The RTX 4070 will be 27% slower than the Ti due to its “low” frequency
The RTX 4070 Ti is being a “success” in sales compared to its older sisters. It is, to this day, the most balanced card within the NVIDIA range and it is being noticed, even against its most optimal price rivals such as the RX 7900 XT. With these visas, those of Huang intend to continue lowering the range with the RTX 4070which has seen its frequency filtered and… Again there will be controversy, since comparatively speaking, with their older sister in mind, they are relatively low.
As they used to say in certain neighborhoods, “you’re going to see yourself in a mess that you don’t know where you’ve gotten into.” Unfortunately there is going to be controversy with performance and frequencies, because NVIDIA is targeting a market niche that… Either the price makes up for it, which is unlikely, or AMD has it really good as soon as it scales its range as described. wait.
NVIDIA RTX 4070, frequencies below expectations
Reference RTX 4070 has a Base clock of 1920 and 2475 boost
first AIB to finish a 4070 is Colorful with their Ultra W OC line, boost of 2505 MHz
— Matthew Smith ᅠ (@T4CFantasy) January 30, 2023
Well, at least given the information and who offers it, so it will be. It has been, once again, Matthew Smithan engineer who works in TechPowerUp and that the one who has slipped the information is in charge of the database of said website for the GPUs.
Specifically and as can be seen, Matthew ensures that the BaseClock is situated in 1920MHzMeanwhile he Boost go up to the 2.475MHz. These data are certainly not as expected, especially since normally at a lower range the frequencies scale as the chips are less “greedy” and smaller, which allows this increase in frequencies, but in this case it will not be like that.
And it will not be because it is a cut version of the RTX 4070 Ti chip, that is, the AD104 that this time will arrive as AD104-250 clipping the 7,680 Shaders to 5,888. The minimum expected would be to match the 2,610 MHz of her older sister, but neither …
A more than considerable distance
What segmenting up in price has is that when you have to do it down in performance, it results in a very wide GAP from card to card. We see it with the RTX 4090 vs RTX 4080 and this one against the RTX 4070 Ti. It will happen again to everyone’s misfortune, since this RTX 4070 and its low frequency will give a moderate performance of 29 TFLOPS (approximately the same as an RTX 3080, saving the architectural distances, which are few as we have already seen), which, when faced with 40 TFLOPS and the same architecture, gives us 27.5% less performance in FP32.
Provided that it complies with the same bus, the same amount of VRAM, the same frequency in it, and therefore, the same effective bandwidth, the scalability in performance will be perfect. The only good thing about all this is that consumption is also reduced to more normal figures, reaching 200 watts for this RTX 4070, all thanks to its lower frequency and Shaders, obviously.
In the absence of knowing the price, given its performance, the card should cost around $650 as MSRP to continue with parity in performance/price/watt scalability, since in consumption the RTX 4070 reduces watts by 30% as TBP . As for the release date, it seems that we will see it at the end of February or mid marchThey are still rumors, but since the RTX 40 Mobile range is launched on the 22nd of next month, everything indicates that NVIDIA will continue to push, shortening times and giving AMD less GAP.