Watch the AMD RX 7600 XT review here! ruclips.net/video/1aopp8XolXE/видео.html Find the written review of the 4070 Ti Super here! gamersnexus.net/gpus/nvidia-geforce-rtx-4070-ti-super-gpu-review-benchmarks-power-efficiency-gaming JUST finished rendering and uploading. Latency comes back for the 4080 Super -- so that'll have our new power efficiency testing, latency, and Cyberpunk (this one re-introduced Cyberpunk). We'll try and get Alan Wake 2 added soon as well! What other games do you want to see? We're replacing some of the tests with new ones. Grab a LIMITED Disappointment PC T-Shirt in Heather (available in all sizes)! store.gamersnexus.net/products/disappointment-pc-2023-heather Or in 100% cotton (running very low on some sizes): store.gamersnexus.net/products/disappointment-pc-2023-cotton
Price/Performance/Efficiency ratio is what i have said is the most ideal for the longest time, thank you for finally presenting the whole picture, thank you!
i love this already 3 seconds in! at the end of the day, per watt is everything! noise, heat, longevity, cost, performance! and please add Apple Silicon here and there!
"joules per frame" means that you can also measure "calories per frame" which means that you can also measure "hamburgers per frame" which means that you can also measure "pizzas per second" as framerate.
"Bigger number better, means better number gets more bigger better, so, that's... better, because its bigger" - It's exactly this kind of hard hitting journalism that keeps me coming back. Don't change GN.
A 4070 that slaps the already somewhat confusing Ti and Super monikers together, while also being based on an entirely different chip from another range of cards. Just leave it to Nvidia to make their line-up even more opaque than it already was. 🙄
@@GamersNexus You are sacrificing braincells however, in an attempt to comprehend nvidias endless skulduggery lmao For that you will have my eternal thanks, that way I dont have to.
Thanks for all your hard work on these reviews. Good stuff. Really good to see my 2080 in a lot of these charts too. Really helps me compare and see how much of a performance boost I might be able to expect if I upgrade to this card.
I got a 2060, looking at the 7900xtx. Will likely go with all AMD build this year unless NVIDIA upcoming 5080 is reasonably priced. Definitely going with Zen 5 X3D. AMD is impressive with their CPUs they deserve my dollars. Never owned an AMD part in 25 years of PC gaming but can't deny their comeback. Hopefully their nexgen GPUs can again challenge the 80 series for cheaper.
I think it would be interesting to see their power efficiency at the same FPS cap. I also think it would be interesting to see how power efficient they were at certain different % power limits, but I think that's a lot of work for a lot of cards.
Thanks Steve... and the Patrick Clone Army and of course the flealess CEO. The grind must be ridiculous right now... waiting impatiently on the 8600g release btw.
It almost seems like Nvidia is intentionally limiting this thing's performance in the driver, but went a little too far in many games and now reviewers are catching it. Let's see if they "fix" it with the next driver. It should be much closer to the base 4080 like the 3080 vs 3080 Ti
So gross. With how many gpus nvidia are selling to meta every goddam day they know they can do basically anything to consumers and then just tap out whenever. Nvidia vs everyone for vending them
I'm waiting on the 4080 Super but was really hoping for a lot more from this card. Geezh, at this rate I might just go for the Sapphire Nitro 7900XTX unless the 4080 Super does something surprising.
Why on Earth would you do that? 4080 Super will be marginally more expensive while giving massive advantages in other areas (ray tracing, encoding, AI, other prof applications), and it will give an edge in rasterized too, all while being much more power efficient. Unless AMD seriously slashes the 7900XTX prices from what they are today, there's absolutely no reason to buy this card over the 4080 Super.
@@SanctusBacchus relax dude amd is fine in all aspect just same not so strong raytracing but either those gpus suck 4090 is minimum for that even that performance just tanks 2much for reflaction gad damn u people are delusenal u take fucking reflection over gameplay no wonder this world has turned into shithole
@@SanctusBacchus It remains to be seen. I don't think I'd buy the XTX without some kind of discount, maybe $900. I am interested in RT for Cyberpunk, but I also play Star Citizen where AMD seems to have an advantage. Also, as you say, power efficiency is slightly important to me. If the 4080 Super can match rasterization with the 7900XTX, then, yeah I'm probably sold on it at that point.
efficiency is a waste of time, you need to be playing for 10 hours every single G-D day to make a significant difference at the end of the year... Most gamers will "save" maaaybe $5 from switching to a "more efficient" card and would have probably spent a lot more money upfront...
@todorkolev7565 it obv depends on conditions and stuff but it is a very important metric, not just as chart and info but also as classification and quality of product. But as example for the CPU video Steve talked about, the i9-14900K *could* cost you 223.1 dollars per year whereas the 7950X3D *could* cost you 73.7 dollars per year comparatively... so yes, it's worth it.
@@todorkolev7565 and while it might not really be fair to compare 3x differences with cpu's to 0.3x differences gpu's i just wanted to show you how important efficiency is (especially when looking at a whole system).
@@todorkolev7565 Not true for me. I pay 0.59€/kwh. So 50 Watt difference makes 170€ over 4 years with a usage of 4h per day. That means that it makes much more sense to consider a RTX4060 (115W) over the 50€ less expensive RX 7600 (165W) Also efficiency correlates with heat and noise, especially important for small builds.
I'm really looking forward to efficiency testing for lower end cards. In most of Europe most cards are way more expensive than in the US and the power's more expensive too, so seeing how cards stack up in power efficiency should make buying a cheaper card easier.
Got a shirt from you guys just now. Been watching Steve with Gamers Nexus for years now I figured the time has come to show some direct support. It's been a joy to watch.
I'm upgrading from a 2070 Super to a 4070 TI Super. Bought myself a 4070 TI last week, canceled the order just in time, and just today bought a 4070 TI Super + Corsair PSU GPU cable for 150 euro less then the previous 4070 TI card. So I'm pretty happy and I'm sure I'll be happy with the upgrade.
I can't state how much I appreciate how quickly you guys drop these videos. People that think being a content creator on RUclips is easy does not understand that you have one of the hardest boss in the world: the viewers of RUclips and the internet, who can be extremely critical and demanding. Your dedication to your work is amazing
Well tech tubers don't have a work schedule, their life is the schedule. Unless we talking about Linus and look how that ended, instead Steve can be proud having one of the highest quality channels on youtube.
Once again, another great video with helpful data, thanks Steve and team. I might be alone in this, but I'd quite like to see a scatter graph showing each cards FPS/W and an average FPS (I'm not smart enough to decide what games would be best suited to make that average). It might not be particularly helpful as a piece of buying data without prices there, but it would be interesting to see the trade-offs/generational improvements in the technology.
Well the video's just dropped, so I can't really comment on it other than that I appreciate your guys' work. Bought a mod mat for a friend (the signed 15 one that apparently says beve on it) and he's been enjoying it! Genuinely, keep up the good work.
Just purchased some more merch fro the GN Store to help support the amazing and in depth testing you all do over there at GN🥰😇😇💪👍. Cheers from Australia 🇦🇺😄
AMD was targeting a +50% uplift in efficiency over RDNA2 before launching these new cards. Unfortunately it does seem true that RDNA3 did not meet these targets. It may be more disproportionately more efficient with lower clocks, but it's already not as competitive as it should be in raw speed (imo) This architecture, the mix of chiplets etc just didn't work out as good as everyone I bet including AMD had hoped
@@Aquaquakekeep in mind that there is a node difference between AMD and Nvidia, which means that for peak clock speeds AMD will be spending much more power to get the top end. Nvidia's dies are also monolithic so they are going to be slightly more efficient from that. Final piece of the puzzle is Nvidia has RT hardware acceleration which makes computing RT tasks more energy efficient than AMD's methods (right now, they will probably add hardware acceleration too next generation or the gen after). Mix that together and you see that ~20-30% efficiency lead. Honestly I am surprised it's not more but sdram isn't scaling well with node shrinks so maybe that's it
I currently have a 4070 Ti and while I would like to have the additional VRAM looks like I'm not missing out on much performance. This card performs about where I expected with the biggest gains coming at 4k.
I think maybe an average efficiency over multiple games could be interesting since the people play multiple games, you would have do eliminate the total fps (so dividing by total fps or something), this way you can maybe identify trends between the gpus a bit better since they are sometimes a bit all over the place between games. Looking forward to see power efficiency in future GPU reviews :)
Would need to figure out how to do the division while retaining data integrity (maybe geomean), but I like that idea! Will play around with math and work on it!
I'm prefacing this with, I don't know jack about squat. With that said, as long as GN published average FPS, and average wattage over a game benchmark, you could do that math yourself and customize it to what games you play (again, as long as those games are included in the GN test suite. Pick the games you play, add the average FPS over those games and divide by the amount of games. Do the same thing for wattage, and then divide the FPS over wattage and you'll have your efficiency score with that card over your own test suite of games. If you don't want the FPS included in that, just don't use FPS in that calculation.
