AMD Radeon 890M vs AMD Ryzen Z2 Go GPU Comparison

Architecture Differences

The AMD Radeon 890M and AMD Ryzen Z2 Go GPU represent two distinct design philosophies from AMD, separated by a full generation of graphics architecture. The 890M is built on the RDNA 3.5 architecture, while the Z2 Go GPU uses the older RDNA 2.0 design. This architectural gap influences nearly every aspect of their performance characteristics.

The 890M is part of the Strix Point chip, manufactured on TSMC's 4 nm process node. The Z2 Go GPU comes from the Rembrandt+ chip and uses a larger 6 nm process. This difference in fabrication technology directly impacts transistor density: the 890M packs 34,000 million transistors into a 233 mm² die, yielding a density of 145.9 million transistors per square millimeter. The Z2 Go GPU contains 13,100 million transistors across a 208 mm² die, for a density of just 63.0 million per square millimeter. The 890M therefore crams more than twice the transistor density into a slightly larger physical package.

The shading resources differ substantially. The 890M carries 1024 shading units, 64 texture mapping units, and 32 raster operation pipelines. It also includes 16 ray tracing cores. The Z2 Go GPU has 768 shading units, 48 TMUs, and 32 ROPs, with 12 ray tracing cores. The 890M holds a 33% advantage in shading units and a 33% advantage in TMUs, while both parts feature identical ROP counts. The ray tracing core count favors the 890M by four cores.

Both GPUs support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so API-level feature support is identical. The 890M is classified as an integrated graphics processor (IGP) in the Navi III IGP generation, while the Z2 Go GPU is categorized as a Console GPU in AMD's product stack. The 890M's predecessor is the Navi II IGP; the Z2 Go GPU has no recorded predecessor.

Specification Differences

The clock behavior separates these two GPUs clearly. The 890M runs at a 400 MHz base clock and boosts to 2900 MHz. The Z2 Go GPU starts at a higher 800 MHz base but boosts to 2700 MHz, which is 200 MHz lower than the 890M's peak. The Z2 Go's memory clock is fixed at 800 MHz with 6.4 Gbps effective transfer, while the 890M uses system-shared memory with no dedicated memory clock.

Memory configuration is one of the starkest contrasts. The 890M has system-shared memory, meaning its capacity, type, bus width, and bandwidth all depend on the host system. The Z2 Go GPU comes with 16 GB of LPDDR5 memory on a 128-bit bus, delivering 102.4 GB/s of bandwidth. This gives the Z2 Go a fixed, dedicated memory pool, whereas the 890M's memory performance is system dependent.

Rasterization throughput slightly favors the 890M. Its pixel rate is 92.80 GPixel/s versus 86.40 GPixel/s for the Z2 Go, a difference of about 7%. The texture rate gap is larger: 185.6 GTexel/s for the 890M compared to 129.6 GTexel/s for the Z2 Go, a 43% advantage. Floating point performance follows a similar pattern. The 890M delivers 5.939 TFLOPS of FP32 compute, while the Z2 Go manages 4.147 TFLOPS, putting the 890M about 43% ahead. The FP16 picture differs because of the rate conventions: the 890M runs FP16 at a 1:1 ratio (5.939 TFLOPS), while the Z2 Go uses a 2:1 ratio and reaches 8.294 TFLOPS, giving it a 40% advantage in half-precision throughput.

Power envelopes also differ. The 890M is rated at 15 W TDP, while the Z2 Go GPU consumes 28 W. The 890M uses a PCIe 4.0 x8 bus interface; the Z2 Go has no recorded bus interface. Display outputs are portable-device dependent for the 890M, while the Z2 Go lists a single USB Type-C output. The 890M has no slot width (it is an IGP), and neither part uses external power connectors.

The 890M was released on 2024-07-14, and the Z2 Go GPU on 2024-12-31. Both are active in production.

FAQ

Q: Which GPU has more shading units?

A: The AMD Radeon 890M has 1024 shading units, compared to 768 on the AMD Ryzen Z2 Go GPU, a 33% advantage.

Q: How does memory differ between the two?

A: The 890M uses system-shared memory with bandwidth dependent on the host system. The Z2 Go GPU includes 16 GB of LPDDR5 memory on a 128-bit bus with 102.4 GB/s of bandwidth.

Q: Which GPU has a higher boost clock?

A: The 890M boosts to 2900 MHz, which is 200 MHz higher than the Z2 Go GPU's 2700 MHz boost clock.

Q: What is the FP32 compute difference?

A: The 890M produces 5.939 TFLOPS of FP32 compute, while the Z2 Go GPU produces 4.147 TFLOPS. The 890M is approximately 43% faster in single-precision throughput.

Q: Do both GPUs support the same graphics APIs?

A: Yes. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: Which GPU has a higher pixel fill rate?

A: The 890M achieves 92.80 GPixel/s, versus 86.40 GPixel/s for the Z2 Go GPU, a difference of roughly 7% in favor of the 890M.

Head-to-Head Benchmarks

The database records no direct head-to-head benchmark results between these two GPUs. However, the available individual benchmark data for the 890M and the specification differences provide a basis for comparison.

