Nvidia has officially unveiled its next-generation RTX 50 series GPUs at CES 2025, based on the Blackwell architecture. The new line-up includes the RTX 5090, 5080, 5070 Ti, and 5070, all promising significant performance improvements over the RTX 40 series. After numerous leaks and rumours let’s see if RTX 50 series is what we thought it would be.
RTX 50 Series Launch Overview.
The initial line-up includes four desktop GPUs: the flagship RTX 5090, the high-end RTX 5080, the mid-range RTX 5070 Ti, and the more budget-friendly RTX 5070. Nvidia is also bringing the RTX 50 series to laptops, with corresponding mobile GPUs for high-performance gaming laptops.
- RTX 5090: The top-of-the-line offering, boasting 32GB of GDDR7 memory, a memory bandwidth of 1,792GB/sec, and 21,760 CUDA cores. Priced at $1,999, available January 30th.
- RTX 5080: Aimed at high-end gamers, featuring 16GB of GDDR7 memory, a memory bandwidth of 960GB/sec, and 10,752 CUDA cores. Priced at $999, available January 30th.
- RTX 5070 Ti: Targeting the mid-range market, with 16GB of GDDR7 memory, a memory bandwidth of 896GB/s, and 8,960 CUDA cores. Priced at $749, available in February.
- RTX 5070: The most affordable option in the initial lineup, offering 12GB of GDDR7 memory, a memory bandwidth of 672 GB/sec, and 6,144 CUDA cores. Priced at $549, available in February.
- The new RTX 50 GPUs feature a 16-pin power connector, PCIe 5.0 interface, GDDR7 memory, and DisplayPort 2.1a. They will also support the new DLSS 4 technology, featuring multi-frame generation
- Laptop variants include the RTX 5090 (24GB GDDR7), RTX 5080 (16GB GDDR7), RTX 5070 Ti (12GB GDDR7), and RTX 5070 (8GB GDDR7), with laptops expected to be available starting in March from various manufacturers.
Flagship Focus: RTX 5090 – A Titan Reborn (and resized?)
The RTX 5090, promises a significant performance uplift over the already powerful RTX 4090.
Beyond the raw performance, the RTX 5090 introduces a surprisingly welcome change: a smaller form factor. The Founders Edition of the 5090 is a two-slot card, a significant departure from the bulky three-slot design of the 4090.
At 304mm long x 137mm high x 2-slot wide, 5090 FE is SFF-Ready Enthusiast GeForce Card capable is a commendable feat from Nvidia engineers. Nvidia says there is new revolutionary new double flow through design in 5090 Founders Edition. I think, even the after market price of the 5090 FE is going to be revolutionary.
Coming back to 5090 general edition.
Here’s a breakdown of the key specifications:
- Memory: 32GB GDDR7
- Memory Bandwidth: 1,792 GB/sec
- CUDA Cores: 21,760
- Total Graphics Power (TGP): 575 watts
- Recommended PSU: 1000 watts
Performance and Benchmarks of RTX 5090
During the launch event, Nvidia showcased Cyberpunk 2077 running on an RTX 5090 with DLSS 4 enabled, achieving a reported 238fps compared to 106fps on an RTX 4090 with DLSS 3.5.
This improved performance can be attributed to the new Blackwell architecture, advancements in DLSS 4, and increased core counts.
Important thing to note here is the improvement over 4090 is not all raw, the 5090 achieved such performance by combined effort of new blackwell architecture and DLSS 4.
The RTX 5090’s combination of raw performance, advanced features like DLSS 4, and a more compact design makes it a highly desirable card for enthusiasts and professionals who demand the best. However, the $1,999 price tag and the need for a high-wattage PSU may inhibit some potential buyers.
While the RTX 5090 grabs the headlines, the RTX 5080, 5070 Ti, and 5070 are not far behind. Nvidia aims to deliver significant performance improvements across the entire lineup, making next-gen gaming experiences more accessible.
RTX 5080
The RTX 5080 is positioned second to RTX 5090 offers excellent performance without the top-tier price of the 5090. Nvidia claims that the 5080 offers double the performance of the RTX 4080, making it a compelling upgrade for those looking to experience high frame rates at 4K resolution and smooth ray tracing.
