Now that the embargo is lifted, I can finally talk about NVIDIA’s newest addition to the Ada Lovelace family, the GeForce RTX 4070 Founders Edition (FE), which I have had running the laps that I always put all GPUs through. There is a fair bit to unpack here, both in terms of its size and design, as well as its performance. That being said, it’s not all that complicated to explain.
Let’s first address the elephant in the room: the size of the RTX 4070 FE. Compared to our RTX 4090 FE, NVIDIA’ top-tier powerhouse GPU dwarfs the newcomer by a staggering scale. Honestly, the vibes that I get are similar to those I had when I first received the RTX 3090 FE and RTX 3070.
Heck, never mind the RTX 4090 FE, the RTX 4070 FE is also shorter than the RTX 3070 Ti, albeit on a far smaller scale. However, despite its diminutive size, the card, as you’ll see in the graphs below, tell a story of a GPU that appears to prioritise power efficiency over performance, in a manner of speaking.
Beneath the hood, the RTX 4070 FE uses the same AD104 architecture as its more powerful TI brethren, but ships out with few CUDA, RT, and Tensor cores, for obvious marketing reasons. Around the outside, NVIDIA has retained the same inverted Dual Axis Flowthrough reference shroud as the rest of its RTX 40 Series cards, which is a little surprising when you recall that the FE versions of the RTX 3070 and 3060 Ti that were cooled with a dual front-fan configuration. Naturally, NVIDIA made a return to form with the Dual Axis Flowthrough design when it launched the RTX 3070 Ti, but even then, its the outline of the FE version of the card resembled the Infinity symbol. Oh, and there’s no RGB LEDs built into the centre of the card.
Other specifications include the same 12GB GDDR6X graphics memory. This again, is another surprise, especially since initial rumours and half-whispers of the RTX 4070 FE had hinted at it being an 8GB, running on the older GDDR6 graphics memory, no less. Unsurprisingly, the card is soldered with NVIDIA’s own 16-pin 12VHPWR PCIe connector, which is a bit of an overkill in my opinion, given the card’s 200W TGP. If you didn’t already know, the 16-pin connector is designed to push as much as 600W, which is understandable when used on the RTX 4090 that demands 450W off the wall.
Lastly, there’s the price of the RTX 4070 FE. At a starting SRP of US$599 (~RM2643), it retails the same as the RTX 3070 Ti when that card initially launched, but in my honest opinion, That is still a little steep for a card that’s supposed to be a cut-down version of the 4070 Ti.
So, to test out the RTX 4070 FE, my testbench comprises the AMD Ryzen 9 7950X, along with 32GB DDR5-6000 RAM from G.Skill. Of course, the motherboard supporting all the components is an X670E motherboard.
For comparison’s sake, I will be benching the RTX 4070 FE against two other GeForce cards: the more powerful Ti variant of this card, and the last generation RTX 3070 Ti FE. Because the testbench is a new system, I also chose to retest the latter graphics card on the same system, so as to ensure a more uniform and fair comparison between the two. Also, I’m limiting the real-world gaming benchmarks to just 1440p and Full HD resolutions. Let’s be real here: if you actually wanted to game at 4K, this card wouldn’t even be a part of your thought and purchasing process.
When it comes to performance, the RTX 4070 FE appears to show two faces: one with the synthetic benchmarks and another within the real-world benchmarks, what we call the gaming portion of our review. With the former, the card is clearly powerful and overtakes the RTX 3070 Ti FE, but not by a margin that the more powerful RTX 40 Series GPUs have been doing since their advent.
The gains are respectable enough, but also an obvious sign that NVIDIA is trying to milk the “mid-range” status of the RTX 4070 FE just enough that it doesn’t the RTX 4070 Ti.
Gaming is where the RTX 4070 FE begins to shine and, once again, as with all Ada Lovelace graphics cards, it benefits from the features and new improvements that NVIDIA has made to the architecture. For the majority of games, the card is either on par with the RTX 3070 Ti FE or just pulling ahead ever so slightly. When it comes to ray-tracing-capable titles, however, that’s where the RTX 3070 Ti begins to show its limitations.
Taking the most obvious ray-tracing-capable title, Cyberpunk 2077, the RTX 4070 FE is obviously able to churn out higher framerates than the RTX 3070 Ti FE, obviously thanks to its support for DLSS 3 and its new Frame Generation technology. Even with the older ray-traced title, Watch Dogs Legion, I can actually see the mid-range Ampere graphics card struggle to maintain the 32 fps average that it scored at 1440p. For the 4070 FE, it barely broke a sweat.
Temperature And Power Consumption
NVIDIA had informed us prior to the review that the RTX 4070 FE would obviously be more power efficient than the previous generation RTX 3070 Ti FE, meaning that the card would also generate considerably less heat. In regards to the heat generation, our card actually generated about the same heat as its predecessor and as you can see in the graph above, it ran about 6°C hotter on average on our final reading, but I have seen the card running cooler, around the 61°C as well.
As for power consumption, I am going to repeat my previous statement: the 16-pin PCIe Gen5 connector is overkill. On average, the card pulls just shy of its full 200W TDP, with some workload and processes even drawing less than that at 170W. On that note, NVIDIA honestly should have just allowed its AiB partners to ship out this card with a standard 8-pin PCIe power port.
NVIDIA’s GeForce RTX 4070 FE is truly the brand’s “bang-for-your-buck” Lovelace GPU, and it’s a graphics card that I believe PC gamers will be more than satisfied to purchase, especially for those that just don’t want to severely cripple their bank account, especially with the base price of components going up.
Note that I said “severely cripple” and not just cripple. That is because, at a starting SRP of US$599 (~RM2643), it is still a rather steep asking price for what is essentially NVIDIA’s only mid-range Lovelace graphics card, at current. Do a quick search on either Lazada or Shopee for an RTX 3060 Ti or RTX 3070, and you’ll see that that card can be bought for around the RM1500 average. Compared to the RTX 4070 FE, that’s an exceedingly good deal, especially if you’re not really psyched up to hitch on to the DLSS 3 bandwagon. That, or you can go and get yourself a Sony PlayStation 5 if all you want to do is game.
Having said all that, the card is still a damn sight cheaper than the other RTX 40 series cards – the RTX 4070 Ti currently sells for about double the price, and the RTX 4080 is nearly three times that. So yeah, if there ever was a time for one to get catch the Ada Lovelace wave on the “cheap”, then the RTX 4070 FE is currently the only card in NVIDIA’s arsenal to do so.
The post NVIDIA GeForce RTX 4070 FE Review: “Bang For Your Buck” Lovelace appeared first on Lowyat.NET.