AMD Instinct MI350X vs AMD Radeon 890M Comparison

Head-to-Head Benchmarks

The recorded data for the AMD Instinct MI350X contains no benchmark scores, while the AMD Radeon 890M has an average benchmark score of 9210. This means the MI350X cannot be compared directly through the standard benchmark suite in the database. The 890M’s percentile ranking sits at 45 among all GPUs, while the MI350X is ranked at the 50th percentile, but that ranking is based on zero measured results for the accelerator.

The Radeon 890M shows its strongest results in Geekbench Vulkan with a score of 40808 and Geekbench OpenCL with 37254. Its Passmark G3D score is 8076, and its Passmark G2D score is 979. The compute-focused Passmark GPU Compute test returns 4157. The 890M’s nearest rivals in the database are AMD Radeon Vega 8 at 9221 (0.1% higher), NVIDIA GeForce GTX 960 at 9273 (0.7% higher), NVIDIA GeForce GTX 465 at 9294 (0.9% higher), and NVIDIA GeForce GTX 850M at 9302 (1% higher). The 890M trails each of these by small margins, confirming that its average score of 9210 places it in the same performance tier as those older discrete and integrated solutions.

The MI350X, by contrast, has no benchmark entries, no average score, and no nearest rivals listed in the database. Its percentile of 50 is a placeholder rather than a measured outcome. Any head-to-head numerical comparison between the two parts must therefore rely on the architectural and specification data, not on actual workload results. The 890M delivers measurable scores across ten tests, while the MI350X delivers none, so the only direct quantitative relationship available is the specification sheet.

Architecture Differences

The MI350X uses the CDNA 4.0 architecture, built on a 3 nm process at TSMC, while the 890M uses RDNA 3.5, built on a 4 nm process also at TSMC. The MI350X is a compute accelerator with 16384 shading units, 1024 texture mapping units, and zero ROPs, reflecting its design for dense parallel workloads rather than rasterization. The 890M is an integrated graphics processor with 1024 shading units, 64 TMUs, 32 ROPs, and 16 ray tracing cores.

The MI350X packs 185,000 million transistors on a die size of 2380 mm², giving a transistor density of 77.7 million per square millimeter. The 890M contains 34,000 million transistors on a 233 mm² die, with a density of 145.9 million per square millimeter. The MI350X’s transistor count is more than five times higher, but its density is roughly half, because the accelerator uses a much larger physical die.

Memory architecture differs fundamentally. The MI350X ships with 288 GB of HBM3e memory on an 8192-bit bus, delivering 8.19 TB/s of bandwidth, with memory clocked at 2000 MHz (8 Gbps effective). The 890M uses system shared memory, with a system-dependent bus width and bandwidth, meaning its performance scales with the host platform’s memory configuration.

Clock speeds also diverge. The MI350X runs at a base clock of 1000 MHz and a boost clock of 2200 MHz. The 890M runs at a base of 400 MHz and a boost of 2900 MHz. The 890M’s higher boost clock reflects its focus on bursty, power-limited graphics tasks, while the MI350X prioritizes sustained throughput.

Where Each One Wins

The MI350X wins decisively in raw compute throughput. Its FP32 performance is 72.09 TFLOPS, and its FP16 performance is also 72.09 TFLOPS at a 1:1 ratio. The 890M delivers 5.939 TFLOPS in both FP32 and FP16, also at 1:1. That puts the MI350X at roughly 12 times the floating-point throughput of the 890M. Texture rate follows the same pattern: the MI350X reaches 2252.8 GTexel/s, while the 890M reaches 185.6 GTexel/s.

The 890M wins in rasterization features. It has 32 ROPs and a pixel rate of 92.80 GPixel/s, while the MI350X has zero ROPs and a pixel rate of 0 MPixel/s. The 890M also supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, whereas the MI350X lists N/A for all three APIs. The 890M includes 16 ray tracing cores; the MI350X has no ray tracing core count listed.

The MI350X wins in memory capacity and bandwidth by an enormous margin: 288 GB versus system shared, 8.19 TB/s versus system dependent. The 890M wins in power efficiency at the specification level. Its TDP is 15 W, while the MI350X draws up to 1000 W. The 890M requires no external power connectors and runs as an IGP, while the MI350X is an OAM module with no power connectors listed but a suggested PSU of 1400 W.

