AMD Radeon R9 370X vs AMD Radeon RX 5700 XT Comparison

The AMD Radeon RX 5700 XT and the AMD Radeon R9 370X represent two distinct eras of AMD graphics, separated by four years of architectural evolution. The data shows a generational chasm in most metrics, yet the R9 370X manages to secure a surprising victory in one specific workload. The RX 5700 XT, built on the 7 nm RDNA 1.0 architecture, is fundamentally a more capable GPU than the 28 nm GCN 1.0-based R9 370X, but the benchmark results reveal that performance is not uniformly better across all test suites, making the comparison more nuanced than a simple generational sweep.

Head-to-Head Benchmarks

The direct comparison between these two cards is stark, with the RX 5700 XT dominating in two of the three shared tests. In Geekbench Vulkan, the RX 5700 XT scores 68,310 against the R9 370X's 9,018, a massive 657.5% delta. This result indicates a fundamental advantage in modern graphics API performance, as the newer architecture's hardware and driver stack are clearly optimized for Vulkan's low-level access. The delta is so large that it suggests the R9 370X is not merely slower but is operating at a completely different performance tier in this workload.

The Geekbench Metal test tells a similar story, though the margin is slightly less extreme. The RX 5700 XT posts a score of 74,647, while the R9 370X manages 12,676, resulting in a 488.9% delta in favor of the newer card. Metal, being Apple's graphics API, is often a strong indicator of a GPU's compute and shading efficiency. The RX 5700 XT's 2,560 shading units and RDNA 1.0 design clearly translate to far superior Metal performance, which aligns with its much higher pixel rate of 121.9 GPixel/s and texture rate of 304.8 GTexel/s.

However, the R9 370X strikes back in Geekbench OpenCL, where it scores 25,893 compared to the RX 5700 XT's 9,524. This is a 63.2% deficit for the newer card. This is a curious anomaly, as OpenCL is a general-purpose compute API. The data suggests that the R9 370X's GCN 1.0 architecture, with its 1,280 shading units, handles this particular OpenCL workload more efficiently than the RDNA 1.0 design. It is plausible that the test itself is either outdated or heavily favors the older compute architecture, but the result stands as a clear win for the R9 370X, giving it a 1-2 record in head-to-head tests.

These results imply that while the RX 5700 XT is overwhelmingly faster in modern gaming and graphics APIs, the R9 370X retains a niche advantage in legacy compute workloads. The average benchmark scores reflect this split, with the RX 5700 XT averaging 16,361 across all its tests, placing it in the 59th percentile of all GPUs. The R9 370X, despite its OpenCL win, averages 15,862, a figure that lands it in the 58th percentile. The proximity of these percentiles is surprising given the head-to-head deltas, but it is explained by the fact that the RX 5700 XT has many more benchmark entries in its record, including lower scores like its Passmark DirectX 12 result of 64, which drag its average down.

Architecture Differences

The architectural gap between these two GPUs is the primary driver of their performance divergence. The RX 5700 XT uses the Navi 10 chip on a 7 nm TSMC process, packing 10,300 million transistors into a 251 mm² die. This yields a transistor density of 41.0M per mm², a figure that highlights the massive manufacturing advantage over the older card. The R9 370X, in contrast, uses the Trinidad chip on a 28 nm TSMC process, with only 2,800 million transistors on a 212 mm² die, giving it a density of just 13.2M per mm². The smaller process node allows the RX 5700 XT to pack nearly four times the transistors into a similar die area.

Core configuration differences are equally pronounced. The RX 5700 XT has 2,560 shading units, 160 texture mapping units (TMUs), and 64 render output units (ROPs). The R9 370X is halved in most respects, with 1,280 shading units, 80 TMUs, and 32 ROPs. These raw counts directly explain the theoretical throughput figures: the RX 5700 XT achieves 9.754 TFLOPS of FP32 performance and 19.51 TFLOPS of FP16 (2:1 ratio), while the R9 370X manages only 2.637 TFLOPS of FP32 and has no listed FP16 capability. Clock speeds also favor the newer card, with the RX 5700 XT boosting to 1905 MHz versus the R9 370X's 1030 MHz boost.

Memory subsystems reflect the generational leap as well. The RX 5700 XT comes with 8 GB of GDDR6 memory on a 256-bit bus, delivering 448.0 GB/s of bandwidth. The R9 370X has 2 GB of GDDR5 on the same 256-bit bus, but only achieves 179.2 GB/s. The effective memory speed is 14 Gbps for the RX 5700 XT versus 5.6 Gbps for the R9 370X. The newer card also supports PCIe 4.0 x16, while the older one is limited to PCIe 3.0 x16. The RX 5700 XT further supports modern display outputs like HDMI 2.0b and DisplayPort 1.4a, whereas the R9 370X is stuck with older DVI and HDMI 1.4a connections.

FAQ

Q: Why does the RX 5700 XT have a much higher average benchmark score than the R9 370X?

A: The RX 5700 XT's average benchmark score of 16,361 is higher than the R9 370X's 15,862, driven by its superior core configuration, faster memory, and newer architecture. However, the difference is smaller than expected because the RX 5700 XT's record includes several low Passmark scores (e.g., 64 in DirectX 12), which pull its average down, while the R9 370X only has three Geekbench results in its record.

Q: Is the R9 370X better than the RX 5700 XT in any way?