I would also be curious to see power metrics with frame rates locked at say 60 FPS. Would be interesting to see how efficient cards are. Not all of us are that interested in playing at 144hz particularly if one lives in a place where energy is expensive. Or in my case off-grid on solar. Power gains from undervolting would also be interesting but that is probably a big ask.
1080 TI still hanging in there ! After seeing the FFIV benchmarks, I decided to check my old scores. Last time I ran it was with my 980 TIs in SLI (89 fps in Heavensward benchmark @4K max settings). Ran my 1080 TIs in SLI (130.7 fps @4K max settings). Downloading the latest benchmark right now (sloow download). Edit: Just installed & ran benchmark ( 107 fps avg, 51 fps minimum).
It would be really cool to see some older generation cards in the power efficiency section, just to see how far we've come. Like 980 Ti, 1080 Ti, 2070 Super, 3060 etc.
They are starting to sound Street fighter special editions! Introducing the 4070 Super turbo extreme mega ultra edition! ( I just got a regular 4070 in my new pc and think it's pretty great coming from a regular 2080)
Honestly, meh is good for me. I have a 2070 super that I'm upgrading from and was keeping an eye on this one. I was told the 2070 super was meh and its been excellent for me. Thanks for all that you guys do. And more importantly, Thanks Steve
ayo same, it's not as high of an uplift as anticipated but this is still the card I think. 4070 super doesn't have the VRAM and AMD doesn't have the RT performance or upscaling tech to match. I finally want to play Cyberpunk and future titles in all the RT glory. If I'm going to be dropping more than $500 on a card, I want the best RT and upscaling tech. Sigh Nvidia.
I've also been waiting for this review as well since I'm still using a previously owned 1650 for my first personal computer build. Really looking forward to having a really nice GPU for once, and while I'll wait to hear what they have to say about the 4080 Super, this one still might be the one for me with my current finances.
Same I've had my 2070 super oc card 4 years now and I haven't had any bad gaming experiences at 1440p , im definitely getting ready with building a pc in the near future and I'm so torn on everything gpu related lol, depending on what happens price wise with the xtx or 7900xt I might pull the trigger.
Hey bro, a 4070s has better value and 4070 tis doesn't use 16gb of ram, only 13gb in one game, 4080 Is faster. all of these cards Will be punched hard when rtx 5000 releases. Got 4070 tis with you, not best value, but what I could afford now, its not bad
Hoping either the 4080 Super isn't falsely marketed (very unlikely) or that the next gen is a good way to get 100% uplift over the 3080 10GB at a reasonable price
did only pay 890€ for my 4080 half a year ago still a shit tier price It was my most expensive gpu purchased in over 28 years and still almost double the price I paid for my last gpu (1080 gtx) rdna3 and ampera did had a really good msrp price tag when it did come to performance but we never got it thanks to miners
Thank you for a well rounded review! It’s very easy to only benchmark the scenarios that favor nVidia (upscaling and heavy RT) and omit everything else. I believe your test suite is very representative of a large number of varying workloads that can be present in real games. The efficiency testing is also a welcome addition for real world applications.
I had the 1080Ti for the longest time and it was such a trooper. I'm considering putting it in a resin block as a display piece to give it the respect it deserves!
Yape, i upgraded my goat EVGA 1080ti last september for a sapphire 7900xt pulse that was cheaper than any 4070ti(-100€ than the cheapest 4070ti at that time)... the performance difference was BIG! First time AMD gpu and no regrets! Long live the GOAT 1080ti!!!
Thank you for including efficiency charts and all the work you always put into these videos. Regarding the possible future additional testing procedures, I would like to see undervoting coming into play. For example, comparing RTX 4090 limited to the 300W vs 4080 300W, trying to approximate the best performance value per given wattage. It may involve too many additional variables like sample quality, but the general idea seems interesting, at least to me.
I swear I remember watching a tech channel a few months ago that was doing this for some video cards, I was specifically looking at the 7800 XT, and they undervolted them to see what power efficiency gains could be had and what it did to the FPS/Gameplay experience. It was quite interesting as they were able to get the energy use down to similar numbers at the 4070 and performance (outside ray-tracing) was also dropped to a similar level as the 4070 so it showed that AMD was pushing the cards to get more performance then Nvidia. I thought it was a great idea and unique way to approach video card reviews.
@@Swarm509 Its also been known since day one that the 4090 especially is far more efficient at around 60% power limit, its one reason I bought one so I could still game in the hot summer without giving up all the performance.
@@Swarm509 There are videos about this topic around the Internet. I feel like there two topics worth mentioning First: Finding the curve for particular GPU, so at which given wattage the GPU should % perform the best. I think is mostly around 80% for current Nvidia generation. By putting the voltage limit to .875-0.95V you should be able to save energy without losing almost any performance. And there are also other benefits like possibly lower coil wine. I am not sure about AMD, but I assume they use similar tactics. Second: Comparison of different GPU using the curve mentioned in first topic.
Thanks for this. I am upgrading to a new PC with a 4070TI Super. Considering that my current PC is rocking a GeForce GTX 1080... I'm hella excited! Also jumping from an i7-7700k to an i9-14900KF
Just some feedback for the comparison between the RX 7900XT and 4070Ti Super (13:05): Perhaps it would be beneficial to keep the direction of the charts in the same direction throughout the review? IE: Positiv number = 4070Ti Super advantage. "Swapping directions" made me do a double take and pause.
@@GamersNexus true But I would prefer to keep the logic based on the review of the 4070 ti super (F@ck it is even annoying to type). I understand that you can't just flip the chart around... 🤔 Perhaps add a picture of each card/company to each side and call it a day maybe, we can infer which ha an advantage. "Bar on this side good for Nvidia, Bar on other side good for AMD. Bigger bar, more betterer oogh!" 😅
Thanks, Steve. So the 7900 XT is still better and also cheaper. Glad I made the right decision by purchasing based on price to performance and not the color and name on the box.
@@3JABO meh. Only if you care about pointless features like ray tracing. Call me old school, but realistic visuals in a game are the least of my concerns. I play games for gameplay. Just give me the best raw performance for the price for smooth and uninterrupted gameplay. If I want realistic rays of light, I’ll go outside and touch grass or open my shades and let the natural ray tracing come in for my eyes and brain to process.
If you just playing, yes. But if you want to use your GPU to anything else (gameplay recording, streaming, video editing, etc.), the red company is not a real option yet. And it makes me sad, because i hate the green company, but i have to use their cards... No one's talk about that, but the best thing in 4070 Ti and Ti Super is the dual NVENC encoder. That's a game changer for content creators.
@@marvin8291 why on earth would you try and play at 4k with a 7800 XT? To each their own, but you’re sabotaging yourself. Just play at 1440p. You’ll get a much better experience.
@@marvin8291 judging you? I’m not judging you at all. I’m just saying that you’re hurting your performance. If 30 fps is enough for you and you’re happy then enjoy it! I just feel running at 1440p would give you an overall smoother and better gaming experience with the hardware you have.
I'm glad I picked up the 7900xt the other day instead of waiting for this card to launch. I saved $70 and I'm getting better raster performance, especially at 1440p
@@quicksilverpr2001 I play at 1440p only so I'm not sure about 4k. I haven't needed to use FSR in any game I've played so far though, so I'm sure you would be more than fine at 4k with fsr... especially more than a 3070 haha
I went from a 3060ti to a 7900 XTX it's a monster. The only game I have where performance is bad for me is Cyberpunk it falls off a cliff(RT path is horrible). However all my mmo's everything else at 4k native just kills it. Even with all my Re titles 4k ultra max with RT settings. 98 percent of my games are max ultra on the 7900 XTX. I don't own alan wake but can tell you cyberpunk is the worst for my 7900XTX OC. I play mmo run AI everything else it's a boss. I also post some of my Comfy UI Stab Dif Ai generated stuff on Skyrim SE artwork page so it does do AI really well. The hard part was finding the right youtube channel that installs on AMD. Drivers I would highly recommend DDU if you don't want crashes if your going from NVIDIA to AMD. Overall best investment in a card for me. Playing 4k on a C2 LG Oled 55inch
@@mynameishaydn3287 thank you for your response. I'm really REALLY tempted to get the 7900xt. I'll wait until tomorrow if the price goes down a little bit more.
Wish you could do a little testing on VR. At least one game would be nice. Lots of VR gamers day you need 16gb of vram, my 4070 works great with 12, would like to have an idea of the real world difference. It's also an interesting technology because frame timing is key.
