Intel Arc A380E vs NVIDIA GeForce RTX 5090 SE Comparison

Where Each One Wins

The Intel Arc A380E and NVIDIA GeForce RTX 5090 SE occupy opposite ends of the GPU spectrum, and the recorded data reflects a complete separation of use cases. The Arc A380E is an end-of-life, single-slot, 75 W part built for embedded or low-power systems. It draws no auxiliary power connectors and requires only a 250 W suggested power supply. The RTX 5090 SE is an active, dual-slot flagship with a 500 W TDP, a single 16-pin connector, and a 900 W suggested power supply. In any metric involving raw throughput, memory capacity, or feature set, the RTX 5090 SE wins outright. The Arc A380E wins only in the narrow sense of power efficiency and physical footprint, where its 75 W envelope and single-slot design allow deployment in chassis that cannot accommodate a 500 W, dual-slot card.

The RTX 5090 SE delivers 66.94 TFLOPS of FP32 compute versus 4.096 TFLOPS for the Arc A380E, a 16.3x gap. Its 24 GB of GDDR7 memory on a 384-bit bus provides 1.34 TB/s of bandwidth, compared to 6 GB of GDDR6 on a 96-bit bus at 186.0 GB/s. The RTX 5090 SE also has 110 RT cores and 440 tensor cores, while the Arc A380E has 8 RT cores and no dedicated tensor cores. Consequently, the RTX 5090 SE is the only viable choice for high-resolution gaming, AI inference, machine learning training, or professional ray-traced rendering. The Arc A380E suits basic display output, legacy workloads, or low-power embedded tasks where its 4x DisplayPort 2.0 outputs and 4.096 TFLOPS are sufficient.

FAQ

Q: Which GPU has more shading units?

A: The RTX 5090 SE has 14,080 shading units, while the Arc A380E has 1,024.

Q: How much memory bandwidth does each card provide?

A: The RTX 5090 SE provides 1.34 TB/s from 24 GB of GDDR7 on a 384-bit bus. The Arc A380E provides 186.0 GB/s from 6 GB of GDDR6 on a 96-bit bus.

Q: What is the power consumption difference?

A: The RTX 5090 SE has a 500 W TDP and requires a 900 W suggested power supply. The Arc A380E has a 75 W TDP and a 250 W suggested power supply.

Q: Which GPU supports ray tracing hardware?

A: Both support ray tracing, but the RTX 5090 SE has 110 RT cores versus 8 RT cores on the Arc A380E.

Q: Are these GPUs from the same architecture family?

A: No. The Arc A380E uses Intel's Xe-HPG architecture on a 6 nm TSMC process, while the RTX 5090 SE uses NVIDIA's Blackwell 2.0 architecture on a 5 nm TSMC process.

Q: What is the release status of each card?

A: The Arc A380E is end-of-life and was released in 2024 with a successor named Battlemage. The RTX 5090 SE is active, released in 2025, with a successor named GeForce 60.

Head-to-Head Benchmarks

The database records no direct benchmark scores for either GPU, but the specification-level comparison provides clear margins. The RTX 5090 SE's FP32 throughput of 66.94 TFLOPS is 16.3x the Arc A380E's 4.096 TFLOPS. Its texture rate of 1,045.9 GTexel/s exceeds the Arc A380E's 128.0 GTexel/s by a factor of 8.2. Pixel throughput stands at 380.3 GPixel/s versus 64.00 GPixel/s, a 5.9x advantage.

Memory bandwidth is the largest proportional gap. The RTX 5090 SE moves data at 1.34 TB/s, which is 7.2x the Arc A380E's 186.0 GB/s. The RTX 5090 SE also has a 4x memory capacity advantage (24 GB vs 6 GB) and uses faster GDDR7 memory at 28 Gbps effective, versus GDDR6 at 15.5 Gbps effective.

In terms of transistor count, the RTX 5090 SE packs 92,200 million transistors on a 750 mm² die, while the Arc A380E has 7,200 million on a 157 mm² die. The RTX 5090 SE's transistor density is 122.9M per mm², compared to 45.9M per mm² for the Arc A380E. The RTX 5090 SE's boost clock of 2377 MHz is higher than the Arc A380E's fixed 2000 MHz. The RTX 5090 SE also has more than double the TMUs (440 vs 64) and 5x the ROPs (160 vs 32). The RTX 5090 SE has a wider bus interface (PCIe 5.0 x16 vs PCIe 4.0 x8). In every measurable performance category, the RTX 5090 SE dominates by a wide margin.

