Pros:
- The highest performance among video cards
- Supreme cooling
- Ideal for professionals in graphics
Cons:
- Super pricey
- Extremely large in size
This card replaces two rivals at once. We’re referring to the specialized RTX Titan and the recent flagship, namely, RTX 2080 Ti. Our review item is a real graphics monster. It has a huge heatsink and proves supremacy in gaming, 3D rendering, and other tasks. It’s all thanks to GDDR6X memory with 24 GB.
Keeping in mind this card’s performance, we assume one thing. Even gamers who prefer playing at the highest settings are not the main target audience. This CPU is more likely to become the choice of 3D animation specialists.
Indeed, the company promotes the device in the professional segment of the graphics market. In this area, it can truly show its best. However, it doesn’t neglect an opportunity to provide steady 60 frames at 8K resolution gaming. At the same time, the product’s versatility gives way to certain discrepancies.
The object of the review — graphics card RTX 3090, which is the flagship of Nvidia company. And, just like in any other GPU generation, it’s supposed to eliminate the market competition. But the sky-high price point hides just a small performance advantage over the RTX 3080.
Therefore, it’s only an appropriate choice for real computer geeks, craving a super powerful GPU. Or else, the same small number of professional artists could be interested in such hardware.
Nvidia GeForce RTX 3090 Price — Amazon
As of now, the video card retails for a pretty high price of $2,277 at Amazon. It’s the first time the company collaborates with the OEMs when producing the Titan-category GPU. As a consequence, there will be even more expensive versions.
Is the card cheaper or more expensive than the last generation? It’s hard to decide. If we compare it to the Titan RTX, it’s more expensive. The rare lucky few bought the Titan RTX for $2,063 at Amazon. On the other hand, there is the RTX 2080 Ti. In fact, it never had a successor and retailed for $1,199 at Amazon.
What do we have as a result? The review card is somewhere in between. The company claims that the GPU is targeted at the consumer market, primarily gamers. However, all the configurations and a price solution put it into the professional segment. Its audience demands rough power for rendering tasks. At the same time, it isn’t willing to shell out for Tesla or Quadro.
Main features: RTX 3090 specs
RTX 3090 vs RTX 3090 TI
The card is built on the same architecture as its predecessor, RTX 3080. They both use Ampere. The 3090 version, however, is upgraded with 82 stream multiprocessors (SM) and 10496 CUDA cores. In addition, it contains 82 RT cores and 382 tensor ones.
At first sight, this doesn’t look like a big difference with Titan RTX’s 72 multiprocessors. But the latter is built on the Turing architecture. So it didn’t provide 2 data channels on each SM that work with FP32 loads. CUDA cores efficiency doubles, making the card a colosseum in the raw rendering segment.
In addition to the doubled multiprocessor performance, we have 24 GB of memory. A bandwidth equals 936GB/s which is nearly 1 TB of data processed every second. Can you imagine that?
Such a combination of video memory speed and incredible volume allows the heaviest 3D rendering. If you work with Blender or Davinci Resolve, you’ll greatly appreciate the speed increase. After all, anything that shortens your working process eventually saves you money. Along with an accessible price (compared to Titan), it appears to be a gold nugget.
There’s one thing we already mentioned when reviewing the RTX 3080. Some serious updates of the RT and tensor cores took place in the series. The manufacturer quite improved them in the last generation. Thus, the RT core bandwidth doubled compared to the previous version.
In games with ray tracing, the multiprocessor passes to the RT cores ray emission calculations. They, in turn, determine the scene’s ray reflections and return the data to the processor. Ray tracing in actual time used to be viewed as almost impossible. It’s because the multiprocessor already had to rasterize the scene and couldn’t do the calculations.
RT cores do a huge part of the whole workload. Nevertheless, ray tracing keeps taking a great amount of processing power. So we can see a lot of wasted performance. For that reason, DLSS is becoming more important in both gaming and software environments.
Tensor cores of the third-generation Ampere GPUs have improved quite a lot. In regards to the Turing ones, speed doubled. But it doesn’t equal performance, since every multiprocessor applies one tensor core. Meanwhile, Turing used two units per processor.
Anyway, DLSS implementation is not the only responsibility of tensor cores. They’re also in charge of Nvidia Broadcast. We consider this feature as one of the most underrated in this generation of GPUs.
We’ve been utilizing it during video conferences and Discord gaming sessions quite a lot recently. So it’s hard to omit incredible microphone audio filtering. Also, with the help of this feature, you can hide the background behind your back.
Now that we spend much time at home, video conferencing is becoming more widely used. So anything that makes it better is of great value. The icing on the cake is the feature’s availability to all RTX card owners.
We’ve talked about the positive sides, now let’s discuss an unpleasant aspect. Power consumption reached 357W at the peak of testing. This is a high number, especially if paired with a strong processor. Nvidia’s recommendation concerning power supply is at least 750W. However, we suggest using a 1000W option. This way you can avoid any possible inconveniences. They can happen in case your power brick doesn’t reach its full capacity.
Read more — https://best.kevin.games/gaming-graphics-cards/nvidia-geforce-rtx-3090