As someone who is tired of the trend of more power for bigger number I really appreciate the inclusion of efficiency testing in your test suite! Electricity isn't free and efficiency can really play into the long term value of a card. Especially when you only upgrade once every half a decade or so
Power efficiency and pricing has to do with node - TSMC makes nearly everything. * RTX 40 - TSMC 4nm * RX 7000 - TSMC 5nm/6nm * RX 6000 - TSMC 6nm/7nm * Intel Arc - TSMC 6nm When a node is new, it's more expensive. RTX 40 used the absolute newest node available, hence it was expensive. Now that the node matured, they can release cheaper products. AMD will use TSMC 4nm for Phoenix APUs (7040 mobile, 8000G desktop) which they designed ages ago, and are just now releasing as the price is getting cheaper. Nvidia is moving into the AI space - so using a bleeding edge node means their cards will be the most efficient and cutting edge, but that it will come at a cost premium at first. AMD isn't in the AI space nearly as much, so they wait for old nodes to come down in price. Before AI, Nvidia used maturer nodes which allowed for cheaper products.
Hi Steve, excellent video as always! Could you review the AMD rx 7900 GRE, AMD launched it in europe now at a price equivalent to 500 USD (€570)! Seems like crazy value to me, especially since the 4070 super is €700 and above here
It seems to perform very similar to Rx 7800 XT. It is too power limited and bandwidth limited to outperform it significantly. It could be interesting if there was a model with unlocked power (or there was MPT for RDNA 3).
@@andersjjensen they didn't actually review it afaik. So this is just theoretical synthetic performance (pretty much tflops). It doesn't translate to gaming results. It is barely faster at all in games. Sometimes even a little slower.
I’ve been really looking forward to this review. I’m about to pull the pin on a gpu and I’m torn between this and the 7900xt. I guess I’ll know which one to pick once I finish the video
It's a hard choice, lemme know what you've picked. I'm personally torn between those two as well, also considering just staying at 1080p for now due to how insanely expensive everything is. I've been feeling eager to have an nvidia card for the first time due to how many positive things people say about it, I'm scared I'd feel left out of nvidia features if I went for AMD
I’m in the same boat, building brand new PC and got all the parts except gpu. Everything in range of my budget is so close in performance. Do i save some money and get the 7900xt that does better without ray tracing? Do i pay $100 more and get the 4070 ti super for that ray tracing performance (is it really that worth it)? Or do i just say screw it, pay a little bit more and get a 7800xtx. 😂 i have no clue what to do even after watching this.
I had a 3060 for bout 2 years and i had lag through out those 2 years. I recently upgraded to this card last night and i'm loving it so far no complaints.
Meanwhile in Australia, the 7900XT is $300 cheaper than the 4070Ti Super ($1199 compared to $1499). The 4070Ti (non super) is $1069 on sale for comparison.
The Ti is about the same price as the Super and the 7900XT, while the Super Ti is the same price as the 7900XTX and only about $100AU less than a 4080. Nvidia's got no idea what they're doing over here
@@MisterFoxton They know exactly what they're doing. They know that only 0.1-1% of their customers will actually make good on the threat to go to AMD for a generation or two.
@andersjjensen I meant why would they price their new products the same as their more powerful products? It's not even upselling and making their more expensive product more appealing (a la 7700XT) because they are practically the same price ($1119 vs $1149)! You get a retailer offering free shipping and you just got a 4070ti cheaper than a Super.
0:56 Make sure to take time for yourself as well. You guys provide such a great wealth of information that's also fun to watch. If anyone deserves good rest it's you guys who work your asses off like this. Last thing I'd want to see is an infamous burnout or quitting RUclips cause of burn out video. If you choose to be done it should be because you want to move on to something else not because you're dead tired of doing it.
I am surprised to see the powerhungry 4090 perform so good in the efficiency test. It makes me wonder if this way of testing is fair. Yes, per frame you might be using less wattages, but at the same time the card pushes out way more frames. How about doing watt/fps but with limiting the fps. Obviously both cards would need to be able to hit a specific framerate. But how would cards compare when you limit fps to 30 / 60 / 90 / 120 and no limit. I wonder if results would still be the same. In addition I would like to see efficientie when just being in Windows. How much is your card using for no reason at all.
I would love to see locked FPS normalized testing. Modern high end GPUs are super power hungry when maxed out but do they use more or less power when running the same workload as a mid-tier GPU?
As a console gamer who's locked to 1080p 30 fps. All these cards seem fine to me. Probably would get the regular 4070 since it does up to 1440p really well.
Yeah, this Super refresh along with AMD price drops just managed to look RX7900XT even better. That is of course unless you love raytraycing. Then Nvidia got you hooked and you got to have their GPU. In reality this means mostly just Alan Wake 2 and CP2077 though. AW2 I can enjoy withou rt and CP2077 was such a mega letdown when it came, that I'm no longer interested in it despite the game being somewhat fixed year after release. Plus Nvidia gpus have surprisingly good results in CP2077 which indicates there were some specific optimalisations done for team green.
@@Hombremaniac Even if you love ray tracing and frame generation, Nvidia's paltry 12gb VRAM won't get you very far, unless you want low textures with performance upscaling.
@@DragonOfTheMortalKombat I´ve bought RX7900XT 2 months ago, so before any Super refresh rumours. I´m happy that my GPU was not made obsolete by 4070ti Super and I can live with the fact that RX7900XT is now some 60eur cheaper where I live (Czech rep).
I recently upgraded from my RTX 3070 to a 4070 Ti Super. I'm not disappointed, the FPS increase is remarkable (ultra settings on most games, 1440p). I've had some problems with the RTX 3070 with my 3 monitors (I use this computer also to work). When multiple fullscreen applications where running, I had some fps drops on the main gaming monitor (165 Hz). But now, everything is really fluent. However, I do have flickering sometimes with HDR on the monitors enabled when starting or exiting games (mostly non-HDR games, I think). May be a monitor thing or driver-related, no idea. It flickers like 2 or 3 times, like switching from HDR to non HDR and back. Also happens in benchmarks. Otherwise I'm really satisfied. But if you're on the 40xx generation already, I wouldn't bother getting it. If you're searching for the 4070 already, just go for the 4070 Ti Super. It has the chip of the 4080 and the Ti Super can be found in the same price range as the 4070 and 4070 Ti. But it's beter than the other two. I'm running it combined with an i7-13700KF, so don't bother getting an i9. The i7-13700KF is usually never above 60/65°C. I haven't seen my GPU going above 80°C in Cyberpunk 77 (again, 1440p, ultra, Raytracing enabled). But then again, I'm not playing on 4K and I don't want to.
would this be a good upgrade over the 3070 in ultrawide 1440p since 8gb of vram is struggling now at this res. Also planning on using frame gen so even more vram is going to be used
I second the yes. I don't know if this is the optimal upgrade but you most definitely need more VRAM and this is one of the options with the other big cards too 3090, 7900xt, etc.
Well the same philosophy when it comes to Apple and bunch of commentary users who's trying their hardest to hide the truth. This is regarding on how cheap to mass produce their IPhone 14.
@@iwantgoat not a good argument. im still using my 2070 super that i bought new in 2019. it was around $500 if i remember right, and i felt like it was overpriced back then. greed with these companies has gotten so damn out of control i that havent wanted to build a new pc out of principle.
@@slumlord2625 Good argument. The only reason you are complaining is because you are poor. Most hobbies costs way way way more than gaming on a computer.
@@iwantgoat poor? nope, i have principles. my pc works just fine. i have no need for a faster video card right now. people that build a new pc with every release need to get a real hobby and touch grass
Always was. AMD's marketing department are experts at eating their crayons and sniffing their glue, but once reality sets in Dr Su comes around and lays down the law. AMD is fighting hard to become TSMCs most liked customer. That means never cancelling wafer orders, no matter what. And it means never "defecting" to other foundries like Nvidia did last gen. So AMD will keep ad-hock adjusting prices to make sure they're the best value, and even adjust after that if the market is simply slow. I bought my 7950X3D at full price, and despite it being the best allround CPU on the market it has already dropped a pretty penny.
@@charmolettafranquestafiestayam Launching FSR3 in a broken state in two unpopular games is an example of what I mean when I say AMD is bad at marketing. Taking petty shots at Nvidia for exceeding 300W with Ampere, then later launching the 6950XT at 355W is another. AMD's marketing division excel at sticking their foot in their mouth.
I'm just looking at upgrading from my long standing 8700k / 1080ti ftw3 setup. It's served me so well... 7800X3D / 4070ti (maybe super) is my next long term build.
love the new efficiency view. another potential metric idea to help guide users to a graphics card: what share of games in the test set hit 30, 60, and 90 FPS at a certain settings threshold (resolution, graphics toggle, upscale/fg setting). it would help people target a card that enables the larger share of games at "at least" a certain playability level. Knowing how cards relatively stack is definitely helpful, but it's hard to know if an arbtrary card achieves playable performance across all or just some of the test suite. this may be a bad idea or a use case you all don't want to focus on. but thought i'd chuck it out there.
