Review of the Nvidia GeForce RTX 3090 Ti.

What are the specifications of the RTX 3090 Ti?

The RTX 3090 Ti likely represents the concluding chapter for the Ampere generation of graphics cards. It is conceivable that the RTX 30-series GPUs will continue to coexist with the forthcoming RTX 40-series Lovelace cards anticipated for release later this year, and there may be opportunities for refreshed versions of the older architecture to fill various niches within the product lineup.

We have extensively discussed the Ampere architecture; however, to provide a brief summary, the key point is that Nvidia has successfully increased the number of floating point units within the GPU’s Streaming Modules (SMs) by enabling the secondary block of ‘cores’ to perform both integer and floating point calculations. This enhancement significantly boosts the gaming performance of these GPUs.

In addition to this, a significant aspect of a Titan-level graphics card is the inclusion of 24GB of GDDR6X memory.

RTX 3090 Ti vs RTX 3090

Header Cell – Column 0	RTX 3090 Ti	RTX 3090
GPU	GA102-350	GA102-300
Architecture	Ampere	Ampere
CUDA Cores	10,752	10,496
Tensor Cores	336	328
RT Cores	84	82
Base clock	1,560	1,395
Boost clock	1,860	1,695
Memory capacity	24GB	24GB
Memory speed	21Gbps	19.5Gbps
TDP	450W	350W

In comparison to the RTX 3090, which this card is intended to succeed, the differences are minimal, featuring merely an additional 256 CUDA cores, two RT cores, and eight Tensor cores. While the memory capacity remains unchanged, Nvidia has opted for faster memory in this iteration, offering 21Gbps as opposed to the previous 19.5Gbps.

The clock speed of the new GPU is significantly greater, reaching a boost of up to 1,860MHz, compared to the 1,695MHz of the RTX 3090 Founders Edition that we have evaluated. It is important to note that these figures represent the rated speeds. During our testing, the RTX 3090 FE operated at an average frequency of 1,766MHz, whereas the Asus TUF RTX 3090 Ti OC that we are currently examining achieved an average of 1,990MHz in our extended 4K Metro Exodus stress test.

When operating at 1080p resolution, the average GPU clock speed exceeds the 2GHz threshold.

Conversely, it necessitates substantial cooling to operate effectively while maintaining acceptable thermal conditions. This requirement is evident in the Asus TUF RTX 3090 Ti, which features a 3.2x slot design, in contrast to the 2.7x slot design of the previous TUF RTX 3090. Essentially, this graphics card is exceptionally large, bordering on unwieldy. Additionally, it demands three 8-pin PCIe GPU power connectors to connect to its triple-headed power adapter.

The RTX 3090 Ti features a total graphics power (TGP) of 450W, which is prominently displayed. In fact, Asus has incorporated a dedicated heatpipe array above the power components on the TUF card’s printed circuit board (PCB). This design acknowledges the necessity of maintaining optimal temperatures to prevent any potential issues.

It seems fitting to address the noise produced by the Asus TUF RTX 3090 Ti at this juncture. The sound it emits can be described as distressing. I am not referring to the fan noise, which remains remarkably subdued even under heavy use; rather, it is the electrical whine generated when the card is subjected to demanding gaming tasks that is quite difficult to endure.

The GPU is under significant strain, and I cannot help but feel remorse for subjecting it to such evident distress merely to obtain a benchmark for Horizon Zero Dawn.

The performance of the Nvidia RTX 3090 Ti.

What is the performance level of the RTX 3090 Ti?

At this juncture, it is essential to understand that the RTX 3090 Ti stands as the most powerful consumer GPU available, surpassing all other gaming graphics cards in performance. Given its base price, it is only reasonable to expect it to significantly outperform competing GPUs.

The concern lies in the fact that as we have progressed through the performance hierarchy, the costs have escalated considerably. Prior to the compounded challenges posed by GPU mining and supply chain disruptions, the RTX 3080 was available, while the RTX 3090 was priced. However, the enhancement in performance did not correspond to the substantial increase in price. With the introduction of the RTX 3090 Ti, which is priced approximately higher than the standard RTX 3090—a 33% rise—there has been only a marginal performance improvement ranging from 5% to 16%.

It is noteworthy that this situation represents nearly an optimal scenario, as the variant of Nvidia’s premier graphics card being evaluated in our tests is an overclocked Asus TUF RTX 3090 Ti. This particular GPU operates at a clock speed exceeding that of the standard Founders Edition, resulting in marginally improved performance metrics.

It is important to note that the performance difference we are discussing is largely consistent in both gaming and productivity contexts. The RTX 3090 was already an exceptional graphics card for creators, providing significantly more capabilities in a consumer GPU than had previously been available, and surpassing the performance of the RTX 3080’s contemporary architecture by a considerable margin.

