Top tech reviewers give their initial reactions to the Nvidia RTX 5090, highlighting its massive architecture jump and memory specs. Along the way it covers Design and Build, Core Specs and Features and Gaming & AI Performance, and more. A final summary evaluating who should buy the graphics card and who should skip it based on gaming versus workstation needs.
/public/creators/rereview-tech/videos/video_Nnuckf9uvRQ/arvp.json
· answer engine at
/public/creators/rereview-tech/moment-agent
Top tech reviewers give their initial reactions to the Nvidia RTX 5090, highlighting its massive architecture jump and memory specs.
On screen: A sleek black NVIDIA GPU, likely an RTX 5090, rests on a dark green surface with a dark background, displaying its specifications in white text above it.
Reviewers examine the physical design of the RTX 5090 Founders Edition, featuring a compact two-slot form factor and dual flow-through cooling layout.
On screen: The 3D render shows a GPU chip being placed onto the main circuit board, surrounded by memory modules, against a background of green lines.
An in-depth breakdown of technical specifications, including DLSS 4 transformer-based AI upscaling and raw FP4 compute capabilities.
On screen: The shot displays a detailed comparison table contrasting the technical specifications of the GeForce RTX 4090 and GeForce RTX 5090, presented against a blurred, light-toned background, with a small watermark for…
Testing real-world gaming performance with multi-frame generation reveals massive frame rate boosts along with latency trade-offs and early software support quirks.
On screen: A man in a black leather jacket stands in a tech-filled room, gesturing with his hands while speaking, with several graphics cards and scattered banknotes on a counter in front of him.
The card is compared against AMD's RX 7900 XTX and previous generation flagship cards like the RTX 4090.
On screen: A screen recording displays a Newegg website showing a list of various AMD Radeon RX graphics cards for sale, with an overlay text "COMPETITION & ALTERNATIVES."
Reviewers debate whether the $2,000 price tag is justified for high-end gaming, video editing, or local AI generation.
On screen: A balding man wearing glasses and a gray hoodie is seated in a dark gaming chair, speaking directly to the camera, with a Shure microphone positioned in front of him.
A final summary evaluating who should buy the graphics card and who should skip it based on gaming versus workstation needs.
On screen: A man with short brown hair and a brown jacket over a white t-shirt speaks directly to the camera, gesturing with his hands, while various exploded and assembled views of a black graphics card with multiple fans and…
0:00 Speaker A: It's here. Speaker B: It is rippin fast. Speaker B: Jeez. Speaker C: There's a big review today. Speaker D: Oh boy, I'm really excited. Speaker E: This one's gonna be interesting. Speaker A: It makes the PS5 Pro look like a toy. Speaker F: Is it possible for a $2,000 graphics card to actually be worth it? Speaker B: Welcome back to RE review where we watch all the reviews so you don't have to Today we're diving deep into the Nvidia RTX 5090. Speaker B: Is it a revolutionary marvel or just an overhyped experiment?
0:25 Speaker D: That. Speaker B: Let's find out straight from the people who've tested it and let's kick things off with their initial reactions. Speaker F: I was actually not going to make this video, but when I was at CES I went to the Nvidia press conference and that keynote was unbelievable. Speaker F: I'm holding 92 billion transistors alongside 32. Speaker G: Gigabytes of memory on the tiniest circuit board that I've ever seen on a flagship gpu. Speaker B: More GPU cores. Speaker H: Boom. Speaker B: More VRAM and faster VRAM Boom.
0:52 Speaker A: Boom. Speaker B: Enhanced RT and AI cores, wider memory bus and PCIe Gen 5. Speaker D: Boom. Speaker D: Boom. Speaker D: Boom. Speaker I: There's a lot we have to talk about and not all of it has to do with gaming. Speaker I: A lot of it has to do with AI, but some of it actually has to do with AI and gaming and frame generation. Speaker E: As generational leaps go, this is as good as it's going to get with a new Black Row Architecture 5080 down, expect lower gen on gen gains. Speaker B: Moving on to the design and build.
1:18 Speaker B: What are the reviewers saying about the 5090 physical form factor? Speaker E: The 5090 is a simply astonishing example of industrial design. Speaker E: Bearing in mind the gigantic processor with its immense hardware requirements. Speaker E: The same level of high quality design and execution as 40 series, but smaller. Speaker C: The RTX 5090 Founders Edition moves to a two slot design and uses that dual flow through configuration. Speaker C: So they've sandwiched the PCB centrally and they've offset the PCIe slots.
