Quick Update: I just noticed that the RTX5080 FE doesn't use LM and the cooler is indeed a bit different from what GN said. I misunderstood this in Nvidias communication
I think you mean "from what GN said, it is indeed a bit different" - if so you could use a comma between "different" and "from", as right now the meaning is ambiguous and could also mean that GN said something that turned out to be incorrect :)
So you would buy a gpu that is 12% slower and has slower memory and none of the new AI stuff for the exact same price ?😂 I get people don’t like the 5080 but that’s just a stupid statement. Unless u can get a 4080 super for 800$ (you can’t) then the 5080 is the next best thing
@ if is available for MSRP. And it will most likely not be. FE will be very limited and sell fast. We will see tomorrow but I expect that most of available ones will be 1150+ with few MSRP placeholder which will be out of stock for months, some of them will never be available. Is this first GPU launch for you? Btw - all you get is these 12% and MFG. Faster memory? That's part of these 12%, it will not be faster tomorrow because of the new memory type. All "new AI" stuff is supported by last gen except MFG but even 2x is barely usable, from reviews I didn't get impression that something changed with 3x/4x. Even on 5090 with much higher base framerates it was so-so. So yes, if new cards end up being $200 more anyone smart will save that money by buying 4080s.
@@Frankguligan256depends normally the 5080 is going to cost more than the traditionally reduced prices of the prior gen, I mean if you want to buy a 5070ti for 5080 prices you do you
I like how hardware unboxed put it. Frame Gen doesn't increase performance, it makes visual smoother. That's a good distinction to make that I think should be widely adopted.
US$829 after rebates and before VAT in Germany (March 2024) seems like daylight robbery nowadays for a 4080; especially as the 1× fake frames are enough for me and the AI bubble kinda is already bursting.
@ Nope, the 40 series cards are already pretty much gone, Nvidia discontinued them last month, all the stock seems to be gone and you can pretty much only get them from 3rd party scalpers.
Why is that a problem? Like, it has the same price as the 4080/4080S had, and its 10/15% better, I dont see any problem here. Great buy if u have a 4070 series card or older gen.
@@CoffeeGameMovieI don't disagree , but I think it should have had the price taken down to $899 at most to make up for the very low uplift - as a new generation. Unless they are planning on launching the 60 series in 18 months time , where the real gen uplifts will be found.
Everything is relative. Sure, 10% faster than 4080 super is very lackluster, but its gonna be 110% faster than my 3070, so why should i really care? If the 5080 was faster than it is it would also cost more. Its already at a good price to performance. And yeah, no shit, if you have a 4000-series you obviously shouldnt upgrade.
I did the exact same thing at the exact same price in August 2023. Legendary purchase. If you have hotspot issues, I highly recommend repasting with PTM7950.
Awesome. I got my SAPPHIRE Pulse 7900xtx in October, open box from MC for 764.00 Drove 200 miles round trip and I will say it was worth it. I have a feeling the XTX is going to be sought after and go up in price after it's all said and done.
maybe you want to upgrade from a 30 or a 20 series card or equivalent, the 5080 is roughly the same price as the 4080S was which of course you cannot find anywhere at the release price now
Since the 40 series is no longer made this would be the only nvidia option. That being said, if you're playing anything 1440p and below the only "reasonable" answer for the top tier is 7900xtx.
Ofcourse people will buy them at msrp? They are decently priced for the performance. Sure, if it was 30% faster than the 4080 super it would be awesome, until you realise it would ATLEAST cost 1200usd instead of 1000usd. Im almost happy that the performance is lackluster, because that increases my chances of getting one at MSRP.
I have the feeling that the Nvidia consumer graphics card "team" is no longer a desirable position in the company. It's all about A.I. now. So the best talent is directed in that direction and what we as consumers see is the best efforts of a "B" team that's understaffed, ignored, and demoralized.
@ When they go 100% it will probably crash, lol Data centers are different from A.I. A.I needs more memory that actual raw GPU power anymore. We've had plenty of A.I power, we don't have enough memory for the data storage size for the models. Theres a reason why Nvidia's got a 96Gbs cards coming out. I'm pretty sure their rack mounts are in the terabytes of memory size now.
I probably have to accept there will never be another 1080ti and buy something :( edit: looking at all these charts, I think convinced myself to "I'd rather buy 7900xtx instead of rewarding Nvidia 5xxx generation"
Thats me! Will prob be pulling trigger on the 5090 and just play everything maxed out without any worries. Going forward I might just sell every generation instead of waiting 8 years like I did with my 1080ti so that my parts still have decent value and can save more when buying the next generation.
You should wait for the 9070XT benchmarks. So far, from the leaks, it looks like it's a bit slower in performance to the 5080 in raster and 5070 level in RT. It'll probably be good value if it's 549.
Looks like the leaks about 5090 being the best case scenario for generational uplift are becoming true. At this rate 5070 Ti and 5070 are going to be even worse in value.
There is a reason they dropped the MSRP $50 from 40 series...They will probably be 10%-15% better than the 40 series counterpart, which is completely ridiculous! Nvidia decided to stay with the same node they used for the 40 series and they are trying to use Multi Frame-gen to make up the difference...
@@brucepreston3927other than the 5070ti and lower, those msrps are fake, their is only 1 msrp card from msi,asus while all they're other cards start at +$200 from msrp, that means they'll have 1 -5 msrp pieces, while all their other cards will have the majority of the stock, so 99/100 will pay at least $200 more than m s r p
@@brucepreston3927 The MSRP on the founders cards doesn't matter that much, since it appears all the AIB cards are gonna be a lot more expensive this time around, and good luck getting an FE card, I have a feeling they're going to be out of stock all the time.
When the best case value for money means linear scaling vs the "theoretical MSRP" the 5000 series seems a little screwed. Maybe Nvidia will be more aggressive in the midrange since that is where AMD is competing. But there isn't much room between the 4070ti super and 5080. Mark my words, looking at the 5070 it will need some aggressive clocks to surpass the 4070 super.
7900xtx bad for streaming av1 sucks and on twitch this is worst. XTX consums more and bad techs but in raster this is the best GPU for the perf and price.
16:35 - Wait, what ? How is a next gen card that delivers barealy any perf improvement "fine" at the same price the previous card was going for ? Isn't the idea behind technological progress to get significantly more for the same price vs last gen or get the same perf at a significantly lower price ? This is barely a refresh-type of upgrade. If we believe that +10% uplift is fine for +10% higher price, then with each gen we are going to get more and more expensive cards but the price/performance metric is going to stay the same. How is that "fine" ? o_O
Generally because chip technologies are bringing less and less relative improvement every generation, and in this specific case because it's on (more or less) the same node as last generation. They're using X% more silicon consuming X% more power to get X% higher performance. 'Whee'.
