AMD Instinct MI355X vs AMD Steam Machine GPU Comparison

Head-to-Head Benchmarks

The recorded data contains no benchmark scores for either the AMD Instinct MI355X or the AMD Steam Machine GPU. The database shows an average benchmark score of 0 for both parts, and the percentile versus all GPUs is identical at 50 for each. With no fill-rate, ray tracing, or synthetic workload results available, the head-to-head comparison rests entirely on architectural specifications and compute capabilities rather than measured performance deltas.

The Instinct MI355X is clearly positioned as a compute accelerator. Its FP32 throughput is listed at 78.64 TFLOPS, and its FP16 throughput matches at 78.64 TFLOPS with a 1:1 ratio. The Steam Machine GPU, by contrast, delivers 17.56 TFLOPS in both FP32 and FP16, also at a 1:1 ratio. That places the MI355X at roughly 4.5 times the raw floating-point throughput of the Steam Machine GPU. Texture rate tells a similar story: the MI355X reaches 2,457.6 GTexel/s, while the Steam Machine GPU manages 274.4 GTexel/s. The pixel rate is where the comparison becomes unusual. The MI355X reports 0 MPixel/s, while the Steam Machine GPU delivers 156.8 GPixel/s. The MI355X has no ROPs listed, which explains the zero pixel rate, a common trait for purpose-built compute accelerators that do not drive displays.

Architecture Differences

The two GPUs belong to completely different design families. The AMD Instinct MI355X uses the MI350 256CU chip built on CDNA 4.0 architecture. It is fabricated on a 3 nm process at TSMC, with 185,000 million transistors packed into a 2380 mm² die. That works out to a transistor density of 77.7 million transistors per square millimeter. The Steam Machine GPU uses the Navi 33 chip built on RDNA 3.0 architecture, with the codename Hotpink Bonefish. It is fabricated on a 6 nm process at TSMC, with 13,300 million transistors on a 204 mm² die, giving a density of 65.2 million transistors per square millimeter. The MI355X therefore uses a smaller process node and a dramatically larger die, with roughly 14 times the transistor count.

The shader configuration differs massively. The MI355X lists 16,384 shading units and 1,024 texture mapping units, but zero ROPs and no ray tracing cores or tensor core counts. The Steam Machine GPU lists 1,792 shading units, 112 TMUs, 64 ROPs, and 28 ray tracing cores. The MI355X is a pure compute device with no graphics rasterization pipeline, which is consistent with its zero pixel rate and lack of display outputs. The Steam Machine GPU is a full graphics processor with a conventional rasterization pipeline, display outputs, and ray tracing support.

Memory architecture is another major divergence. The MI355X carries 288 GB of HBM3e memory on an 8192-bit bus, with memory clocked at 2000 MHz and 8 Gbps effective, producing 8.19 TB/s of bandwidth. The Steam Machine GPU carries 8 GB of GDDR6 memory on a 128-bit bus, with memory clocked at 2250 MHz and 18 Gbps effective, producing 288.0 GB/s of bandwidth. The MI355X has 36 times the memory capacity and roughly 28 times the memory bandwidth. The Steam Machine GPU counters with a much higher effective memory clock speed, but the narrow bus and small capacity keep it far behind in aggregate bandwidth.

Power and physical design also differ sharply. The MI355X has a TDP of 1400 W, a suggested PSU of 1800 W, and comes as an OAM Module with no power connectors listed, since OAM carriers provide power through the socket. It measures 102 mm in length and 165 mm in width. The Steam Machine GPU has a TDP of 110 W, no suggested PSU listed, and measures 156 mm in length, 152 mm in height, and 162 mm in width. The Steam Machine GPU is a compact console-class part, while the MI355X is a high-power accelerator module.

Where Each One Wins

The MI355X wins decisively in compute throughput, memory capacity, memory bandwidth, and transistor scale. Its 78.64 TFLOPS in FP32 and FP16 makes it suitable for large-scale numerical workloads, and the 288 GB HBM3e frame buffer is an enormous pool for models and datasets that exceed the 8 GB capacity of the Steam Machine GPU. The 8.19 TB/s bandwidth is in a different class from the 288.0 GB/s of the Steam Machine GPU, which matters for memory-bound kernels. The 3 nm process and 185,000 million transistors indicate a part built for maximum compute density, not efficiency per watt in a consumer context.

The Steam Machine GPU wins in areas tied to graphics and practical integration. It has 64 ROPs and a pixel rate of 156.8 GPixel/s, while the MI355X has none. It supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, while the MI355X lists N/A for all three APIs. The Steam Machine GPU also provides 1x HDMI 2.1a and 1x DisplayPort 2.1 outputs, while the MI355X has no display outputs at all. The Steam Machine GPU includes 28 ray tracing cores, which the MI355X does not list, making it the only one of the two with any ray tracing capability. Its 110 W TDP is far lower than the 1400 W TDP of the MI355X, and its compact dimensions (156 mm length, 152 mm height, 162 mm width) allow it to fit in a console chassis, whereas the MI355X is a 102 mm by 165 mm OAM module designed for server racks.

The Steam Machine GPU also has a higher base clock at 1720 MHz versus 1000 MHz, and a slightly higher boost clock at 2450 MHz versus 2400 MHz. Its game clock is listed at 2250 MHz, a figure the MI355X does not report. Clock speed alone does not compensate for the massive difference in shading units and memory bandwidth, but it does indicate a part tuned for lighter, latency-sensitive workloads.

