AMD Instinct MI308X vs AMD Steam Machine GPU Comparison

AMD
RADEON

AMD Instinct MI308X

CORE STATE Aqua Vanjaram
VRAM 192 GB
CLOCK SPEED 2100 MHz
TDP 750 W
BUS WIDTH 8192 bit
ARCHITECTURE CDNA 3.0
nm
PROCESS 5 nm
LAUNCH DATE 2023
VS
AMD
RADEON

Steam Machine GPU

CORE STATE Navi 33
VRAM 8 GB
CLOCK SPEED 2450 MHz
TDP 110 W
BUS WIDTH 128 bit
ARCHITECTURE RDNA 3.0
nm
PROCESS 6 nm
LAUNCH DATE 2026

Analysis: AMD Instinct MI308X vs AMD Steam Machine GPU

The Verdict

The database comparison between the AMD Instinct MI308X and the AMD Steam Machine GPU reveals two entirely different compute devices serving distinct purposes. The MI308X is a data center accelerator built on the CDNA 3.0 architecture, while the Steam Machine GPU is a compact console graphics processor using RDNA 3.0. Neither device has recorded benchmark scores in the database, so both sit at the 50th percentile among all GPUs with an average benchmark score of zero. The MI308X targets massive parallel compute workloads with 19,456 shading units and 192 GB of HBM3 memory, while the Steam Machine GPU is designed for gaming with 1,792 shading units and 8 GB of GDDR6 memory. The data indicates the MI308X is for server deployments where raw throughput matters, while the Steam Machine GPU belongs in a small form factor console chassis. For anyone needing a headless compute accelerator with enormous memory capacity, the MI308X is the clear choice. For a console GPU with display outputs and standard graphics APIs, the Steam Machine GPU is designed for that role.

Architecture Differences

The two processors come from different architectural families within AMD's lineup. The MI308X uses CDNA 3.0, AMD's compute-focused architecture, built on a 5 nm process at TSMC with the chip codename Aqua Vanjaram. It belongs to the Instinct (MIx) generation and has a transistor count of 153,000 million across a 1017 mm² die, resulting in a transistor density of 150.4M per mm². The Steam Machine GPU uses RDNA 3.0, a graphics-oriented architecture, built on a 6 nm process at TSMC with the Navi 33 chip and the codename Hotpink Bonefish. It belongs to the Console GPU (Valve) generation and contains 13,300 million transistors on a 204 mm² die, giving a transistor density of 65.2M per mm².

The MI308X has 19,456 shading units, 1,216 texture mapping units, and zero ROPs, with no dedicated ray tracing cores or tensor cores listed. Its pixel rate is recorded as 0 MPixel/s, while its texture rate reaches 2,553.6 GTexel/s. The Steam Machine GPU has 1,792 shading units, 112 texture mapping units, and 64 ROPs, plus 28 ray tracing cores. Its pixel rate is 156.8 GPixel/s, and its texture rate is 274.4 GTexel/s. The MI308X has no display outputs and exposes no graphics APIs, with DirectX, OpenGL, and Vulkan all marked as N/A. The Steam Machine GPU provides 1x HDMI 2.1a and 1x DisplayPort 2.1 outputs, and supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The MI308X uses a PCIe 5.0 x16 bus interface, while the Steam Machine GPU has no listed bus interface, reflecting its soldered console design.

Clock behavior differs sharply between the two. The MI308X runs at a 1000 MHz base clock and 2100 MHz boost clock, with memory at 1300 MHz (5.2 Gbps effective). The Steam Machine GPU runs at 1720 MHz base, 2450 MHz boost, and 2250 MHz game clock, with memory at 2250 MHz (18 Gbps effective). The MI308X's memory configuration uses HBM3 across an 8192-bit bus, yielding 5.32 TB/s bandwidth. The Steam Machine GPU uses GDDR6 across a 128-bit bus, producing 288.0 GB/s bandwidth. The MI308X has no power connectors and an OAM Module slot width, with a TDP of 750 W and a suggested PSU of 1150 W. The Steam Machine GPU also has no power connectors, has a 110 W TDP, and no suggested PSU listed.