ровно то же самое я говорю уже давно, примерно с выхода 3000 серии. единственный правильный пример работы этого вашего "технического прогресса" - это выход Паскалей, когда 1080ти стоила дешевле 980ти и была лучше во всем. а полный издец начался с выхода 2000 серии, когда дурачье схавало лучики и начало его пиарить, тем самым поставив крест на гетеросексуальных гтх-картах! все еще жду выход карты 1690 супер ти. или обновленную серию 2000-х карт: гтх 2065 или гтх 2085)
2025, you will be satisfied because we tell yo u to be satisfied. 2024 Gamers need to get used to not owning games 2023 the more you buy the more you save
Jensen needs to prioritize their higher margin and cash cow AI chips production so they greatly reduce gpu production and raised prices to make up for that reduction...
The PCIe connection is most probably the problem. There are really two things that can cause problems, the length of the signal path and signal contamination depending on the extra connector between the PCIe riser and the actual graphics card. That connector is a lot smaller than the PCIe connector on the motherboard. The pins are tiny and packed closer together. this means the signal quality is likely to suffer. But it's probably also very dependent on the signal quality of the PCIe slot on the motherboard. So these problems might be more common when using some certain make and model of motherboards. So the PCIe "raiser" on the FE cards can be a problem, and it might be different between cards. And the PCIe signal quality of the motherboard might be causing these problems. Someone will surely figure out what is really causing the problems.
4080 Super Boost or 4080 Super+ kinda like the 650 Ti Boost. This thing also has the same performance as the 4090 D which should have been called 4080 Ti.
I had a similar issue like this on my X79 system when PCIe 3.0 was beta qualified for my Sabertooth board. It took a few BIOS revisions to my the HD7970 play nice, especially when I picked up another one for Crossfire a few months later. PCIe 2.0 was a stability option to use but eventually a few BIOS updates later it worked as expected. Hopefully that's all that is needed here.
If I recall correctly, I beleive he got some prop used on sets.. Like some kind of golden head.. Blurry memory sorry Edit: Yeah it was on a video 2y ago!
Bad cooling performance might be due to a cheaper cooler being used, e.g. vapor chamber vs copper plate like they did with the GTX 1080 FE and GTX 1070 FE. And yeah having PCIe Gen 5 running through a flimsy ribbon cable and tiny connector right next to a 100°C heat source is terrifying. Even if it works now, what happens if the tiny contacts start degrading after constant heat cycles?
Anyone who got a 4090 near MSRP a couple years ago apparently got something of a deal. What a GPU market. Poor 10 series hangers-on waiting to upgrade and now starting to be between a rock and a hard place.
I can't believe I got a 4090 for $1600 by accident. Seems like a great deal. If I found a 5090 for $2000, and nothing more, I would consider it but even at MSRP 5000 series looks meh.
Been using my base 4080 for 2 years now and am very happy with it. Got it for 980 usd equivalent before taxes. It was 1800 SGD after conversion (1350 USD after 29% taxes)
4080 was terrible price/performance but it was perfect tdp/performance for me, got it late 2023 when the price dropped to MSRP undervolted and capped to 144hz it's usually 200-250W, only CP2077 goes to 270-280W so pretty happy with it 4080S never was 999 in Europe anyway, it was 1150-1250 so pretty much equal to 4080 1199 MSRP and what I paid for mine
@@HeadBassVTEC Totally agree. Never would have bought the 4080 if it wasn't for the offer for only $850. Even though it was the super version. Did it 4 times)
@@HeadBassVTEC I just run my MSI RTX 4080 Gaming X Trio with stock settings as an upgrade from my barely usable 2070 Super which was getting annoyingly slow esp for MSFS 2020 and wukong due to VRAM limitations
Thank you for making it clear the 4080 was also tested on the exact same rig. Can i expect those 1% low gains on the 5080 ( over a 4080 ) on a 7800x3d, i understand the lows will be lower but will that number still scale roughly ?
In a 4080 + 7800x3d vs a 7800x3d and 5080 situation yes, youll see the same imrpovement. Just dont compare to his 1% because its different due to his R7 9800x3d
I bought a gigabyte 5080 gaminc oc and... crash after crash on valorant (that is the game i basically play)! The most i managed to do was a deathmatch... I updated MB bios, installed fresh drivers from nvidia; even re-installed valorant 2 times! Then i remembered i saw this video concerning gen 5 PCIe issues... Just watched your video again! Will check bios config... Very frustrating!
Thanks for the review! Very interesting that the 5th gen PCI-E is causing so many issues. Seems strange that nvidia went for it considering how little there was to gain, but I guess every percentage unit counts when you're doing a minimal performance gain generation. I wasn't expecting there to be this level of issues from the beginning with it, more down the line as the connections and screws get more lax over time etc. Less than two minutes in, and I noticed that you're comparing different cards for % gains (4080 STRIX / 4080 Aorus). I don't think that's very representative, and it seems weirdly lazy to do that. Maybe we'll see later on that both of those cards perform exactly the same? - Edit: no we didn't. Also, there are a lot of unnecessary "NVIDIA FE" texts on the Cyberpunk RT + PT, and SW Outlaws 4K Max charts. Maybe an idea you were playing with but forgot to remove due to being in a hurry? :) Again, thank you for the review and for sharing the struggles, those do matter a lot for anyone considering purchasing these.
Thanks for the hint for solving the PCIe Issue. I just got a gigabyte rtx 5080 gaming oc. And it was not work on my Asus ROG X870e motherboard. Changing to the PCI to Gen4 make it run. But the mobo is started to running in a loop showing different Q-Code in the range 30-70. :-| Expect there will pop something up on the internet in the upcoming days if it is related to the RTX 5080.
@SansAppellation My bad, what I mean is he did the review in Ultra Wide. So if you watch it on an Ultra Wide monitor it takes up the whole monitor instead of having borders on the side like Linus's review.
Was having the same issue when I installed my 5080 FE, in my case I was using a gen 4 riser, went into the bios and changed from Auto to gen 4 and it fixed all my problems.
The RTX 5080/5090 will be gone a few minutes after launch as thousands of scalpers would be buying them by the thousands and selling them for $5K+ each
Hi Roman, i read your post mentioning you are working on new RTX 5090 waterblocks both for the Asus and Nvidia reference PCBs. Can you bring the gap over the Vram down to 0.5mm to reduce thermal resistence? GDDR7 seem to scale exceptionally well, i guess keeping them at ambient temp could give them an additional boost. Bye.
You're the best! Until seeing your video, i was going to buy the 5080 FE but not now. Does the 5090 FE also have the Pcie 5 problem? I need the 50mm dimension for the new Antec pro 1M case (which Amazon will now ship on 19 Feb - order date was Jan 6).
Another great review, and another nail in the coffin for PCIe 5 having any use or benefit whatsoever on consumer platforms. All it does is drive up prices. Yay, my box has a higher number on it, and my SSD can wow me with benchmark numbers with zero real world benefit, wooo! Such wow, much excite. Just another sign of how the tech industry has been slowing down in the past decade or so (which, to be clear, isn't a bad thing it's just how things go as technologies mature and markets saturate), and how desperate they are to force a continuation of their previous levels of growth. This, crypto, NFTs, VR, AR, AI, none of it is the panacea that the industry wants or wanted them to be. And no new idea will be.
