if you have seen the ounders edition of rtx 4090 you know how thick and bulky it looks. That GPU is absolute monster in apperance but this news RTX 5090 is a lot slimmer than that.
In the Unboxing experience I have see.
This time the box is very simple and environment friendly This is the box of latest RTX 5090 FE.
The box itself look like a enlarged version of GPU itself. It’s a very unique box to be honest.
Now if you don’t know them RTX 5090 uses the PCIe gen g 12V 2×6 adaptors.
its a 4 8 pin connector and it will be able to reach full 600W
cokapred to last time the cable now is not soft and delicate it can bend easily. It feels much softer too.
So of you want the RTX 5090 your psu must be at least ATX 3.0 power supply of 1000W
for my liking I prefer msi meg ai1300p and ashs tuf 1000W ATX 3.0 power supply.
wgich can directly use PCIe gen5 12V 2×6 specifications.
Now back to the box the design is very interesting.itscompletely different style from the past.
The entire box is like a GPU itself. Or a prototype.
i also think the box material is also very special and unique. You can see there are many layers of
Wow, today we are going to unbox something that all of you players, including myself, are looking forward to. This should be NVIDIA’s recently launched GeForce RTX 5090 graphics card. Before the release, we are going to unbox it for you first. This is the shipping box I just received it in. I can’t find any information about this product outside, so I’m not sure if it’s a GeForce RTX 5090. Maybe it’s a GeForce RTX 5080. OK, but this The size of the shipping box is too slim. If you have bought the Founder Edition of the 40 series, you can take a look at this. This is the shipping box of the GeForce RTX 4090. Now, after the GeForce RTX 5090, you see! Its size is only It should be about half. This founding edition has a very unique design in the RTX 40 series, but relatively speaking, its size is really too huge. Let’s unbox it now. In fact, I don’t know what’s inside. It looks like it should be torn from the top. Wow, how about this box packaging? It has a minimalist and environmentally friendly style. Ladies and gentlemen, this is the outer box of the latest GeForce RTX 5090 founding edition. It looks like an enlarged version. The appearance of a graphics card and the design of the entire box are very unique. Let’s take a look at the transport box first.
00:01:57 Because I think that the shipping boxes for the RTX40 series are very unique. Every structural design of the box is very ingenious and it is minimalist and environmentally friendly. Any plastic filler or any material other than paper should be part of its accessories. OK, this time the RTX 5090 still uses the PCIe Gen5 (12V 2×6) specification adapter. The number can be seen that it has reached 4 8PIN directly. This should be the limit of 600 watts. And the wire part is softer and more delicate than the last time. But I think I should buy it. RTX5090 users will have the latest ATX3.0 power supply, which can directly use the native PCIe Gen5 (12V 2×6) specification. I think the box design this time is very interesting and completely different from the past. You can see that the entire box is designed with the shape of the RTX graphics card as the prototype. I also think the material is very special. You can see that it is very multi-layered. This should be corrugated paper stacked up. It is super environmentally friendly and the material used is very close to the RTX40 series, but this time no inkjet is used. It is all original.
The prototype of the paper should be environmentally friendly to some extent.
The way to open the box should be to pull out the two sides. After pulling it out, what should it be? Open it directly? You can see it. This is the body of RTX 5090. Wow, it is really only two slots thick. The design of this radiator is really incredible. This time, you have to know that the GeForce RTX 5090 is Its TGP is higher than that of GeForce RTX 4090. This is GeForce RTX 4090. Its TGP is 450W. The whole volume is quite compact. It adopts a three-slot dual-fan structure. This time, GeForce RTX 5090 is even more It’s amazing that its TGP is 575W, which is 125W more than the 4090. In this case, the radiator is made smaller and it is a very standard dual-slot card. This should be in line with If you pay attention to the structure of the RTX 50 series founding edition, you will find that its design is full of ingenuity from the PCB to the entire heat dissipation structure. In order to maximize the airflow effect, the PCB adopted a solution that we have never seen before.
