my gpu is, plus enable hi-res textures

[@Thomas_Angelo]

The leak is not the max settings. It's using only 11 gbs of vram. Unless 9070 has got some magical neural rendering equivalent too, the article is wrong.

[@zyzU400Kg]

i was looking for the same. CoD uses alot more that vram on 4k max settings

[@Dazzxp]

Could also be Z-buffer too in how the textures are compressed to the VRAM. That's why VRAM usage between generations and manufactures are not the same typically every generation it's improved.

[@Thomas_Angelo]

@Dazzxp I assure you it's just not max graphics. They won't offer 4090 level of performance in a 9070.

[@blackface-b1v]

​@@Thomas_Angelowhy so pessimistic

[@Dazzxp]

@@Thomas_Angelo Well a 7900GRE shouldn't be smoking my 4070Ti Super but it is in this game! Not only that but it's very close to a RTX4080!!!

[@X30_]

Kinda confused should i buy 7900 gre or wait for the launch of 9070 / 9070 xt, since i am getting 7800 gre for 500$ here.

[@tguit-fiddler5692]

Personally at this point I'd wait and see all the 3rd party benchmarks

[@gabber_]

you really should wait, no matter where you are in the world. at best 2 months and all the gpus will be available worldwide then you can choose the best one you can afford. old gen prices will not get higher.

[@X30_]

@gabber_ I see, won't like that was exactly my worries 😅, thanks for the clarification. Time to wait I guess

[@ultramasterultra5724]

none just get the 5070

[@vinnime8768]

I would say wait and see if the price is less than $400 then you can consider 9070

[@hiddedevries8853]

Maybe its possible to based on the Vram usage estimate the texture settings?

[@darkkingastos4369]

I'd be curious to see someone run a 7800xt with different settings until they reach these framerates

[@PREDATEURLT]

Texture settings make tiny difference in performance as long as vram is not overflown.

[@hiddedevries8853]

@@PREDATEURLT Huh? Look up any comparison between extreme and low setting in COD BO6 4k

[@PREDATEURLT]

@@hiddedevries8853 Textures or overall setting? Because textures alone means nothing to performance on any 16GB card.

[@Psi-Storm]

@@PREDATEURLT I doubt someone would use custom settings on a show floor pc. They probably set it to 4k medium so the framerate looks better. So the memory usage is directly linked to the quality preset.

[@synonys]

28% generational improvement doesn’t sound out of the realm of possibility.

[@WayStedYou]

it does when you are cutting out 33% of the CUs and its basically the same node.

[@andersjjensen]

@@WayStedYou Exactly. This was always designed to be the midrange die. AMD just axed the elaborate top end chiplet approach because it flopped with RNDA3.

[@zenmonk5403]

Except it isn't a full new generation, as the naming suggests. There will be 9800 and 9900(XT(X))

[@Varil92]

In this era, it does.

[@mynickisalreadytaken]

@@zenmonk5403 No, 9070XT is max for now for this Gen. No High End Cards this time. This is normal for Radeon. They do it every couple gens.

[@xYarbx]

AMD was talking about improved texture ai compression so the lower reported vram usage might also signal that it's out and working on RDNA 4. We have no indication but that's just the first idea that popped into my head when I saw those numbers.

[@FLMKane]

@@xYarbx yo wtf, it'd literally be cheaper for them to just give us more Vram than to develop that junk!

[@flatfish-d9o]

Once again here to request a look at the history of generational GPU price:performance improvement over the years 🙏

[@РаЫо]

Cry

[@pablosanmartinvarela1773]

He just did this?

[@AbzRaider]

bro he already made a video on this

[@JBrinx18]

No, make it yourself

[@alarik95]

Lmao bro

[@pixelschubser2556]

6:27 This calculation here, along with your explanation, actually teached me for the first time how you can measure performance increases in therms of FPS in games. Like taking the higher number, devide it by the lower one and the first two numbers after the comma - or the dot in american systems - is the percentage that got increased. I'm by far not a programmer or really good at math, but this one i got really fast. Good to know ngl.