I'm learning that one just can't upgrade with every new generation, and one has to choose very carefully when to upgrade. The RTX40x0 cards featured a new node, and was a great choice in terms of performance per watt. The Intel 13th gen was a good CPU, whilst the current issue is disappointing. Based on DerBauers review, I bought a Inno3D water cooled 4090, and now that seems like a good choice. Super quiet obviously and not a day of trouble.
The 5080 is literally the worst generational performance improvement of the last decade... maybe even the last 2 decades. The performance is in line with what would be expected of a 70 class gpu. So instead of increasing the price, nvidia shifted the performance tiers again. The 5080 is actually slower than the 4090... For the first time in the last decade the new 80 class cannot beat the previous gen flagship. This is 70 class performance. This is a 600$ GPU for 1000$. I've seen multiple reviews now where even the 7900XTX in some titles actually beats the 5080. That is the 5080 getting smashed by the PREVIOUS GENERATION flagship of the competitor... I would be ashamed if I put out a product this bad for this much money. I'm actually surprised Nvidia hasn't already "unlaunched" this product. Multi framegen is a joke as a feature. On the high end, you don't really need it, on the low end, the use is limited because most of the time your base framerate is too low. There has literally never been a larger gap between the 80 class and flagship gpu. For 1000$ this card should be avoided at all cost
People often compare this to the previous gen, and say it's not worth upgrading... big surprise. However if you were to recommend a High End (1440p) gaming card right now for a completely new build, which would you recommend? The 5080 doesn't seem a bad purchase today, especially if the new frame gen and other features get good support in the future.
In that side by side comparison at 13:30 are those max power readings right???? 855w for the 5080 and 2057w for the 5090????? WTF, tell me those are some sort or sofware error and not real transient spikes.... Or maybe I'm misinterpreting the readings and that is not the actual power drawn.
@@mycosys The 5090 AIB cards will be £2500 or so, you watch. There's no way that's going to be good value, I'm waiting until AMD's next flagship. Nvidia is taking the piss honestly.
@@mirarid Nvidia is taking the piss? Do you not understand that the whole reason why nvidia has been able to do this is because AMD has been terrible on the gpu side they cant even get their drivers right lol
The reviews for the 5080 are if anything making the 7900xt and xtx look good by comparison. At least in anything that doesn't involve a bunch of ray tracing.
Just food for thought from a sample size of one. I have had nothing but problems with what i think is both pcie gen5 and ddr5 on my msi b650-p motherboard with a 7700x and 1 16gb stick of ram running whatever minimum default spec is like 4300 or whatever. The board is super not stable with expo and two ram sticks. Gpu likes to disconnect and crash all the time in addition. Its a gtx 1080. Forcing pcie gen down to 3 on everything has helped a lot. I really think signal integrity on both pcie gen 5 and ddr5 is hell on low and mid range motherboards. Wouldn't surprise me if that extra pcb is effecting signal integrity. These higher speed interfaces just dont leave as much room for error in design and it seems a fair number of manufacturers are having issues getting it right. I really dont want to spend $500 on a motherboard to hope for better stability but it seems that is going to be the way going forward.
I encourage you guys to watch frame chasers video he put up today. Blackwell overclocks way better than ada. With less voltage it goes to 3300 mhz and shows almost 4090 oc performance once you do so. Take everything with a grain of salt. Blackwell is better architecture, faster clock speeds, faster ram, and faster bus speeds. firmware hasn't been optimized yet, but when it does, it's gonna be much better than these entry tests that are base tests.
Uhuh and how much of the population that buys these GPUs bothers with extreme OCing ? And as the other guy said, wonder what could be the reason for having such large headroom, definitely not for a Super refresh 12 months later.
No surprise that the partner cards built the "normal" way have less issues with PCIe 5 signal integrity than the FE which pretty much has riser built in as you said. And PCIe 5 is way more sensitive to signal integrity, just see how much more expensive/better built are the PCIe 5 risers and U.2 drive adapters vs PCIe 4 or even PCIe 3 ones.
I have 2070s. I was going to build new rig with 5080... Going to re-evaluate my options and wait for amd. Let's hope they deliver!!! I have my doubts with them too 😢 don't want to spend on 5090
To be able split 8x 5.0 for graphics card and 8x 4.0 for fast NIC on AM5 was only reason for me to buy. May be models from AIB partners will be better.
I have the same problem with MSI Tomahawk x870 and MSI 5080. On PCI-e 5.0 I have lags and stutters every 2-5 seconds, system just freezes for a second every time. Launching games = automatic PC restart. After changing to 4.0 mode - everything works well. Had no problems with "Auto" setting with my GTX 1070.
how can you justify price increases based on pure performance gains? Where are the days of generational improvements at the same/marginally higher price?
Yeah they removed the hotspot sensor in the 50 series for some reason, so right now the software is picking up a null signal. Should be an update to remove that metric from 50 series cards soon (it's in beta)
i would hypothesize that this is caused by the weird built-in pcie "adapter" that plugs in to the pcb. whether its a physical contact/seating issue, internal pcb flexion, or just the break in the traces for the connector...something is degrading the signal integrity
I also had a poblem with some 8x pcix gpus like 4060 ti... that was not recognized by the mb when in bios the lanes was set to auto. I had to set it to 8x + 8x to get it recognised ... The coincidence is that it was an amd am5 mobo.
14:59 360 and 575 watts is with the vram, and both cards don't have the same amount. So you can't do this simple division You have to substract memory power first
Honestly, I'm almost glad it's not so much of an improvement. As somebody who's looking to upgrade from my hot and loud 3080Ti to 5000 series, it seems like I won't have a ton of competition from people looking to upgrade from their 4000 series cards - and that should hopefully keep the prices down somewhat. Not holding my breath for picking on up at MSRP any time soon, but I can dream.
Same situation for me. MY 3080 Ti is already gone, so I need one of these one way or another, but I'm not paying over MSRP, and I'm sure as heck not paying over what the 4080 Super was at $999. The 5080 is not impressive.
I upgraded to a 4070Ti Super when my old 1080 could barely run FF16, a game I had been looking forward to for a while. I felt a bit of buyer's remorse in the last month or two, thinking maybe I should have been patient and waited for the 5000 series. Nowadays, I don't feel so bad lol
I had maybe same problem with Asus TUf B550 board and EVGA rtx3070. Board wasn´t cooperating, and I have to run PCI-e 3.0 x 16 or PCI-e 4.0 x 8 . After change for MSI board it was OK. It is AMD 😕 I never had this problem on Intel platform. Or Asus problem :D (It wouldn´t be first time, right) Did you have same power consumption? In my case I had feeling that something is "jamming" data signal going through PCI-e. And looking on fps and power consumption it was like GPU hasn´t data to compute.