00:04:41 It is placed directly in the center. You can see that the left and right sides are completely blown through by the heat sink fins. The PCB is in the middle, so the airflow from the dual fans can get the best The air intake efficiency is good. The I/O port is separated from the PCIe interface. So the design structure of the entire founding edition is super complex. Let’s take a look at the GeForce RTX 40 series. Part of the heat is exhausted from this baffle, but the GeForce RTX 50 series emphasizes the structure of blowing through the entire radiator, so it seals this baffle, so all the heat is exhausted from this The back panel at the back shows the CPU. Haha, this time the GeForce RTX 50 series also seems to use liquid metal, so it can make its overall heat dissipation efficiency better. I may not be able to do it here at the moment. First of all, I don’t have any replacement materials at hand, and due to time constraints, I hope to show you the test content of the GeForce RTX 5090 this week. If you are interested in its entire structure, I highly recommend that you go and see that Gamers Nexus released a disassembly video of the founding edition a few days ago. Well, since the performance has not been released yet,
00:05:53 So we can only quickly share with you the unboxing and appearance of the GeForce RTX 5090 Founder Edition. The specific on-machine testing and evaluation will have to wait until 1 After 10pm on the 24th, the major media outlets started publishing the results. I will try my best to show you our actual test results as soon as possible, and then ask if you would like to see related tests. If you want to know the content, you can leave a message under this video or leave a message in our Discord. It seems that the sale in Taiwan is delayed until the Chinese New Year, so the supply may be stable after the Chinese New Year. During the Chinese New Year period, there is a long window of time. I will share with you some details that were not mentioned in the first video or some content that I personally think is worth paying attention to. You can subscribe to our channel and follow us. For the latest information, please join our Discord. Although the performance has not been unlocked yet, the GeForce RTX 50 series actually adds a lot of new technologies and functions. First of all, you should know that this time the GeForce RTX 50 series uses a brand new Blackwell architecture. In fact, Blackwell architecture was introduced before GeForce RTX 50 was launched.
00:06:54 It has already been used in common AI or server data centers. This architecture is now introduced to consumer-grade GeForce GPUs for the first time. Of course, NVIDIA is targeting general users and consumers. Some adjustments or specializations are made to make this graphics card a gaming-specific GPU. There is a premise that everyone should know that this time the GeForce RTX 50 series (Blackwell) still uses the same GPU as the GeForce RTX 40 series. The ADA architecture is the same as the TSMC 4N process node. If the process has not been upgraded, then why can Huang hype up the RTX 5070 at CES this time, saying that the RTX 4090 has such an exaggerated performance increase? The performance of the RTX 4090 is only $549! OK All of this comes from the new DLSS4 upgrade exclusive to the GeForce RTX 50 series. DLSS 4 can generate twice as many frames as DLSS 3, all of which are calculated by AI. Strange. Of course, if you have watched our previous DLSS3 detailed analysis video, you may have a question. The previous DLSS 3 produced some extra frames with some flaws or resolution reduction in some details. DLSS4 produces twice as many frames. Number of frames
00:08:03 Wouldn’t that result in more flaws and a greater loss of resolution? OK, I have to say that the principle of generating extra frames in DLSS4 is completely the same as that in DLSS 3. The first difference is that it uses a brand new model. In the past, games used the traditional CNN model (Convolution Neural Network). This time, with DLSS 4, a major update is the new Transformer model. This model has been used in the past. Over the past two years, it has been widely used in major production AIs, including the currently common GPT. Of course, all of these use NVIDIA’s higher-end GPUs for training. Now, let’s talk about our current gaming field. For example, everyone has used ChatGPT, right? You can throw a very long file into it and it will generate a summary that is very concise and focuses on the key points. What kind of application can it bring to games? For example, In these complex scenes in games, this new Transformer model can directly find the special places that our eyes will notice and then focus resources on processing instead of wasting time processing every detail evenly. After using this model, it may not only bring Improving image quality can also reduce system resources, which ultimately leads to an increase in overall FPS.