1:45 Speaker C: That 40 foot is offset to the side of the PCB and it's down as a result. Speaker C: That also means that Nvidia needs a separate PCB for the IO at the back, which feeds the monitors connected via a Flex cable to the main pcb with the GPU. Speaker F: Looking at the 5090, it is clear that this is an expensive piece of hardware. Speaker F: Unlike traditional graphics cards, which are, I'll say, a little bit more straightforward in their design, this actually has flow through on both front and back.
2:12 Speaker F: So if you look on the inside, it actually has all the actual GPU bits on the middle so the power delivery, the actual GPU itself, the memory and it actually has a little daughterboard for not only the PCI slots but also the rear IO. Speaker C: The FE model does a lot of small things to improve performance like exhausting the air out of the top of the card. Speaker C: You can see that here in our Schlieren imaging that with again that dual flow through. Speaker C: A lot of the benefit from this two slot design comes from pushing the air through surrounding the pcb.
2:39 Speaker B: It's a compact two slot design with innovative airflow. Speaker B: Nvidia has put a lot of thought into thermal management diving into the core specs and performance. Speaker B: What are the numbers telling us and what new features stand out? Speaker E: RTX 5090 it's an even bigger processor on the same base technology. Speaker E: A gigantic 744 square millimeters consuming even more power than the 4090. Speaker E: Here's how the specs compare between 5090 and 4090 33% increase in CUDA core count while a gigantic 512 bit memory interface provides a 77% boost to memory bandwidth.
3:17 Speaker E: Plus we're moving up to 32 gigs of memory from 24. Speaker E: Maximum power consumption increases from 450 watts on the 4090 to 575 watts on 5090, a 28% increase while boost clocks actually drop. Speaker E: The 2.52 GHz average from 4090 drops to 2.41 GHz on 5090. Speaker F: The basic concept hasn't really changed the the GPU renders a lower resolution image and then uses specialized tensor cores to upscale it to your target resolution.
3:49 Speaker F: What is new at DLSS4 is how it handles this upscaling. Speaker F: The RTX 50 series uses what Nvidia calls a Transformer based AI model which is theoretically similar to what powers AI tools such as ChatGPT. Speaker F: This is significantly smarter in understanding the entire frame as opposed to small parts of it. Speaker F: In theory this means better temporal stability AKA less flickering and artifacts when things are moving and particularly big improvements when it comes to ray traced content.
4:16 Speaker I: I'm not really sure what gamers are going to do. Speaker I: I'm not sure. Speaker I: I'm not sure how they're going to take it. Speaker I: But this is a card that can do so much more than gaming. Speaker I: It seems like that it's the best. Speaker G: Of all worlds biggest feature about this card. Speaker G: AI capabilities. Speaker G: It's got so much AI power it's ridiculous. Speaker G: For those people who know how they can utilize this for local LLMs or or if they want to do some generative AI.
4:43 Speaker G: It's mental. Speaker G: Let me put it in perspective in FP4 this 5090 can produce over 3300 tops. Speaker G: So for reference, the RTX 4090 is 1300, so that's more than double the performance. Speaker G: But let's put some other guys in the mix there. Speaker G: I've got an M4 Max in there.
5:08 Speaker G: The fastest chip that Apple offers right now and that one offers 38 tops. Speaker G: The Neural Engine. Speaker G: The NPU in there is 38 tops. Speaker G: That's over 88 times slower than the RTX 5090. Speaker G: We're looking at some of the other cards from Intel. Speaker G: The fastest card available Right now, the Arc B580 is around 288 tops compared to the 3300 with this.
5:37 Speaker B: Let's get practical. Speaker B: How does the 5090 translate to real world gaming? Speaker B: And what about the new multi frame generation? Speaker E: There are also more limited use case scenarios, so it's pretty awesome to see frame rates in the hundreds, right? Speaker E: But I'd consider a 4K 240Hz monitor a must for this class of hardware. Speaker H: The previous version of Frame gen, it uses AI to generate in between frames based on optical flow data and rendered frames.
6:02 Speaker H: But multi frame gen now allows users to generate up to three in betweens rather than just one. Speaker H: Boosting frame rates to up to four times native. Speaker A: But the feeling of the frame rate is still going to feel like the base frame rate. Speaker A: So for instance, in Cyberpunk 2077 the base frame rate before turning on frame gen is 60.2 FPS. Speaker A: Even though we are getting a displayed frame rate of 201, the input is going to feel like you're still at 60fps, so you're going to get what feels like kind of a jello y slightly jello y feel to it.