*The 5090 and 5080 have a huge gap bwcause there will be a 5080 Ti releasing in October. 24GB of GDDR7 and a 384-bit bus width.* *Nvidia knows people will not be happy with the 4080 Super and 5080 gap. It's called upselling. It forces people to buy the next tier up.* *The 5080 would be more like a 4080 Ti SUPER. Consider the Ti usually only has an 11% fps gain over the non Ti variant in games. But the Ti SUPER would have a 16% increase in fps performance over the non Ti or SUPER variant.* *When you do the math, the 5070 Ti matches the 4080 SUPER. The 5080 matches what would be a 4080 SUPER +. (Aka a Ti SUPER). The 5090 is equal to what WOULD be a 4090 Ti.*
Quick question for anyone who knows. IF the GPU TDP is 500W are we really dissipating 500W of heat? I don't think that can be true. If it were true the card would be totally inefficient. If I understand correctly the heat dissipated by the card cannot be 500W it must be lower if the card is actually using some of the power to actually do the work. So the heat that is dissipated is waste heat from it being inefficient. So how much power is it actually using to do work and how much power is just wasted as heat?
Can't wait untill they start making it blatant and just release cards with the same exact performance as a previous iteration but tack on a mystery technology that you don't need or more fake frames.
Looking to upgrade from a 1060 and my local microcenter has had people camping for 4 days now. It's the day before launch. All I can do is be prepared and fight like hell. lol
NVIDIA fai. The proof is in the fact the partner cards don’t have the problem. I can see the future. I seen a lot of 50 series founders failing prematurely due to the pcb extension slot.
Is the 50 series more optimized towards the Transformer DLSS model? Because on my 4090 Transformer uses more power compared to CNN, worth about one step on the quality slider in CP77. So Transformer Performance eats about the power of CNN Quality.
@b127_1 Whatever happens under the hood, the new model uses as much power as the old model at 1 step higher quality. And the new model currently introduces very serious (dithering?) noise in some parts.
it seems that there is no specific optimisation for the Transformer model. And the new model is about 4-5% slower than the older one. at least it looks better.
@@mariuspuiu9555 Whether it looks better had to be compared based on its performance/power usage. Currently it seems to me like I have to compare Transformer to CNN by pushing CNN's quality slider 1 step higher!?
you should test with intel series like 12900ks or 12900kf with z690 with pcie 5.0 motherboard, i think some z690 have pcie 5.0 to make sure if there no problem in intel lga 1700
so what i have been thinking is what if the connection between the pci and board of the 4080 has some drop out. and with them makeing the bus lower than the 5090 is what is causing it. cause if the 5090 is not having an issue but the 5080 is haveing issues then what has changed everything is their but the bus speed.
SOOOO happy i bought a 4090. Got 2 years of the worlds best card, and now i get at least 2 more years of the world's 2nd best card and all for 1400 which isn't bad spread over 4-5 years.
Quick Update: I just noticed that the RTX5080 FE doesn't use LM and the cooler is indeed a bit different from what GN said. I misunderstood this in Nvidias communication
FYI GN isn’t always accurate
I think you mean "from what GN said, it is indeed a bit different" - if so you could use a comma between "different" and "from", as right now the meaning is ambiguous and could also mean that GN said something that turned out to be incorrect :)
I doubt gn would say it's not liquid metal and open it to check before declaring such a statement
@@itsuadmanSteve explicitly said the 5080 doesn't have LM, and that it has some changes on the cooler, OP is probably just missing a comma there.
So can you confirm that moving to PCIe 4.0 has made it fully stable, or did you see issues at 4.0 too?
Nvidia has already taken care of that "you can just buy 4080s instead" advice and removed them from the market a month ago.
There are a couple on Amazon for insane prices.
Plenty still here in Australia at same prices. Not sure how much longer for though after these Meh reviews.....
So you would buy a gpu that is 12% slower and has slower memory and none of the new AI stuff for the exact same price ?😂 I get people don’t like the 5080 but that’s just a stupid statement. Unless u can get a 4080 super for 800$ (you can’t) then the 5080 is the next best thing
@ if is available for MSRP. And it will most likely not be. FE will be very limited and sell fast. We will see tomorrow but I expect that most of available ones will be 1150+ with few MSRP placeholder which will be out of stock for months, some of them will never be available. Is this first GPU launch for you?
Btw - all you get is these 12% and MFG. Faster memory? That's part of these 12%, it will not be faster tomorrow because of the new memory type. All "new AI" stuff is supported by last gen except MFG but even 2x is barely usable, from reviews I didn't get impression that something changed with 3x/4x. Even on 5090 with much higher base framerates it was so-so.
So yes, if new cards end up being $200 more anyone smart will save that money by buying 4080s.
@@Frankguligan256depends normally the 5080 is going to cost more than the traditionally reduced prices of the prior gen, I mean if you want to buy a 5070ti for 5080 prices you do you
I like how hardware unboxed put it. Frame Gen doesn't increase performance, it makes visual smoother.
That's a good distinction to make that I think should be widely adopted.
considering the variety of reviews and games w/in those reviews, I would agree for the most part.
In fact, it can make performance worse if you cap FPS to your monitors max (like 120 or 165, not so much for 240 models).
HU does good work. Digital Foundry goes way deeper if you want to go down that rabbit's hole
I mean does it matter if I get an enjoyable experience?
I also like their point that it's like an advanced form of motion blur that doesn't suck.
Buying a 40 series card a few months ago seems smart now
(STRIX 4070 Ti Super OC for $900 US)
US$829 after rebates and before VAT in Germany (March 2024) seems like daylight robbery nowadays for a 4080; especially as the 1× fake frames are enough for me and the AI bubble kinda is already bursting.
Its more smart to buy after 5xxx series launch.
4080S seems great value atm
@@chronyk743
1) availability
2) tariffs
@ Nope, the 40 series cards are already pretty much gone, Nvidia discontinued them last month, all the stock seems to be gone and you can pretty much only get them from 3rd party scalpers.
this is why they discontinued the older 40x0 sereis cards to force us to buy this crap
crap is right
Why is that a problem? Like, it has the same price as the 4080/4080S had, and its 10/15% better, I dont see any problem here. Great buy if u have a 4070 series card or older gen.
@@CoffeeGameMovie because those older 40x0 cards would be more than 10-15% cheaper new.
@ cause it’s a new generation not a ti launch and it’s garbage
@@CoffeeGameMovieI don't disagree , but I think it should have had the price taken down to $899 at most to make up for the very low uplift - as a new generation. Unless they are planning on launching the 60 series in 18 months time , where the real gen uplifts will be found.
Thumbs up for honest reviews. I'm definitely having second thoughts on 5080 after seeing the data today. It just doesn't seem so great.
This gen looks like a big skip. Without a node shrink or real architectural overhaul the gains are just not here, 5090 included.
Depends what you are coming from. Lol awkward. If you have a 4080 or 4080s this card was never for you anyway. Lol
Roman and gamers nexus do some of the best reviews and both are not impressed at all.
Don't worry, Jensen said the 5070 is the fast card.
Everything is relative. Sure, 10% faster than 4080 super is very lackluster, but its gonna be 110% faster than my 3070, so why should i really care? If the 5080 was faster than it is it would also cost more. Its already at a good price to performance. And yeah, no shit, if you have a 4000-series you obviously shouldnt upgrade.
The PCIe 5 issues are very interesting. Well done for investigating those in detail.
Just picked up a 7900xtx open box from microcenter for $720. Looks like that was a deal. Passed 30+ people camping out for 5000 series
I did the exact same thing at the exact same price in August 2023. Legendary purchase. If you have hotspot issues, I highly recommend repasting with PTM7950.