[@TXHuyIcefrzzy]

The Vram usage of your 7900xt in BO6's Buildin Benchmarks is around 13GB while 7900xtx is around 16gb, and both has 20gb Vram available. So there is something hidden that cause the different in Vram usage of these two gpu. If the new gpu - x9070 somehow use other "technique" to deal with Vram usage and can be use that much less Vram to works with the same graphic setting, it can be the way. Update: there is reply of my comment remind me that 7900xtx has 24gb Vram, so it just make me more "believe" in the theory more Vram use more, less Vram use less :D

[@Ouss-2077]

The XT has 20gb of vram while the XTX has 24

[@TXHuyIcefrzzy]

@@Ouss-2077 So it just make my "theory" more sense. More Vram, use more, less Vram, use less :))

[@Psi-Storm]

@@TXHuyIcefrzzy Yes, allocation scales with usable memory. Probably because it wants to avoid memory overflows in longer gaming sessions, so it leaves a percentage free at level load. Best approach would be to test the 7800XT, which also has 16GB, and then try different game quality profiles until you end up at the same 9 GB allocation. Too bad he doesn't have just tested 4k medium setting to see where the XT would end up there.

[@nightmarepotato5000]

Yep, someone else in this comment section mentioned something like COD always allocating some 70% of available VRAM for itself. So might be something like that

[@TheTallRaver]

Fantastic insight, great analysis and testing! I much appreciate your methodological approach! That’s why I keep coming back to your channel!💪👍😉 Keep it up!

[@Nite-Lite-Gamers]

The driver version is 24.30.0, is this a beta driver for the 9070 series????

[@haukikannel]

To run 9070… yeah there has to be beta for it!

[@grandpagibz]

The vram usage can be explained by the diff versions of vram used by each card.

[@ChibataMan]

i checked an old benchmark i had with the 4080 at the start of season 1 and i got 9.85gb of vram at 4K dlss quality (1440p) and 6.79 gb of ram with a 32 gb system so a difference of 0.88gb of vram with the same amount of vram at a lower resolution and 1.5 gb difference in ram, so i really think these aren't max settings.

[@WutTheHekk]

and Nvidia's gpu generally allocated less vram than AMD does

[@ChibataMan]

Did some testing and got 9.9 with ultra and extreme settings, this seems to be the nearest i can get with the presets, as high gives me around 6, variable rate shading didnt affect vram usage and i found no correlation between any settings and ram usage

[@R1b3z]

I think it's not about the restart but more if they had enabled VRS because that makes a HUGE difference in bo6 and it doesn't show on this bench result, and it should, but still if it has VRS enabled this is an amazing result at 4k

[@blowmage]

Came here to say this.

[@guy197]

Here's something to ask Nvidia. Notice that all the gaming benchmarks they've shared on their benchmark charts are RT games. No raster. Not one. And looking at relative spec tables (TFLOP numbers), it seems RT gains this gen, of perhaps 30-50% over last gen, will be much higher than raster gains. The good news is that it could mean we'll be taking a much smaller FPS hit when switching RT on, but it looks like raster gains this gen will be very modest. Possibly 10-20%. It does feel like Nvidia is hiding raster benchmarks. As an example, in shader TFLOPS the 5080 will gain 14% over the 4080, but in RT TFLOPS it will gain 51%. Now, I know TFLOP numbers don't translate 1:1 to FPS gain, but if a 51% RT TFLOPS gain means the 5080 will be 25-30% faster in Far Cry 6 RT and A Plague Tale Requiem in DLSS+RT, as Nvidia's charts show, it stands to reason that a 14% shader TFLOP gain will give a much smaller FPS gain when these games run with RT off.

[@silvy7394]

They do it because AMD cant ray trace worth a crap. Despite it being the next generation method of lighting in games.

[@TreFeKsGaming]

Look at their Cuda Cores for their Gpus... the 5070 even hes 14% less Cuda cores then for example the 4070 super which was only 7% more expensive

[@guy197]

@@silvy7394 Regardless. I now expect we'll possibly get very modest raster gains this Nvidia gen.

[@th3summoner]

@@guy197 Video about AMD benchmarks: "NvIDiA's hIDiNG beHInD UpScaLInG" Holy shit Nvidia is rent free in ur head.

