AMD Radeon 8060S vs NVIDIA CMP 90HX Comparison

The NVIDIA CMP 90HX and AMD Radeon 8060S represent two entirely different approaches to graphics hardware, one a dedicated mining card built on a desktop-class GPU and the other a mobile integrated processor. The recorded benchmark data shows a clear split: the AMD Radeon 8060S wins the only shared test, the Geekbench OpenCL score, by 18.5% over the NVIDIA CMP 90HX (84,626 versus 69,000). However, the NVIDIA card holds a higher overall percentile ranking at 90% of all GPUs compared to 87% for the AMD part, a discrepancy driven by the fact that the AMD card’s average benchmark score of 55,757 is dragged down by its additional tests, including a 3DMark Steel Nomad DX12 score of 2,162 and a Geekbench Vulkan score of 80,483. The data implies that raw compute throughput favors the AMD part, while the NVIDIA card’s narrower benchmark profile gives it a better percentile position.

The Verdict

The database shows two products with different intended lifetimes and performance profiles. The NVIDIA CMP 90HX is an end-of-life product released on July 27, 2021, built for compute workloads without display outputs, while the AMD Radeon 8060S is an active mobile solution released on January 5, 2025, with portable-device-dependent display outputs and system-shared memory. For users prioritizing raw OpenCL compute performance, the AMD Radeon 8060S is the stronger option, delivering an 18.5% higher score in that specific test. For users who need a card with a higher percentile ranking relative to the entire GPU market, the NVIDIA CMP 90HX sits at the 90th percentile, three points above the AMD part’s 87th percentile. The NVIDIA card also has a larger power envelope at 320 W TDP versus 55 W TDP for the AMD part, which suggests that the AMD integrated solution achieves its competitive compute results at a fraction of the power draw. The AMD Radeon 8060S is the better choice for mobile or power-constrained systems, while the NVIDIA CMP 90HX offers a dedicated, higher-percentile compute card for desktop mining or compute tasks, assuming the system can supply 700 W via the suggested PSU and accommodate a 285 mm dual-slot card with two 8-pin connectors. The data does not support a single winner across all metrics; the choice depends on whether the user values the AMD part’s OpenCL lead and low power draw or the NVIDIA part’s higher market percentile and dedicated memory subsystem.

Architecture Differences

The two GPUs come from different foundries and process nodes. The NVIDIA CMP 90HX uses the GA102 chip built on Samsung’s 8 nm process, while the AMD Radeon 8060S uses the Strix Halo chip on TSMC’s 4 nm process. The NVIDIA chip is physically larger at 628 mm² with 28,300 million transistors, giving it a transistor density of 45.1 million per square millimeter. The AMD chip measures 308 mm², but its transistor count is listed as unknown, so no density calculation is possible. The NVIDIA architecture is Ampere, belonging to the Mining GPUs generation, while the AMD architecture is RDNA 3.5 from the Navi Mobile (RX 8000M) generation. The AMD part has a predecessor listed as Polaris Mobile, while the NVIDIA card has no predecessor or successor in the database.

Compute resources differ sharply. The NVIDIA CMP 90HX packs 6,400 shading units, 200 texture mapping units, 80 render output units, 50 ray tracing cores, and 200 tensor cores. The AMD Radeon 8060S has 2,560 shading units, 160 TMUs, 64 ROPs, and 40 ray tracing cores, with no tensor core count listed. Despite having fewer shading units, the AMD part achieves a higher pixel rate of 185.6 GPixel/s versus 136.8 GPixel/s for the NVIDIA card, and a higher texture rate of 464.0 GTexel/s versus 342.0 GTexel/s. The NVIDIA card still leads in raw FP32 throughput at 21.89 TFLOPS versus 14.85 TFLOPS for the AMD part, and both offer FP16 at a 1:1 ratio with their FP32 figures. Clock behavior explains part of this: the AMD part has a lower base clock of 1295 MHz but a much higher boost clock of 2900 MHz, while the NVIDIA card runs at 1500 MHz base and 1710 MHz boost. Memory architecture is fundamentally different. The NVIDIA CMP 90HX has 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s of bandwidth and a memory clock of 1188 MHz (19 Gbps effective). The AMD Radeon 8060S uses system-shared memory, with the type, bus width, and bandwidth all listed as system dependent. The NVIDIA card also has no display outputs, while the AMD part’s outputs are portable-device dependent. The bus interface differs as well: the NVIDIA card lists PCIe 1.0 x4, which is unusual for a mining card, while the AMD part uses PCIe 5.0 x16. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Where Each One Wins

