We see what to expect when comparing the entry level cards of the series’
For yet another comparison, we see what to expect for the RTX 4070 vs 3070. The lower-end choices for the initial release of the graphics card series.
The series has become more popular with a slightly cheaper price tag on it. While also still providing good performance on top of it.
For release dates, the RTX 3070 has been around for a couple of years now. It launched in October 2020 a bit further down the line from the 20 series launch.
Although plagued with the GPU shortage there wasn’t much stock of the card. With many not able to buy the card but now there has been some improvement.
Whilst the RTX 4070 is rumored to launch at the end of 2022, if not early 2023. As there has not been any confirmation and possible delays from current overstock.
Even with that, we can compare what we know so far to know what to expect from your next best graphics card.
ASUS ROG Strix NVIDIA GeForce RTX 3070 Gaming
Clock Speed
1755MHz Boost Clock (OC)
Thermal Design
Axial-tech Triple Fan
Size
318mm x 140mm x 57mm
Architecture
Being in two different generations there is a change in how they are built. Nvidia has moved from the Ampere architecture in the 30 series to the Ada build in the RTX 4000 series.
Specifically, the 3070 is using the GA104-300 GPU variant. Built with Samsungs 8nm process, it has a die of 392mm² featuring 17.4 billion transistors.
Whilst the 4070 is likely an AD104-275 GPU variant. Based on TSMCs 5nm process it has a die size of 300mm² with yet an unknown transistor count.
Although it is likely the newer chip will contain more. With a more compact build possible it could fit more in.
Specifications
| RTX 4070 | RTX 3070 | |
|---|---|---|
| GPU | AD104-275 | GA104-300 |
| GPU process | TSMC 5nm | Samsung 8nm |
| CUDA cores | 7,680 | 5,888 |
| SMs | 60 | 46 |
| Base clock | 2,310 MHz | 1,500 MHz |
| Boost/Max clock | 2,610/2,800 MHz | 1,750/1,890 MHz |
| Memory | 12GB GDDR6X | 8GB GDDR6 |
| Memory clock | 21 Gbps | 14 Gbps |
| Memory bus | 192-bit | 256-bit |
| Bandwidth | 504 GB/s | 448 GB/s |
| TDP | 285W | 240W |
These specifications are based on leaks and rumors for the 4070. So they may not be final and can change up until the launch.
But we can get a basic understanding of what to expect between the RTX 4070 vs 3070. While also updating with the latest when it comes, we take a deep dive into what the specs mean.
CUDA cores
Compute Unified Device Architecture is Nvidia’s parallel processing system. It is an ecosystem to improve the processing of graphics on the card.
It allows for the GPU cores to work together splitting the workload and improving the speed of processing with what it can achieve.
Although we can compare the numbers available, the efficiency and how it works change between the generations.
The 3070 itself contains 5,888 CUDA cores inside of its core. While the 4070 is rumored to contain 7,680.
A significant increase between generations especially if their power also increases. Which will lead to a more powerful card but also a more power-hungry one.
Frequency
The frequency or clock rate is the oscillation of silicon crystals inside the wafer. It determines how fast the GPU can process instructions.
With faster rates able to get through the data and create the graphics to see your final image.
As there is some control to the spec it is a target for overclocking. This allows you to push the rates above but at the cost of power and heat.
If the new specs are correct, the new generation will see an increase in the overall clock rate. The 4070 has a supposed 2,310 MHz base clock and a 2,610 MHz boost and even a 2,800 MHz max boost clock.
A big improvement over the previous gen and closer to what AMD had to offer on its Radeon cards.
As we see the 3070 with a base clock of 1,500 MHz. Whilst reaching a boost clock of 1,750 MHz and a max boost clock of 1,890 MHz.
Nearly a 1GHz improvement over the last gen. Which will further improve the ability of the cards.
VRAM
In terms of video memory, there are also some improvements. But first, we see what it actually does.
Its purpose of it is as a frame buffer and pixel store. So that you can experience smooth framerates and lower latency.
Especially useful for higher resolutions like 4k. As they have much larger pixel requirements.
Between the RTX 4070 vs 3070, there is a change in capacity and speed. The 40 card bearing 12GB of GDDR6X memory, clocked at 21 Gbps across a 192-bit bus gives a 504 GB/s bandwidth.
Whereas the 3070 is fitted with 8GB of GDDR6 memory. Clocked at 14 Gbps across a 256-bit bus it has a bandwidth of 448 GB/s.
Another area in which the new card has an advantage.
TDP
TDP is the thermal design power of the card. It is an indicator of much energy the GPU will consume at max utilization.
As well as giving you an estimate of how much heat it will give off. Since the work put into running that chip is converted to waste. So it gives an idea of your power bill as well as how much of a radiator it will be.
With the increased power of the new card, we also see an increase in power consumption. As the 4070 is said to have a total card power of 285W.
Whereas, the 3070 sits with a 240W TDP. So although the newer card does require more power it should bring more performance. Therefore we will have to rely on performance per Watt and price comparison.
RTX 4070 vs 3070 performance
We can’t really compare their game performance yet but we do have some preliminary data. With a look at what to expect to see from 3DMark’s Time Spy Extreme benchmark.
It is unlikely to give us a straight answer to the performance it is useful for comparisons. As it is a well-established database with lots of users.
With that, we see the rumors of the graphics score, and the 4070 achieves around 10,000. So that lets us see what it may be like compared to the old.
The 3070 achieves an average of 6,333 in TSE. That means a 58% uplift in the score. Although not final is a significant amount for 19% more power.
Price comparison
Comparing the RTX 4070 vs 3070 on the price we have to do some speculation on it. As it is one of the last things to be confirmed on the card.
So we look back at what previous offerings have to offer and likely increase above that. But the RTX 3070 initially came out with an MSRP of $499 USD/£469.
This would likely be the base price of the 4070 but we could expect a price increase. So we speculate it will cost in the range of $499-$649USD/£469-599.
A likely cause front the cost inflations as well as the rising performance of the series.
MSI Gaming GeForce RTX 3070 Gaming X Trio
Clock Speed
1830MHz Boost Clock (OC)
Thermal Design
TORX Fan 3.0 cooling
Size
323mm x 140mm x 56mm
Is the 4070 worth it over 3070?
It will be best to wait for reviews to come out before deciding on the RTX 4070 vs 3070. This will give us a better understanding of what to expect between the two.
But from the current specifications and performance, it is easy to say the 4070 is the better option. With the likely higher performance but coming at the cost of price and power.
RTX 4070 vs 3070 FAQs
How much better is 4070 than 3070?
Based on Time Spy Extremes’ rumored scores, the 4070 is 68% better than the 3070. This may not be the final result but it is a good indicator. Although it does come at the cost of 19% more power consumption.
How much faster is 4070 than 3070?
From the rumored specs the 4070 is nearly 1 GHz faster than the 3070. That is across the base, boost, and max clock speeds. Specifically, it has a 2,310 MHz base clock, 26,10 MHz boost clock, and 2,800 MHz max boost clock. Whereas the 3070 has a base clock of 1,500 MHz, a 1,750 MHz boost clock, and a max boost of 1,890 MHz.
