GPU Comparison
AMD Radeon RX 570
CMP 70HX
PERFORMANCE BENCHMARKS
Analysis: AMD Radeon RX 570 vs NVIDIA CMP 70HX
Where Each One Wins
The benchmark data paints a starkly one-sided picture. The AMD Radeon RX 570 wins both head-to-head tests recorded in the the benchmark database, leaving the NVIDIA CMP 70HX with zero wins. This is not a close contest; the RX 570 leads by double-digit margins in both compute and graphics API workloads.
Looking at the broader benchmark landscape, the CMP 70HX's average benchmark score of 30,476 places it in the 75th percentile of all GPUs, while the RX 570's average of 28,766 sits at the 74th percentile. Despite the RX 570 winning the direct comparisons, the CMP 70HX holds a slight edge in overall average score, a nuance that suggests the CMP 70HX may perform better in other unlisted workloads. However, in the two tests where both cards appear, the RX 570 is decisively ahead.
The CMP 70HX is a mining-focused product with no display outputs, so its wins are entirely absent from the recorded data. The RX 570, by contrast, is a conventional consumer card with display outputs, and it wins on both compute throughput (OpenCL) and graphics API performance (Vulkan). For any workload represented by these two tests, the RX 570 is the clear choice.
Architecture Differences
The architectural gap between these two GPUs is generational and fundamental. The NVIDIA CMP 70HX uses the GA104 chip built on Ampere architecture, manufactured on Samsung's 8 nm process. It packs 17,400 million transistors into a 392 mm² die, yielding a transistor density of 44.4 million per square millimeter. The AMD Radeon RX 570 uses the Polaris 20 chip based on GCN 4.0, built on GlobalFoundries' 14 nm process. It contains 5,700 million transistors on a 232 mm² die, with a density of 24.6 million per square millimeter.
The CMP 70HX is the more complex silicon by a wide margin, roughly three times the transistor count and nearly double the die area. It features 3,840 shading units, 120 texture mapping units, and 64 raster operations pipelines. The RX 570 counters with 2,048 shading units, 128 TMUs, and only 32 ROPs. The NVIDIA card also includes 30 RT cores and 120 tensor cores, hardware that the AMD card lacks entirely.
Memory subsystems differ sharply. The CMP 70HX uses 8 GB of GDDR6X memory on a 256-bit bus, delivering 608.3 GB/s of bandwidth. The RX 570 uses 4 GB of GDDR5 on the same 256-bit bus width, but bandwidth drops to 224.0 GB/s. Clock speeds also diverge: the CMP 70HX runs at a 1365 MHz base and 1395 MHz boost, while the RX 570 operates at 1168 MHz base and 1244 MHz boost. The CMP 70HX's memory runs at 1188 MHz (19 Gbps effective), compared to the RX 570's 1750 MHz (7 Gbps effective).
API support reflects the generational divide. The CMP 70HX supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, while the RX 570 is limited to DirectX 12 (12_0) and Vulkan 1.3. Both support OpenGL 4.6. The CMP 70HX has no display outputs by design, whereas the RX 570 provides 1x DVI, 1x HDMI 2.0b, and 3x DisplayPort 1.4a. Power delivery also differs: the CMP 70HX uses a single 12-pin connector with a 200 W suggested PSU, while the RX 570 uses a 6-pin connector with a 450 W suggested PSU and a 150 W TDP.
Head-to-Head Benchmarks
The two recorded head-to-head tests both favor the AMD Radeon RX 570, and the margins are substantial. In Geekbench OpenCL, the RX 570 scores 32,084 against the CMP 70HX's 25,135. That is a 21.7% advantage for the AMD card. The NVIDIA part trails by this significant margin despite having more shading units, higher clocks, and vastly more memory bandwidth. The data suggests the CMP 70HX's mining-oriented design does not translate into compute performance that matches its raw specifications.
In Geekbench Vulkan, the gap narrows but remains firmly in AMD's favor. The RX 570 scores 42,752, while the CMP 70HX manages 35,817. This is a 16.2% lead for the RX 570. Even in a graphics API workload, the mining card cannot overcome the architectural and driver-level advantages of the consumer Polaris part. The RX 570's nearest rivals include the AMD Radeon RX 6800M (delta -0.4%) and the AMD Radeon RX 470 (delta -0.8%), which places it in a competitive mid-range tier. The CMP 70HX, by contrast, sits near the NVIDIA Tesla M60 (delta 0%) and the AMD Radeon RX 6700 (delta 0.1%).