In the only direct head-to-head benchmark, the AMD Radeon 8060S wins Geekbench OpenCL with a score of 84,626 against the NVIDIA CMP 90HX’s 69,000, a delta of 18.5% in favor of AMD. This makes the AMD part the clear choice for OpenCL compute workloads as measured by that test. The AMD part also has additional benchmark coverage, including a Geekbench Vulkan score of 80,483 and a 3DMark Steel Nomad DX12 score of 2,162, though the NVIDIA card has no corresponding scores in those tests, so no direct comparison is possible. The NVIDIA CMP 90HX wins on market percentile, sitting at the 90th percentile versus 87th for the AMD part. It also has dedicated memory with 10 GB of GDDR6X and 760.3 GB/s of bandwidth, which is likely advantageous for workloads that require consistent, high-bandwidth memory access rather than system-shared memory. The NVIDIA card’s higher FP32 throughput of 21.89 TFLOPS versus 14.85 TFLOPS for the AMD part suggests it should win in raw FP32 compute tasks, though the recorded OpenCL benchmark shows the opposite result, indicating that architecture efficiency or driver optimizations play a role. The AMD part wins on power efficiency in the data: 55 W TDP versus 320 W, and it requires no power connectors, while the NVIDIA card needs two 8-pin connectors and a 700 W suggested PSU. For mobile or compact systems, the AMD Radeon 8060S is the only viable option given its IGP slot width, while the NVIDIA card is dual-slot and 285 mm long.

FAQ

Q: Which GPU has the higher Geekbench OpenCL score?

A: The AMD Radeon 8060S scores 84,626, which is 18.5% higher than the NVIDIA CMP 90HX’s 69,000.

Q: How do their overall market percentiles compare?

A: The NVIDIA CMP 90HX ranks at the 90th percentile of all GPUs, while the AMD Radeon 8060S ranks at the 87th percentile.

Q: What are the TDP differences between the two?

A: The NVIDIA CMP 90HX has a 320 W TDP, while the AMD Radeon 8060S has a 55 W TDP.

Q: Does the NVIDIA CMP 90HX have display outputs?

A: No, the NVIDIA CMP 90HX has no display outputs at all. The AMD Radeon 8060S has outputs that are portable-device dependent.

Q: What memory does each GPU use?

A: The NVIDIA CMP 90HX uses 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth. The AMD Radeon 8060S uses system-shared memory, with bandwidth and bus width listed as system dependent.

Q: Which card is listed as active in the database?

A: The AMD Radeon 8060S is listed as active, released on January 5, 2025. The NVIDIA CMP 90HX is end-of-life, released on July 27, 2021.