[@guy197]

@@th3summoner Just trying to get Daniel's attention, so he might look into this, by posting in his latest video. Posting a comment in an older video won't do that.

[@johnsmithee6541]

What i want to know most is, at what settings a 7900 GRE lands at around 100 AVG and with these settings, where does the 7900 XT stand?

[@jordanmackay6746]

Hey Daniel you might want to double check your claim that graphics settings dont apply until after a restart. I just tested it in warzone and it defiantly applies without restarting. Not sure about the benchmark mode though.

[@Osprey850]

Daniel moved to a house across the street from the school so that he could put out these videos in the morning and still get to school on time. What dedication to us!

[@rENEGADE666JEDI]

AMD is doing very well in rasterization, where the problem is in RT, If this card will have similar results in Cyberpunk as 4080, it will be a small success for AMD and an indication that the next generation of consoles will be awesome.

[@RandallRoman]

So, looking at TechPowerUp's chart with custom scene, the 7900 XT was 75 fps, and your card was 67 fps on the in-game benchmark, suggesting the TPU custom scene is an easier workload. So you should multiply any values there by about 0.9 to get the built-in benchmark value. That would take the 4090 to about 90fps. I'm having a hard time believing that the 9070 (non-XT!) is going to give 10% better raster performance than a 4090, even in an AMD friendly game like COD. I think you've hit it spot on - they don't have the right graphics settings.

[@aymanelbahtary1540]

If the benchmark is real this would be crazy

[@laszlozsurka8991]

I doubt. I think it's exactly what Daniel was saying... IGN didn't restart the game after applying the new settings and the game ran at whatever default settings was set at.

[@andersjjensen]

Temper expectations. Given what we know about the chip size, memory bus width, memory speed and CU count this is flat out impossible. And if it outran the XTX it would have been called the 9080/XT for sure....

[@Eveningbreeze78]

@danielowentech .. Did you notice you also had half the RAM allocated on the Rx9070Xt time around 7:45 so.. you might be right, could be OS optimisations as well .. or (driverlevel) afmf or FSR4 enabled?

[@slickrounder6045]

Amd lowkey sitting on a Sleeper with Rdna 4. It was wise of Amd to not reveal their "Hand" and their Cards right away. Nvidia likely felt pressured by Amd since they know Amd will have a bigger gen on gen leap per Core (going from 5nm to 4nm, meanwhile Nvidia is on a similar node), not to mention the massive (already confirmed with the Ps5 Pro) RT uplift that will get close to evening out the disparity, which was the only real hardware deficit Amd had. As such Nvidia went more aggressive than anyone expected with the $999 price 5080 (rumors had it at $1199-1300ish), $749 5070ti (rumors had it at $799 to $899), and $549 5070 (rumors had it at $599 to $699). There is a very good chance that the 5070ti will be strongly pressured by the 9070xt which likely can be nearly on par with Raster performance while costing much less (I'd expect $599 if it will reach the heights many now believe is possible). Meanwhile the 9070 (non-xt) could very well end up being outright better than the 5070, on top of having the clear and significant 16gb vram versus 12gb vram advantage. Oh and all but surely it will be $50-100 cheaper to boot (i'd expect $449 to $499 price point on it).

[@Daniel.AJ.R]

Keep dreaming

[@MyBestBuddiesForever]

Here's hoping and i hope that will make the market more competitive and it's gonna be a win win situation for us Consumers.

[@deathtoinfidelsdeusvult2184]

Theres a reason why amd didnt even show a pricepoint in their cards.

[@MyBestBuddiesForever]

Here's hoping and i hope that'll happens so the market will be more competitive and it's gonna be a win win situation for us.

[@15ag34]

I don't believe AMD to be a threat to NVidia unfortunately (Market share seems to confirm this). I believe Nvidia kept the price the same because they faced significant backlash after the initial prices of the 40 series. And they want people to buy these cards, reason being they are going to release the higher Vram models later this year (likely 24gb 5080, 20gb 5070 ti S)

[@LamyCND]

You are an absolute Chad for this kind of reporting. Thanks!