A: Yes, in the Geekbench OpenCL test, the R9 370X scores 25,893, which is 63.2% higher than the RX 5700 XT's 9,524. This specific compute workload appears to favor the GCN 1.0 architecture, but it does not translate to wins in other tests.

Q: What is the transistor density difference between the two cards?

A: The RX 5700 XT has a transistor density of 41.0M per mm², while the R9 370X has only 13.2M per mm². This is due to the RX 5700 XT's 7 nm process versus the R9 370X's 28 nm process, allowing the newer card to fit 10,300 million transistors on a 251 mm² die.

Q: Which card has more memory bandwidth?

A: The RX 5700 XT has significantly more memory bandwidth at 448.0 GB/s, compared to the R9 370X's 179.2 GB/s. This is due to the RX 5700 XT using faster GDDR6 memory at 14 Gbps effective, versus GDDR5 at 5.6 Gbps effective, even though both use a 256-bit bus.

Q: How do the shading unit counts compare?

A: The RX 5700 XT has 2,560 shading units, exactly double the 1,280 shading units found on the R9 370X. This doubling is consistent across TMUs (160 vs 80) and ROPs (64 vs 32), explaining the RX 5700 XT's higher theoretical fill rates.

Q: What are the power connector requirements?

A: The RX 5700 XT requires a 1x 6-pin and 1x 8-pin power connector setup, while the R9 370X uses 2x 6-pin connectors. The suggested PSU for the RX 5700 XT is 550 W, compared to 450 W for the R9 370X.

The Verdict

The benchmark data is unambiguous for most use cases: the RX 5700 XT is the superior GPU. Its wins in Geekbench Metal (74,647 vs 12,676) and Geekbench Vulkan (68,310 vs 9,018) are so large that no modern gaming or graphics workload would reasonably favor the R9 370X. The RX 5700 XT's 8 GB of GDDR6 memory, 2,560 shading units, and 9.754 TFLOPS of FP32 performance provide a foundation for contemporary applications that the R9 370X's 2 GB of GDDR5 and 2.637 TFLOPS simply cannot match.

The R9 370X's sole claim to relevance is its OpenCL score of 25,893, which is 63.2% ahead of the RX 5700 XT. This suggests that for very specific, likely older compute tasks, the GCN 1.0 architecture has an efficiency advantage. However, this single win does not offset the RX 5700 XT's massive leads elsewhere. The data also shows the RX 5700 XT's percentile ranking of 59 versus the R9 370X's 58, indicating that despite the RX 5700 XT's clear wins, the overall performance gap as measured by average score is not as wide as the head-to-head tests suggest. For any user prioritizing modern graphics, gaming, or general GPU acceleration, the RX 5700 XT is the only rational choice.

Specification Differences

The two cards differ in nearly every major specification category. The process node is 7 nm for the RX 5700 XT versus 28 nm for the R9 370X. Transistor count is 10,300 million versus 2,800 million, and die size is 251 mm² versus 212 mm². The RX 5700 XT has a base clock of 1605 MHz and a boost clock of 1905 MHz, while the R9 370X operates at 980 MHz base and 1030 MHz boost. Memory configuration differs completely: 8 GB GDDR6 at 14 Gbps effective versus 2 GB GDDR5 at 5.6 Gbps effective, with bandwidth at 448.0 GB/s versus 179.2 GB/s.

Core counts are doubled on the RX 5700 XT, with 2,560 shading units, 160 TMUs, and 64 ROPs, versus 1,280 shading units, 80 TMUs, and 32 ROPs on the R9 370X. Pixel rate is 121.9 GPixel/s versus 32.96 GPixel/s, texture rate is 304.8 GTexel/s versus 82.40 GTexel/s, and FP32 throughput is 9.754 TFLOPS versus 2.637 TFLOPS. The RX 5700 XT has a TDP of 225 W versus 180 W, uses a PCIe 4.0 x16 interface versus PCIe 3.0 x16, and has different display outputs (1x HDMI 2.0b and 3x DisplayPort 1.4a versus 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2). The RX 5700 XT also supports Vulkan 1.4, while the R9 370X supports Vulkan 1.2.170, and the DirectX support is 12 (12_1) versus 12 (11_1). The RX 5700 XT's launch MSRP is 399 USD, while the R9 370X's is 199 USD.

Where Each One Wins

The RX 5700 XT wins decisively in all modern graphics scenarios. Its Geekbench Vulkan score of 68,310 is 657.5% higher, making it the clear choice for any Vulkan-based game or application. Its Metal score of 74,647 is 488.9% higher, which is critical for macOS or iOS development environments. The RX 5700 XT also has the hardware headroom for future workloads, with 8 GB of memory and PCIe 4.0 support. For gaming at any resolution above 1080p, the RX 5700 XT's higher texture rate of 304.8 GTexel/s and pixel rate of 121.9 GPixel/s provide the necessary throughput.

The R9 370X wins only in the Geekbench OpenCL test, where it beats the RX 5700 XT by 63.2%. This suggests that for legacy OpenCL compute tasks, particularly those that are not optimized for RDNA architectures, the R9 370X may perform adequately. Its lower TDP of 180 W also means it requires less power, and its shorter length of 221 mm (8.7 inches) versus the RX 5700 XT's 272 mm (10.7 inches) makes it easier to fit in compact cases. However, these physical advantages do not translate into performance wins outside of the single OpenCL benchmark. The data supports the conclusion that the R9 370X is only a niche choice for specific compute workloads, while the RX 5700 XT is the superior all-around GPU.