AMD Radeon R7 M370 vs NVIDIA GeForce RTX 5070 Ti SUPER Comparison

The AMD Radeon R7 M370 is an end-of-life mobile GPU from 2015, built on GCN 1.0 architecture and a 28 nm process, while the NVIDIA GeForce RTX 5070 Ti SUPER is an active desktop part from the GeForce 50-series, using Blackwell 2.0 architecture and a 5 nm node. The R7 M370’s average benchmark score is 6764, placing it in the 38th percentile of all GPUs, while the RTX 5070 Ti SUPER’s average score is 6270, placing it in the 36th percentile. These are different classes of hardware, but the benchmark data reveals a counterintuitive comparison that hinges on which specific tests are run.

Head-to-Head Benchmarks

The R7 M370 posts an average benchmark score of 6764, which is 7.3% higher than the RTX 5070 Ti SUPER’s average of 6270. That raw average is dominated by the R7 M370’s two Geekbench results: a score of 7063 in Geekbench OpenCL and 6465 in Geekbench Vulkan. In contrast, the RTX 5070 Ti SUPER’s only listed benchmark is a 3DMark Steel Nomad DX12 score of 6269.5. The R7 M370’s OpenCL result alone is 12.7% above the RTX 5070 Ti SUPER’s 3DMark score, and its Vulkan score is 3.1% higher. On the basis of these specific tests, the R7 M370 wins outright in both of its recorded disciplines.

However, the RTX 5070 Ti SUPER’s 3DMark Steel Nomad DX12 score of 6269.5 is nearly identical to its average of 6270, and this test is a modern DX12 workload that stresses features the R7 M370 cannot even access, such as ray tracing hardware. The R7 M370 has no RT cores and no tensor cores, while the RTX 5070 Ti SUPER has 70 RT cores and 280 tensor cores. The R7 M370 supports DirectX 12 (11_1), whereas the RTX 5070 Ti SUPER supports DirectX 12 Ultimate (12_2). The benchmark data does not include a shared test between the two, so direct head-to-head comparisons are limited to their respective average scores and percentile placements.

Looking at nearest rivals, the R7 M370 sits 1% behind the AMD FirePro M5100 (avg 6830) and 1% ahead of the NVIDIA GeForce GT 1010 (avg 6698). It is 2.3% ahead of the AMD Radeon R7 M460 (avg 6612) and 2.6% behind the NVIDIA GeForce GTX 675M (avg 6946). The RTX 5070 Ti SUPER, by contrast, is exactly tied with the NVIDIA GeForce RTX 4070 Ti SUPER AD102 (avg 6270), 0.2% ahead of the NVIDIA Quadro K620 (avg 6282), 0.8% ahead of the AMD FirePro W600 (avg 6223), and 0.9% behind the AMD Radeon R7 M350 (avg 6327). These deltaPct values show that the RTX 5070 Ti SUPER’s score is tightly clustered with older, lower-end GPUs in this benchmark pool, which is unusual for a 2025 flagship-tier part.

The performance gap becomes clearer when examining raw specifications that drive compute. The R7 M370 has 384 shading units, 24 TMUs, and 8 ROPs, delivering 737.3 GFLOPS FP32, 23.04 GTexel/s, and 7.680 GPixel/s. The RTX 5070 Ti SUPER has 8960 shading units, 280 TMUs, and 96 ROPs, delivering 43.94 TFLOPS FP32, 686.6 GTexel/s, and 235.4 GPixel/s. That is roughly 59.6 times higher FP32 throughput, 29.8 times higher texture rate, and 30.7 times higher pixel rate. Yet the RTX 5070 Ti SUPER’s benchmark score is lower. This discrepancy suggests that the 3DMark Steel Nomad DX12 test may be bottlenecked by something other than raw compute, or that the R7 M370’s Geekbench tests are more favorable to its architecture.