Reminder that the AMD price drops are also just Temporary. And in some countries made no real difference, people will still go Nvidia. Why buy a 7900XT when you can get a 4070 Super for €200 less that performs about the same as a 4070Ti with an OC.
This honestly just shows what an amazing card the 4080 is, if only it was priced properly. It would be the contender to the best gpu Nvidia has ever launched. Very unfortunate that pricing has hit down so hard on something that would otherwise have knocked the 1080/ti off that throne
Its pricing is pretty ridiculous but being able to grab them for around $900 has been the norm for almost a year. It looks like they are still holding about that value so at least there is that.
@@John-df3lm Yep. Here in Denmark the 4080 price has recently dropped even further to 5-10% below what the 7900 XTX sells for, which is pretty crazy compared to what it was at launch. People got burned bad by the initial price
Thanks for the great review. The efficiency charts are a most welcome addition. If I may suggest yet another metric, I think a fps capped test like you did with the CPUs would be a good addition as well. Limiting to 120fps, or 60 on very demanding games/settings, would give a good idea of power consumption on typical usage scenarios, and with higher gaps in the values themselves, specially in those demanding single player games where the highest FPS possible is not that much more useful than in competitive games. Consider an example: if I'm interested in buying a RTX 40 series card and I'm going to play games at the top temporal and spatial resolution of 1440p@120fps, because that's the display I have, and I don't play competitive shooters, it'd be good to know the gap in efficiency between having a 4090 at more comparatively "relaxed" workload, or having, say, a 4070 Ti Super working near capacity while gaming at that max 120 fps.
I feel like you guys tend to lean towards negative takes and have generally a toxic vibe to your videos. If something is doing really well you don’t bother reviewing it. If it’s something you can speak negatively about, you tend to prioritize those videos.
@@dialatedtechnology Value sucks, especially without frame gen, as obvioused by all their marketing shenanigans. Let the market cool and force them to offer something good.
Great Kudos for the power efficiency part! Would like to see two further metrics in that department: minium FPS (0.1% ot 1%, whichever you usually work with) vs power consumption and (again minimum) FPS vs full system power consumption. As I am not 100% sure that the cards use the CPU/System RAM the same.
Thanks for the fast review. Wanted to buy one, but went for a used 7900xtx instead. Same price, but as you showed much better performance. (Raytracing aside.) But now I need a new monitor...
The thumbnail says it all, although I usually watch most of what you put out there, I didn't watch this or any nVidia related content this last year. Thanks Steve !
Thanks for your review....I spent 10 hours benchmarking and ocing the 70 super for just gpu mining alone, about to start gaming rn...I understand how much time you guys spend on these reviews and thank you so much for the good work you do for the community.
Watch the AMD RX 7600 XT review here! ruclips.net/video/1aopp8XolXE/видео.html
Find the written review of the 4070 Ti Super here! gamersnexus.net/gpus/nvidia-geforce-rtx-4070-ti-super-gpu-review-benchmarks-power-efficiency-gaming
JUST finished rendering and uploading. Latency comes back for the 4080 Super -- so that'll have our new power efficiency testing, latency, and Cyberpunk (this one re-introduced Cyberpunk). We'll try and get Alan Wake 2 added soon as well! What other games do you want to see? We're replacing some of the tests with new ones. Grab a LIMITED Disappointment PC T-Shirt in Heather (available in all sizes)! store.gamersnexus.net/products/disappointment-pc-2023-heather
Or in 100% cotton (running very low on some sizes): store.gamersnexus.net/products/disappointment-pc-2023-cotton
Price/Performance/Efficiency ratio is what i have said is the most ideal for the longest time, thank you for finally presenting the whole picture, thank you!
Good job, Steve. How tired are you?
You need to test some UE 5 games, because a lot of newer games will be using it for many years to come. Maybe RoboCop: Rogue City.
i love this already 3 seconds in!
at the end of the day, per watt is everything!
noise, heat, longevity, cost, performance!
and please add Apple Silicon here and there!
Avatar would be a good addition to the tests
"joules per frame" means that you can also measure "calories per frame" which means that you can also measure "hamburgers per frame" which means that you can also measure "pizzas per second" as framerate.
Pizza Time?
what about tacos?
@@Not_a_lier if you want.
"Bigger number better, means better number gets more bigger better, so, that's... better, because its bigger" - It's exactly this kind of hard hitting journalism that keeps me coming back. Don't change GN.
I literally read this comment AS he was saying it 😅😂
Harder, Better, Faster, Stronger 😂
Ah yes, the Super Tie.
That's Tie Super, to you!
@@GamersNexus
I prefer a Windsor Tie myself, fancy but approachable.
Super 👔
4070 T I S
If the fans scream like a TIE Fighter then I guess I’m switching back to nvidia.
I thought the XT/XTX naming bugged me a little, but the Non-TI, TI, Super and TI Super convention roasts my almonds, lol. Great testing and info!
Just wait until they refresh the Ti Super after that poor review and it becomes the 4070 Ti Super Mega OK Tuf Gaming OC White Edition
A 4070 that slaps the already somewhat confusing Ti and Super monikers together, while also being based on an entirely different chip from another range of cards.
Just leave it to Nvidia to make their line-up even more opaque than it already was. 🙄
Numbers are cool. Just make it "4070, 4070.1, 4070.2" or some shit.
Who cares
wait till they release the 4070 ti hyper and then follow up with the 4070 ti ultra
My toxic trait is I watch review videos after buying the product lol
Lol me too!! Just bought it.
Im about to buy it
I have the $800 in my hand watching this in front of the store
same
I'm happy with it
I think Nvidia must be using an AI name generator for all their cards now,
4 variants of a 70 series is ridiculous
Chat GPT, generate a name for a video card by only using a mix of these letters or words: "4070," "T," "i," and "Super." Go!
@@GamersNexus don't forget FE, OC, and a random number of X's.
@@GamersNexus "My name is Vidia. Nvidia Sup4 Tier 007."
I would buy that card.
@@GamersNexus ChatGPT: Sure, how about the name "T4070i Super"?
Sounds good to me
Nah, AI would probably come up with RTX4071, 4072, 4073, & 4074.
I salute you for your sacrifice, Steve. Godspeed.
Some sleep sacrifice every now and then isn't the worst. It's a fun job! Love doing this stuff.
@@GamersNexusI hope you have a mattress to sleep on, some food to eat and water to drink. Thank you for your efforts. 🙂
@@GamersNexus You are sacrificing braincells however, in an attempt to comprehend nvidias endless skulduggery lmao For that you will have my eternal thanks, that way I dont have to.
@@unclej3910 What do you think they use those disappointment PCs for?
You've got such a cool job, man, I love what you guys are doing. Keep up the good work! @@GamersNexus
Thanks steve
"Thanks steve" is how you know an early comment isn't a bot.
@@GamersNexusYou can literally see it!
Gracias Steve
lmao your profile picture is absolutely hilarious
Thanks james
Thanks for all your hard work on these reviews. Good stuff. Really good to see my 2080 in a lot of these charts too. Really helps me compare and see how much of a performance boost I might be able to expect if I upgrade to this card.
I got a 2060, looking at the 7900xtx. Will likely go with all AMD build this year unless NVIDIA upcoming 5080 is reasonably priced. Definitely going with Zen 5 X3D. AMD is impressive with their CPUs they deserve my dollars. Never owned an AMD part in 25 years of PC gaming but can't deny their comeback. Hopefully their nexgen GPUs can again challenge the 80 series for cheaper.
I think it would be interesting to see their power efficiency at the same FPS cap. I also think it would be interesting to see how power efficient they were at certain different % power limits, but I think that's a lot of work for a lot of cards.
Just to let you know earth is hell
The amount of 70s card this gen is overwhelming
We all aren't prepared for the 7th gen 7070 ti super (XxX Version)
and all over them overpriced
@@traiges414Don’t forget the 7770 super ti. It gets edged out by the 7070 ti super XXX (but it’s $5 cheaper!)
When they ramped up the prices, the 70 is where the market is. The 4070 is what the 80 was to the 20 series.
@@xRaptorScreamx Yeah but it hurts less buying a 4070 Super OC than it does buying a non super 4070 TI for a lot more.
Thanks Steve... and the Patrick Clone Army and of course the flealess CEO.
The grind must be ridiculous right now... waiting impatiently on the 8600g release btw.
You're killing it with these videos man. Great work
It almost seems like Nvidia is intentionally limiting this thing's performance in the driver, but went a little too far in many games and now reviewers are catching it. Let's see if they "fix" it with the next driver. It should be much closer to the base 4080 like the 3080 vs 3080 Ti
So gross. With how many gpus nvidia are selling to meta every goddam day they know they can do basically anything to consumers and then just tap out whenever. Nvidia vs everyone for vending them
Yea this thing is weaker than I thought it'd be tbh. Thought it would handle the XT fairly easy
Assuming Nvidia isn't cheating, it could also mean the smaller L2 cache and power limit are hampering performance on this versus the 4080
@@miguelaraneta3144I think it's a power issue.
