Is the NVIDIA GeForce RTX 5090 Worth Its Jaw-Dropping Price? Find Out Now!

At long last, I present my RTX 5090 review! Interestingly, the NVIDIA Founders Edition of this latest GeForce graphics card sports a distinct design compared to its RTX 4090 FE forerunner.

In this updated version, it boasts a sleeker design accompanied by an innovative fan layout, while improvements to its power connector aim to eliminate any potential issues related to melting cables, as experienced with the previous models.

Nonetheless, the setup utilizing a 600W 12V2x6 hookup is noteworthy given its potential maximum board power of 575W. However, there remains uncertainty regarding NVIDIA’s capacity to manage the inevitable rise in temperature, even with an advanced thermal solution and the cunning DLSS 4 techniques.

Despite boasting a setup that can deliver up to 575W, it’s unclear if NVIDIA can handle the heat increase, even with a new cooling system and smart tricks like DLSS 4.

I’m putting the RTX 5090 FE graphics card to the test using both synthetic benchmarks and gaming trials, with upscaling and frame generation turned on and off, to determine its capabilities and whether it’s truly necessary for anyone to have such immense power under the hood.

RTX 5090: Price, availability, and specs

On January 30, 2025, the top-tier graphics card by NVIDIA, the RTX 5090, will hit the market. The version I’m discussing in this review is the Founders Edition model, which, as customary, comes in a limited number. In addition to this, you can anticipate various models from NVIDIA’s collaborating partners such as ASUS, Colorful, Galax, Gigabyte, Inno3D, MSI, PNY, Zotac, and others to also be available.

The starting price for NVIDIA’s Founders Edition model of their GPU is fixed at $1,999. However, the pricing for custom versions from NVIDIA’s partners is yet to be disclosed, with a likely range spanning both affordable and premium options. In the United States, you can find Founders Edition cards available at both NVIDIA’s official store and Best Buy. Meanwhile, listings for reference cards from other brands have appeared on Newegg and B&H.

On the British side, NVIDIA is providing the Founders Edition, while online retailers such as Overclockers and Scan are listing the reference cards supplied by their hardware partners.

👉 How to buy an RTX 5090 and RTX 5080 on launch day

Compared to the initial price of $1,999, NVIDIA’s RTX 5090 is their costliest consumer graphics card yet, outpacing the launch price of its predecessor, the RTX 4090, by a substantial difference. Let’s take a look at the raw specifications of the RTX 5090 and its previous models.

RTX 4090 vs. RTX 5090 specs

Who is the target audience?

The RTX 4090 has long been considered an expensive, luxury hardware item, and this perception holds even stronger with the introduction of the RTX 5090. However, its steep price tag might not justify the investment for most computer users, whether you’re a gamer or a content creator. Given that it costs $400 more than its predecessor and equals the price of a decent gaming laptop or PC, this is not a decision to be made impulsively.

As an observer, it appears that NVIDIA is primarily aiming its latest top-tier GPU, the RTX 5090, at a diverse group of individuals who can truly leverage its advanced hardware specifications. This group includes enthusiasts of high-definition gaming in 4K, artists and designers seeking to push visual boundaries, scientists and researchers delving into complex computational tasks, engineers working on intricate simulations, and professionals specializing in artificial intelligence. The RTX 5090 stands out as the most potent GPU available today, boasting an unprecedented number of Tensor cores (for AI applications), ray tracing cores, shader cores, and CUDA cores compared to its predecessors.

In the 5090 model, ray tracing performance has significantly increased, reaching 318 teraflops as opposed to the RTX 4090’s 191 teraflops. It boasts over 5,000 additional CUDA cores compared to its previous version, and it also upgrades the video memory to 32GB of GDDR7, moving away from the slower GDDR6X’s 24GB.

The computational power of Tensor cores, a key metric measuring AI tasks’ trillion operations per second capacity, has significantly increased in the new RTX 5090. Specifically, it can now perform approximately 3,352 TOPS as opposed to 1,321 TOPS in the RTX 4090. With its enhanced VRAM, the RTX 5090 is anticipated to be the top-tier GPU for AI applications, making it a worthwhile investment given its superior performance.

NVIDIA’s 50-series graphics cards offer an extra advantage: Multi Frame Generation (MFG) of DLSS 4. Although enhancements in Ray Reconstruction, Super Resolution, and Deep Learning Anti-Aliasing provided by DLSS 4 are beneficial for older RTX cards starting from the 20-series, MFG is currently available only on the RTX 5000 GPUs.