Micro Center is the GOAT
Xtx still has issues with some game engines..
@@GlennsHardWired 7900 XTX is my first AMD GPU and most stable GPU I ever owned.
Awesome. I got my SAPPHIRE Pulse 7900xtx in October, open box from MC for 764.00 Drove 200 miles round trip and I will say it was worth it.
I have a feeling the XTX is going to be sought after and go up in price after it's all said and done.
I hope scalpers get these cards and get stuck with them because I can't see people even wanting to buy these at msrp 😂😂😂
maybe you want to upgrade from a 30 or a 20 series card or equivalent, the 5080 is roughly the same price as the 4080S was which of course you cannot find anywhere at the release price now
Since the 40 series is no longer made this would be the only nvidia option. That being said, if you're playing anything 1440p and below the only "reasonable" answer for the top tier is 7900xtx.
Ofcourse people will buy them at msrp? They are decently priced for the performance. Sure, if it was 30% faster than the 4080 super it would be awesome, until you realise it would ATLEAST cost 1200usd instead of 1000usd. Im almost happy that the performance is lackluster, because that increases my chances of getting one at MSRP.
@@garypinholster1962 thats delusional. DLSS is still extremely worth it at 1440p.
Super easy to sell it to EU for big profit.
I have the feeling that the Nvidia consumer graphics card "team" is no longer a desirable position in the company. It's all about A.I. now. So the best talent is directed in that direction and what we as consumers see is the best efforts of a "B" team that's understaffed, ignored, and demoralized.
Compared to 2019, AI now accounts for 78% (from 25%) of Nvidia's revenue while gaming only 17% (from 53%).
@@jaywye Theres always the golden rule of never putting your eggs in one basket.
Nvidia will learn later one day.
Do you think data center profits are going anywhere? lol I'd put that much in ai too if I was the cutting edge tech leader atm
@ When they go 100% it will probably crash, lol Data centers are different from A.I. A.I needs more memory that actual raw GPU power anymore. We've had plenty of A.I power, we don't have enough memory for the data storage size for the models. Theres a reason why Nvidia's got a 96Gbs cards coming out. I'm pretty sure their rack mounts are in the terabytes of memory size now.
I hope that Deepseek will make them reconsider
I probably have to accept there will never be another 1080ti and buy something :(
edit: looking at all these charts, I think convinced myself to "I'd rather buy 7900xtx instead of rewarding Nvidia 5xxx generation"
Thats me! Will prob be pulling trigger on the 5090 and just play everything maxed out without any worries. Going forward I might just sell every generation instead of waiting 8 years like I did with my 1080ti so that my parts still have decent value and can save more when buying the next generation.
@@saiyaman9000 No, that isnt you.
You should wait for the 9070XT benchmarks. So far, from the leaks, it looks like it's a bit slower in performance to the 5080 in raster and 5070 level in RT. It'll probably be good value if it's 549.
@@HypoIReMIA Rasterized is a dying breed.
@@TheDeeGeeNL dude there are only a few titles requiring RT and everything else is literally rasterization, I wouldn't call it a dying breed...
Looks like the leaks about 5090 being the best case scenario for generational uplift are becoming true. At this rate 5070 Ti and 5070 are going to be even worse in value.
1day Review too on 5070.. says all
There is a reason they dropped the MSRP $50 from 40 series...They will probably be 10%-15% better than the 40 series counterpart, which is completely ridiculous! Nvidia decided to stay with the same node they used for the 40 series and they are trying to use Multi Frame-gen to make up the difference...
@@brucepreston3927other than the 5070ti and lower, those msrps are fake, their is only 1 msrp card from msi,asus while all they're other cards start at +$200 from msrp, that means they'll have 1 -5 msrp pieces, while all their other cards will have the majority of the stock,
so 99/100 will pay at least $200 more than m s r p
@@brucepreston3927 The MSRP on the founders cards doesn't matter that much, since it appears all the AIB cards are gonna be a lot more expensive this time around, and good luck getting an FE card, I have a feeling they're going to be out of stock all the time.
When the best case value for money means linear scaling vs the "theoretical MSRP" the 5000 series seems a little screwed.
Maybe Nvidia will be more aggressive in the midrange since that is where AMD is competing. But there isn't much room between the 4070ti super and 5080.
Mark my words, looking at the 5070 it will need some aggressive clocks to surpass the 4070 super.
We finally got the 4080 Ti.
Great review, again!!! Thank you!
This only shows how great 7900xt/x are, and how much underestimated and unappreciated they were.
*Pats 7900xtx *
What?! No.
7900xtx bad for streaming av1 sucks and on twitch this is worst. XTX consums more and bad techs but in raster this is the best GPU for the perf and price.
16:35 - Wait, what ? How is a next gen card that delivers barealy any perf improvement "fine" at the same price the previous card was going for ? Isn't the idea behind technological progress to get significantly more for the same price vs last gen or get the same perf at a significantly lower price ? This is barely a refresh-type of upgrade. If we believe that +10% uplift is fine for +10% higher price, then with each gen we are going to get more and more expensive cards but the price/performance metric is going to stay the same. How is that "fine" ? o_O
Generally because chip technologies are bringing less and less relative improvement every generation, and in this specific case because it's on (more or less) the same node as last generation. They're using X% more silicon consuming X% more power to get X% higher performance. 'Whee'.
@@RN1441 Only because these companies are chasing the next bubble instead of throwing large amounts of capital at innovation.
Nvidia has milked each generation more and more over time. Being super rich is not enough for CEOs these days. They want unlimited money.
ровно то же самое я говорю уже давно, примерно с выхода 3000 серии. единственный правильный пример работы этого вашего "технического прогресса" - это выход Паскалей, когда 1080ти стоила дешевле 980ти и была лучше во всем. а полный издец начался с выхода 2000 серии, когда дурачье схавало лучики и начало его пиарить, тем самым поставив крест на гетеросексуальных гтх-картах! все еще жду выход карты 1690 супер ти. или обновленную серию 2000-х карт: гтх 2065 или гтх 2085)
2025, you will be satisfied because we tell yo u to be satisfied.
2024 Gamers need to get used to not owning games
2023 the more you buy the more you save
Jensen needs to prioritize their higher margin and cash cow AI chips production so they greatly reduce gpu production and raised prices to make up for that reduction...
hopefully deepseek makes Nvidia crash and burn due to the overinvestment in AI chips
The PCIe connection is most probably the problem. There are really two things that can cause problems, the length of the signal path and signal contamination depending on the extra connector between the PCIe riser and the actual graphics card. That connector is a lot smaller than the PCIe connector on the motherboard. The pins are tiny and packed closer together. this means the signal quality is likely to suffer. But it's probably also very dependent on the signal quality of the PCIe slot on the motherboard. So these problems might be more common when using some certain make and model of motherboards. So the PCIe "raiser" on the FE cards can be a problem, and it might be different between cards. And the PCIe signal quality of the motherboard might be causing these problems. Someone will surely figure out what is really causing the problems.
