NVIDIA GeForce MX570 A vs NVIDIA P104-100 Comparison

# Head-to-Head Benchmarks

The database places the NVIDIA GeForce MX570 A and the NVIDIA P104-100 in adjacent performance territory, but the comparison is not a clean sweep. Across the two recorded benchmark suites, the P104-100 takes both wins, with the MX570 A finishing behind in each test. In the Geekbench OpenCL test, the P104-100 scores 52,368 against 39,780 for the MX570 A, a delta of 24% in favor of the NVIDIA mining card. The Vulkan test shows a similar pattern: 45,165 for the P104-100 versus 37,601 for the MX570 A, a 16.7% advantage for the P104-100. The data records zero wins for the MX570 A and two wins for the P104-100, so the head-to-head record is firmly in favor of the latter.

Architecture Differences

The two accelerators are from different NVIDIA families and their internal designs are quite different. The MX570 A is built on the GA107SB chip, using the Ampere architecture, and is manufactured at Samsung on an 8 nm process. The P104-100, on the other hand, uses the GP104 chip, belongs to the older Pascal architecture, and is made by TSMC on a 16 nm node. That process difference gives the MX570 A a considerable transistor density advantage: the GA107SB packs 8,700 million transistors into a 200 mm² die, for a density of 43.5 million transistors per square millimeter. The P104-100's GP104 integrates 7,200 million transistors across a larger 314 mm² die, which yields a lower density of 22.9 million per square millimeter.

The memory subsystems also diverge sharply. The MX570 A uses 2 GB of GDDR6 memory on a 64-bit bus, with a 96.00 GB/s of bandwidth. The P104-100 steps up to 4 GB of GDDR5X memory, also on a 256-bit bus, for a memory bandwidth of 320.3 GB/s. The P104's wider memory bus and higher capacity memory are key contributors to its benchmark advantage.

Other compute resources are not equal either. The MX570 A includes 2,048 shading units, 64 texture mapping units, 32 ROPs, 16 ray tracing cores, and 64 tensor cores. The P104-100 has 1,920 shading units, 120 TMUs, 64 ROPs, and none of the specialized tensor or RT cores of the Ampere part. The clocks also differ: the MX570 runs at a base of 832 MHz and a boost of 1155 MHz, while the P104-100 is set to run at 1607 MHz base and 1733 MHz boost.

FAQ

Q: Which GPU has more shading units?

A: The MX570 A has 2,048 shading units, while the P104-100 has 1,920. The MX570 A has the higher count, with a 128-unit advantage.

Q: Is the MX570 A compatible with DirextX 12 Ultimate?

A: Yes, the MX570 A supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: What is the memory bandwidth difference?

A: The P104-100 has a memory bandwidth of 320.3 GB/s, which is significantly higher than the 96.0 GB/s of the MX570 A. The P104-100's 256-bit bus and GDDR5X memory contribute to this bandwidth advantage.

Q: Which card has a higher pixel rate?

A: The P104-100 has a higher pixel rate at 110.9 GPixel/s, against the MX570 A's 36.96 GPixel/s. The texture rate is also higher: 208.0 GTexel/s versus 73.92 GTexel/s. The P104-100 has the advantage in both fill-rate metrics.

Specification Differences

The most obvious differences are in the memory and bus. The MX570 A has 2 GB GDDR6 over a 64-bit interface, while the P104-100 has 4 GB of GDDR5X over a 256-bit interface. The P104 has a dual-slot cooler, while the MX570 A is a single-slot IGP. The P104-100 requires a single 8-pin power connector, the MX570 A has no power connector. The P104-100 uses a PCIe 1.0 x4 bus interface, and the MX570 A uses a PCIe 4.0 x8 interface. Display outputs: the MX570 A outputs are portable device dependent, the P104-100 has no outputs. The MX570 is rated for a 25 W TDP, and the P104-100 has no TDP in the database. The MX570 A was launched in the end-of-life status, and the P104-100 is also end-of-life.

The Verdict

The data indicates a clear separation in use cases. The NVIDIA P104-100 is the better choice for compute-bound workloads that stress raw memory bandwidth, pixel and texture fill rates, and FP32 compute. In OpenCL and Vulkan tests it takes both wins and has the higher memory capacity and bandwidth. The MX570 A is not without its strengths: its architecture is newer (Ampere with 8 nm versus Pascal with 16 nm), it has a higher transistor density (43.5M per mm² vs 22.9M), and it brings tensor and RT cores to the table, plus higher shading unit and TMU counts. The MX570 A is the one to get for portability (single-slot, no external power, laptop IGP), but if the goal is raw GPU compute, the P104-100 should be chosen. The P104-100 wins both head-to-head tests, so the measured data is favorable.

Where Each One Wins

The P104-100 is the winner in compute-focused tasks. It wins the OpenCL benchmark by 12,588 points (24%) and the Vulkan test by 7,564 points (16.7%). Its larger memory (4 GB vs 2 GB), wider bus (256-bit vs 64-bit), and higher memory bandwidth (320.3 GB/s vs 96 GB/s) give it a decisive advantage in fill-rate and compute-bound synthetic loads.

The MX570 A is for the user who needs an integrated, power-efficient, low-profile part with no external power connectors and no display outputs. It is the only one with a PCIe 4.0 x8 interface, which is faster than the P104-100's PCIe 1.0 x4. It also comes with modern API support (Vulkan 1.4, OpenGL 4.6, DirectX 12 Ultimate), while the P104-100 goes up to Vulkan 1.4, OpenGL 4.6, and DirectX 12 (12_1). For pure compute performance, the P104-100 is the one to choose. For a laptop IGP that can drive displays, the MX570 A is the only option here.