In simpler terms, during the CES 2025 keynote, NVIDIA stated that the RTX 5070 is as powerful as the RTX 4090, thanks in large part to the advancements in Machine Learning Framework (MFG) technology. This tech, which generates frames, puts less strain on hardware other than Tensor cores, thus improving overall performance. However, while NVIDIA hasn’t completely discarded MFG for its older GPUs, the RTX 5090 will still offer the best performance for MFG due to its high number of cores.

Remember that the power consumption of the RTX 5090 has increased to a 575-watt TGP, which is quite a leap from the 450-watt TGP of the RTX 4090. NVIDIA advises using a 1,000-watt power supply for the RTX 5090, while the RTX 4090 was content with an 850-watt one. The additional cores in the RTX 5090 don’t come without their price tag, which isn’t just monetary.

RTX 5090: Design

Instead of performance testing, I got an early look at the unboxing experience from NVIDIA this week, which gave me insights into how they managed to reduce the size of their premier GeForce GPU. To be clear, the reduction was in width, not insignificantly so. Compared to its NVIDIA reference predecessor, the RTX 4090 Founders Edition, the new RTX 5090 Founders Edition is now a two-slot GPU, having been halved in width.

Of course, it also sheds some weight. The RTX 5090 FE is a noteworthy 6.51% lighter, tipping the scales at just 1.82 kg (4.01 lbs), compared to the RTX 4090 FE at 2.18 kg (4.80 lbs) and the RTX 3090 FE, which weighed equally heavy at 2.18 kg before it. To ensure your PCIe 5.0 socket stays safe from harm, I’d suggest using a GPU sag mount as a precaution. Frankly, I’m thrilled by this sleek design and find it refreshing that it buckles the trend of earlier community assumptions.

A more important change than the sleek design of the NVIDIA RTX 5090 is the shift in the power connector’s position, a move aimed at decreasing cable bends and enhancing safety during use. Given that this massive GPU requires up to 575W of power, there’s no room for risk regarding melted cables, a concern carried over from the RTX 4090. This revamped 12V2x6 port addresses the most significant problems associated with its 12VHPWR counterpart by modifying the internal pins within the RTX 5090 itself.

If you have a 600W-rated cable that comes in a power supply capable of delivering over 1000W, you probably won’t need to replace it for the cable shape match. However, using NVIDIA’s adapter that requires four 8-pin PCIe connectors from your PSU could pose cable management challenges. Unfortunately, my 450W 12VHPWR cable isn’t sufficient for the RTX 5090, so I had to find and use four additional cables from my modular power supply.

In simpler terms, when your computer case doesn’t have many openings for cables, a four-way adapter can look messy. For instance, if you’re using a Corsair 4000D Airflow case, the rear panel is already filled with thick cables, leaving little room for additional connections. However, the RTX 5090 has essential connectors that don’t obstruct its “Double Flow Through” fans, which are mounted on the bottom side. In essence, if you’re considering purchasing an RTX 5090, it might be more convenient to buy a new power supply unit (PSU) with a 12V2x6 cable rated for 600W, which will provide sufficient connections and eliminate this issue.

In summary, the new power input placement on the RTX 50 Series graphics cards is an important improvement that will result in cleaner, safer builds within high-power desktop systems. When you combine a top-tier processor like the AMD Ryzen 7 9800X3D with the RTX 5090, it significantly increases the overall power consumption of your custom gaming setup. Therefore, to minimize potential risks related to inferior cabling, it’s advisable to utilize a single 600W-rated 12V2x6 connection and appreciate the streamlined design of the new RTX 50 Series.

RTX 5090: Performance benchmarks

As a passionate tech enthusiast, I was eager to compare the Nvidia RTX 5090 Founders Edition with its predecessor, the RTX 4090 Founders Edition, in GPU-focused benchmarks. The difference in performance between these two powerhouses, considering a 25% increase in price, surpassed my expectations. In synthetic tests, I observed that the RTX 5090 delivers approximately 33% better results than its predecessor. However, it’s important to note that this enhanced performance comes at the expense of higher temperatures during operation.

The performance gains vary across different benchmarks. For instance, in Blender’s 4.2.0 ‘junkshop’ test, the RTX 5090 outperformed its predecessor by an impressive +43.14%. Similarly, in 3DMark’s Steel Nomad, the RTX 5090 demonstrated a significant boost of +49.81%, though it should be noted that this test skips ray tracing capabilities.

As an enthusiast, I’d say: I noticed that for a 25% markup on the Manufacturer’s Suggested Retail Price, the new RTX 5090 delivers approximately 33% improved performance over its RTX 4090 predecessor in synthetic benchmarks. However, it’s important to note that this comes with higher operating temperatures.