This is intended to be a modest jump from the 4070 ti not a real improvement. Classic “super” “refresh”
'The new Nvidimeh RTX 4070 TI Super Tie'
"Nvidimeh 4070 Ti Super Tie" is actually genius
I'm waiting on the 4080 Super but was really hoping for a lot more from this card. Geezh, at this rate I might just go for the Sapphire Nitro 7900XTX unless the 4080 Super does something surprising.
Why on Earth would you do that? 4080 Super will be marginally more expensive while giving massive advantages in other areas (ray tracing, encoding, AI, other prof applications), and it will give an edge in rasterized too, all while being much more power efficient.
Unless AMD seriously slashes the 7900XTX prices from what they are today, there's absolutely no reason to buy this card over the 4080 Super.
@@SanctusBacchusAMD will most surely slash the price of XTX, 100$ perhaps
@@SanctusBacchus relax dude amd is fine in all aspect just same not so strong raytracing but either those gpus suck 4090 is minimum for that even that performance just tanks 2much for reflaction gad damn u people are delusenal u take fucking reflection over gameplay no wonder this world has turned into shithole
@@SanctusBacchus It remains to be seen. I don't think I'd buy the XTX without some kind of discount, maybe $900. I am interested in RT for Cyberpunk, but I also play Star Citizen where AMD seems to have an advantage. Also, as you say, power efficiency is slightly important to me. If the 4080 Super can match rasterization with the 7900XTX, then, yeah I'm probably sold on it at that point.
If I wouldnt want VR I’d absolutely go for the xtx, but alas.
love the efficiency charts, please keep them in the future GPU reviews if you can :)
efficiency is a waste of time, you need to be playing for 10 hours every single G-D day to make a significant difference at the end of the year... Most gamers will "save" maaaybe $5 from switching to a "more efficient" card and would have probably spent a lot more money upfront...
@@todorkolev7565
Heat and noise.
@todorkolev7565 it obv depends on conditions and stuff but it is a very important metric, not just as chart and info but also as classification and quality of product. But as example for the CPU video Steve talked about, the i9-14900K *could* cost you 223.1 dollars per year whereas the 7950X3D *could* cost you 73.7 dollars per year comparatively... so yes, it's worth it.
@@todorkolev7565 and while it might not really be fair to compare 3x differences with cpu's to 0.3x differences gpu's i just wanted to show you how important efficiency is (especially when looking at a whole system).
@@todorkolev7565 Not true for me. I pay 0.59€/kwh.
So 50 Watt difference makes 170€ over 4 years with a usage of 4h per day.
That means that it makes much more sense to consider a RTX4060 (115W) over the 50€ less expensive RX 7600 (165W)
Also efficiency correlates with heat and noise, especially important for small builds.
Thanks for the test. Because of that, I just ordered a Sapphire 7900 XT. Should arrive at the end of the week.
Same, I ordered mine last week when the price dropped to $720
Yeah enjoy the stuttering and black screen driver crashing.
I'm really looking forward to efficiency testing for lower end cards. In most of Europe most cards are way more expensive than in the US and the power's more expensive too, so seeing how cards stack up in power efficiency should make buying a cheaper card easier.
I appreciate the work you do
Thank you!
@@GamersNexus no u
@@GamersNexusI love u homi
For all the work you and the team have to do I’d like to say, Thanks Steve.
I love the Gamers Nexus team and all the work they do is brilliant
Got a shirt from you guys just now. Been watching Steve with Gamers Nexus for years now I figured the time has come to show some direct support. It's been a joy to watch.
I'm upgrading from a 2070 Super to a 4070 TI Super.
Bought myself a 4070 TI last week, canceled the order just in time, and just today bought a 4070 TI Super + Corsair PSU GPU cable for 150 euro less then the previous 4070 TI card.
So I'm pretty happy and I'm sure I'll be happy with the upgrade.
I can't state how much I appreciate how quickly you guys drop these videos. People that think being a content creator on RUclips is easy does not understand that you have one of the hardest boss in the world: the viewers of RUclips and the internet, who can be extremely critical and demanding. Your dedication to your work is amazing
A job's a job, it's less about easy or hard and more about pay.
"one of the hardest boss in the world" lmaoo
Well tech tubers don't have a work schedule, their life is the schedule. Unless we talking about Linus and look how that ended, instead Steve can be proud having one of the highest quality channels on youtube.
It's just a job, chill
Once again, another great video with helpful data, thanks Steve and team. I might be alone in this, but I'd quite like to see a scatter graph showing each cards FPS/W and an average FPS (I'm not smart enough to decide what games would be best suited to make that average). It might not be particularly helpful as a piece of buying data without prices there, but it would be interesting to see the trade-offs/generational improvements in the technology.
Good idea. I'll mess around with that. Won't make it into the next review (probably), but adding to my list!
Well the video's just dropped, so I can't really comment on it other than that I appreciate your guys' work. Bought a mod mat for a friend (the signed 15 one that apparently says beve on it) and he's been enjoying it!
Genuinely, keep up the good work.
Thank you so much for buying one! Glad he likes it. And thank you for the support!
I love focusing on FPS per watt, in pure rasterization, no DLSSX gimmicks. Thank you Steve and team!
Just purchased some more merch fro the GN Store to help support the amazing and in depth testing you all do over there at GN🥰😇😇💪👍. Cheers from Australia 🇦🇺😄
The Furmark benchmark shows what a beast the 4090 is when overclocked.
Loved the efficiency testing. Please show more transient spikes data of more GPU models, thanks.
these efficiency benchmarks are what i've always wanted to see
Yep, performance per watt and price is the most important thing to me as that suggest real progress.
But how come the AMD cards are so terribly inefficient? There surely must be something wrong with testing metrics, this cannot be happening.
because AMD is inefficient @@Aquaquake
AMD was targeting a +50% uplift in efficiency over RDNA2 before launching these new cards.
Unfortunately it does seem true that RDNA3 did not meet these targets.
It may be more disproportionately more efficient with lower clocks, but it's already not as competitive as it should be in raw speed (imo)
This architecture, the mix of chiplets etc just didn't work out as good as everyone I bet including AMD had hoped
@@Aquaquakekeep in mind that there is a node difference between AMD and Nvidia, which means that for peak clock speeds AMD will be spending much more power to get the top end. Nvidia's dies are also monolithic so they are going to be slightly more efficient from that. Final piece of the puzzle is Nvidia has RT hardware acceleration which makes computing RT tasks more energy efficient than AMD's methods (right now, they will probably add hardware acceleration too next generation or the gen after). Mix that together and you see that ~20-30% efficiency lead. Honestly I am surprised it's not more but sdram isn't scaling well with node shrinks so maybe that's it
I currently have a 4070 Ti and while I would like to have the additional VRAM looks like I'm not missing out on much performance. This card performs about where I expected with the biggest gains coming at 4k.
Until you hit the 12gb mark
@@lifemocker85 Hasn't happened yet.
@@richardfarmer6570 no longevity with only 12gb
The efficiency testing is great, thanks for the hard work. With power being expensive here this is a consideration.
making my $799 7900xtx purchase last month feel pretty gooooood.
👍which one? Pretty good price
XFX SPEEDSTER MERC 310@@diwiak
I got extremely lucky. It was that price for about 90 minutes on amazon for a lightning deal. @@surfer8438
XTX? Damn, that was a great deal!
XFX Speedster MERC310@@diwiak
I think maybe an average efficiency over multiple games could be interesting since the people play multiple games, you would have do eliminate the total fps (so dividing by total fps or something), this way you can maybe identify trends between the gpus a bit better since they are sometimes a bit all over the place between games. Looking forward to see power efficiency in future GPU reviews :)
Would need to figure out how to do the division while retaining data integrity (maybe geomean), but I like that idea! Will play around with math and work on it!
@@GamersNexus Maybe something like the geomean of fps/watt?
I'm prefacing this with, I don't know jack about squat.
With that said, as long as GN published average FPS, and average wattage over a game benchmark, you could do that math yourself and customize it to what games you play (again, as long as those games are included in the GN test suite. Pick the games you play, add the average FPS over those games and divide by the amount of games. Do the same thing for wattage, and then divide the FPS over wattage and you'll have your efficiency score with that card over your own test suite of games.
If you don't want the FPS included in that, just don't use FPS in that calculation.
I would also be curious to see power metrics with frame rates locked at say 60 FPS. Would be interesting to see how efficient cards are. Not all of us are that interested in playing at 144hz particularly if one lives in a place where energy is expensive. Or in my case off-grid on solar. Power gains from undervolting would also be interesting but that is probably a big ask.