00:09:09 Of course, this feature is not limited to the RTX 50 series, but can be used by RTX 40 series or older GPUs. It is just the efficiency part. The performance of each generation is different, and it brings different performance growth. So this time, the RTX 50 The exclusive upgrade feature brought by the series is the Multi Frame Generation function. Remember in the previous DLSS3 single frame generation analysis video, we mentioned that the RTX40 series’ ADA architecture needs to rely on hardware to generate each supplementary frame. The optical flow accelerator (Optical Flow Accelerator) of the volume and the depth motion vector in the game are used for calculation. Because the generation of each supplementary frame requires additional independent calculations, although DLSS3 can double the FPS, in this process It also consumes hardware resources. But if we want to generate multiple frames through DLSS3, then each new picture needs to be repeatedly used by the Optical Flow Accelerator and the deep motion vector for calculation. This means that the graphics card needs to handle twice as much computing demand. Obviously, this method puts a lot of pressure on the hardware. In order to break through this bottleneck, this time DLSS4 first introduced a new multi-screen generation architecture. AI optical flow model to replace the hardware optical flow detector
00:10:12 The new AI model is not only faster and more efficient, but it only needs to perform one AI operation on each native frame to generate up to 3 additional frames. Another major focus this time is The Transformer-based neural network architecture model allows the GPU to be in more complex scenes and generate more accurate details in rapidly changing images. At the same time, it also makes the transition between frames more natural and smooth, which avoids In the past, there may be a problem of local object flickering. The last one is that Blackwell introduced an enhanced Flip Metering technology at the hardware level. This technology can transfer the rhythm logic of frame generation from the CPU to the GPU display engine. Let the entire generated picture be more evenly distributed, making each picture more stable and smooth. Of course, is it as powerful as NVIDIA’s official display? The GeForce RTX 50 series has greatly increased its FPS, but of course for us gamers In terms of the details, I am more concerned about whether the game screen is damaged or there are some flaws when zooming in, or as we mentioned before, DLSS3 often encounters flickering, etc. Is there any improvement in the problem? This is also the part I am very concerned about and want to focus on in the next test.
00:11:18 The above are probably the main core upgrades brought by the RTX50 series this time. Next, let’s talk about what upgrades this time’s hardware brings. Although the process has not been upgraded this time, But in addition to the manufacturing process, I think NVIDIA has really made improvements in many areas this time. For example, the PCIe interface has finally been upgraded to Gen5. For example, our current mainstream X870(E) Z890 and other host versions are all Switched to PCIe Gen5 specification. This time, it finally switched to the fifth generation PCIe interface. And in the display output part, it finally upgraded to the new DP2.1 (UHBR20). Since the GeForce RTX 50 series Emphasis on multi-screen generation function, so 4K 240 FPS DP2.1 (UHBR20) should be no problem. For creators, this time the RTX 50 series brings a very heavy upgrade in the audio and video editing part. This time BlackWell The upgraded codec can support 4K 10 Bit 4:2:2 specifications, and the RTX 5090 even supports up to three encoders, which is really a big plus for YouTubers or professional editors. This is a great news. I can’t wait to use this RTX5090 to edit movies (can I edit?)
00:12:24 Yes. In addition to the above, in the AI part (Tensor Core) supports lower precision FP4 calculations. This is why you see NVIDIA’s RTX 50 series performance being promoted. The AI TOPS indicator used is calculated using the FP 4 standard. 3352 AI TOP with FP4 OK. As for the specific performance test, we will wait until the ban is lifted and share it with you in detail. I know everyone has been waiting for this card for a long time, so there should be a lot of questions. If you have any questions or want to ask, you can leave a message under the video or add our Discord to discuss with us. Discussion Well, this video has come to an end. Thank you all for watching to the end. Goodbye for this review. Bye bye for this review.