In simple terms, both Speed Way and Port Royal showed impressive improvements over the RTX 4090 FE when it comes to real-time ray tracing, with scores of +39.93% and +38.49% respectively. However, in an OpenCL rendering test using Geekbench 6, the increase was lower at +18.84%. It’s important to note that while the Geekbench 6 GPU test focuses on traditional rendering, its Vulkan equivalent performed much better at +35.81%. Given that many modern PC games are now adopting the Vulkan rendering API, this statistic remains significant.

In this extended trial, I repeatedly performed the Speed Way benchmark using 3DMark and closely monitored the real-time statistics of the RTX 5090 FE graphics card. On average, there was a boost in frame rate by approximately 40 FPS, but this improvement came at a lower frequency compared to the RTX 4090 FE. With an additional 150W of power consumption, it resulted in a 6°C increase in temperature. This trend seems consistent when considering gaming performance as well. If NVIDIA reduced the size of its top-tier case without compromises, it appears that this adjustment is reflected by higher temperatures within the same setting.

In essence, the RTX 5090 FE offers impressive synthetic performance, but it comes at a cost: its compact and lightweight design. Whether this trade-off is justified for an extreme GPU that teeters on extravagance is open to discussion. However, many consumers might find it concerning that a fresh card could reach temperatures close to 80°C. If a larger, triple-slot size had resulted in cooler temperatures, it would have been more fitting for a GPU that borders on a luxury item.

RTX 5090 temperature and fan noise

3DMark’s Speed Way test kept the RTX 5090 FE at its maximum, causing hot spots to reach 66.4°C. This temperature is consistent with the dual fans expelling heat via the top-mounted heatsink fins. For comparison, hotspots on the RTX 4090 FE were at 54.7°C when situated in the chassis center and within its own exhaust fan. This represents a significant rise in temperature, which seems to be a result of the smaller chassis used for the RTX 50 Series GPU. Despite various cooling technologies, temperatures have generally risen. This issue might be addressed by future triple-fan cards from third-party manufacturers.

The sound of the fan on the RTX 5090 FE hasn’t become extraordinarily loud, even though a decibel meter reads a peak of 50.5 dBA when the card runs near 80°C. The Double Flow Through fans are operating at maximum capacity, but they produce about as much noise as possible with the PC case side panel removed. Fortunately, my specific model doesn’t suffer from persistent coil whine and sounds similar to the RTX 4090 FE when everything is enclosed for optimal airflow.

RTX 5090: Gaming benchmarks

In a comparison between the RTX 5090 FE and its predecessor RTX 4090 FE, both demonstrate comparable generational advancements when DLSS frame generation is either enabled or disabled. In this test, classic games like Silent Hill 2 (2024), Cyberpunk 2077, Black Myth: Wukong, and Baldur’s Gate 3 run conventionally. However, S.T.A.L.K.E.R. 2 is experimenting with AI frames but no upscaling. The real-time game rendering showcases a performance boost of approximately +25% that aligns with the new RTX 5090’s MSRP. With maximum settings at 1440p resolution, even the “brute force rendering” maintains a minimum frame rate of 60 FPS. Keep in mind that game engines can be temperamental, so further testing will be conducted post-review to account for variations in performance among different titles.

The real-time game graphics perform approximately 25% better compared to the RTX 5090’s new market price, and under “maximum power rendering” conditions, it maintains a minimum of 60 frames per second at a 1440p resolution with Epic/Ultra settings.

If you’re considering playing high-end AAA games without DLSS frame generation, whether it’s using DLSS 4 MFG or not, your frame rate will typically be capped at around 60 frames per second (FPS) when running at maximum settings below native 4K. Although visuals may approach the realm of placebo effects, with many improvements becoming imperceptible to the human eye, sacrificing some unnecessary extras can provide an exceptional 4K gaming experience or even enable DLSS upscaling to 8K when available. Currently, the Nvidia RTX 5090 cards are the only GPUs offering top-tier performance, so this is now the standard for maximum output.

Instead of synthetic benchmarking, PC games tend to operate at similar frequencies, but they consume more power. This increased speed often results in higher temperatures in certain games, usually causing an approximately +6°C rise in exchange for a smoother gameplay without frame drops. In the case of S.T.A.L.K.E.R. 2, choosing frame generation appears to maintain stable temperatures, aligning with NVIDIA’s marketing emphasis for its revamped Multi-Frame Sampling (MFS) technology.