While the raw rendering times may not significantly differ among the Ampere GPUs, the 24GB frame buffer of the RTX 3090 facilitates real-time manipulation of large render models in applications such as Blender and Autodesk Maya. This capability extends beyond merely achieving the final render; it encompasses the numerous adjustments made prior to the substantial acceleration provided by the RTX 3090.

The RTX 3090 Ti does not advance significantly beyond its predecessor.

An examination of the Nvidia RTX 3090 Ti.

What implications does the RTX 3090 Ti hold for the gaming community?

This resembles a critique of an Intel processor from the era preceding Alder Lake. While it is true that the RTX 3090 Ti currently stands as the most powerful consumer GPU available, one must question its significance. The diminishing returns in overall gaming performance suggest that a price tag of  is excessive for an individual merely seeking to enjoy some casual polygon rendering.

The current pricing landscape presents a significant challenge, as the two grand price point appears somewhat justifiable within the market context. This is particularly evident when considering the RTX 3080 and RTX 3080 Ti, priced, respectively. However, this scenario is unlikely to persist. With stock levels increasing and prices beginning to decline, the RTX 3080 Ti will eventually return to a more acceptable, albeit still somewhat excessive, manufacturer’s suggested retail price (MSRP). Consequently, the RTX 3090 Ti will seem even more exorbitantly priced in comparison.

The recent price reductions have already influenced the RTX 3090 Ti. Manufacturers have indicated that the cards would have been priced approximately higher just a month ago. While it is improbable that prices will decrease further, the postponed launch has resulted in the RTX 3090 Ti being less financially burdensome than originally anticipated.

Indeed, this represents a somewhat unnecessary luxury for the exceptionally affluent gamer, providing minimal additional performance compared to the graphics cards it aims to supersede. However, what can be said about the professional and creator-oriented aspects? In truth, the RTX 3090 was already capable of handling demanding tasks, establishing itself as a formidable graphics card that could compete with significantly pricier Quadro GPUs. The RTX 3090 Ti appears to be merely a enhancement—it may offer marginally faster rendering, but the significant improvements in workflow efficiency have already been achieved.

 

I remain personally opposed to the use of the GeForce designation. Nvidia emphasizes that this decision is beneficial, as refraining from labeling the RTX 3090/Ti as a Titan card has enabled the company to produce a greater variety of models and distribute them through third-party vendors, rather than solely offering the Founders Edition cards from Nvidia. The reviewer’s guide implies that these cards serve as “an ideal solution for content creators who require substantial GPU memory but do not necessitate the extra professional software optimizations available on TITAN and Quadro GPUs.”

The initial rationale for Titan cards was fundamentally similar, targeting creators who did not require the enhancements or assistance provided by Quadro GPUs. Furthermore, it is my belief that if Nvidia had desired to label it as a Titan and permit the availability of third-party cards, they certainly had the capability to do so.

Referring to it as a GeForce card inherently attracts the attention of gamers, as it is a well-established gaming brand. Consequently, despite Nvidia’s intention to position the RTX 3080 as its flagship Ampere card at the time of its release, the RTX 3090 emerged as the premier choice that gamers aspired to, a status that now extends to the RTX 3090 Ti. Had it been designated as a Titan, it is likely that fewer gaming enthusiasts—those with disposable income but perhaps less familiarity with the intricacies of the graphics card market—would have considered it for their high-end gaming setups. Such products would likely be featured in a less prominent section of pre-built PC retail websites.

This strategy appears to be astute and may have allowed Nvidia to increase sales of its high-end GPUs beyond what would have been possible under the Titan branding. However, I would prefer a more transparent explanation of the underlying rationale. We understand that this is capitalism at work.

Evaluation of the Nvidia RTX 3090 Ti.

Is it advisable to purchase the RTX 3090 Ti?

The Asus TUF RTX 3090 Ti OC Edition, part of the RTX 3090 Ti series, stands out as the most powerful consumer graphics card available on the market. It surpasses all competitors in speed, features, and overall size. Additionally, it comes with a significantly higher manufacturer’s suggested retail price. While this may seem less relevant in the current unpredictable GPU market, I believe it will hold greater significance in the future.

This card is not primarily designed for gamers, nor does it offer significant advancements for creators. While the improved memory speed and increased clock speed do reduce final rendering times, the RTX 3090 had already revolutionized the experience with its real-time interaction capabilities.

If you have a budget of for a GPU at this time, this is the optimal choice. However, it is important to recognize that this represents the maximum capability of the Ampere architecture. The upcoming Lovelace and Radeon RX 7000-series graphics cards are expected to significantly outpace the RTX 3090 Ti, potentially rendering it comparable to the Radeon VII in a short period.