If you want to go the extra mile, you can use the PCIe adapter from the 5090 on the 5080 (if they are the same) to see if the issue persists.
16GB for $1000 is stupid.
It does not even deserve to be called a Ti
4080 Super Boost or 4080 Super+ kinda like the 650 Ti Boost. This thing also has the same performance as the 4090 D which should have been called 4080 Ti.
its 4080 MFG edition
I had a similar issue like this on my X79 system when PCIe 3.0 was beta qualified for my Sabertooth board. It took a few BIOS revisions to my the HD7970 play nice, especially when I picked up another one for Crossfire a few months later. PCIe 2.0 was a stability option to use but eventually a few BIOS updates later it worked as expected. Hopefully that's all that is needed here.
I've only just noticed the 'Stargate' tattoo
If I recall correctly, I beleive he got some prop used on sets.. Like some kind of golden head.. Blurry memory sorry
Edit: Yeah it was on a video 2y ago!
I did recently too, Hes a huge stargate fan, Bought multiple props from the show, even made a video on unboxing them.
Yes and last year had even the chance to meet Richard Dean Anderson. That was absolutely amazing.
@@Asdosie That makes the both of us. I am a fan of Stargate too.
@@der8auer-en In the middle of my backswing!
Bad cooling performance might be due to a cheaper cooler being used, e.g. vapor chamber vs copper plate like they did with the GTX 1080 FE and GTX 1070 FE.
And yeah having PCIe Gen 5 running through a flimsy ribbon cable and tiny connector right next to a 100°C heat source is terrifying. Even if it works now, what happens if the tiny contacts start degrading after constant heat cycles?
Possible damage during transport to the internal “riser” for the PCI-E issue?
Anyone who got a 4090 near MSRP a couple years ago apparently got something of a deal. What a GPU market. Poor 10 series hangers-on waiting to upgrade and now starting to be between a rock and a hard place.
They had their chance to upgrade and chose not to
Going from 10 series to any 40 or 50 series is a huge upgrade
I can't believe I got a 4090 for $1600 by accident. Seems like a great deal. If I found a 5090 for $2000, and nothing more, I would consider it but even at MSRP 5000 series looks meh.
Bought one in 23 and im a happy camper, the release of the 50 series just made the 40 series stronger 💪
11:56 Im pretty sure GN reported theyre not the same and the 80 uses regular paste, not liquid metal
You are 100% correct!
Been using my base 4080 for 2 years now and am very happy with it.
Got it for 980 usd equivalent before taxes.
It was 1800 SGD after conversion (1350 USD after 29% taxes)
4080 was terrible price/performance but it was perfect tdp/performance for me, got it late 2023 when the price dropped to MSRP
undervolted and capped to 144hz it's usually 200-250W, only CP2077 goes to 270-280W so pretty happy with it
4080S never was 999 in Europe anyway, it was 1150-1250 so pretty much equal to 4080 1199 MSRP and what I paid for mine
@@HeadBassVTEC Totally agree. Never would have bought the 4080 if it wasn't for the offer for only $850. Even though it was the super version. Did it 4 times)
@@HeadBassVTEC I just run my MSI RTX 4080 Gaming X Trio with stock settings as an upgrade from my barely usable 2070 Super which was getting annoyingly slow esp for MSFS 2020 and wukong due to VRAM limitations
Thank you for making it clear the 4080 was also tested on the exact same rig.
Can i expect those 1% low gains on the 5080 ( over a 4080 ) on a 7800x3d, i understand the lows will be lower but will that number still scale roughly ?
In a 4080 + 7800x3d vs a 7800x3d and 5080 situation yes, youll see the same imrpovement. Just dont compare to his 1% because its different due to his R7 9800x3d
I bought a gigabyte 5080 gaminc oc and... crash after crash on valorant (that is the game i basically play)! The most i managed to do was a deathmatch... I updated MB bios, installed fresh drivers from nvidia; even re-installed valorant 2 times! Then i remembered i saw this video concerning gen 5 PCIe issues... Just watched your video again! Will check bios config... Very frustrating!
I've got the same problem with Gainward phantom 5080. Freezing and crashing while gaming. Switching to pcie4 didn't help in my case
It seems to be an iracing issue - limited FPS on and stops freezing. We will see
Edit
Or it depends on FPS - freezes were over 200 fps
Thanks for the review!
Very interesting that the 5th gen PCI-E is causing so many issues. Seems strange that nvidia went for it considering how little there was to gain, but I guess every percentage unit counts when you're doing a minimal performance gain generation. I wasn't expecting there to be this level of issues from the beginning with it, more down the line as the connections and screws get more lax over time etc.
Less than two minutes in, and I noticed that you're comparing different cards for % gains (4080 STRIX / 4080 Aorus). I don't think that's very representative, and it seems weirdly lazy to do that. Maybe we'll see later on that both of those cards perform exactly the same? - Edit: no we didn't.
Also, there are a lot of unnecessary "NVIDIA FE" texts on the Cyberpunk RT + PT, and SW Outlaws 4K Max charts. Maybe an idea you were playing with but forgot to remove due to being in a hurry? :)
Again, thank you for the review and for sharing the struggles, those do matter a lot for anyone considering purchasing these.
your channel has great review with many details with unbiased . keep up the good work 👍☝👍
Thanks for the hint for solving the PCIe Issue. I just got a gigabyte rtx 5080 gaming oc. And it was not work on my Asus ROG X870e motherboard. Changing to the PCI to Gen4 make it run.
But the mobo is started to running in a loop showing different Q-Code in the range 30-70. :-| Expect there will pop something up on the internet in the upcoming days if it is related to the RTX 5080.
These ultrawide reviews are awesome by the way!
Where was the ultrawide data? Scrubbing but can see anything
@SansAppellation My bad, what I mean is he did the review in Ultra Wide. So if you watch it on an Ultra Wide monitor it takes up the whole monitor instead of having borders on the side like Linus's review.
@@GregoryShtevensh ah, gotcha! Thanks.
Your graphs are very nicely done. That must have taken a lot of work.
Was having the same issue when I installed my 5080 FE, in my case I was using a gen 4 riser, went into the bios and changed from Auto to gen 4 and it fixed all my problems.
After buying my 4080S last Dec, this is great. 😅
The RTX 5080/5090 will be gone a few minutes after launch as thousands of scalpers would be buying them by the thousands and selling them for $5K+ each
Hi Roman, i read your post mentioning you are working on new RTX 5090 waterblocks both for the Asus and Nvidia reference PCBs. Can you bring the gap over the Vram down to 0.5mm to reduce thermal resistence? GDDR7 seem to scale exceptionally well, i guess keeping them at ambient temp could give them an additional boost. Bye.
MFG needs a cap option. Like if it's 120fps once it reaches the cap, it can reduce the frames generated to ensure the best image quality.
You're the best! Until seeing your video, i was going to buy the 5080 FE but not now. Does the 5090 FE also have the Pcie 5 problem? I need the 50mm dimension for the new Antec pro 1M case (which Amazon will now ship on 19 Feb - order date was Jan 6).