Clock speeds also differ drastically. The R7 M370 runs at 875 MHz base and 960 MHz boost, with memory at 900 MHz (3.6 Gbps effective). The RTX 5070 Ti SUPER runs at 2295 MHz base and 2452 MHz boost, with memory at 1750 MHz (28 Gbps effective). The RTX 5070 Ti SUPER’s memory bandwidth is 896.0 GB/s versus 57.60 GB/s for the R7 M370, a 15.6-fold advantage. Its 16 GB GDDR7 memory is 8 times the capacity of the R7 M370’s 2 GB GDDR5. The R7 M370’s pixel rate of 7.680 GPixel/s is less than one-thirtieth of the RTX 5070 Ti SUPER’s 235.4 GPixel/s.

The Verdict

The data shows that the RTX 5070 Ti SUPER is the superior hardware by nearly every measurable specification, but the benchmark scores tell a different story. The R7 M370 wins on average benchmark score (6764 vs 6270), Geekbench OpenCL (7063 vs none recorded), and Geekbench Vulkan (6465 vs none recorded). The RTX 5070 Ti SUPER wins on every architectural and memory metric: FP32 compute is 59.6 times higher, memory bandwidth is 15.6 times higher, memory capacity is 8 times higher, and it has dedicated RT and tensor cores. The RTX 5070 Ti SUPER also supports modern APIs including DirectX 12 Ultimate and Vulkan 1.4, while the R7 M370 is limited to DirectX 12 (11_1) and Vulkan 1.2.170.

For a builder making a choice strictly from this data, the RTX 5070 Ti SUPER is the obvious pick for any modern workload. Its 3DMark Steel Nomad DX12 score of 6269.5 may be lower than the R7 M370’s Geekbench scores, but that test is designed to stress current-generation features like ray tracing and mesh shaders, which the R7 M370 cannot handle. The R7 M370’s scores come from older Geekbench workloads that do not leverage those features. The RTX 5070 Ti SUPER’s launch MSRP is 749 USD, and it is an active product with a 350 W TDP, dual-slot cooler, and a 16-pin power connector. The R7 M370 is end-of-life with no TDP listed, no slot width, and no launch MSRP.

The R7 M370 is a legacy part that only makes sense in a retro or low-power context, but even then, its benchmark wins are marginal. It is 1% faster than the NVIDIA GeForce GT 1010 and 2.3% faster than the AMD Radeon R7 M460, but those are all ancient mobile GPUs. The RTX 5070 Ti SUPER ties the RTX 4070 Ti SUPER AD102 and edges out the Quadro K620, which are also older parts, but it does so in a modern DX12 test. The verdict is clear: the RTX 5070 Ti SUPER is the only rational choice for any current or future application, despite its lower benchmark average.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon R7 M370 has an average benchmark score of 6764, which is 7.3% higher than the NVIDIA GeForce RTX 5070 Ti SUPER’s average of 6270.

Q: Does the RTX 5070 Ti SUPER support ray tracing?

A: Yes, the RTX 5070 Ti SUPER has 70 RT cores and 280 tensor cores, while the R7 M370 has none of either.

Q: What is the memory capacity difference?

A: The RTX 5070 Ti SUPER has 16 GB of GDDR7 memory, while the R7 M370 has 2 GB of GDDR5 memory.

Q: Which GPU has a higher boost clock?

A: The RTX 5070 Ti SUPER has a boost clock of 2452 MHz, compared to the R7 M370’s boost clock of 960 MHz.

Q: How do their nearest rivals compare?

A: The R7 M370 is 1% behind the AMD FirePro M5100 and 1% ahead of the NVIDIA GeForce GT 1010. The RTX 5070 Ti SUPER is tied with the RTX 4070 Ti SUPER AD102 and 0.2% ahead of the Quadro K620.

Q: Which GPU has a higher pixel rate?

A: The RTX 5070 Ti SUPER has a pixel rate of 235.4 GPixel/s, which is about 30.7 times higher than the R7 M370’s 7.680 GPixel/s.

