AMD Steam Machine GPU vs Intel Arc Pro B390 Comparison

Where Each One Wins

The recorded data for the AMD Steam Machine GPU and the Intel Arc Pro B390 shows no benchmark entries, no head-to-head benchmark results, and no wins for either part. With an empty benchmark suite and zero wins on each side, a conventional use-case split based on measured performance cannot be derived from the database. What the data does reveal is a clear functional separation: the AMD Steam Machine GPU is a discrete, self-contained graphics card with dedicated video memory, while the Intel Arc Pro B390 is an integrated graphics processor (IGP) that shares system memory. Those structural differences define where each part is intended to operate.

The AMD Steam Machine GPU carries 8 GB of GDDR6 memory on a 128-bit bus, delivering 288.0 GB/s of dedicated bandwidth. This makes it suitable for workloads where local memory access and sustained bandwidth matter, such as gaming or GPU-accelerated rendering on a Valve console platform. Its 110 W TDP and lack of power connectors indicate it draws power from the slot or the host board, but it remains a discrete component with its own memory subsystem. The Intel Arc Pro B390, by contrast, uses "System Shared" memory with bandwidth listed as "System Dependent," meaning its performance scales with the host platform's memory configuration. Its 80 W TDP and "IGP" slot width confirm it is built into the processor package, targeting portable or low-power devices where a separate card is not feasible.

The generation labels reinforce this split. The AMD part is listed under "Console GPU (Valve)," pointing to a fixed, embedded gaming use case. The Intel part sits in "Arc Graphics-WM (Panther Lake)," a mobile or workstation-oriented integrated solution. Neither part has a launch MSRP in the database, so any price-based positioning is absent. The data therefore frames the choice not as one of measured speed, but of platform fit: the AMD Steam Machine GPU for a dedicated console-style card, the Intel Arc Pro B390 for an integrated mobile graphics solution.

FAQ

Q: Which GPU has more shading units?

A: The AMD Steam Machine GPU has 1792 shading units, while the Intel Arc Pro B390 has 1536 shading units. The AMD part leads by 256 units.

Q: How does the memory configuration differ?

A: The AMD Steam Machine GPU uses 8 GB of GDDR6 with a 128-bit bus and 288.0 GB/s bandwidth. The Intel Arc Pro B390 uses "System Shared" memory with a "System Shared" bus and bandwidth listed as "System Dependent."

Q: What are the boost clock speeds?

A: The AMD Steam Machine GPU boosts to 2450 MHz, with a game clock of 2250 MHz and a base clock of 1720 MHz. The Intel Arc Pro B390 boosts to 2500 MHz with a base clock of 300 MHz.

Q: Which part has a higher FP32 throughput?

A: The AMD Steam Machine GPU delivers 17.56 TFLOPS of FP32, while the Intel Arc Pro B390 delivers 7.680 TFLOPS. The AMD part is more than double the Intel part in this metric.

Q: Do both support the same API feature set?

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

Q: What is the process node difference?

A: The AMD Steam Machine GPU uses a 6 nm process from TSMC, while the Intel Arc Pro B390 uses a 3 nm process from Intel.

Head-to-Head Benchmarks

The database contains no head-to-head benchmark results for these two GPUs. With `headToHeadBenchmarks` empty and both `winsA` and `winsB` at zero, there are no measured scores to compare directly. The absence of benchmark data means any quantitative comparison must rely on the recorded specification fields, which do provide several notable deltas.

The most significant gap appears in FP32 compute. The AMD Steam Machine GPU delivers 17.56 TFLOPS, while the Intel Arc Pro B390 delivers 7.680 TFLOPS. That is a 2.29x advantage for the AMD part in single-precision floating-point throughput. The FP16 figures are closer in raw terms: the AMD part lists 17.56 TFLOPS (1:1) while the Intel part lists 15.36 TFLOPS (2:1). The Intel part's 2:1 ratio means it halves its FP32 rate to achieve that FP16 number, whereas the AMD part maintains a 1:1 ratio. In effect, the AMD part matches its FP32 performance in FP16, while the Intel part requires a doubled rate to reach a similar figure.