Another great review, and another nail in the coffin for PCIe 5 having any use or benefit whatsoever on consumer platforms. All it does is drive up prices. Yay, my box has a higher number on it, and my SSD can wow me with benchmark numbers with zero real world benefit, wooo! Such wow, much excite. Just another sign of how the tech industry has been slowing down in the past decade or so (which, to be clear, isn't a bad thing it's just how things go as technologies mature and markets saturate), and how desperate they are to force a continuation of their previous levels of growth. This, crypto, NFTs, VR, AR, AI, none of it is the panacea that the industry wants or wanted them to be. And no new idea will be.
I'm learning that one just can't upgrade with every new generation, and one has to choose very carefully when to upgrade. The RTX40x0 cards featured a new node, and was a great choice in terms of performance per watt. The Intel 13th gen was a good CPU, whilst the current issue is disappointing. Based on DerBauers review, I bought a Inno3D water cooled 4090, and now that seems like a good choice. Super quiet obviously and not a day of trouble.
Well I am glad you learned that lesson, now remember you need to apply that to EVERYTHING in life.
The grass is not always greener on the other side.
You have to skip a generation if you are always going high end. If you buy every generation, your gains aren't worth it.
Yes and the 4090 is so efficient. I could upgrade but the efficiency on the 50 series just isnt there. My setup pulls such little power and heat
Been upgrading from flagship to flagship for decades. The RTX50 is the first time Im not bothering, its pure nonsense.
The 5080 is literally the worst generational performance improvement of the last decade... maybe even the last 2 decades. The performance is in line with what would be expected of a 70 class gpu. So instead of increasing the price, nvidia shifted the performance tiers again.
The 5080 is actually slower than the 4090... For the first time in the last decade the new 80 class cannot beat the previous gen flagship. This is 70 class performance. This is a 600$ GPU for 1000$.
I've seen multiple reviews now where even the 7900XTX in some titles actually beats the 5080. That is the 5080 getting smashed by the PREVIOUS GENERATION flagship of the competitor... I would be ashamed if I put out a product this bad for this much money. I'm actually surprised Nvidia hasn't already "unlaunched" this product.
Multi framegen is a joke as a feature. On the high end, you don't really need it, on the low end, the use is limited because most of the time your base framerate is too low.
There has literally never been a larger gap between the 80 class and flagship gpu. For 1000$ this card should be avoided at all cost
Thankyou for providing PUBG benchmarks! PUBG has a big player base and rarely makes peoples benchmark charts anymore.
People often compare this to the previous gen, and say it's not worth upgrading... big surprise. However if you were to recommend a High End (1440p) gaming card right now for a completely new build, which would you recommend? The 5080 doesn't seem a bad purchase today, especially if the new frame gen and other features get good support in the future.
In that side by side comparison at 13:30 are those max power readings right???? 855w for the 5080 and 2057w for the 5090????? WTF, tell me those are some sort or sofware error and not real transient spikes.... Or maybe I'm misinterpreting the readings and that is not the actual power drawn.
Great video,We appreciate all your hard work !! Can you do also 1080p? more interest in pubg low settings
We are missing something in between the 5090 and the 5080. Something slightly faster than the 4090
That was supposed to be the 5080's place, but nvidia butchered the card too much, now it's a joke.
@@mirarid yeah, this card gets blown away by the 4090... What a useless card.
I suspect thats deliberate - to make you spend more
@@mycosys The 5090 AIB cards will be £2500 or so, you watch. There's no way that's going to be good value, I'm waiting until AMD's next flagship. Nvidia is taking the piss honestly.
@@mirarid Nvidia is taking the piss? Do you not understand that the whole reason why nvidia has been able to do this is because AMD has been terrible on the gpu side they cant even get their drivers right lol
Thanks for the real review 👍
Yeah I remember Jay's 5090 review video. There was a comment about pcie signalling and the 2nd pcb
The reviews for the 5080 are if anything making the 7900xt and xtx look good by comparison. At least in anything that doesn't involve a bunch of ray tracing.
too late for that i started lookin for one after a few reviews the 7900 xtx around price and more
Just food for thought from a sample size of one. I have had nothing but problems with what i think is both pcie gen5 and ddr5 on my msi b650-p motherboard with a 7700x and 1 16gb stick of ram running whatever minimum default spec is like 4300 or whatever. The board is super not stable with expo and two ram sticks. Gpu likes to disconnect and crash all the time in addition. Its a gtx 1080. Forcing pcie gen down to 3 on everything has helped a lot. I really think signal integrity on both pcie gen 5 and ddr5 is hell on low and mid range motherboards. Wouldn't surprise me if that extra pcb is effecting signal integrity. These higher speed interfaces just dont leave as much room for error in design and it seems a fair number of manufacturers are having issues getting it right. I really dont want to spend $500 on a motherboard to hope for better stability but it seems that is going to be the way going forward.
I encourage you guys to watch frame chasers video he put up today. Blackwell overclocks way better than ada. With less voltage it goes to 3300 mhz and shows almost 4090 oc performance once you do so. Take everything with a grain of salt. Blackwell is better architecture, faster clock speeds, faster ram, and faster bus speeds. firmware hasn't been optimized yet, but when it does, it's gonna be much better than these entry tests that are base tests.
U need to think why they leave that clock headroom in the gpu?
Uhuh and how much of the population that buys these GPUs bothers with extreme OCing ? And as the other guy said, wonder what could be the reason for having such large headroom, definitely not for a Super refresh 12 months later.
This gen of Nvidia GPU's is pretty much a joke, but their fanbase will probably defend and buy them anyway, sadly.
No surprise that the partner cards built the "normal" way have less issues with PCIe 5 signal integrity than the FE which pretty much has riser built in as you said. And PCIe 5 is way more sensitive to signal integrity, just see how much more expensive/better built are the PCIe 5 risers and U.2 drive adapters vs PCIe 4 or even PCIe 3 ones.
RX 9070 XT for me, depending on price?
Start to seems that 9070XT is good buy at $750…
So lets see…
Because most AIB versions of 5070ti will be close $999…
I hope it will f' up the 5080, but we know AMD, its prob. not gonna kick its ass
I have 2070s. I was going to build new rig with 5080... Going to re-evaluate my options and wait for amd. Let's hope they deliver!!! I have my doubts with them too 😢 don't want to spend on 5090
The new Antec case comes with a Pcie 4 riser but i ordered an AVA5 PCIE 5.0) Right Angle (v2) riser 20cm.
To be able split 8x 5.0 for graphics card and 8x 4.0 for fast NIC on AM5 was only reason for me to buy. May be models from AIB partners will be better.
Have any other reviewers indicated PCIe issues? If not, could it mean der8auer's card might have an issue with the PCB connector?
I already know before the launch, 5080 is just a 5070 rename, Nvidia "Unlaunch" the real 5080 again.
I have the same problem with MSI Tomahawk x870 and MSI 5080. On PCI-e 5.0 I have lags and stutters every 2-5 seconds, system just freezes for a second every time. Launching games = automatic PC restart. After changing to 4.0 mode - everything works well. Had no problems with "Auto" setting with my GTX 1070.
I am very happy with my 4090. Not planning on switching anytime soon!