Specification Differences

The two GPUs differ on nearly every listed specification. The R7 M370 uses 2 GB of GDDR5 memory on a 128-bit bus with 57.60 GB/s bandwidth, while the RTX 5070 Ti SUPER uses 16 GB of GDDR7 on a 256-bit bus with 896.0 GB/s bandwidth. The R7 M370 has 384 shading units, 24 TMUs, and 8 ROPs, whereas the RTX 5070 Ti SUPER has 8960 shading units, 280 TMUs, and 96 ROPs. The R7 M370’s clocks are 875 MHz base and 960 MHz boost, with memory at 900 MHz (3.6 Gbps effective); the RTX 5070 Ti SUPER runs at 2295 MHz base and 2452 MHz boost, with memory at 1750 MHz (28 Gbps effective). The R7 M370 has a PCIe 3.0 x8 interface, while the RTX 5070 Ti SUPER uses PCIe 5.0 x16. The R7 M370 has no listed TDP, slot width, power connectors, or display outputs; the RTX 5070 Ti SUPER has a 350 W TDP, is dual-slot, uses one 16-pin connector, and offers 1x HDMI 2.1b and 3x DisplayPort 2.1b outputs. The RTX 5070 Ti SUPER measures 304 mm in length, 137 mm in height, and 48 mm in width; the R7 M370 has no listed dimensions.

Architecture Differences

The R7 M370 is built on GCN 1.0 architecture with a chip codenamed Litho, manufactured on a 28 nm process by TSMC. It contains 950 million transistors on a 77 mm² die, yielding a transistor density of 12.3M per mm². The RTX 5070 Ti SUPER uses Blackwell 2.0 architecture with a GB203 chip, also from TSMC but on a 5 nm process. It packs 45,600 million transistors on a 378 mm² die, for a density of 120.6M per mm². The R7 M370 has no RT cores or tensor cores, while the RTX 5070 Ti SUPER has 70 RT cores and 280 tensor cores. The R7 M370 supports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170; the RTX 5070 Ti SUPER supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The R7 M370’s FP32 throughput is 737.3 GFLOPS, while the RTX 5070 Ti SUPER achieves 43.94 TFLOPS FP32 and the same 43.94 TFLOPS FP16 (1:1). The R7 M370’s production status is end-of-life, with a release date of 2015-05-04, while the RTX 5070 Ti SUPER is active, released 2025-12-31.

Where Each One Wins

The R7 M370 wins in the two Geekbench tests recorded in the data: Geekbench OpenCL (7063) and Geekbench Vulkan (6465). It also wins on average benchmark score (6764 vs 6270) and percentile ranking (38th vs 36th). These wins are narrow and come from older workloads that do not utilize modern GPU features. The R7 M370 is also 2.3% ahead of the AMD Radeon R7 M460 and 1% ahead of the NVIDIA GeForce GT 1010 among its nearest rivals, showing it is competitive within its own ancient class.

The RTX 5070 Ti SUPER wins decisively on every raw performance metric: FP32 compute (43.94 TFLOPS vs 737.3 GFLOPS), texture rate (686.6 GTexel/s vs 23.04 GTexel/s), pixel rate (235.4 GPixel/s vs 7.680 GPixel/s), memory bandwidth (896.0 GB/s vs 57.60 GB/s), and memory capacity (16 GB vs 2 GB). It also wins on clock speeds (2452 MHz boost vs 960 MHz boost), process node (5 nm vs 28 nm), transistor count (45,600 million vs 950 million), and die size (378 mm² vs 77 mm²). The RTX 5070 Ti SUPER supports newer APIs, has RT and tensor cores, and is an active product with a 350 W TDP. In any modern gaming, rendering, or compute scenario that leverages DX12 Ultimate features, the RTX 5070 Ti SUPER’s 3DMark Steel Nomad DX12 score of 6269.5, while lower in absolute terms, represents a workload the R7 M370 simply cannot run. The R7 M370 only makes sense for legacy software that relies on Geekbench-style OpenCL or Vulkan 1.2 paths, where its scores are surprisingly competitive.