The pattern is consistent: the RX 570 wins both tests by double-digit percentages, and the CMP 70HX fails to secure a single victory. The average benchmark scores tell a slightly different story, the CMP 70HX edges ahead at 30,476 versus 28,766, but this is driven by a different set of tests not shared between the two products. In the tests where they directly compete, the RX 570 is the faster card.
FAQ
Q: Which GPU wins the Geekbench OpenCL test?
A: The AMD Radeon RX 570 wins with a score of 32,084, which is 21.7% higher than the NVIDIA CMP 70HX's 25,135.
Q: How large is the Vulkan performance gap?
A: The RX 570 leads with 42,752 points versus the CMP 70HX's 35,817, a 16.2% advantage.
Q: Does the NVIDIA card win any head-to-head benchmark?
A: No. The data shows zero wins for the CMP 70HX and two wins for the RX 570 across the recorded head-to-head tests.
Q: What are the memory specifications of each card?
A: The CMP 70HX has 8 GB of GDDR6X on a 256-bit bus with 608.3 GB/s bandwidth. The RX 570 has 4 GB of GDDR5 on a 256-bit bus with 224.0 GB/s bandwidth.
Q: Do both cards support the same DirectX version?
A: No. The CMP 70HX supports DirectX 12 Ultimate (12_2), while the RX 570 supports DirectX 12 (12_0).
Q: Which card has a higher average benchmark score?
A: The CMP 70HX has an average benchmark score of 30,476, placing it in the 75th percentile. The RX 570 averages 28,766, in the 74th percentile.
Specification Differences
The two cards differ across nearly every specification field. The CMP 70HX uses an 8 nm process from Samsung, whereas the RX 570 uses a 14 nm process from GlobalFoundries. Transistor counts are 17,400 million versus 5,700 million, and die sizes are 392 mm² versus 232 mm². Transistor density is 44.4M per mm² for the CMP 70HX and 24.6M per mm² for the RX 570.
Base clocks are 1365 MHz versus 1168 MHz, and boost clocks are 1395 MHz versus 1244 MHz. Memory clocks are 1188 MHz (19 Gbps effective) versus 1750 MHz (7 Gbps effective). Memory capacity is 8 GB of GDDR6X versus 4 GB of GDDR5. Bandwidth is 608.3 GB/s versus 224.0 GB/s.
Shading units number 3,840 versus 2,048. TMUs are 120 versus 128. ROPs are 64 versus 32. The CMP 70HX has 30 RT cores and 120 tensor cores; the RX 570 has neither. Pixel rates are 89.28 GPixel/s versus 39.81 GPixel/s. Texture rates are 167.4 GTexel/s versus 159.2 GTexel/s. FP32 compute is 10.71 TFLOPS versus 5.095 TFLOPS.
The CMP 70HX has no TDP listed, while the RX 570 has a 150 W TDP. Power connectors are 1x 12-pin versus 1x 6-pin. Suggested PSU is 200 W versus 450 W. Bus interfaces are PCIe 1.0 x4 versus PCIe 3.0 x16. Display outputs are none versus 1x DVI, 1x HDMI 2.0b, and 3x DisplayPort 1.4a. The CMP 70HX supports DirectX 12 Ultimate (12_2) and Vulkan 1.4; the RX 570 supports DirectX 12 (12_0) and Vulkan 1.3. Dimensions are 267 mm versus 241 mm in length, with heights of 112 mm versus unspecified.
The Verdict
The data is unambiguous for anyone choosing between these two GPUs for general compute or graphics workloads. The AMD Radeon RX 570 wins both head-to-head benchmarks, by 21.7% in OpenCL and 16.2% in Vulkan. It also offers display outputs, a conventional PCIe 3.0 x16 interface, and a lower 150 W TDP. The CMP 70HX, despite its newer architecture, larger transistor count, and 608.3 GB/s of memory bandwidth, cannot translate those advantages into a single benchmark win in the recorded tests.
The CMP 70HX does hold a higher average benchmark score (30,476 versus 28,766) and a better percentile ranking (75th versus 74th). This suggests that in other workloads not covered by the head-to-head tests, the NVIDIA part may perform competitively. Its mining-focused design, no display outputs, a PCIe 1.0 x4 interface, and a single 12-pin power connector, makes it unsuitable for standard desktop use. The RX 570 is a fully featured consumer card with a release date of April 17, 2017, and a launch MSRP of 169 USD.
For any user whose workload resembles the Geekbench OpenCL or Vulkan tests, the RX 570 is the superior choice. For a mining-specific deployment where display output and conventional interface support are irrelevant, the CMP 70HX's higher average score and newer architecture may offer advantages, but the recorded data does not substantiate that claim. The verdict from the benchmarks alone: the RX 570 wins on measured performance, while the CMP 70HX's case rests on unrecorded potential.