The presentations are done, and yesterday the mystery surrounding everything related to the RTX 40 SUPER series was finally solved. Interestingly, it raised another question we are addressing today: Why has NVIDIA equipped the RTX 4070 Ti SUPER with 112 ROPs? The reason is because it is aimed directly at competing with AMD’s RX 7900 XT, forcing AMD to release a refresh to maintain their strong sales.
There is no doubt about the data, and we will present it in percentages for better understanding of NVIDIA’s intentions with this launch and how almost everyone has overlooked the fact that this RTX 4070 Ti SUPER is actually a disguised, more affordable, and lower consumption RTX 4080.
The NVIDIA RTX 4070 Ti SUPER will have 112 ROPs; is it the RX 7900 XT killer?
It certainly seems so. The strategy is very well-thought-out by the green team, and we will explain it in detail and specificity. There are three GPUs in play: RTX 4080, RTX 4070 Ti SUPER, and RTX 4070 Ti, so we will make two comparisons on paper:
RTX 4070 Ti vs RTX 4070 Ti SUPER
RTX 4070 Ti SUPER vs RTX 4080
Starting with the first comparison, specifications, and percentage data:
RTX 4070 Ti | RTX 4070 Ti SUPER | Percentage Difference
Shaders: 7680 | 8448 | +10%
SM: 60 | 66 | +10%
TMU: 240 | 264 | +10%
ROP: 80 | 112 | +40%
Tensor Cores: 240 | 264 | +10%
RT Cores: 60 | 66 | +10%
L2: 48 MB | 49 MB | +2.08%
Base Clock: 2310 MHz | 2340 | +1.29%
Boost Clock: 2610 MHz | 2610 | 0%
Bus: 192 bits | 256 bits | +33.33%
GDDR6X Memory: 12 GB | 16 GB | +33.33%
VRAM Speed: 21 Gbps | 21 Gbps | 0%
TGP: 285 watts | 285 watts | 0%
MSRP: $799 | $799 | 0%
Looking at the table and percentage values, it is clear that something doesn’t add up, and that something is the ROPs. The Render Output Units, also called Raster Operation Pipelines by NVIDIA, or ROPs, play a crucial role in the final rendering phase. Without going into too much detail, the ROPs perform various functions, such as checking depth, merging different framebuffers, swapping information with RAM, and ultimately assigning color and depth to each pixel after the shaders’ prior work.
Therefore, the ROP is responsible for writing to the framebuffer to send the completed FPS to the monitor, and as they have been separate and independent units from the TMUs for years, NVIDIA can balance the number of ROPs with only two trade-offs: physical space on the die and a higher power consumption as more ROPs are added.
To achieve this, NVIDIA modifies the Raster Engine within each GPC, allowing it to maintain intact SMs and balance the number of ROPs needed.
What is NVIDIA aiming for with the 112 ROPs on the RTX 4070 Ti SUPER and the +40% increase?
They are attempting to address a scaling issue that the RTX 4070 Ti currently faces. This is crucial because NVIDIA wants to compete in 4K against the RX 7900 XT. Including more ROPs is essential when increasing the bus and the amount of VRAM while also boosting the rest of the specifications.
It is directly linked to the interface and the amount of memory since it will have to write many more times to the framebuffer and work much more intensively with the system’s RAM to push performance with high-load textures. The original RTX 4070 Ti competed very well against the RX 7900 XT at 1080p and 2K resolutions, but its performance fell significantly short at 4K.
This was intentional because NVIDIA wanted this GPU to focus on players with monitors that support up to 2K resolution. However, seeing AMD’s strong performance with their GPU, the green team had to adjust their strategy. Knowing that the RTX 4080 already has 112 ROPs, the easiest solution was to use the same AD103, limit its capabilities, and maintain the ROP count while reducing the number of SMs.
Is it a disguised RTX 4080?
RTX 4070 Ti SUPER | RTX 4080 | Percentage Difference
Shaders: 8448 | 9728 | -13.15%
SM: 66 | 76 | -13.15%
TMU: 264 | 304 | -13.15%
ROP: 112 | 112 | 0%
Tensor Cores: 264 | 304 | -13.15%
RT Cores: 66 | 76 | -13.15%
L2: 49 MB | 64 MB | -23.43%
Base Clock: 2340 | 2205 MHz | +6.12%
Boost Clock: 2610 | 2505 MHz | +4.19%
Bus: 256 bits | 256 bits | 0%
GDDR6X Memory: 16 GB | 16 GB | 0%
VRAM Speed: 21 Gbps | 22.4 Gbps | -6.25%
TGP: 285 watts | 320 watts | -10.93%
MSRP: $799 | $1,199 | -33.36%
If we look at the table above, it appears so. As we can see, the biggest percentage difference is in the L2, which is 23.43% lower in the new NVIDIA GPU, but the key specifications are only 13.15% worse, and the average frequency scales up by almost 5%.
Therefore, comparing the RTX 4070 Ti SUPER to the current RTX 4080 gives us a very close performance at 1080p and 2K resolutions, and a slightly larger gap at 4K in favor of the more powerful of the two models, heavily influenced by its larger cache, which is necessary given the 13.15% increase in units and its higher VRAM speed.
However, in terms of ROPs, they are evenly matched, proving that NVIDIA has indeed created the RX 7900 XT killer. AMD’s GPU suffers from the same issue as the original RTX 4070 Ti: poor scaling as resolution increases. The RX 7900 XT fails to keep pace with the RTX 4080 between 1080p, 2K, and 4K resolutions, losing steam.
The RTX 4070 Ti SUPER with its 112 ROPs will address this issue, not only delivering performance similar to AMD’s offering but also providing more consistent, versatile performance, which could attract a larger number of users, as they have similar launch prices.
The pride of NVIDIA and AMD’s sales: why didn’t the RTX 4080 price drop in response to poor results?
Clearly, unless AMD lowers the price of the RX 7900 XT to soften the blow, which would also be good news. In any case, NVIDIA will have higher profitability, using the binned AD103 and leftover chips from the RTX 4080 (the worst-selling GPU in history due to its high price) to create the AD103-275, making good money from it. After all, if the RTX 4080s in stock don’t sell, the lower-priced RTX 4070 Ti SUPER certainly will.
As was recently stated on the internet: “NVIDIA will not lower prices. It will sell more powerful new GPUs at lower prices than the originals, with better performance/cost ratios, but will not offer discounts on existing models.” It appears to be a matter of pride and determination not to admit a mistake. In the end, they release a new model, almost as fast and at a lower price, to avoid admitting they were wrong.
It is a welcome development, of course, but AMD has claimed sales throughout the process. It’s undoubtedly an absurd situation, but it seems to be the green team’s stance. Regardless, the RTX 4070 Ti SUPER is an option worth considering for 4K and high-end gaming; some might say it is what the original RTX 4080 should have been, and they might not be wrong.
The entry NVIDIA goes after the AMD RX 7900 XT: the RTX 4070 Ti SUPER has 112 ROPs, like the RTX 4080 first appeared on El Chapuzas Informático.
Need Help?
Can't Find What Your Looking For? Just fill out our form and we'll ask our trusted distributors