Cyberpunk 2077: DLSS 4 MFG on vs. off

A key aspect of NVIDIA’s unveiling for the RTX 5090 included a bold statement about DLSS 4 Multi Frame Generation significantly boosting performance within Cyberpunk 2077. In my own testing, using a resolution of 2560 x 1440, fullscreen mode exclusively, with the ‘Ray Tracing Overdrive’ graphics test profile selected and DLSS scaling turned off but frame generation set to 4x in ‘Transformer’ mode, the RTX 5090 FE consistently delivered an average of 228 frames per second (FPS) during Cyberpunk 2077’s built-in benchmark.

In Cyberpunk 2077, there’s no discernible lag when NVIDIA Reflex is turned on. Furthermore, the DLSS 4 feature, which generates multiple frames simultaneously at 4x, maintained a consistent 228 frames per second in gameplay, significantly increasing from its original rate of 68 FPS.

Reducing frame generation and DLSS scaling while maintaining all other settings results in approximately a 63 FPS average. This might seem satisfactory when looking at raw numbers, but it suggests that 4K gaming could benefit from NVIDIA’s Tensor Cores assistance. In this generation, DLSS 4 is key for NVIDIA, and surpassing 200 FPS is an impressive feat, considering the debate surrounding “artificial frames.

During gameplay, how does Cyberpunk 2077 perform? With all graphical settings maxed out at 1440p resolution, the game shows no perceptible lag when NVIDIA Reflex is activated, a feature that minimizes latency. In a bustling cityscape, DLSS 4’s multi-frame generation at 4x maintained a consistent 228 frames per second (FPS), jumping from an initial 68 FPS. Interestingly, temperatures remained lower when frame generation was activated, but this is insignificant compared to the impact of DLSS upscaling. The technology is remarkable and offers a viable 4K120Hz gaming experience at maximum settings.

RTX 5090: Competition

On the day of its launch, NVIDIA’s RTX 5090 FE holds an unusual advantage – it has no direct competitors. Instead, other companies like ASUS, MSI, and ZOTAC produce third-party versions that are larger in size to enhance cooling capabilities with a triple-fan setup. Meanwhile, AMD’s Radeon 9000 cards have been delayed until March and aren’t expected to challenge NVIDIA’s top model at this time.

For everyday consumers who desire top-tier performance, it’s wiser to opt for the RTX 5080 with a $999 price tag instead. This graphics card is more suitable for avid gamers and the price gap is substantial. We will provide side-by-side comparisons as soon as possible, but purchasing either GPU at their listed prices remains a rarity. Instead of stretching your budget to reach the RTX 5090 FE, consider waiting for its cheaper sibling, the RTX 5080.

RTX 5090: Should you buy it?

You should buy this if …

✅ You’re in a professional creative field

The RTX 5090 FE exceeds the requirements of 4K and even 8K gaming, finding a closer fit with professional users who typically purchase in large quantities, rather than individuals dealing with increased prices due to resellers.

✅ You won’t settle for anything but the best

Regardless of budget constraints, the RTX 5090 FE would undeniably shine in a stylish, self-designed setup on a content creator’s or professional streamer’s desk. If you manage to secure it at a reasonable cost, its price tag aligns with the performance improvement it offers.

You should avoid this if …

❌ You already own a high-end 40 Series card

Even though the improvement in performance per dollar at the manufacturer’s suggested retail price (MSRP) is reasonable, it’s unlikely that many consumers will be able to find the RTX 5090 FE at its intended price. If you’re not upgrading from an RTX 30 Series or older graphics card, the benefits of DLSS 4 alone may not seem worth the investment.

In simpler terms, the NVIDIA RTX 5090 FE is an extremely powerful GPU that comes with a hefty price tag. Unless you’re gaming at ultra-high resolutions like 4K or 8K, or have expensive additional hardware, it might not be necessary to spend so much. For most dedicated gamers, the RTX 5080 priced at $999 would be a more suitable choice. As for professionals, while the benefits of the RTX 5090 for AI work are still emerging, it should meet most requirements. It’s a costly card, but if it boosts your work efficiency, it could potentially pay for itself in a professional setting, provided you avoid the inflated prices due to scalpers and market manipulation.

1. The RTX 5090 has become the top-tier consumer graphics card globally, with its equivalent versions from other brands pushing their internal components for enhanced performance. However, I stand by my statement that it’s not necessary for gaming; the RTX 5080 is a better high-end choice in that regard.

2. The redesign of the RTX 5090 is stunning, reducing the former giant into a more streamlined and practical device. However, this upgrade results in higher temperatures and a substantially increased power consumption.

3. DLSS 4 is an impressive technology, but it doesn’t quite persuade me to abandon the RTX 4090. This is especially true if NVIDIA decides to bring it back to the 40 Series.

2025-01-23 17:10