@@GamersNexusI love u, please bless me with Ur long hair so I can get a 4080 super on release thanks
How many tries did it take you to get the "bigger better bigger" line? Lol it was glorious
hahaha, 3 and I still am not sure exactly what I said!
@@GamersNexus extremely confusing for me as non-native speaker lol
felt like I advanced to C2 level, thank you for that
6:04
@elvisevillive as a native speaker, it was still gibberish. 😂
1080 TI still hanging in there ! After seeing the FFIV benchmarks, I decided to check my old scores. Last time I ran it was with my 980 TIs in SLI (89 fps in Heavensward benchmark @4K max settings). Ran my 1080 TIs in SLI (130.7 fps @4K max settings). Downloading the latest benchmark right now (sloow download).
Edit: Just installed & ran benchmark ( 107 fps avg, 51 fps minimum).
My 1080 is starting to get artifacting so I'll need to underclock it a bit more :P
I need a new card soon 👀
It would be really cool to see some older generation cards in the power efficiency section, just to see how far we've come. Like 980 Ti, 1080 Ti, 2070 Super, 3060 etc.
Combining the Ti and Super badges was a weird decision.
Time for Ultra branding.
At first doing a mixing of peanut butter and banana feel the same to me. And then i fall in love. and then i'm fat.
Imagine if they had stuck with naming it 4080, we'd have two 4080 supers now
Next geneation we will have
RTX™ 5070® Ti Super Duper Tie Ultra© for mere $999.99/month for first 3 months.*
*Therms and consition may apply read all document carefully before subscribing.
They are starting to sound Street fighter special editions! Introducing the 4070 Super turbo extreme mega ultra edition! ( I just got a regular 4070 in my new pc and think it's pretty great coming from a regular 2080)
Honestly, meh is good for me. I have a 2070 super that I'm upgrading from and was keeping an eye on this one. I was told the 2070 super was meh and its been excellent for me. Thanks for all that you guys do. And more importantly, Thanks Steve
ayo same, it's not as high of an uplift as anticipated but this is still the card I think. 4070 super doesn't have the VRAM and AMD doesn't have the RT performance or upscaling tech to match. I finally want to play Cyberpunk and future titles in all the RT glory. If I'm going to be dropping more than $500 on a card, I want the best RT and upscaling tech. Sigh Nvidia.
@@slickvibez16 same, although i'm upgrading from a 2060 super
I'm also upgrading from 2070 super to 4070 Ti Super, it will definitely be a beast for 1440p!
I've also been waiting for this review as well since I'm still using a previously owned 1650 for my first personal computer build. Really looking forward to having a really nice GPU for once, and while I'll wait to hear what they have to say about the 4080 Super, this one still might be the one for me with my current finances.
I'm still on a 1070 ti, so even the non-super 4070 ti would've been a pretty huge jump for me.
Thank you for including older cards! I upgrade infrequently and seeing my 2080 on the graphs really helped!
Same! I'm torn between this card and a 4080/super. I only got 650 watt power supply tho so unsure
Same I've had my 2070 super oc card 4 years now and I haven't had any bad gaming experiences at 1440p , im definitely getting ready with building a pc in the near future and I'm so torn on everything gpu related lol, depending on what happens price wise with the xtx or 7900xt I might pull the trigger.
I just purchased the 4070ti super .. and im happy with it.. getting punching distance of a 4080 for less than 800$ .. I can’t get mad at that.
Hey bro, a 4070s has better value and 4070 tis doesn't use 16gb of ram, only 13gb in one game, 4080 Is faster. all of these cards Will be punched hard when rtx 5000 releases. Got 4070 tis with you, not best value, but what I could afford now, its not bad
7900xt it is! Especially for us that just care about online esport gaming.
Shoutout to the first gen 40X0 buyers.. keeping those double dividends alive!
Hard regrets not buying into it a few years ago, or even 1 year ago. Could afford on of their cards if I did!
Hoping either the 4080 Super isn't falsely marketed (very unlikely) or that the next gen is a good way to get 100% uplift over the 3080 10GB at a reasonable price
did only pay 890€ for my 4080 half a year ago still a shit tier price
It was my most expensive gpu purchased in over 28 years and still almost double the price I paid for my last gpu (1080 gtx)
rdna3 and ampera did had a really good msrp price tag when it did come to performance but we never got it thanks to miners
double dividents are thanks to nvidia AXXXX cards. 6k for a card, easily
It does hurt having paid $900 for my 4080 about a year ago but at least the prices haven’t dropped too much.
cant wait for your coverage on the 4080 super. really hoping it can fill the 80 slot after the disappointment of the base 4080
I just hope we can get a 4080 Super FE on release
Don't expect more than 5% over the regular 4080.
Thank you for a well rounded review! It’s very easy to only benchmark the scenarios that favor nVidia (upscaling and heavy RT) and omit everything else. I believe your test suite is very representative of a large number of varying workloads that can be present in real games. The efficiency testing is also a welcome addition for real world applications.
Your Advanced Summary Conclusion on 32:10 sound so GREEK! Thumbs up Steve!
I'm very happy to see FPS per Watt (Frames/J) for GPUs as well ! :)
Thank you for including the 1080ti. I think it might almost be time for the GOAT to be put to eBay when the 4080 Super lands.
I had the 1080Ti for the longest time and it was such a trooper. I'm considering putting it in a resin block as a display piece to give it the respect it deserves!
Yape, i upgraded my goat EVGA 1080ti last september for a sapphire 7900xt pulse that was cheaper than any 4070ti(-100€ than the cheapest 4070ti at that time)... the performance difference was BIG! First time AMD gpu and no regrets!
Long live the GOAT 1080ti!!!
I’m worried about 4080 super stock since it seems that is what everyone is going for now.
Thank you for including efficiency charts and all the work you always put into these videos.
Regarding the possible future additional testing procedures, I would like to see undervoting coming into play.
For example, comparing RTX 4090 limited to the 300W vs 4080 300W, trying to approximate the best performance value per given wattage.
It may involve too many additional variables like sample quality, but the general idea seems interesting, at least to me.
I swear I remember watching a tech channel a few months ago that was doing this for some video cards, I was specifically looking at the 7800 XT, and they undervolted them to see what power efficiency gains could be had and what it did to the FPS/Gameplay experience. It was quite interesting as they were able to get the energy use down to similar numbers at the 4070 and performance (outside ray-tracing) was also dropped to a similar level as the 4070 so it showed that AMD was pushing the cards to get more performance then Nvidia. I thought it was a great idea and unique way to approach video card reviews.
@@Swarm509 Its also been known since day one that the 4090 especially is far more efficient at around 60% power limit, its one reason I bought one so I could still game in the hot summer without giving up all the performance.
@@Swarm509 There are videos about this topic around the Internet. I feel like there two topics worth mentioning
First: Finding the curve for particular GPU, so at which given wattage the GPU should % perform the best. I think is mostly around 80% for current Nvidia generation. By putting the voltage limit to .875-0.95V you should be able to save energy without losing almost any performance. And there are also other benefits like possibly lower coil wine. I am not sure about AMD, but I assume they use similar tactics.
Second: Comparison of different GPU using the curve mentioned in first topic.
Despite the price, this card is actually really good. A Massive upgrade from a 3070. I can play anything on max, including HellBlade 2.
I am planning on upgrading to this from 1660 TI, just a little bit more and Ill have them money saved
Thanks for listening and adding back modern graphically intensive games alongside the suite of well-tested and efficient games on a short notice.
Thanks for this. I am upgrading to a new PC with a 4070TI Super. Considering that my current PC is rocking a GeForce GTX 1080... I'm hella excited!
Also jumping from an i7-7700k to an i9-14900KF
Just some feedback for the comparison between the RX 7900XT and 4070Ti Super (13:05):
Perhaps it would be beneficial to keep the direction of the charts in the same direction throughout the review? IE: Positiv number = 4070Ti Super advantage.
"Swapping directions" made me do a double take and pause.
Yeah, that's the challenge with it. The options are keep the math the same as much as possible or keep the usability the same.
@@GamersNexus true
But I would prefer to keep the logic based on the review of the 4070 ti super (F@ck it is even annoying to type). I understand that you can't just flip the chart around... 🤔
Perhaps add a picture of each card/company to each side and call it a day maybe, we can infer which ha an advantage.
"Bar on this side good for Nvidia, Bar on other side good for AMD. Bigger bar, more betterer oogh!" 😅
Thanks, Steve. So the 7900 XT is still better and also cheaper. Glad I made the right decision by purchasing based on price to performance and not the color and name on the box.
Raw power, 100%. Dlss is such a game changer though.