Pixel and texture rates further separate the two. The AMD Steam Machine GPU posts 156.8 GPixel/s and 274.4 GTexel/s, compared to the Intel Arc Pro B390's 60.00 GPixel/s and 120.0 GTexel/s. The AMD part leads by 2.61x in pixel rate and 2.29x in texture rate. These ratios align closely with the FP32 difference, suggesting the AMD part's larger shader and texture unit counts drive the gap. The AMD part has 112 TMUs and 64 ROPs, while the Intel part has 48 TMUs and 24 ROPs.

Ray tracing hardware also differs. The AMD Steam Machine GPU includes 28 RT cores, while the Intel Arc Pro B390 includes 12 RT cores. Without benchmark scores, the data cannot state how these translate into real-world ray-traced performance, but the 16-core difference indicates a broader RT pipeline on the AMD side.

Clock behavior is distinct. The Intel Arc Pro B390 has a higher boost clock at 2500 MHz versus 2450 MHz, but its base clock is drastically lower at 300 MHz versus 1720 MHz. The AMD part also lists a game clock of 2250 MHz, which the Intel part does not provide. The AMD part's narrower clock range suggests it runs at a more sustained high frequency, while the Intel part likely ramps up and down depending on thermal and power limits within an integrated package.

Memory bandwidth is another clear divider. The AMD Steam Machine GPU has 288.0 GB/s of dedicated bandwidth, while the Intel Arc Pro B390's bandwidth is "System Dependent," meaning it cannot be quantified in the database. The AMD part's memory runs at 2250 MHz with 18 Gbps effective, while the Intel part's memory clock is listed as "System Shared," further confirming its reliance on host memory.

Specification Differences

The two GPUs differ across nearly every measured field. The AMD Steam Machine GPU uses a 6 nm TSMC process, while the Intel Arc Pro B390 uses a 3 nm Intel process. The AMD part has 13,300 million transistors on a 204 mm² die with a transistor density of 65.2M / mm²; the Intel part lists transistor count, die size, and density as unknown.

Memory differs fundamentally. The AMD part has 8 GB of GDDR6 on a 128-bit bus with 288.0 GB/s bandwidth. The Intel part uses "System Shared" memory, bus width, and bandwidth, with the memory clock also listed as "System Shared."

Compute resources differ in count. The AMD part has 1792 shading units, 112 TMUs, 64 ROPs, and 28 RT cores. The Intel part has 1536 shading units, 48 TMUs, 24 ROPs, and 12 RT cores. Neither part lists tensor cores.

Rates differ as well. The AMD part delivers 156.8 GPixel/s, 274.4 GTexel/s, 17.56 TFLOPS FP32, and 17.56 TFLOPS FP16 (1:1). The Intel part delivers 60.00 GPixel/s, 120.0 GTexel/s, 7.680 TFLOPS FP32, and 15.36 TFLOPS FP16 (2:1).

Power and form factor diverge. The AMD part has a TDP of 110 W, no power connectors, and no listed slot width or bus interface. The Intel part has a TDP of 80 W, no power connectors, and is listed as "IGP" for both slot width and bus interface.

Physical dimensions exist only for the AMD part: 156 mm length, 152 mm height, and 162 mm width. The Intel part has no listed dimensions. Display outputs differ too: the AMD part lists 1x HDMI 2.1a and 1x DisplayPort 2.1, while the Intel part lists "Portable Device Dependent."

Release dates differ by five months. The AMD Steam Machine GPU has a release date of 2026-06-28, while the Intel Arc Pro B390 has a release date of 2026-01-26. The Intel part has a predecessor listed as "HD Graphics-WM," while the AMD part has no predecessor or successor. Both parts are listed as "Active" in production status.

The Intel part's architecture is Xe3-LPG on the Panther Lake chip, while the AMD part uses RDNA 3.0 on the Navi 33 chip with the codename "Hotpink Bonefish." The AMD part has no codename listed for the Intel side.

Architecture Differences

The architectural split is clear from the recorded fields. The AMD Steam Machine GPU uses RDNA 3.0, built on the Navi 33 chip with the codename "Hotpink Bonefish." The process is 6 nm from TSMC, with 13,300 million transistors on a 204 mm² die. The Intel Arc Pro B390 uses Xe3-LPG, built on the Panther Lake chip, with a 3 nm process from Intel. Transistor count and die size are unknown for the Intel part.

