Nvidia Mobile RTX 4070, 4060, 4050 Benchmarks Arrive, Die Sizes Confirmed
Many changes and improvements have been made to the Nvidia Ada Lovelace architecture to power some of the best graphics cards. And the RTX 4090, 4080, 4070, 4060, and 4050 variants are starting to ship, powering some of the best gaming laptops out there. The last three are especially interesting, even if you’re not into gaming laptops. It also gives you hints that you might get from an underlying desktop PC.
Computerbase posted a benchmark (opens in new tab) Mobile 4070, 4060, and 4050 photos. At the same time, someone took a picture of the AD106 and AD107 chips that power these GPUs to confirm the die size.sky juice (of Anglonomics (opens in new tab) fame for providing the first definite details of AMD’s GPU chiplet architecture) murmured (opens in new tab) AD106 is 186mm^2 and AD107 is 156mm^2, slightly smaller and larger than previous estimates respectively. (opens in new tab).
AD106: 186mm² 36SMAD107: 156mm² 24SMTime Spy Graphics Curve vs GA104: 4050 stops scaling at 95W 4060 stops scaling at 105W 4070 stops scaling at 105WFebruary 21, 2023
As expected, benchmarks show that these mobile GPUs are hitting their power limits, resulting in limited performance. As we’ve seen with the 160W RTX 4090 and 4080 mobile parts, there’s only so much you can do without increasing power consumption. One of the reasons the RTX 4090 desktop card has a TGP of 450W is that it doesn’t need to use that much power, but if you do, it can offer significantly more performance than the 320W RTX 4080.
Here’s a quick summary of the specifications of the various mobile parts (although some factors such as transistor count remain unconfirmed).
GPUs | RTX 4090 laptop | RTX 4080 Laptop | RTX 4070 laptop | RTX 4060 laptop | RTX 4050 laptop |
---|---|---|---|---|---|
architecture | AD103 | AD104 | AD106 | AD107 | AD107 |
process technology | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N |
Transistor (billion) | 45.9 | 35.8 | ? | ? | ? |
Die size (mm^2) | 378.6 | 294.5 | 186 | 156 | 156 |
streaming multiprocessor | 76 | 58 | 36 | twenty four | 20 |
GPU core (shader) | 9728 | 7424 | 4608 | 3072 | 2560 |
Boost Clock (MHz) | 1455–2040 | 1350–2280 | 1230–2175 | 1470–2370 | 1605–2370 |
Tensor cores | 304 | 232 | 144 | 96 | 80 |
Ray Tracing Core | 76 | 58 | 36 | twenty four | 20 |
TMU | 304 | 232 | 144 | 96 | 80 |
ROP | 112 | 80 | 48 | 32 | 32 |
L2 Cache (MB) | 64 | 48 | 32 | 32 | 12 |
memory interface (bits) | 256 | 192 | 128 | 128 | 96 |
Memory speed (GT/s) | 18-20 years old | 18-20 years old | 16 | 16 | 16 |
TGP (Watts) | 80-150 | 60 to 150 | 35–115 | 35–115 | 35–115 |
All mobile chips have a fairly wide range of power options depending on the specific laptop configuration and implementation. The 4090 and 4080 can also exceed the 150W TGP by up to 25W in some cases with “Maximum Dynamic Boost 2.0”. The maximum number of clocks also depends on power constraints and workload. Most chips may use clocks hundreds of MHz lower than desktop chips for typical gaming workloads.
Computerbase tested performance using Clevo (XMG) and Gigabyte laptops for the RTX 4070, 4060 and 4050 mobile variants. The XMG model used the maximum configuration with a maximum power of 140W, while the Gigabyte Aero 14 OLED limits the RTX 4050 to a maximum power of 45W (10W via Dynamic Boost).
Overall, the RTX 4070 laptop averaged 44 fps at 2560×1440 across the seven-game test suite, while the RTX 4060 dropped to 37 fps. With ray tracing enabled, performance dropped to 34 fps on 4070 and 30 fps on 4060. The RTX 4050 was not tested at 1440p. With only 6GB of VRAM, performance could be very slow (with or without ray tracing). ).
At 1080p, things are much better, with the RTX 4070 delivering an average of 76 fps, the 4060 65 fps, and the RTX 4050 (low power envelope) 38 fps. With ray tracing, these scores drop to 59 fps on the 4070 and 51 fps on the 4060, and the 4050 has not been tested with ray tracing enabled.
It does not have DLSS 2/3 (if supported), so you can get better performance while targeting higher resolutions. Computerbase shows his DLSS scaling with and without frame generation. spiderman remastered at 1440p. DLSS performance mode with frame generation nearly doubled performance for both 4070 and 4060, as expected.
Overall, the low-tier RTX 40-series mobile GPUs look good, but what you actually get will depend on other components like CPU and display, as well as the specific configuration of the GPU. It’s also an interesting change of pace for Nvidia, launching the 4070, 4060 and 4050 in laptops first and still waiting for their respective desktop implementations. No doubt it is at least partly due to this.
Current rumors point to a desktop RTX 4070 likely coming in April, but it’s unclear when the 4060 and 4050 will arrive. It could launch on Computex in June, but it could very much depend on how quickly the existing parts inventory clears up and how much demand Nvidia expects.