@@3JABO meh. Only if you care about pointless features like ray tracing. Call me old school, but realistic visuals in a game are the least of my concerns. I play games for gameplay. Just give me the best raw performance for the price for smooth and uninterrupted gameplay. If I want realistic rays of light, I’ll go outside and touch grass or open my shades and let the natural ray tracing come in for my eyes and brain to process.
If you just playing, yes. But if you want to use your GPU to anything else (gameplay recording, streaming, video editing, etc.), the red company is not a real option yet. And it makes me sad, because i hate the green company, but i have to use their cards... No one's talk about that, but the best thing in 4070 Ti and Ti Super is the dual NVENC encoder. That's a game changer for content creators.
@@marvin8291 why on earth would you try and play at 4k with a 7800 XT? To each their own, but you’re sabotaging yourself. Just play at 1440p. You’ll get a much better experience.
@@marvin8291 judging you? I’m not judging you at all. I’m just saying that you’re hurting your performance. If 30 fps is enough for you and you’re happy then enjoy it! I just feel running at 1440p would give you an overall smoother and better gaming experience with the hardware you have.
I'm glad I picked up the 7900xt the other day instead of waiting for this card to launch. I saved $70 and I'm getting better raster performance, especially at 1440p
who's perform in 4K? I'm really tempting to get the 7900Xt, switching from a 3070ti.
@@quicksilverpr2001 I play at 1440p only so I'm not sure about 4k. I haven't needed to use FSR in any game I've played so far though, so I'm sure you would be more than fine at 4k with fsr... especially more than a 3070 haha
I went from a 3060ti to a 7900 XTX it's a monster. The only game I have where performance is bad for me is Cyberpunk it falls off a cliff(RT path is horrible). However all my mmo's everything else at 4k native just kills it. Even with all my Re titles 4k ultra max with RT settings. 98 percent of my games are max ultra on the 7900 XTX. I don't own alan wake but can tell you cyberpunk is the worst for my 7900XTX OC. I play mmo run AI everything else it's a boss. I also post some of my Comfy UI Stab Dif Ai generated stuff on Skyrim SE artwork page so it does do AI really well. The hard part was finding the right youtube channel that installs on AMD. Drivers I would highly recommend DDU if you don't want crashes if your going from NVIDIA to AMD. Overall best investment in a card for me. Playing 4k on a C2 LG Oled 55inch
@@NephilimGodMusic same I play on a C2 OLED 48. So far my 3070ti has been Great but... It's time to get something more powerful.
@@mynameishaydn3287 thank you for your response. I'm really REALLY tempted to get the 7900xt. I'll wait until tomorrow if the price goes down a little bit more.
Wish you could do a little testing on VR. At least one game would be nice.
Lots of VR gamers day you need 16gb of vram, my 4070 works great with 12, would like to have an idea of the real world difference. It's also an interesting technology because frame timing is key.
Thanks Steve!
Seriously, I appreciate all the work you & your team do. Power efficiency benchmarks look great
As someone who is tired of the trend of more power for bigger number I really appreciate the inclusion of efficiency testing in your test suite! Electricity isn't free and efficiency can really play into the long term value of a card. Especially when you only upgrade once every half a decade or so
Power efficiency and pricing has to do with node - TSMC makes nearly everything.
* RTX 40 - TSMC 4nm
* RX 7000 - TSMC 5nm/6nm
* RX 6000 - TSMC 6nm/7nm
* Intel Arc - TSMC 6nm
When a node is new, it's more expensive. RTX 40 used the absolute newest node available, hence it was expensive. Now that the node matured, they can release cheaper products.
AMD will use TSMC 4nm for Phoenix APUs (7040 mobile, 8000G desktop) which they designed ages ago, and are just now releasing as the price is getting cheaper.
Nvidia is moving into the AI space - so using a bleeding edge node means their cards will be the most efficient and cutting edge, but that it will come at a cost premium at first.
AMD isn't in the AI space nearly as much, so they wait for old nodes to come down in price.
Before AI, Nvidia used maturer nodes which allowed for cheaper products.
Hi Steve, excellent video as always! Could you review the AMD rx 7900 GRE, AMD launched it in europe now at a price equivalent to 500 USD (€570)! Seems like crazy value to me, especially since the 4070 super is €700 and above here
It seems to perform very similar to Rx 7800 XT. It is too power limited and bandwidth limited to outperform it significantly. It could be interesting if there was a model with unlocked power (or there was MPT for RDNA 3).
@@matejnemec3022 Techpowerup has the 7800XT at 86% the performance of the 7900GRE, which translates to 16% faster so "it's not nothing".
@@andersjjensen they didn't actually review it afaik. So this is just theoretical synthetic performance (pretty much tflops). It doesn't translate to gaming results. It is barely faster at all in games. Sometimes even a little slower.
I’ve been really looking forward to this review. I’m about to pull the pin on a gpu and I’m torn between this and the 7900xt. I guess I’ll know which one to pick once I finish the video
So? Which one did you go with
It's a hard choice, lemme know what you've picked. I'm personally torn between those two as well, also considering just staying at 1080p for now due to how insanely expensive everything is. I've been feeling eager to have an nvidia card for the first time due to how many positive things people say about it, I'm scared I'd feel left out of nvidia features if I went for AMD
@@hedzersb8523if you plan on staying 1080p this card is way OP...a 4070 Super is enough...
I’m in the same boat, building brand new PC and got all the parts except gpu. Everything in range of my budget is so close in performance. Do i save some money and get the 7900xt that does better without ray tracing? Do i pay $100 more and get the 4070 ti super for that ray tracing performance (is it really that worth it)? Or do i just say screw it, pay a little bit more and get a 7800xtx. 😂 i have no clue what to do even after watching this.
@@connorreid8440 we're all like that mate
Thanks for these reviews guys. All this work is much appreciated!
I had a 3060 for bout 2 years and i had lag through out those 2 years. I recently upgraded to this card last night and i'm loving it so far no complaints.
@GamersNexus if ever there was a clip that deserves to become a YT Short or TikTok video, it's the "Bigger Number Better" piece.
Meanwhile in Australia, the 7900XT is $300 cheaper than the 4070Ti Super ($1199 compared to $1499). The 4070Ti (non super) is $1069 on sale for comparison.
It is even worse in Brazil, where 4070Ti is almost 60% more!
weird prices. anyway 4070Ti Super is releasing tomorrow so $1499 price tag might be overblown and price may decline in a few days to MSRP
The Ti is about the same price as the Super and the 7900XT, while the Super Ti is the same price as the 7900XTX and only about $100AU less than a 4080.
Nvidia's got no idea what they're doing over here
@@MisterFoxton They know exactly what they're doing. They know that only 0.1-1% of their customers will actually make good on the threat to go to AMD for a generation or two.
@andersjjensen I meant why would they price their new products the same as their more powerful products? It's not even upselling and making their more expensive product more appealing (a la 7700XT) because they are practically the same price ($1119 vs $1149)! You get a retailer offering free shipping and you just got a 4070ti cheaper than a Super.
0:56 Make sure to take time for yourself as well. You guys provide such a great wealth of information that's also fun to watch. If anyone deserves good rest it's you guys who work your asses off like this. Last thing I'd want to see is an infamous burnout or quitting RUclips cause of burn out video. If you choose to be done it should be because you want to move on to something else not because you're dead tired of doing it.
Wft do you genuinely know about these people. Everyone works hard. Stop boot lickng.
I am surprised to see the powerhungry 4090 perform so good in the efficiency test.
It makes me wonder if this way of testing is fair.
Yes, per frame you might be using less wattages, but at the same time the card pushes out way more frames.
How about doing watt/fps but with limiting the fps.
Obviously both cards would need to be able to hit a specific framerate.
But how would cards compare when you limit fps to 30 / 60 / 90 / 120 and no limit.
I wonder if results would still be the same.
In addition I would like to see efficientie when just being in Windows. How much is your card using for no reason at all.
Thank you so much for the time and effort put into all these videos!
I would love to see locked FPS normalized testing. Modern high end GPUs are super power hungry when maxed out but do they use more or less power when running the same workload as a mid-tier GPU?
Excelente vídeo! Estou torcendo pra Intel e AMD avançarem cada vez mais
How old is that keyboard and mouse on the desk behind you?😎😎
Hahaha, I'm sure someone can chime in with that!
Looks like a Logitech 3 three button mouse.
Most awaited review by me. Thanks, Steve, and keep up the good work!
As a console gamer who's locked to 1080p 30 fps. All these cards seem fine to me. Probably would get the regular 4070 since it does up to 1440p really well.
Truly was a "Tie"
Why no 3090 Ti? Would be good to know if 4070/4070 Ti Super finally matches/outperforms the 3090 Ti in 4K
The 7900XT is finally a good value. I've been very pleased with the MERC edition in one of my builds.