Head-to-Head Benchmarks

The database contains no recorded benchmark results for either device, so direct performance comparisons rely on the specification data. In raw compute throughput, the MI308X delivers 81.72 TFLOPS for both FP32 and FP16 (1:1 ratio). The Steam Machine GPU delivers 17.56 TFLOPS for both FP32 and FP16 (1:1 ratio). This means the MI308X provides approximately 4.65 times the floating-point throughput of the Steam Machine GPU in both precisions, based on the listed figures.

Memory capacity and bandwidth show an even larger gap. The MI308X holds 192 GB of HBM3 memory, which is 24 times the Steam Machine GPU's 8 GB of GDDR6. The MI308X's 5.32 TB/s bandwidth is roughly 18.5 times the Steam Machine GPU's 288.0 GB/s. The memory bus width difference is substantial as well: 8192 bits versus 128 bits, a 64-fold difference. These figures indicate the MI308X is designed to hold massive datasets and move them quickly, while the Steam Machine GPU's memory subsystem suits typical console game loads.

Texture processing rates also favor the MI308X heavily. It achieves 2,553.6 GTexel/s compared to 274.4 GTexel/s for the Steam Machine GPU, a ratio of about 9.3 to 1. However, the Steam Machine GPU has a functional pixel rate of 156.8 GPixel/s, while the MI308X is listed at 0 MPixel/s because it lacks ROPs entirely. This reflects the MI308X's compute-only role, where pixel output is not applicable. The Steam Machine GPU's 64 ROPs enable actual rasterization, which the MI308X cannot perform.

The transistor and die size differences illustrate the scale gap. The MI308X uses 153,000 million transistors on a 1017 mm² die, while the Steam Machine GPU uses 13,300 million on 204 mm². The MI308X die is roughly five times larger in area and contains about 11.5 times more transistors. Despite the larger process node for the Steam Machine GPU at 6 nm versus 5 nm for the MI308X, the MI308X's transistor density is still more than double at 150.4M per mm² versus 65.2M per mm².

Power consumption reveals the intended deployment environments. The MI308X draws 750 W, while the Steam Machine GPU draws 110 W. The MI308X requires a suggested PSU of 1150 W, while the Steam Machine GPU has no suggested PSU listed, consistent with its console integration where the system power supply is fixed. The MI308X's physical format is an OAM Module, designed for server racks, whereas the Steam Machine GPU has dimensions of 156 mm length, 152 mm height, and 162 mm width, suitable for a compact console shell.

Specification Differences

The two devices differ across nearly every measurable specification field. The MI308X uses a 5 nm process node, while the Steam Machine GPU uses 6 nm. Transistor counts are 153,000 million versus 13,300 million. Die sizes are 1017 mm² versus 204 mm². Transistor densities are 150.4M per mm² versus 65.2M per mm². Base clocks are 1000 MHz versus 1720 MHz. Boost clocks are 2100 MHz versus 2450 MHz. The Steam Machine GPU has a game clock of 2250 MHz, while the MI308X has none listed. Memory clocks are 1300 MHz (5.2 Gbps effective) versus 2250 MHz (18 Gbps effective). Memory sizes are 192 GB versus 8 GB. Memory types are HBM3 versus GDDR6. Bus widths are 8192 bit versus 128 bit. Memory bandwidth is 5.32 TB/s versus 288.0 GB/s.

Shading units number 19,456 versus 1,792. Texture mapping units number 1,216 versus 112. ROPs are 0 versus 64. Ray tracing cores are absent on the MI308X and 28 on the Steam Machine GPU. Pixel rates are 0 MPixel/s versus 156.8 GPixel/s. Texture rates are 2,553.6 GTexel/s versus 274.4 GTexel/s. FP32 and FP16 performance are both 81.72 TFLOPS on the MI308X versus 17.56 TFLOPS on the Steam Machine GPU. TDP is 750 W versus 110 W. Slot width is OAM Module versus not listed. Suggested PSU is 1150 W versus not listed. Bus interface is PCIe 5.0 x16 versus not listed. Display outputs are none versus 1x HDMI 2.1a and 1x DisplayPort 2.1. DirectX support is N/A versus 12 Ultimate (12_2). OpenGL is N/A versus 4.6. Vulkan is N/A versus 1.4. Dimensions are not listed for the MI308X, while the Steam Machine GPU measures 156 mm by 152 mm by 162 mm. Release dates are 2023-12-05 for the MI308X and 2026-06-28 for the Steam Machine GPU. Production status is not listed for the MI308X and Active for the Steam Machine GPU. The MI308X's predecessor is Radeon Instinct, while the Steam Machine GPU has no predecessor listed. The architectures differ as CDNA 3.0 versus RDNA 3.0, and the generations differ as Instinct (MIx) versus Console GPU (Valve).