Head-to-Head Benchmarks

The single recorded head-to-head benchmark between these two parts is Geekbench OpenCL. The AMD Radeon 8060S wins with a score of 84,626, while the NVIDIA CMP 90HX scores 69,000. The delta percentage is 18.5% in favor of the AMD part. This is a substantial lead, especially considering the NVIDIA card’s higher FP32 rating of 21.89 TFLOPS and its dedicated GDDR6X memory with 760.6 GB/s of bandwidth. The AMD part achieves this win despite having fewer shading units, 2,560 versus 6,400, and a lower FP32 throughput of 14.85 TFLOPS. The data suggests that the AMD RDNA 3.5 architecture on the 4 nm TSMC process is more efficient at converting its compute resources into OpenCL performance, or that driver and software optimizations favor the newer mobile part. The NVIDIA card’s closest rivals in the database are the Intel Arc A770 at 68, average score 68,809, delta 0.3% relative to the NVIDIA part, and the AMD Radeon Pro WX 8200 at 69, average score 69,870, delta -1.2% relative to the NVIDIA part. For the AMD Radeon 8060S, its nearest rival is the AMD Radeon RX 6750 GRE 12 GB, average score 55,698, delta 0.1% relative to the AMD part, and the AMD Radeon Pro W5700X at 54, average score 54,828, delta 1.7% relative to the AMD part. The AMD Radeon 8060S. The AMD part also has a Geekbench Vulkan score of 80,483, which is higher than its OpenCL score of 84,626, indicating that the card performs well across multiple API frameworks, though the NVIDIA card lacks a Vulkan score for comparison. The AMD 3DMark Steel Nomad DX12 score of 2,162 rounds out the AMD card’s benchmark profile, and its average benchmark score across all tests is 55,757, which is lower than the NVIDIA card’s average of 69,000, due to the inclusion of the 3DMark test dragging the average down. The NVIDIA card’s pixel rate of 136.8 GPixel/s and texture rate of 342.0 GTexel/s are both lower than the AMD card’s 185.6 GPixel/s and 464.0 GTexel/s, even though the AMD part has lower base clocks at 1295 MHz versus 1500 MHz. The NVIDIA card’s boost clock of 1710 MHz is lower than the AMD part’s 2900 MHz boost, which likely contributes to the AMD card’s higher pixel and texture rates. In the end, the recorded data clearly favors the AMD Radeon 8060S in OpenCL, and the absence of any wins for the NVIDIA card in the head-to-head table confirms that the AMD part is the winner in the direct comparison, with one win out of one test.

Specification Differences

The specification table reveals where these two cards diverge across every major category. The NVIDIA CMP 90HX chip is the GA102 on Samsung 8 nm, while the AMD Radeon 8060S chip is Strix Halo on TSMC 4 nm. The NVIDIA die size difference is substantial: 628 mm² versus 308 mm². The NVIDIA transistor count is 28,300 million, while the AMD transistor count is unknown. The AMD part has no transistor density listed, but the NVIDIA card has a density of 45.1 million per square millimeter. The NVIDIA base clock is 1500 MHz, the AMD base clock is lower at 1295 MHz. The NVIDIA boost clock is 1710 MHz, the AMD boost clock is 2900 MHz. The memory clock differs massively, with the NVIDIA at 1188 MHz with 19 Gbps effective, and the AMD memory is system shared. The memory size is 10 GB for the NVIDIA, and system shared for the AMD, with the AMD bandwidth listed as system dependent and the NVIDIA bandwidth at 760. The NVIDIA shading units number 6,400, and the AMD shading units number 2,560. The TMUs are 200 for the NVIDIA, 160 for the AMD. The ROPs are 80 and 64 respectively. The RT cores are 50 for the NVIDIA and 40 for the AMD. The NVIDIA tensor cores number 200 while the AMD tensor cores are null. The pixel rate for the NVIDIA 136. The AMD pixel rate is 185. The NVIDIA texture rate for the NVIDIA 342. The AMD texture rate is 464. The FP32 the NVIDIA is 21. The AMD FP32 the AMD 14. The FP16 both the 21. The NVIDIA TDP 320 W 55 W. The NVIDIA slot width dual-slot versus IGP for the AMD. The NVIDIA power connectors two 8-pin versus none for the AMD. The suggested PSU 700 W for the NVIDIA and null for the AMD. The bus interface for the NVIDIA is PCIe 1.0 x4. The AMD uses PCIe 5.0 x16. Display outputs are absent for the NVIDIA and portable-device dependent for the AMD. Both share the same API support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. The NVIDIA dimensions are 285 mm in length and 112 mm in height, while the AMD has no listed dimensions. The production status is end-of-life for the NVIDIA and active for the AMD. The release dates are July 27, 2021 for the NVIDIA and January 5, 2025 for the AMD. The AMD has a predecessor, Polaris Mobile, while the NVIDIA has none. Neither has a launch MSRP listed. The power connectors, slot width, and PSU requirements highlight the NVIDIA card as a desktop add-in board and the AMD as an integrated part.