Specification Differences

The two parts differ across nearly every measurable specification. The MI350X uses a 3 nm process; the 890M uses 4 nm. The MI350X has 185,000 million transistors versus 34,000 million for the 890M. Die size is 2380 mm² versus 233 mm². Transistor density is 77.7M per mm² versus 145.9M per mm².

Base clock: 1000 MHz versus 400 MHz. Boost clock: 2200 MHz versus 2900 MHz. Memory clock: 2000 MHz (8 Gbps effective) versus system shared. Memory size: 288 GB HBM3e versus system shared. Bus width: 8192 bit versus system shared. Bandwidth: 8.19 TB/s versus system dependent.

Shading units: 16384 versus 1024. TMUs: 1024 versus 64. ROPs: 0 versus 32. Ray tracing cores: not listed versus 16. Pixel rate: 0 MPixel/s versus 92.80 GPixel/s. Texture rate: 2252.8 GTexel/s versus 185.6 GTexel/s. FP32: 72.09 TFLOPS versus 5.939 TFLOPS. FP16: 72.09 TFLOPS versus 5.939 TFLOPS.

TDP: 1000 W versus 15 W. Slot width: OAM module versus IGP. Power connectors: none for both, but the MI350X suggests a 1400 W PSU while the 890M lists none. Bus interface: PCIe 5.0 x16 versus PCIe 4.0 x8. Display outputs: none versus portable device dependent.

DirectX: N/A versus 12 Ultimate (12_2). OpenGL: N/A versus 4.6. Vulkan: N/A versus 1.4. Dimensions: the MI350X measures 102 mm by 165 mm; the 890M has no listed dimensions. Release date: June 11, 2025 for the MI350X, July 14, 2024 for the 890M. Production status: not listed for the MI350X, active for the 890M. Predecessor: Radeon Instinct for the MI350X, Navi II IGP for the 890M.

FAQ

Q: Which GPU has higher FP32 performance?

A: The AMD Instinct MI350X delivers 72.09 TFLOPS FP32, while the AMD Radeon 890M delivers 5.939 TFLOPS. The MI350X is roughly 12 times higher in raw floating-point throughput.

Q: Can the MI350X output video to a display?

A: No. The MI350X lists no display outputs, while the 890M lists portable device dependent outputs. The MI350X also has zero ROPs and a pixel rate of 0 MPixel/s, confirming it is not designed for display or rasterization tasks.

Q: What is the power draw difference?

A: The MI350X has a TDP of 1000 W and a suggested PSU of 1400 W. The 890M has a TDP of 15 W and no suggested PSU listed. The 890M is an IGP, while the MI350X is an OAM module.

Q: Does the 890M support ray tracing?

A: Yes. The 890M includes 16 ray tracing cores. The MI350X lists no ray tracing core count.

Q: How much memory does each GPU have?

A: The MI350X has 288 GB of HBM3e on an 8192-bit bus with 8.19 TB/s bandwidth. The 890M uses system shared memory with a system dependent bus width and bandwidth.

Q: Which GPU has a higher boost clock?

A: The 890M boosts to 2900 MHz, while the MI350X boosts to 2200 MHz. The 890M also has a lower base clock at 400 MHz versus 1000 MHz.

The Verdict

The data shows two completely different classes of hardware. The AMD Instinct MI350X is a compute accelerator with massive FP32 and FP16 throughput, 288 GB of HBM3e, and a 1000 W TDP. Its architecture, CDNA 4.0, targets dense parallel computation, not graphics. The AMD Radeon 890M is an integrated GPU for mobile devices, with 15 W power draw, 16 ray tracing cores, and full support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

The MI350X has no benchmark scores in the database, so its 50th percentile ranking is unverified. The 890M has a measured average score of 9210, placing it at the 45th percentile, just behind the Vega 8, GTX 960, GTX 465, and GTX 850M. Those rivals outscore it by 0.1% to 1%, which places the 890M in a tight performance band near those older parts.

For compute workloads, the MI350X is the only choice based on the recorded specifications. Its 72.09 TFLOPS FP32, 8.19 TB/s memory bandwidth, and 288 GB capacity are orders of magnitude beyond the 890M. For graphics and display tasks, the 890M is the only option, as the MI350X has no ROPs, no display outputs, and no graphics API support.

A builder selecting between these two should base the decision on the workload. The MI350X suits server or datacenter compute environments requiring high throughput and large memory. The 890M suits portable devices needing integrated graphics with ray tracing and modern API support. The database records no overlap in their measured capabilities, and the specification sheets confirm they serve distinct roles with no meaningful head-to-head competition.