The AMD part belongs to the "Console GPU (Valve)" generation, indicating a fixed-function console design. The Intel part belongs to "Arc Graphics-WM (Panther Lake)," with a predecessor of "HD Graphics-WM," showing lineage from Intel's integrated graphics line. The Intel part is explicitly an IGP, while the AMD part is a discrete GPU with its own memory.

Memory architecture differs by design. The AMD part has dedicated GDDR6, while the Intel part shares system memory. That fundamental difference affects bandwidth predictability: the AMD part has a fixed 288.0 GB/s, while the Intel part's bandwidth is "System Dependent."

The shader and fixed-function unit counts reflect different design philosophies. The AMD part has more of everything: 1792 shading units versus 1536, 112 TMUs versus 48, 64 ROPs versus 24, and 28 RT cores versus 12. The Intel part compensates with a higher boost clock (2500 MHz versus 2450 MHz) but a much lower base clock (300 MHz versus 1720 MHz). The AMD part's 2250 MHz game clock indicates a sustained performance level that the Intel part does not specify.

FP16 ratios differ. The AMD part lists 17.56 TFLOPS FP16 with a 1:1 ratio, meaning it does not double its rate for half precision. The Intel part lists 15.36 TFLOPS FP16 with a 2:1 ratio, meaning it doubles its rate relative to FP32. This suggests the Intel architecture is tuned for workloads that benefit from half-precision throughput, while the AMD part provides consistent throughput across both formats.

API support is identical: both list DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The AMD part's display outputs are fixed at 1x HDMI 2.1a and 1x DisplayPort 2.1, while the Intel part's outputs are "Portable Device Dependent," reflecting its integrated nature in portable devices.

The power envelope differs by 30 W. The AMD part is rated at 110 W, the Intel part at 80 W. Both use no power connectors, which is expected for the Intel IGP and indicates the AMD part draws power from its slot or board interface. The AMD part's physical dimensions (156 mm, 152 mm, 162 mm) show a standalone card form factor, while the Intel part has no dimensions because it is integrated.

The Verdict

The data indicates two different products for two different platforms. The AMD Steam Machine GPU is a discrete console GPU with dedicated memory, larger compute resources, and higher throughput across every measured rate. It delivers 17.56 TFLOPS FP32, 156.8 GPixel/s, and 274.4 GTexel/s, backed by 8 GB of GDDR6 and 288.0 GB/s of bandwidth. Its 110 W TDP and no power connectors suggest a slot-powered design for a Valve console. The Intel Arc Pro B390 is an integrated GPU with 7.680 TFLOPS FP32, 60.00 GPixel/s, and 120.0 GTexel/s, relying on system memory with bandwidth that varies by host. Its 80 W TDP and "IGP" classification point to a mobile or low-power embedded solution.

For users or systems that need a dedicated graphics card with fixed memory bandwidth and higher raw compute, the AMD Steam Machine GPU is the choice based on the specification deltas. Its FP32 output is more than double the Intel part, and its pixel and texture rates are similarly ahead. The RT core count is also higher at 28 versus 12. For a platform that requires an integrated solution with no separate card, the Intel Arc Pro B390 is the only option, and its 3 nm process and higher boost clock (2500 MHz) indicate a modern, power-efficient design.

The absence of benchmark scores means no measured performance verdict can be stated. The database records no wins for either part and no head-to-head results. What the data does support is a structural verdict: the AMD Steam Machine GPU is built for dedicated throughput, the Intel Arc Pro B390 for integrated efficiency. The 30 W TDP difference and the memory architecture difference are the decisive factors. A system with room for a discrete card gets the AMD part's dedicated 288.0 GB/s and 17.56 TFLOPS. A system without that room gets the Intel part's integrated 7.680 TFLOPS and system-dependent bandwidth. Neither part has a launch MSRP in the database, so no cost-based guidance is available. The recorded data directs each part to its own platform, with the AMD part offering more raw resources and the Intel part offering integration.