FAQ

Q: Which GPU has more shading units?

A: The AMD Instinct MI308X has 19,456 shading units, compared to 1,792 for the AMD Steam Machine GPU.

Q: What is the memory capacity difference between the two?

A: The MI308X has 192 GB of HBM3 memory, while the Steam Machine GPU has 8 GB of GDDR6 memory. The MI308X also has a 8192-bit memory bus versus a 128-bit bus, and 5.32 TB/s bandwidth versus 288.0 GB/s.

Q: Does the Steam Machine GPU support ray tracing?

A: Yes, the Steam Machine GPU has 28 ray tracing cores. The MI308X has no ray tracing cores listed, and its graphics APIs are marked as N/A.

Q: What display outputs does each GPU provide?

A: The Steam Machine GPU provides 1x HDMI 2.1a and 1x DisplayPort 2.1. The MI308X has no display outputs and is a headless compute accelerator.

Q: How do the power requirements compare?

A: The MI308X has a TDP of 750 W and a suggested PSU of 1150 W. The Steam Machine GPU has a TDP of 110 W and no suggested PSU listed.

Q: What are the release dates for these products?

A: The MI308X was released on 2023-12-05, and the Steam Machine GPU is scheduled for release on 2026-06-28. The Steam Machine GPU has an Active production status, while the MI308X has no production status listed.

DETAILED SPECIFICATIONS

SPECIFICATION
Instinct MI308X
Steam Machine GPU
Core Specs
Shading Units
19,456
1,792 -90.8%
Shaders
19,456
1,792 -90.8%
TMUs
1,216
112 -90.8%
ROPs
0
64 +∞%
Compute Units
304
28 -90.8%
Clocks
Base Clock
1000 MHz
1720 MHz
Boost Clock
2100 MHz
2450 MHz
Game Clock
2250 MHz
Memory Clock
1300 MHz 5.2 Gbps effective
2250 MHz 18 Gbps effective
Memory
Memory Size
192 GB
8 GB
VRAM (MB)
196,608
8,192 -95.8%
Memory Type
HBM3
GDDR6
Memory Bus
8192 bit
128 bit
Bandwidth
5.32 TB/s
288.0 GB/s
Cache
L1 Cache
16 KB (per CU)
128 KB per Array
L2 Cache
16 MB
2 MB
L3 Cache
256 MB
32 MB
L0 Cache
32 KB per WGP
Performance
Pixel Rate
0 MPixel/s
156.8 GPixel/s
Texture Rate
2,553.6 GTexel/s
274.4 GTexel/s
FP32 (TFLOPS)
81.72 TFLOPS
17.56 TFLOPS
FP64 (TFLOPS)
40.86 TFLOPS (1:2)
548.8 GFLOPS (1:32)
FP16 (TFLOPS)
81.72 TFLOPS (1:1)
17.56 TFLOPS (1:1)
AI/RT
RT Cores
28
Matrix Cores
1,216
56 -95.4%
Power
TDP
750 W
110 W
TDP (W)
750
110 -85.3%
Suggested PSU
1150 W
Power Connectors
None
None
Architecture
Architecture
CDNA 3.0
RDNA 3.0
GPU Name
Aqua Vanjaram
Navi 33
Codename
Hotpink Bonefish
Generation
Instinct (MIx)
Console GPU (Valve)
Process Size
5 nm
6 nm
Transistors
153,000 million
13,300 million
Die Size
1017 mm²
204 mm²
Foundry
TSMC
TSMC
Density
150.4M / mm²
65.2M / mm²
AMD MCM
MCM
2
API Support
DirectX
12 Ultimate (12_2)
OpenGL
4.6
Vulkan
1.4
OpenCL
3.0
2.2
Shader Model
6.9
Physical
Slot Width
OAM Module
Length
156 mm 6.1 inches
Height
152 mm 6 inches
Outputs
No outputs
1x HDMI 2.1a1x DisplayPort 2.1
Bus Interface
PCIe 5.0 x16
Other
Production
Active
Predecessor
Radeon Instinct
View Instinct MI308X Details View Steam Machine GPU Details