The 890M's benchmark suite shows a broad range of results. In 3DMark Steel Nomad DX12, it scores 590 points. Geekbench results are stronger: 37,254 in OpenCL and 40,808 in Vulkan. Passmark results show 8,076 in G3D, 979 in G2D, and 4,157 in GPU compute. The older DirectX API tests show 97 in DirectX 9, 73 in DirectX 11, 37 in DirectX 12, and 33 in DirectX 10. Its average benchmark score is 9,210.

The Z2 Go GPU has no recorded benchmark scores in the database. Its average benchmark score is listed as 0, and it has no nearest rivals. The percentile ranking, however, places the Z2 Go at the 50th percentile of all GPUs, while the 890M sits at the 45th percentile. This percentile data suggests the Z2 Go is positioned slightly higher in overall GPU rankings despite lacking individual benchmark entries.

The 890M's nearest rivals provide context for its performance class. The AMD Radeon Vega 8 scores 9,221 on average, which is 0.1% above the 890M. The NVIDIA GeForce GTX 960 averages 9,273, 0.7% higher. The GTX 465 averages 9,294, 0.9% higher, and the GTX 850M averages 9,302, a full 1% higher. The 890M sits in a tightly packed performance band where all four rivals are within 1% of its average score.

The specification comparison indicates where each GPU would likely dominate in a direct test. The 890M's 43% advantage in texture rate and FP32 compute would show strongly in texture-heavy workloads and general compute tasks. The Z2 Go's dedicated 102.4 GB/s memory bandwidth could help in scenarios where memory latency and bandwidth consistency matter more than raw shader throughput.

The Z2 Go's FP16 output of 8.294 TFLOPS exceeds the 890M's 5.939 TFLOPS by 40%, which would favor the Z2 Go in half-precision compute workloads, such as certain machine learning inference tasks. The 890M's higher boost clock of 2900 MHz versus 2700 MHz suggests better sustained single-threaded graphics performance.

Where Each One Wins

The AMD Radeon 890M wins in scenarios that demand raw shading throughput. Its 1024 shading units, 64 TMUs, and 5.939 TFLOPS of FP32 compute position it ahead of the Z2 Go GPU in conventional 3D rendering, particularly at higher resolutions where pixel and texture throughput matter. The 890M's 185.6 GTexel/s texture rate is 43% higher than the Z2 Go's 129.6 GTexel/s, so texture-heavy scenes would show a clear advantage. Its 16 ray tracing cores versus 12 on the Z2 Go also gives it an edge in ray-traced workloads.

The Z2 Go GPU wins in memory-dependent scenarios. Its dedicated 16 GB of LPDDR5 memory on a 128-bit bus with 102.4 GB/s bandwidth is a fixed resource, whereas the 890M's system-shared memory is only as fast as the host system allows. In systems where shared memory bandwidth is constrained, the Z2 Go would maintain consistent performance. The Z2 Go's 28 W TDP, while higher than the 890M's 15 W, indicates a more power-hungry but potentially more capable memory subsystem.

The Z2 Go also wins in FP16 compute. Its 8.294 TFLOPS of half-precision throughput, achieved through a 2:1 ratio, is 40% higher than the 890M's 1:1 FP16 output. Applications that rely heavily on half-precision math, such as certain AI inference or image processing pipelines, would favor the Z2 Go.

The 890M wins in power efficiency. At 15 W TDP, it draws nearly half the power of the 28 W Z2 Go GPU while delivering more FP32 performance and higher clock speeds. For thermally constrained portable devices, the 890M's lower power draw is a decisive advantage.

The Z2 Go's 50th percentile ranking versus the 890M's 45th percentile suggests that in aggregate GPU performance, the Z2 Go may hold a slight edge despite lacking benchmark scores. The practical interpretation is that the Z2 Go is better suited to standalone console-like devices with fixed memory, while the 890M fits into systems where shared memory and lower power are priorities.

The Verdict

The data supports a clear split based on deployment context. The AMD Radeon 890M is the stronger choice for systems where power consumption and shading throughput are the primary concerns. Its 15 W TDP, 2900 MHz boost clock, 1024 shading units, and 5.939 TFLOPS of FP32 compute make it the more capable part for traditional graphics rendering within a tight power budget. The 890M's 45th percentile ranking and average score of 9,210, sitting within 1% of rivals like the GTX 960 and GTX 850M, confirm it as a mid-pack integrated solution.

The AMD Ryzen Z2 Go GPU is the better option for devices that need a fixed, dedicated memory configuration. Its 16 GB LPDDR5 pool with 102.4 GB/s bandwidth removes the variability of system-shared memory. The 50th percentile ranking, despite zero recorded benchmarks, indicates the database positions it above the 890M in overall GPU standings. Its 8.294 TFLOPS FP16 output and 28 W power envelope suggest a part designed for sustained compute workloads rather than power-constrained integrated graphics.

Builders targeting thin-and-light portables with shared memory should select the 890M. Developers working on console-style hardware with dedicated memory should select the Z2 Go GPU. The 890M's 43% lead in texture rate and FP32 compute serves traditional gaming and rendering. The Z2 Go's dedicated bandwidth and half-precision throughput serve compute-oriented tasks. Neither part is universally faster; the choice depends entirely on the memory architecture and power constraints of the host device.