NVIDIA GeForce RTX 2070 Mobile Refresh

The NVIDIA GeForce RTX 2070 Mobile Refresh is a GeForce 20-series mobile GPU built on the Turing architecture. It uses the TU106B chip manufactured by TSMC on a 12 nm process, with 10,800 million transistors on a 445 mm² die and a transistor density of 24.3M / mm². Its base clock is 1260 MHz, its boost clock is 1455 MHz, and its memory runs at 1375 MHz with 11 Gbps effective transfer. The GPU has 8 GB of GDDR6 on a 256-bit bus, yielding 352.0 GB/s of bandwidth. It belongs to the GeForce 20 Mobile generation, was released on 2020-03-03, and is currently marked end-of-life. In the database, the percentile vs all GPUs is 50, while the benchmarks array is empty and the average benchmark score is 0.

Who Should Consider It

The data positions this GPU at the 50th percentile across all GPUs in the database, which places it at the midpoint of the overall performance distribution rather than at the top end. Because no benchmark scores are stored for this specific refresh, resolution and settings guidance must be inferred from the listed memory and compute specifications.

The memory side is defined by 8 GB of GDDR6 and 352.0 GB/s of bandwidth, both of which are directly relevant to high-resolution rendering. Larger frame buffers help hold the render targets, texture sets, and depth buffers that become heavier as output resolution climbs. The 256-bit bus width provides the path over which those memory contents move. The compute side contains 2304 shading units, 144 texture mapping units, and 64 ROPs, with a peak FP32 throughput of 6.705 TFLOPS. These figures describe a mobile part intended for moderately high detail levels in a portable system.

The form factor is a decisive factor for potential buyers. The slot width is listed as MXM Module, and the power connectors are None, which indicates a module designed to be integrated into a laptop or portable chassis rather than installed as a desktop card. The 115 W TDP gives the power envelope for that mobile context. Because the production status is end-of-life, this is not a current-generation product; the successor is GeForce 30 Mobile. Someone considering this GPU would need to be working with a system that accepts a Turing-generation MXM module and that can deliver power through the module interface. The 1260 MHz base clock and 1455 MHz boost clock define the sustained and maximum frequency range, and the pixel rate of 93.12 GPixel/s and texture rate of 209.5 GTexel/s provide the throughput ceilings for high fill-rate workloads. Without measured results, the strongest statement supported by the data is that this GPU has a mid-range global ranking, a mobile form factor, and enough memory bandwidth for demanding frame-buffer workloads.

Memory Subsystem

The memory subsystem consists of 8 GB of GDDR6, a 256-bit bus width, and 352.0 GB/s of bandwidth. The memory clock is 1375 MHz, which is stated as 11 Gbps effective. The effective data rate is the figure that matters when moving data across the bus. At higher resolutions, the capacity of the frame buffer and the speed at which it can be read determine whether a GPU can keep textures and render targets resident. The 8 GB capacity is the amount of memory available for those tasks, while 352.0 GB/s is the theoretical bandwidth limit.

The interaction between memory and the rest of the GPU is visible in the shading and texturing rates. With 2304 shading units and 144 TMUs, the GPU can issue a large volume of texture operations; the texture rate of 209.5 GTexel/s is the peak rate for those operations. Feeding that texture rate requires the 256-bit bus to deliver data quickly. Likewise, the pixel rate of 93.12 GPixel/s sets the maximum speed at which pixels can be written to the frame buffer, and high pixel counts at high resolutions place direct pressure on memory bandwidth. The listed GDDR6 type is the current memory technology in this entry, and the bus is 256 bits wide. No benchmark data is available to translate these specifications into playable settings, but the memory subsystem numbers describe a configuration that balances capacity and bandwidth for a mobile Turing GPU.

Ray Tracing and Feature Set

This GPU includes 36 RT cores and 288 tensor cores within the Turing architecture. Those core counts are the hardware feature-set facts present in the data, along with the API list: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. DirectX 12 Ultimate (12_2) support indicates a modern API feature level in the database’s classification, and Vulkan 1.4 is the listed Vulkan version. OpenGL 4.6 is the supported OpenGL version.

The FP16 performance is 13.41 TFLOPS, listed with a 2:1 ratio to the FP32 figure of 6.705 TFLOPS. The tensor cores are present in the count of 288, and the RT cores are present in the count of 36, giving the Turing media a set of specialized hardware blocks. The bus interface is PCIe 3.0 x16, which is the host connection for the module. Display outputs are listed as portable-device dependent, meaning that the actual connectors depend on the laptop’s implementation rather than on the GPU module itself. The feature set is therefore defined by the Turing core arrangement and the API support specified in the data.

How It Compares

The nearestRivals array in the data is empty, so there are no named rival GPUs and no rival scores or deltaPct values to report. This means a per-rival comparison cannot be constructed from the FACT PACK. The only global ranking fact is the 50th percentile vs all GPUs. The average benchmark score is 0, which aligns with the empty benchmarks list rather than indicating a measured result.

The entry still has a positional context through its generation and production status. Its predecessor is GeForce 10 Mobile, and its successor is GeForce 30 Mobile, placing the RTX 2070 Mobile Refresh between those two mobile generations. It is part of the GeForce 20-series and specifically the GeForce 20 Mobile generation. The chip is TU106B, a Turing die. Because the nearestRivals list is empty, no direct paragraph can be written for any competitor; the only quantitative comparison available is the percentile field, which puts this GPU at the midpoint of all GPUs in the database.

Benchmark Performance

The benchmarks array contains no entries, so no game or synthetic benchmark results are attached to this GPU. The average benchmark score is 0, which in the absence of any benchmark entries is not a meaningful performance measurement. The percentile vs all GPUs is 50, indicating a median position among all GPUs in the database. This is the only performance-relative metric present.

Because nearestRivals is empty, there are no deltaPct values, and exact percentage differences versus competitor GPUs cannot be derived. The theoretical throughput figures listed in the data are the FP32 rate of 6.705 TFLOPS, the FP16 rate of 13.41 TFLOPS with a 2:1 ratio, the pixel rate of 93.12 GPixel/s, and the texture rate of 209.5 GTexel/s. These are peak specification limits rather than measured benchmark outputs. The boost clock of 1455 MHz is the maximum clock listed, the memory effective rate is 11 Gbps, and the memory bandwidth is 352.0 GB/s. In short, the performance data available for this entry reduces to a 50th percentile ranking; the benchmark field and average score do not supply additional measurement points.

FAQ

Q: When was the NVIDIA GeForce RTX 2070 Mobile Refresh released?

A: It was released on 2020-03-03, and its production status is end-of-life.

Q: What memory configuration does it have?

A: It has 8 GB of GDDR6 on a 256-bit bus, with a memory clock of 1375 MHz, an effective data rate of 11 Gbps, and bandwidth of 352.0 GB/s.

Q: What APIs are supported?

A: The GPU supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: How many RT and tensor cores does it have?

A: It has 36 RT cores and 288 tensor cores, within the Turing architecture.

Q: What are its compute unit counts?

A: It has 2304 shading units, 144 TMUs, and 64 ROPs, with FP32 performance of 6.705 TFLOPS and FP16 performance of 13.41 TFLOPS at a 2:1 ratio.

Q: Where does it sit in the database’s performance ranking?

A: The percentile vs all GPUs is 50, while the benchmarks array is empty and the average benchmark score is 0.