[@michaelthomas5433]

I remember AMD on the night before they launched the 7900 XTX and 7900 XT and saying how their cards would destroy Nvidia and do it with much less power. We know how that turned out.

[@Ted_Kenzoku]

they do

[@lennartj.8072]

@@Ted_Kenzoku with more power though

[@Ted_Kenzoku]

@@lennartj.8072 who cares about that you won't see the difference in your bill

[@lennartj.8072]

@@Ted_Kenzoku True

[@andersjjensen]

AMD was under the impression that they were fighting a driver bug and that the XTX would outperform the 4090 because they had seen it do that in 2 games. It would turn out that it wasn't a driver bug but a hard wall in the "high performance fan-out" that connects the MCDs to the GCD.

[@Hriskataaa]

great review / compare! Can you do a quick follow-up on what would it take to achieve similar results with the 7900XT? Like, drop the resolution or switch to med/high settings?

[@shk0014]

Its pretty clear from the infographic that they are trying to specify the price bracket, on which planet would they place 7900xtx lower than 4080 being AMD. Anyway that content is just all over the place.

[@lennartj.8072]

Does that make sense though? The 9070 XT will at best deliver performance between the 7900 XT and XTX with better RT, so if the infographic was about pricing having it on the same tier as the XT, ergo the same price, would make it a DoA card. Doubly so since the 7900 XT had such an "interesting" price history.

[@WayStedYou]

@@lennartj.8072 i guess 5070 is DOA as well since its a 4070ti without dlss 4.0

[@WayStedYou]

no its not the 9070 is much cheaper to make than a 7900XTX

[@lennartj.8072]

@@WayStedYou No it's not, because it's a Nvidia card and DLSS 4 will be a huge selling point. Also a lot of less knowledgable people will believe that it's actually as fast as a 4090 because Jensen said so. Besides, it's still gonna be $250 less than the 4070 Ti at launch.

[@lennartj.8072]

@@WayStedYou Okay, but if it's just a slightly better 7900 XT with the same price as a 7900 XT nobody will buy it. Manufacturing costs don't matter. The graphs OP was talking about are 100% not about pricing, unless the benchmark in the video is actually accurate (extremely doubtful) and the 9070 is somehow as fast as a 4090, which I suppose would justify it being the same price as a 7900 XT.

[@dsrocks6905]

Believe it or not, BO6 does use ray tracing at extreme settings (check the game config). That performance uplift may be from the dedicated and reworked RT cores that are coming with RDNA 4 actually reducing the load on the main pipeline for rendering, which may allow the ridiculous raw performance if a card like say the 7900xtx to actually really shine if it is just a 7900xtx with better RT performance.

[@SinisterPuppy]

Press X for doubt. I'm remember when AMD (ATi) completely trashed NVidia with their ATI Radeon 9700 in the early 2000s; I can dream we're going to get a repeat.

[@OrlandoCaba]

Yes, dx 9/Pixel shader 2.0 9700pro :D So I can play UT2003

[@arch1107]

and for that gpu we didnt get another 9700, we got 9070

[@miguelmaganromero5035]

All I want to say is, the kids who have you as a teacher are very lucky. You are awesome at explanations!

[@JorgeAlarconL]

I want to add that COD Black Ops 6 even with a "mere" RTX 3080 defaults at extreme as "recommended" setting, so the game was probably running on Extreme. Also this game had the bug of settings not applying, but since Season 1 Reloaded patch that's no more the case, and this game is 100% online so it has to be an updated build running this card. I don't get why AMD is getting these numbers at all. They are sandbagging again like they do with CPUs before release, or this is not a 9070 but a 9080 and they didn't tell to anyone and it's the best secret to be revealed.

[@irishRocker1]

In any situation, it's usually a good idea to test before and after to see if changes actually make a difference or if changes have actually been made. They probably got a few minutes on the pc and decided to do the benchmark and should have checked it as is first, I guess with fsr enabled and then without. They'd have seen the results were the same and change hadn't been taken effect

[@adink6486]

Daniel: I have to teach first period math guys. He's already teaching 0 period math here.