Key Specifications of RTX 5080:
- Memory: 16GB GDDR7
- Memory Bandwidth: 960 GB/sec
- CUDA Cores: 10,752
- Total Graphics Power (TGP): 360 watts
- Recommended PSU: 850 watts
Performance and Benchmarks of RTX 5080
Nvidia claims it offers double the performance of the RTX 4080.
The RTX 5080’s 360W TGP is a noticeable increase over the RTX 4080’s 320W, but it’s significantly lower than the 5090’s 575W. This makes it a potentially more power-efficient option for high-end gaming. The $999 price tag places it in the premium segment, but if the performance claims hold true, it could offer excellent value for those seeking a top-tier gaming experience without breaking the bank.
RTX 5070 Ti
This card is aimed at the mid-range market, offering a compelling balance of performance and price. It features 16GB of GDDR7 memory, a memory bandwidth of 896GB/s, and 8,960 CUDA cores.
Key Specifications of RTX 5070 Ti:
- Memory: 16GB GDDR7
- Memory Bandwidth: 896 GB/s
- CUD Cores: 8,960
- Total Graphics Power (TGP): 300 watts
- Recommended PSU: 750 watts
Nvidia claims it’s twice as fast as the RTX 4070 Ti. At $749, it’s expected to deliver excellent performance at 1440p and solid performance at 4K.
With a TGP of 300 watts, the 5070 Ti is more power-efficient than the higher-end 5080 and 5090, making it a potentially attractive option for gamers concerned about power consumption. The $749 price tag positions it competitively in the mid-range market. If the performance claims hold true, the RTX 5070 Ti could become a popular choice for gamers seeking a balance of performance and affordability.
RTX 5070
The RTX 5070 represents the most affordable entry point into the RTX 50 series, aiming to bring next-generation gaming experiences to a broader audience. Nvidia has made a particularly bold claim about this card, stating that it will deliver “RTX 4090 performance at $549.” However, it’s crucial to understand that this claim is heavily contingent on the use of DLSS 4 and likely doesn’t reflect raw rasterization performance.
Key Specifications of RTX 5070:
- Memory: 12GB GDDR7
- Memory Bandwidth: 672 GB/sec
- CUDA Cores: 6,144
- Total Graphics Power (TGP): 250 watts
- Recommended PSU: 650 watts
Performance and Benchmarks of RTX 4070:
Nvidia has made the bold claim that the 5070 will deliver “RTX 4090 performance at $549,” though this is undoubtedly tied to the use of DLSS 4 and not raw rasterization performance. This card aims to provide a next-gen gaming experience at a more accessible price point.
The claim of RTX 4090-level performance at a much lower price is undoubtedly enticing. But the reality is inactual 4090 is still a better performer if we look at complete picture, which you can understand by comaprison below.
The above comparison clearly exposes the over marketing done by Nvidia.
With a TGP of 250 watts, the RTX 5070 is the most power-efficient card in the initial RTX 50 series lineup. This, combined with its $549 price tag, makes it a potentially attractive option for budget-conscious gamers. However, the reliance on DLSS 4 to achieve high-end performance means that the actual experience will depend heavily on game support for this technology.
Architecture and Features: The Engine Room of Performance
The RTX 50 series is built on Nvidia’s new Blackwell architecture, which brings several key improvements over the previous Ada Lovelace architecture. While detailed specifications are still being released, we can focus on the key features and technologies that define this generation:
Blackwell Architecture:
While specific details are still emerging, Blackwell is expected to bring advancements in core design, memory architecture, and overall efficiency. These improvements contribute to the significant performance gains seen across the RTX 50 lineup. Key areas of improvement are expected to include increased SM (Streaming Multiprocessor) counts, enhanced cache hierarchies, and improved ray tracing capabilities.
DLSS 4: The AI-Powered Frame Rate Booster:
As discussed previously, DLSS 4 is a major highlight. This new iteration of Deep Learning Super Sampling uses advanced AI algorithms to generate multiple frames for each traditionally rendered frame. This can significantly boost frame rates, potentially multiplying them by up to 8x, according to Nvidia. DLSS 4 also incorporates new techniques like Multi Frame Generation and real-time application of transformers to improve image quality, reduce ghosting, and add higher detail in motion. It’s also worth noting that some of these DLSS 4 improvements are being backported to older RTX cards via driver updates, which is a significant benefit for existing RTX owners.