4090 is the daddy still :)
how can you justify price increases based on pure performance gains? Where are the days of generational improvements at the same/marginally higher price?
great video! thank you :)
So rtx 5000 is just a refresh gen? Seems fine if someone upgrades from something old (if for unscalped price of course)
Is RTX5080 actually 40% faster than RTX3080 or RTX5080 has worse price/performance ratio than 4 years old card?
Did that hotspot temp say 255C?
Yeah they removed the hotspot sensor in the 50 series for some reason, so right now the software is picking up a null signal. Should be an update to remove that metric from 50 series cards soon (it's in beta)
Is the latency significantly lower on newer PCIE gens? Or is it mostly just throughput that's increasing?
Have you tried using an intel setup using pcie5.0? to see if it has the same problems?
Or just not an Asus board...
i would hypothesize that this is caused by the weird built-in pcie "adapter" that plugs in to the pcb. whether its a physical contact/seating issue, internal pcb flexion, or just the break in the traces for the connector...something is degrading the signal integrity
Buying a RX 6700 XT made so much sense for me 1 year after launch. It's still a solid mid range card on Linux.
I also had a poblem with some 8x pcix gpus like 4060 ti... that was not recognized by the mb when in bios the lanes was set to auto. I had to set it to 8x + 8x to get it recognised ... The coincidence is that it was an amd am5 mobo.
14:59 360 and 575 watts is with the vram, and both cards don't have the same amount. So you can't do this simple division
You have to substract memory power first
Wo genau kommen die 389 Watt Stromverbrauch beim Cyberpunk Raster Test her? Die 7900XTX zieht auf Stock nur 350, 389 ist doch PL+15%, nicht Stock.
What a joke. I’ll continue to use my 3080ti until the next generation of consoles
Honestly, I'm almost glad it's not so much of an improvement. As somebody who's looking to upgrade from my hot and loud 3080Ti to 5000 series, it seems like I won't have a ton of competition from people looking to upgrade from their 4000 series cards - and that should hopefully keep the prices down somewhat. Not holding my breath for picking on up at MSRP any time soon, but I can dream.
Same situation for me. MY 3080 Ti is already gone, so I need one of these one way or another, but I'm not paying over MSRP, and I'm sure as heck not paying over what the 4080 Super was at $999. The 5080 is not impressive.
they arent going to have stock i feel so it will have huge price increases
I upgraded to a 4070Ti Super when my old 1080 could barely run FF16, a game I had been looking forward to for a while. I felt a bit of buyer's remorse in the last month or two, thinking maybe I should have been patient and waited for the 5000 series. Nowadays, I don't feel so bad lol
I had maybe same problem with Asus TUf B550 board and EVGA rtx3070. Board wasn´t cooperating, and I have to run PCI-e 3.0 x 16 or PCI-e 4.0 x 8 . After change for MSI board it was OK. It is AMD 😕 I never had this problem on Intel platform. Or Asus problem :D (It wouldn´t be first time, right) Did you have same power consumption? In my case I had feeling that something is "jamming" data signal going through PCI-e. And looking on fps and power consumption it was like GPU hasn´t data to compute.
*The 5090 and 5080 have a huge gap bwcause there will be a 5080 Ti releasing in October. 24GB of GDDR7 and a 384-bit bus width.*
*Nvidia knows people will not be happy with the 4080 Super and 5080 gap. It's called upselling. It forces people to buy the next tier up.*
*The 5080 would be more like a 4080 Ti SUPER. Consider the Ti usually only has an 11% fps gain over the non Ti variant in games. But the Ti SUPER would have a 16% increase in fps performance over the non Ti or SUPER variant.*
*When you do the math, the 5070 Ti matches the 4080 SUPER. The 5080 matches what would be a 4080 SUPER +. (Aka a Ti SUPER). The 5090 is equal to what WOULD be a 4090 Ti.*
I recently built my first pc keeping in mind the rtx 5000 for 4k gaming, should i still go with this?
Have you tried the card on multiple motherboards?
Is there still a 2% performance hit using Pcie 4 on the RTX 5080 or is it lower/higher?
Quick question for anyone who knows. IF the GPU TDP is 500W are we really dissipating 500W of heat? I don't think that can be true. If it were true the card would be totally inefficient. If I understand correctly the heat dissipated by the card cannot be 500W it must be lower if the card is actually using some of the power to actually do the work. So the heat that is dissipated is waste heat from it being inefficient. So how much power is it actually using to do work and how much power is just wasted as heat?
Can't wait untill they start making it blatant and just release cards with the same exact performance as a previous iteration but tack on a mystery technology that you don't need or more fake frames.
Looking to upgrade from a 1060 and my local microcenter has had people camping for 4 days now. It's the day before launch. All I can do is be prepared and fight like hell. lol
both Bestbuy and Canada computers didn't get any stock here in Canada
Well, 5080 crashes on PUBG, works on everything else i have testet. Any fix for this?
switch to pci gen 4
@@322-Dota2 Did that, and it did infact not work
Hey Derbaur. Will you be testing Spider-Man 2? It comes out on PC tomorrow.
NVIDIA fai. The proof is in the fact the partner cards don’t have the problem. I can see the future. I seen a lot of 50 series founders failing prematurely due to the pcb extension slot.
What are the motherboards on the wall behind him?
Tested pci 5.0 with intel? Amd always have dodgy IO but if the 5090 worked fine dunno
Is the 50 series more optimized towards the Transformer DLSS model? Because on my 4090 Transformer uses more power compared to CNN, worth about one step on the quality slider in CP77. So Transformer Performance eats about the power of CNN Quality.
I wouldn't be surprised if the new model simply uses the tensor hardware more.
@b127_1 Whatever happens under the hood, the new model uses as much power as the old model at 1 step higher quality. And the new model currently introduces very serious (dithering?) noise in some parts.
it seems that there is no specific optimisation for the Transformer model. And the new model is about 4-5% slower than the older one. at least it looks better.
@@mariuspuiu9555 Whether it looks better had to be compared based on its performance/power usage. Currently it seems to me like I have to compare Transformer to CNN by pushing CNN's quality slider 1 step higher!?
Transformer uses a different part of the card than CNN does. That is why you see the difference in power
you should test with intel series like 12900ks or 12900kf with z690 with pcie 5.0 motherboard, i think some z690 have pcie 5.0 to make sure if there no problem in intel lga 1700
so what i have been thinking is what if the connection between the pci and board of the 4080 has some drop out. and with them makeing the bus lower than the 5090 is what is causing it. cause if the 5090 is not having an issue but the 5080 is haveing issues then what has changed everything is their but the bus speed.
SOOOO happy i bought a 4090. Got 2 years of the worlds best card, and now i get at least 2 more years of the world's 2nd best card and all for 1400 which isn't bad spread over 4-5 years.
Well in charts this 9800xtx looks pretty good with this 24g vram 😅 and price but no RT no dlss features only fsr but it cost less
I have not had any PCIE5 issues on my 2 year old intel 13th gen mobos from Asus, interesing if folks are hitting this on specific chipsets?
will nvidia have the same problem than intel?