[@bumpahhi]

They for sure use some different graphic driver likely not available for public at the moment. Not that it should cause drastic difference but some for sure.

[@Cine95]

4:43 bro use an ad blocker

[@Indjinnity]

He can't. Those sites could sue him.

[@denxx56]

​@@Indjinnity I'm no lawyer but there's no damn way that you could get sued for using an ad blocker

[@YaBoiMemo313]

Bruh imagine giving companies the legal right to shove ads up yo bottom

[@swilleh_]

@@Indjinnity mate, FBI recommends using an AD blocker

[@androidfixer7052]

Or just use brave browser

[@veyron984]

Noticed the system ram is also lower on IGN's test, and they have more Ram available on that system, that could also indicate lower actual settings right? something definately seems to not be quite right

[@tomtomkowski7653]

Just test it again but this time 4K medium and low settings. This way we will learn even more I think because it looks like this was not Extreme settings really.

[@Yoshimatsu414]

RDNA excels in the COD game engine. The 7900XTX is faster than an RTX 4090 in Call of Duty, I'm pretty sure. Not surprised it continues to be the case here with RDNA4. That's really good though 👍🏿 i did see that bench mark you pulled up had the 4090 faster but I've definitely seen the 7900XTX faster in the past, maybe Nvidia did a tweak or game engine was updated in BO6.

[@edhorseman2237]

I think AMD just learned from past mistakes and aren't exaggerating performance this round. People are actually surprised that a company could learn from past failures? Understating performance is actually good, it provides better presss and makes the value proposition more appealing.

[@broose5240]

amd gets Nvidia hype/hate bots. This way amd says nothing , all bots get exposed

[@mattblyther5426]

dont people realize that cod uses a desired vram target percentage scale from i believe 50 is the low end to 95%. this is why vram usage is all of the place. for example just looking at the images and figuring the percentages the 9070 was about 70% vram target and for daniels XTX it was pretty much 75% i do agree with daniel and there was a miss applied settings change though and the 9070's fps is being miss represented ( hopefully on accident lol).

[@muiwols6709]

Can't wait for a 9060 with 8gb 128 bit and 8 lines of pcie.

[@christophermullins7163]

Nvidia will likely wait for 3gb modules so it's 12gb on 128bit. Which is fine especially on gddr7. Problem is that AMD does not have the same option. Guessing they'll do 128bit 16gb for the lowest sku just like 7600xt. If they do the 8gb thing.. itll have to be $199 and be at 7600 performance. Gonna look bad beside the 5060 12gb for $299.

[@MarianPV20]

on techpowerup show rx 9060xt with 12gb

[@ultrademigod]

We don't know what the specs for the 9060 will be, but it could have 12gb.

[@MarianPV20]

@@muiwols6709 they said "9060 series" so 9060 non xt 8gb and 9060 xt 12 gb.

[@andersjjensen]

@@christophermullins7163 I think they'll do a 9060XT with clamshell 16GB and a cut down variant of the die with 8GB. If they have any sense they'll call it the 9050XT and price it at $200. That would be a worthy successor to the RX 6600 in both name and price, in a price bracket where an extra $100 is out of the question.

[@elu5ive]

absolutely! their settings change didn't apply. RAM and VRAM loads look more like 4k at low settings

[@WayStedYou]

or 1440p extreme>4k extreme

[@TheTechNerd-s7g]

I have a feeling this gpus are going to be extremely underwhelming . My prediction 9070XT=7900 GRE, 9070=7800XT

[@CrazO9111]

I think 7900 XT = 9700XT

[@andersjjensen]

@@CrazO9111 I think in HUBs eventual 40 game average at 1080p, 1440p and 4k it's going to slot in somewhere between the XT and XTX in raster. I have an XTX and it "doesn't care" about memory bandwidth. I can underclock my memory by 15-20% before I see performance drops exceeding 3%, depending on the game. The GRE, on the other hand, benefits massively from memory overclocking. So with a bit of napkin math, I'm slightly optimistic that it will edge ahead of the XT, and I think we'll get a bit of RDAN2's "scales well with lowered resolution" propensity back, as the cache has been moved back inside the chip.