Yeah, this Super refresh along with AMD price drops just managed to look RX7900XT even better. That is of course unless you love raytraycing. Then Nvidia got you hooked and you got to have their GPU. In reality this means mostly just Alan Wake 2 and CP2077 though. AW2 I can enjoy withou rt and CP2077 was such a mega letdown when it came, that I'm no longer interested in it despite the game being somewhat fixed year after release. Plus Nvidia gpus have surprisingly good results in CP2077 which indicates there were some specific optimalisations done for team green.
@@Hombremaniac Even if you love ray tracing and frame generation, Nvidia's paltry 12gb VRAM won't get you very far, unless you want low textures with performance upscaling.
@@DragonOfTheMortalKombat I´ve bought RX7900XT 2 months ago, so before any Super refresh rumours. I´m happy that my GPU was not made obsolete by 4070ti Super and I can live with the fact that RX7900XT is now some 60eur cheaper where I live (Czech rep).
@@DragonOfTheMortalKombat but the TI Super has 16GB
I recently upgraded from my RTX 3070 to a 4070 Ti Super. I'm not disappointed, the FPS increase is remarkable (ultra settings on most games, 1440p).
I've had some problems with the RTX 3070 with my 3 monitors (I use this computer also to work). When multiple fullscreen applications where running, I had some fps drops on the main gaming monitor (165 Hz). But now, everything is really fluent.
However, I do have flickering sometimes with HDR on the monitors enabled when starting or exiting games (mostly non-HDR games, I think). May be a monitor thing or driver-related, no idea. It flickers like 2 or 3 times, like switching from HDR to non HDR and back. Also happens in benchmarks. Otherwise I'm really satisfied. But if you're on the 40xx generation already, I wouldn't bother getting it.
If you're searching for the 4070 already, just go for the 4070 Ti Super. It has the chip of the 4080 and the Ti Super can be found in the same price range as the 4070 and 4070 Ti. But it's beter than the other two.
I'm running it combined with an i7-13700KF, so don't bother getting an i9. The i7-13700KF is usually never above 60/65°C. I haven't seen my GPU going above 80°C in Cyberpunk 77 (again, 1440p, ultra, Raytracing enabled). But then again, I'm not playing on 4K and I don't want to.
would this be a good upgrade over the 3070 in ultrawide 1440p since 8gb of vram is struggling now at this res. Also planning on using frame gen so even more vram is going to be used
yes
Don't buy Super refreshes. Super refreshes are done only if a gen performed badly. You'll probably be regretting that decision in a few months
I second the yes. I don't know if this is the optimal upgrade but you most definitely need more VRAM and this is one of the options with the other big cards too 3090, 7900xt, etc.
3080 ti is enough for 1440p for now. It's not future proof but this card isn't either
@@ij6708-That would make the 1600, 2000, 3000 and 4000 series as "poorly performing" generations. Think before you speak.-
$1,000 for a 4080 is still way to much . greedy executives suck more than leaches
Well the same philosophy when it comes to Apple and bunch of commentary users who's trying their hardest to hide the truth. This is regarding on how cheap to mass produce their IPhone 14.
$1000 for a gpu is nothing. You use it for 5 years+
@@iwantgoat not a good argument. im still using my 2070 super that i bought new in 2019. it was around $500 if i remember right, and i felt like it was overpriced back then. greed with these companies has gotten so damn out of control i that havent wanted to build a new pc out of principle.
@@slumlord2625 Good argument. The only reason you are complaining is because you are poor. Most hobbies costs way way way more than gaming on a computer.
@@iwantgoat poor? nope, i have principles. my pc works just fine. i have no need for a faster video card right now. people that build a new pc with every release need to get a real hobby and touch grass
7900XT it is
Always was. AMD's marketing department are experts at eating their crayons and sniffing their glue, but once reality sets in Dr Su comes around and lays down the law. AMD is fighting hard to become TSMCs most liked customer. That means never cancelling wafer orders, no matter what. And it means never "defecting" to other foundries like Nvidia did last gen. So AMD will keep ad-hock adjusting prices to make sure they're the best value, and even adjust after that if the market is simply slow. I bought my 7950X3D at full price, and despite it being the best allround CPU on the market it has already dropped a pretty penny.
@@charmolettafranquestafiestayam Launching FSR3 in a broken state in two unpopular games is an example of what I mean when I say AMD is bad at marketing. Taking petty shots at Nvidia for exceeding 300W with Ampere, then later launching the 6950XT at 355W is another. AMD's marketing division excel at sticking their foot in their mouth.
I'm just looking at upgrading from my long standing 8700k / 1080ti ftw3 setup. It's served me so well... 7800X3D / 4070ti (maybe super) is my next long term build.
love the new efficiency view. another potential metric idea to help guide users to a graphics card: what share of games in the test set hit 30, 60, and 90 FPS at a certain settings threshold (resolution, graphics toggle, upscale/fg setting). it would help people target a card that enables the larger share of games at "at least" a certain playability level. Knowing how cards relatively stack is definitely helpful, but it's hard to know if an arbtrary card achieves playable performance across all or just some of the test suite. this may be a bad idea or a use case you all don't want to focus on. but thought i'd chuck it out there.
Hope this is the point where prices will finally drop…
Spoiler: they won’t.
Optimistic thinking but I’m a realist
Reminder that the AMD price drops are also just Temporary. And in some countries made no real difference, people will still go Nvidia. Why buy a 7900XT when you can get a 4070 Super for €200 less that performs about the same as a 4070Ti with an OC.
This honestly just shows what an amazing card the 4080 is, if only it was priced properly. It would be the contender to the best gpu Nvidia has ever launched. Very unfortunate that pricing has hit down so hard on something that would otherwise have knocked the 1080/ti off that throne
Its pricing is pretty ridiculous but being able to grab them for around $900 has been the norm for almost a year. It looks like they are still holding about that value so at least there is that.
@@John-df3lm Yep. Here in Denmark the 4080 price has recently dropped even further to 5-10% below what the 7900 XTX sells for, which is pretty crazy compared to what it was at launch. People got burned bad by the initial price
The naming scheme is dumb. 4070, 4075, 4077 would make much more sense than 4070, 4070 Ti, 4070 Ti Super
Thanks for the great review. The efficiency charts are a most welcome addition.
If I may suggest yet another metric, I think a fps capped test like you did with the CPUs would be a good addition as well. Limiting to 120fps, or 60 on very demanding games/settings, would give a good idea of power consumption on typical usage scenarios, and with higher gaps in the values themselves, specially in those demanding single player games where the highest FPS possible is not that much more useful than in competitive games.
Consider an example: if I'm interested in buying a RTX 40 series card and I'm going to play games at the top temporal and spatial resolution of 1440p@120fps, because that's the display I have, and I don't play competitive shooters, it'd be good to know the gap in efficiency between having a 4090 at more comparatively "relaxed" workload, or having, say, a 4070 Ti Super working near capacity while gaming at that max 120 fps.
I'm looking to pair the 4070 ti Super with an i5 13600k and 32GB DDR5 6000Mhz RAM is this a good pair or will I suffer bottle neck with it?
Should be good.
I feel like you guys tend to lean towards negative takes and have generally a toxic vibe to your videos. If something is doing really well you don’t bother reviewing it. If it’s something you can speak negatively about, you tend to prioritize those videos.
If you feel like it’s a negative take then don’t watch the whole video and don’t even watch the rest of what Steve has to say.
Wait for the 50 series
Why?
@@dialatedtechnology Value sucks, especially without frame gen, as obvioused by all their marketing shenanigans. Let the market cool and force them to offer something good.
@@C_C-u can't force nvidia to anything
They do what they want
Lol why would 50 serie value be better?
Delusion and "wait for next gen" is why it'll be cheaper
Great Kudos for the power efficiency part! Would like to see two further metrics in that department: minium FPS (0.1% ot 1%, whichever you usually work with) vs power consumption and (again minimum) FPS vs full system power consumption. As I am not 100% sure that the cards use the CPU/System RAM the same.
Pretty impressive how well the 7900xtx does and how close it gets to the 4090 🤔
Thanks for the fast review. Wanted to buy one, but went for a used 7900xtx instead. Same price, but as you showed much better performance. (Raytracing aside.) But now I need a new monitor...
The thumbnail says it all, although I usually watch most of what you put out there, I didn't watch this or any nVidia related content this last year. Thanks Steve !
Thanks for your review....I spent 10 hours benchmarking and ocing the 70 super for just gpu mining alone, about to start gaming rn...I understand how much time you guys spend on these reviews and thank you so much for the good work you do for the community.
25:00 - Why is the 4090 not listed in this benchmark?! Literally came here for this.
12:53 I choked on my lunch after the golden sample punchline. Thanks, Steve (i guess?!).
Looking at doing a build next month. Should I get the 4070 ti super or 7900xt
Great review. Thanks!