6:38 Speaker E: This Cyberpunk clip actually looks okay in that respect. Speaker E: Finally we have latency. Speaker E: The big bone of contention is a game truly more performant if input lag is actually higher with frame generation active,. Speaker D: This is flawless already. Speaker D: If you want to play like this, it's still going to look fantastic on a high refresh rate panel. Speaker C: It's. Speaker D: It's smooth, it doesn't stutter at all. Speaker D: Frame time graph is a little bit thicker than I'd like, but I can't really feel any micro stuttering issues at all.
7:07 Speaker D: It's just great of an experience. Speaker D: Just for a quick test here with frame generation, this is the old frame gen as well. Speaker D: Enabling that over native res gives you 160s 150s instead of the 120s and 130s that we were seeing here. Speaker D: It also makes the frame time graph function flatter, although it doesn't really matter because it was very very smooth still previously. Speaker D: Look at that 160s now in the city area instead of getting down into the lower one hundreds.
7:35 Speaker H: It's an awesome experience here in Horizon. Speaker D: Forbidden west this is super enjoyable. Speaker G: This is not really working particularly well in a lot of the applications and a lot of the Adobe applications especially don't have full support for the 50 series GPUs, which means that what you see in there is a little bit screwed and a little bit broken. Speaker G: What I mean by that is you're not going to get the full benefits of the gpu, so the AI capabilities are most likely not going to be shown.
8:05 Speaker G: So I'm actually very excited to learn and utilize more AI or generative AI softwares and the more they're going to come out, the more this supports multi. Speaker B: Frame generation delivers impressive smoothness, particularly in demanding titles. Speaker B: However, increased latency might be noticeable, especially in competitive games so how does the 5090 stack up against the competition and even its predecessor, the 4090, the RX7900XTX.
8:31 Speaker C: Is AMD's closest competitor. Speaker C: Pricing is around 870 bucks or so to $900. Speaker C: The company also bowed out of the high end race. Speaker F: The only real competition for the RTX 5090 is Nvidia themselves. Speaker F: Now the 4090 has been the undisputed performance king ever since it launched back in 2022. Speaker E: You can almost view the 5090 as a kind of titan like GPU that would slot in above the 4090.
8:58 Speaker E: Ah, circa 25 to 30% boost is pretty much in line with the kind of delineation typically seen between products in the same lineup. Speaker C: So if you're on 30 series and still not in the mega mega baller income bracket, I guess all we can. Speaker D: Do is wait and see. Speaker B: The 5090 clearly leads in raw power, but it's OPER in a different category than previous generations. Speaker B: AMD has bowed out of the high end race for now, leaving the 5,090 without a direct competitor.
9:24 Speaker B: Alright, the big question is, is the 5090 price tag justified? Speaker B: Let's see what the consensus is on value. Speaker F: When Nvidia announced the $2,000 price tag for the 5090 it was met with a general even though it amounts to a $400 price hike. Speaker C: Nvidia also has the 5070 Ti and the 5070 launching in presumably February for $750 and $550. Speaker G: If you're not doing AI and if you're not doing video editing, then probably for 3D, the best bang for buck 4090 is going to be amazing.
9:55 Speaker G: If you are doing AI video editing or you're interested in generative AI, then the 5090 is in a whole nother league. Speaker E: The trend is clear, the more you want, the more you'll need to Pay. Speaker E: And with RTX 5090 there's a broad equivalency between how much you pay and and how many extra frames you get versus the 4090. Speaker B: So after all is said and done, where do we stand with the RTX 5090?
10:21 Speaker B: It's a powerhouse, no doubt pushing the boundaries of what's possible, especially at 4K. Speaker B: The new AI enhancements, particularly the multi frame generation, are genuinely impressive even with limitations. Speaker B: But let's be honest, this isn't a gaming card anymore. Speaker B: It's evolved into something more, targeting a broader audience of content creators, AI researchers and professionals who demand ultimate performance no matter the cost. Speaker B: The roughly 30% performance uplift over the 4090 might not seem groundbreaking considering the price increase.
10:51 Speaker B: However, for those who can harness the AI capabilities and a substantial 32 GB of VRAM, the 5090 offers a unique experience. Speaker B: If you're a gamer with a 4090, an upgrade might be hard to justify. Speaker B: If you're on an older card and need the best, or if you can leverage those AI features professionally, the 5090 might just be worth the investment. Speaker B: Ultimately, the RTX 5090 is a statement piece, a glimpse into the future where AI plays an increasingly significant role in our computing experience.
11:20 Speaker B: It's not for everyone, but for those who need its blend of power and cutting edge features, it stands alone. Speaker B: That's it for the ultimate roundup review of the RTX 5090. Speaker B: Check out the links for full review videos below and subscribe for more re reviews where we watch all the reviews so you don't have to.