[@Mugenz0]

so the main point here is "they showed in BRANDING slide that the positioning of 9070 at 7900 gre" so new card should perform at same level as are, or what? and look at allocation on vram for 16/20/24g, they increase pretty linear, so don't see any indications of "not properly set settings" I think that something really can be wrong, but not in the parts that are discussed in video, such huge uplift based on licked specs does not seem real, but who knows, who knows, 23 of January is close

[@artemca]

So AMD is better than Nvidia. Nvidia has only 5070, but amd already have 9070, it is 4 generations ahead!

[@djuice1701]

You can also check VRAM usage, on your bench, it was using 12.93GBs, and on the IGN one is using 8.97Gbs. So definitely they were probably benching on Medium Texture Settings, without a game restart.

[@ohanski]

3:20 I think you mean nvidia underperforms at cod XDD

[@broose5240]

most games are optimized for Nvidia

[@nameless.0190]

@@broose5240 no hes right here, AMD always performs way better in CoD.

[@Mizt_Plays]

best PC news channel ngl

[@zooteddd]

Maybe this is why they were waiting for Nvidia to announce their gpus. Perhaps they were expecting a larger generational uplift for blackwell over Ada. Hopefully they will not give it dumb prices again.

[@hochhaul]

That would be awesome but I don't think so. I think the bench was actually running at 1440p extreme.

[@ianukytd2852]

i seen some things about gaming on cod on steam vs battlenet i seen a video saying the if you use the battlenet launcher youll get more frames?

[@victoremorrose9027]

as always ,excellent !!!

[@InfernoTrees]

Might be able to lock down what settings they used by checking what vram usage is like on each setting. Then maybe we can get a better comaprison

[@РаЫо]

How did Daniel become such a gigachad?

[@joewills7582]

By getting a 4090. I got offered head 4 times on the subway back from Best Buy when I got mine, for example.

[@nour_z]

did you have SAM enabled? this could be disabled if u had an nvidia gpu in before

[@mindrover777]

And the point being

[@nour_z]

@@mindrover777 SAM can give a slightly favourable bump in performance that may have been missed in this instance, meaning it would not be a perfect apples to apples comparison

[@WayStedYou]

@@nour_z most likely thing is the game was running at 1440p extreme and they set it to 4k extreme and the res didnt adjust

[@andersjjensen]

It's enabled in BIOS by default. Putting an Nvidia GPU in, and swapping back, does not change that.

[@nour_z]

@ it’s not on by default in most cases

[@DragonBane299]

Here's hoping the 9070XT is under 400... but knowing AMD, they're gonna price it at 499 cos the 5070 is 549.....

[@xyr3s]

500 would be very reasonable for the power of a 7900xt. I think they might be aiming for 600 or even 650. At 400 it would be a miracle lol.

[@haukikannel]

$499 makes Sense. Little bit less, with more vram. Very typical approach!

[@Ted_Kenzoku]

the 7700xt was 450 it will probably be around that. the 5700xt was 399$ in 2019. they aren't going below the price of cards from 6 years ago. inflation happened.

[@HunterTracks]

I would not expect it to cost $400 or less, because it's their current top offering. It basically leaves no space for their cheaper cards, and probably sets the margins on the entire line lower than they'd be comfortable with. My best bet is $449-469.

[@Will-zi6fn]

Might as well ask for it for free lol

[@PCrealitys]

AMD got Stomped Out by Nvidia and its their fault, its what happens when you chant " We Strive To Be #2 We Strive To Be #2 " Guess what, They succeeded !!!!

[@anakinskywalker7079]

Try to adjust your graphic setting to obtain the same quantity of VRAM. It can be a good indication.

[@weirdodude1173]

These numbers had me wondering, too. If AMD really can make 9070 XT compete with a 4080 and cost

[@tda0626]

It won't according to their own slides that Gamers Nexus showed. It would be a contender with a 4070TI. AMD launch of their flagship card is a dud. Even a 7900xtx outperforms this card. it isn't a good sign

[@Will-zi6fn]

@@tda0626that’s such a dumb take. AMD isn’t chasing the high end so having less performance than a 7900 xtx is irrelevant, price to performance is what’s important

[@tda0626]

@Will-zi6fn Dumb, no. I was responding to the OP comment or did you not read it?