Specification Differences

The two GPUs differ in nearly every field. The Arc A380E uses the DG2-128 chip with Xe-HPG architecture, fabricated on a 6 nm TSMC process with 7,200 million transistors and a 157 mm² die. The RTX 5090 SE uses the GB202 chip with Blackwell 2.0 architecture, fabricated on a 5 nm TSMC process with 92,200 million transistors and a 750 mm² die.

Memory: 6 GB GDDR6 on a 96-bit bus at 186.0 GB/s versus 24 GB GDDR7 on a 384-bit bus at 1.34 TB/s. Memory clock: 1937 MHz (15.5 Gbps effective) versus 1750 MHz (28 Gbps effective). Shading units: 1,024 versus 14,080. TMUs: 64 versus 440. ROPs: 32 versus 160. RT cores: 8 versus 110. Tensor cores: none versus 440. Pixel rate: 64.00 GPixel/s versus 380.3 GPixel/s. Texture rate: 128.0 GTexel/s versus 1,045.9 GTexel/s. FP32: 4.096 TFLOPS versus 66.94 TFLOPS. FP16: 8.192 TFLOPS (2:1) versus 66.94 TFLOPS (1:1). TDP: 75 W versus 500 W. Slot width: single-slot versus dual-slot. Power connectors: none versus 1x 16-pin. Suggested PSU: 250 W versus 900 W. Bus interface: PCIe 4.0 x8 versus PCIe 5.0 x16. Display outputs: 4x DisplayPort 2.0 versus 1x HDMI 2.1b and 3x DisplayPort 2.1b. Dimensions: 254 mm length, 127 mm height, 20 mm width versus 267 mm length, 111 mm height, 40 mm width. Production status: end-of-life versus active. Release date: 2024 versus 2025. Predecessor: Xe Graphics versus GeForce 40. Successor: Battlemage versus GeForce 60.

Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so API compatibility is identical. The RTX 5090 SE has a launch MSRP of 1,499 USD.

Architecture Differences

The architecture gap is fundamental. Intel's Xe-HPG architecture in the Arc A380E is designed for efficiency and modest performance, with a 2:1 FP16 ratio that halves throughput for FP16 work. NVIDIA's Blackwell 2.0 architecture in the RTX 5090 SE delivers 1:1 FP16 performance, matching its FP32 rate at 66.94 TFLOPS, which indicates full-rate tensor and compute execution for AI workloads.

The RTX 5090 SE includes 440 tensor cores, a feature entirely absent from the Arc A380E, enabling dedicated matrix math for deep learning and neural shading. Its 110 RT cores provide substantially more ray tracing hardware than the Arc A380E's 8 RT cores, which is critical for path-traced rendering and modern game engines that rely on hardware-accelerated ray tracing.

The RTX 5090 SE uses GDDR7 memory, a newer standard than GDDR6, and runs it at a much higher effective speed of 28 Gbps. Its 384-bit bus width is four times the Arc A380E's 96-bit bus, and its bandwidth of 1.34 TB/s is in a different class entirely. The Arc A380E's memory clock of 1937 MHz is higher than the RTX 5090 SE's 1750 MHz base memory clock, but the effective data rate and bus width completely overwhelm that advantage.

The RTX 5090 SE's transistor density of 122.9M per mm² reflects a more advanced 5 nm process and a much larger die. The Arc A380E's 45.9M per mm² density on 6 nm is typical for a small, power-limited part. The RTX 5090 SE also supports PCIe 5.0 x16, doubling the bandwidth of the Arc A380E's PCIe 4.0 x8 interface, which matters for data transfer in compute or multi-GPU scenarios.

The power architecture differs as well. The Arc A380E operates entirely from the PCIe slot with no auxiliary connectors and a 75 W TDP, while the RTX 5090 SE requires a dedicated 16-pin connector and 500 W TDP. The suggested power supply figures, 250 W versus 900 W, underscore the scale of the RTX 5090 SE's power delivery requirements. The physical design follows the same pattern: the Arc A380E is a single-slot card that is 20 mm thick, while the RTX 5090 SE is dual-slot at 40 mm thick, though the RTX 5090 SE is shorter (267 mm vs 254 mm) and lower in height (111 mm vs 127 mm). The Arc A380E has no tensor cores, no dedicated AI acceleration, and a narrower memory subsystem, making it suitable only for workloads that fit within its 4.096 TFLOPS and 186.0 GB/s envelope. The RTX 5090 SE, with 66.94 TFLOPS, 1.34 TB/s, tensor cores, and 110 RT cores, is designed for the highest-end compute and rendering tasks.