RTX Neural Shaders and Neural Faces:
These AI-powered features aim to enhance the visual fidelity of games. RTX Neural Shaders can compress textures more efficiently, potentially reducing storage requirements for games without sacrificing visual quality. RTX Neural Faces uses generative AI to improve the realism of in-game character faces, leading to more immersive and believable characters. These features demonstrate Nvidia’s continued investment in using AI to enhance the gaming experience.
PCIe Gen 5 and DisplayPort 2.1:
The RTX 50 series supports the latest PCIe Gen 5 standard, offering increased bandwidth for data transfer between the GPU and the CPU. This can be particularly beneficial for demanding workloads and future-proofing systems. The inclusion of DisplayPort 2.1 connectors enables support for higher resolutions and refresh rates, including 8K at 165Hz, catering to users with the latest high-end displays. This is important for those pushing the boundaries of display technology.
Power Consumption and Pricing : The Cost of Performance
The RTX 50 series, while promising significant performance gains (but not as wild as claimed by Nvidia), also raises concerns about power consumption and pricing.
Power Consumption – The RTX 5090’s Total Graphics Power (TGP) of 575 watts is a substantial increase compared to the RTX 4090’s 450 watts. This higher power draw will require users to have robust cooling solutions and high-wattage power supplies (Nvidia recommends a 1000W PSU for the 5090).
While Nvidia claims improved power efficiency per watt with the Blackwell architecture, real-world testing will be necessary to confirm this.
The power consumption of the lower-tier cards (5080, 5070 Ti, and 5070) has also increased, though not as dramatically as the 5090. This trend towards higher power consumption might raise some concerns about the overall thermal footprint of high-end gaming PCs.
Pricing and the “CUDA Tax” making RTX 40 valuable
The RTX 50 series carries premium price tags, with the RTX 5090 priced at $1,999, the RTX 5080 at $999, the RTX 5070 Ti at $749, and the RTX 5070 at $549.
These prices place the RTX 50 series firmly in the high-end segment of the market.
The lack of strong competition from AMD in this segment allows Nvidia to maintain these premium prices.
Furthermore, the high demand for Nvidia GPUs in AI and professional workloads, due to their strong CUDA ecosystem, contributes to what some call the “CUDA tax.” This means consumers are often willing to pay a premium for Nvidia cards even if comparable performance might be available elsewhere, simply for the advantages of the CUDA platform.
The continued availability of RTX 40 series cards at potentially lower prices will also be a factor consumers consider.
Availability
While Nvidia has stated its intention for bulk production, the actual availability of the RTX 50 series remains a concern, especially given the past few years of GPU shortages. If demand significantly outstrips supply, we could see inflated prices from scalpers and retailers, making these cards even less accessible.
A Powerful but Pricey Proposition as usual
The Nvidia RTX 50 series represents a significant leap forward in graphics technology, offering substantial performance gains, innovative AI-powered features like DLSS 4, and a surprisingly compact design for the flagship RTX 5090. The Blackwell architecture and advancements in AI are poised to redefine gaming experiences and accelerate workflows in professional applications.
However, these advancements come at a cost. The increased power consumption, particularly for the RTX 5090, raises concerns about electricity usage and cooling requirements. The premium pricing of the entire RTX 50 lineup, coupled with the potential for limited availability and the ongoing “CUDA tax,” makes these cards a significant investment for consumers.
The RTX 5070’s claim of reaching RTX 4090 performance through DLSS 4 is an interesting proposition, but independent benchmarks will be crucial to determine the true extent of this claim and the performance differences in raw rasterization.
Ultimately, the RTX 50 series offers a compelling combination of performance and features, but the high price and power consumption will likely limit its accessibility to enthusiasts and professionals with deep pockets. The availability of the RTX 40 series at potentially lower prices might also influence purchasing decisions. The true value proposition of the RTX 50 series will become clearer once independent benchmarks are released and real-world availability is established. For now, the RTX 50 series is a powerful but pricey glimpse into the future of PC gaming and AI.
Leave a Reply
You must be logged in to post a comment.