[@WayStedYou]

@@tda0626 a 500 dollar card beating a 900 dollar card isnt a good sign? I guess the 5070 being a 4070ti with DLSS 4.0 isnt a good sign either then

[@tda0626]

@@WayStedYou Way to deflect. We aren't talking about Nvidia chief.

[@gringoboy701]

IGN and tech competency have never been linked.

[@S1B1C1]

I wonder if they ran the game with VRR enabled and or if they put shader on High and did not reboot the game to recompile shaders, if thats then the game will be using Low or mid shaders their also a few setting in the config of call of duty that even if you change the preset does not effect some stuff in the config, ( LOD/relighting update stages ) and a few others.

[@AlexZaglushaka]

0:51 nice ad on screen 😊 rssa must loose

[@MarianPV20]

even tough this is a rx 9070 non xt test not the xt version

[@Franics_G]

omg legit just bought the 7900xt getting rly bad buyers remorse

[@joewills7582]

How much?

[@Void-uj7jd]

you dont need the latest current thing, that's drone mindset. be happy.

[@MidnightMuzic]

You'll be fine. We got a couple of leaked benchmarks which don't mean much until the cards release and the actual reviews come out. If you feel that bad about it sell the thing for close to what you bought it for or just keep it as a back up. It'll probably last longer then the 9070 anyway due to the amount of vram among other things

[@thunderkrac5998]

My XFX 7900xt OC/UV and a 7700x cpu, without any upscaling on extreme graphical settings, hits my tv limit of 120fps 4k 50 inch

[@joshbracken5450]

That's an awesome and strong gpu and for a fair price. We've seen new releases never having the pricing they list. It's always higher.

[@TheRealHypervizor]

Does that setting have any ray tracing in it? RDNA4 is supposed to be way better at RT than RDNA3. Maybe that's why the frame rate is a good bit higher than expected? Just wondering, I stopped playing COD games after BLOPS III.

[@Struct.3]

Il wait for the Cyberpunk PT test

[@Acconda]

The game where even the 5090 can only do 28fps without all the fake crap added ? 👍🏼

[@Struct.3]

@@Acconda well im not playing at 4k no dlss, and i dont think its crap, its actually pretty good at higher base framerate

[@SALTINBANK]

It don't apply settings > each time i played the game > nvdia DLAA wasn't active anymore (my prior choice) then game switched to NIS technique upscaling every time > awfull rendering ... Some other LOD settings were reseted also > Draw distance They updated the title two weeks ago to fix this issue but still on my game settings for upscaling techniques are reseted every launch

[@christophermullins7163]

AMD may be sandbagging but at the same time that gpu is NOT getting 30% faster than xtx. Hopefully its 4080 and theyre sadbagging a bit to get nvidia to releasw at higher pricing. If so, it didnt work. 9070 xt will have to $499 at XT perf and $600 at 4080 performance. At MOST.

[@Moshenokoji]

It's going to have to do better then $600 at 4080 performance. 5070 at $550 is already the ceiling and they absolutely know they have to undercut Nvidia, they lost about 10% market share in a year. Reference model will have to come in at $450 to make sense.

[@joshbracken5450]

There is no way any of the nvivda gpus will release at the price they are saying. They have always been higher. In my own country the 4090 is nearly 2.5k anyway.

[@jony_tough]

Shout out to the fellow educator! I'm an art teacher myself, so cool find out that u're teaching math.

[@emrahcoban4488]

English teacher here 👋🏻

[@jony_tough]

@@emrahcoban4488 Thank you for teaching people how to speak.

[@emrahcoban4488]

@@jony_tough and likewise, thank you for teaching a core aspect of what makes humanity and civilization!

[@sebastian_manuells]

Greetings & new year wishes from a fellow gamer & teacher to another. 🙌🎉

[@EABeatsInTheMix]

Free RDNA4

[@kelpmaxxing]

It's not as improbable as it seems, the rdna 3 cards have a lot of clock headroom, so presuming that amd reduced power consumption considerably and/or increased the relative power limit, an extra 20-30% from that alone is well within possibility , coupled with faster memory, and you start approaching those numbers , without any significant architectural changes.