Specification Differences

• Process node: MI355X is 3 nm, Steam Machine GPU is 6 nm.
• Transistors: MI355X has 185,000 million, Steam Machine GPU has 13,300 million.
• Die size: MI355X is 2380 mm², Steam Machine GPU is 204 mm².
• Transistor density: MI355X is 77.7M / mm², Steam Machine GPU is 65.2M / mm².
• Base clock: MI355X is 1000 MHz, Steam Machine GPU is 1720 MHz.
• Boost clock: MI355X is 2400 MHz, Steam Machine GPU is 2450 MHz.
• Game clock: MI355X has none listed, Steam Machine GPU is 2250 MHz.
• Memory size: MI355X is 288 GB, Steam Machine GPU is 8 GB.
• Memory type: MI355X is HBM3e, Steam Machine GPU is GDDR6.
• Memory bus width: MI355X is 8192 bit, Steam Machine GPU is 128 bit.
• Memory bandwidth: MI355X is 8.19 TB/s, Steam Machine GPU is 288.0 GB/s.
• Shading units: MI355X has 16,384, Steam Machine GPU has 1,792.
• TMUs: MI355X has 1,024, Steam Machine GPU has 112.
• ROPs: MI355X has 0, Steam Machine GPU has 64.
• Ray tracing cores: MI355X has none listed, Steam Machine GPU has 28.
• Pixel rate: MI355X is 0 MPixel/s, Steam Machine GPU is 156.8 GPixel/s.
• Texture rate: MI355X is 2,457.6 GTexel/s, Steam Machine GPU is 274.4 GTexel/s.
• FP32: MI355X is 78.64 TFLOPS, Steam Machine GPU is 17.56 TFLOPS.
• FP16: MI355X is 78.64 TFLOPS, Steam Machine GPU is 17.56 TFLOPS.
• TDP: MI355X is 1400 W, Steam Machine GPU is 110 W.
• Suggested PSU: MI355X is 1800 W, Steam Machine GPU has none listed.
• Slot width: MI355X is OAM Module, Steam Machine GPU has none listed.
• Bus interface: MI355X is PCIe 5.0 x16, Steam Machine GPU has none listed.
• Display outputs: MI355X has none, Steam Machine GPU has 1x HDMI 2.1a and 1x DisplayPort 2.1.
• DirectX support: MI355X is N/A, Steam Machine GPU is 12 Ultimate (12_2).
• OpenGL support: MI355X is N/A, Steam Machine GPU is 4.6.
• Vulkan support: MI355X is N/A, Steam Machine GPU is 1.4.
• Dimensions: MI355X is 102 mm by 165 mm, Steam Machine GPU is 156 mm by 152 mm by 162 mm.
• Production status: MI355X has none listed, Steam Machine GPU is Active.
• Release date: MI355X is 2025-06-11, Steam Machine GPU is 2026-06-28.
• Predecessor: MI355X is Radeon Instinct, Steam Machine GPU has none listed.
• Codename: MI355X has none listed, Steam Machine GPU is Hotpink Bonefish.
• Generation: MI355X is Instinct (MIx), Steam Machine GPU is Console GPU (Valve).

FAQ

Q: Which GPU has more memory bandwidth?

A: The AMD Instinct MI355X has 8.19 TB/s of bandwidth from HBM3e memory on an 8192-bit bus. The AMD Steam Machine GPU has 288.0 GB/s from GDDR6 on a 128-bit bus.

Q: Does the AMD Steam Machine GPU support ray tracing?

A: Yes, the Steam Machine GPU lists 28 ray tracing cores. The AMD Instinct MI355X does not list any ray tracing cores.

Q: Which GPU supports modern graphics APIs?

A: The AMD Steam Machine GPU supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The AMD Instinct MI355X lists N/A for DirectX, OpenGL, and Vulkan.

Q: Can the AMD Instinct MI355X output video to a display?

A: No, the MI355X has no display outputs listed and a pixel rate of 0 MPixel/s. The Steam Machine GPU has 1x HDMI 2.1a and 1x DisplayPort 2.1 outputs.

Q: What is the power requirement difference between the two?

A: The MI355X has a TDP of 1400 W and a suggested PSU of 1800 W. The Steam Machine GPU has a TDP of 110 W and no suggested PSU listed.

Q: How do their release dates compare?

A: The AMD Instinct MI355X has a release date of 2025-06-11. The AMD Steam Machine GPU has a release date of 2026-06-28.

The Verdict

The recorded data separates these two GPUs into entirely different product categories. The AMD Instinct MI355X is a compute accelerator with 78.64 TFLOPS of FP32 throughput, 288 GB of HBM3e memory, 8.19 TB/s of bandwidth, and a 1400 W TDP. It has no rasterization pipeline, no display outputs, and no graphics API support. The AMD Steam Machine GPU is a compact console GPU with 17.56 TFLOPS of FP32 throughput, 8 GB of GDDR6 memory, 288.0 GB/s of bandwidth, 64 ROPs, 28 ray tracing cores, and a 110 W TDP. It supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

For anyone selecting a part for large-scale compute workloads, the MI355X is the only choice from this data. Its memory capacity is 36 times larger, its bandwidth is roughly 28 times higher, and its FP32 throughput is about 4.5 times higher. For anyone building a graphics-oriented system, the Steam Machine GPU is the only option, because the MI355X cannot render frames or drive a display. The Steam Machine GPU also wins on efficiency per watt, since it delivers its 17.56 TFLOPS at 110 W, while the MI355X requires 1400 W to reach 78.64 TFLOPS. Neither part has benchmark scores in the database, so the verdict rests on specifications alone. The data indicates two specialized tools with no meaningful overlap in intended use.