[@PohnnyRico]

i think you are wrong. Thats the finger pistol from one piece. :D Why dont use call of duty? because you and basically everybody uses cyberpunk that was made extra for nvidia gpus (and nobody talks about it) and you and others use it for Benchmark comperisons...its always the same, doubt amd but eat everything nvidia throws. still like your videos because you are explaining things good

[@Antares-dw9iv]

Cyberpunk actually tends be fairly representative when you're benchmarking raster performance, if anything AMD's 7000 series performs disproportionately well. Obviously not once you turn on ray tracing and it doesn't excuse the poor FSR implementation but the game definitely doesn't favor NVIDIA when it comes to rasterization whereas AMD's performance in Call of Duty is a massive outlier.

[@TheSoxor123]

Doesn't matter. Remember Avatar Frontier of Pandora is an AMD sponsored game but ran significantly better on Nvidia GPUs.

[@DerBaerserker]

i would thinking the same that the power of the 9070 XT is similar to the 7900XT but other people in tech forum told that this is not what the slide is showing and performance considerings from this would be highly speculative. maybe they are right, and this slide is only for showing the new naming

[@broose5240]

this is 5d chess by amd....

[@atnfn]

They've only pretended they don't go for the high end! In reality the 9070XT is actually as fast as AMD expects the nVidia RTX 9070 to be in roughly 8 years from now!!!.

[@broose5240]

@@atnfn that was the non xt

[@rallyscoot]

Your driver version is different too.. They used 24.30 drivers... And yours seems to be older / or latest which amd had released.

[@RobbieKhan]

CoD benching is a cop out to showcase on because CoD has always been AMD optimised/biased.

[@broose5240]

amd did not do anything here blame tech tubers

[@RobbieKhan]

@@broose5240 Well yes they are still part of the blame as they only had CoD on display for people to play there. If there were 5 other games then sure it would be a balanced take.

[@Momonga32]

Then you could say the same with nvidia

[@Cobinja]

And how exactly does that affect this AMD vs. AMD benchmark?

[@HunterTracks]

​@@RobbieKhanYou do understand that they didn't actually intend people to benchmark the cards, right?

[@SergeiSugaroverdoseShuykov]

10:26 that fact that they aren't chancing for high end does not means that their new GPU should be worse or exactly the same as previous generation, yet it might be just game-specific reaction to positive changes in architecture

[@hinatashoyo2999]

If the card isn't 399$ Then its not worth it

[@LeeMc007]

Great video, I do enjoy your explanations, I think you're right about the settings unfortunately, AMD being more competitive would benefit us all though. Here's some Math for ya Daniel 😉 101k views + 4k likes = ouch 😕

[@davidepannone6021]

Lol sure. And then pigs fly 😂😂😂😂

[@shk0014]

your mom does

[@davidepannone6021]

​@@shk0014 yours taught her how

[@kjellbeats]

​@davidepannone6021 weakest comeback :(

[@davidepannone6021]

@@kjellbeats lemme show you the fucks I care:

[@nameless.0190]

@@shk0014 bro is rattled.

[@Akksuli]

You can see in the photo that the card inside that benchmark PC is in fact RX 9070 XT reference card and not RX 9070.

[@olivur_1459]

It's clearly stated 9070 non xt. I hope you have drunk enough water for all the PJSalt you're experiencing right now. It's better to stay hydrated in these extreme salt situations ^

[@Akksuli]

@olivur_1459 Did you even check the photo? On the right side there is the AMD 9070 XT reference card inside the case. 9070 reference card is slimmer. It is 9070 XT. That IGN journalist thought for some reason that its 9070, or got somehow confused when writing the article.

[@E.Cristian7]

Maybe as fast as the RX 7900 XT? For a lower price, would be nice. Well I always stick to the budget stuff as the RX 9060 and I love Radeon Image Sharpening can't live without it anymore. I'm Gonna Wait