AMD Instinct MI325X vs Intel Arc B390 Comparison

Where Each One Wins

The recorded data places these two processors at opposite ends of the compute spectrum. The AMD Instinct MI325X is a data center accelerator with no benchmark entries in the database, while the Intel Arc B390 is an integrated graphics processor with a single recorded 3DMark Steel Nomad DX12 score of 1482. The MI325X therefore has zero recorded benchmark wins, and the Arc B390 has zero recorded benchmark wins in a head-to-head context, as the head-to-head benchmark table is empty. What the data does show is where each part is designed to operate.

The AMD Instinct MI325X wins in raw compute capacity. It delivers 81.72 TFLOPS of FP32 performance and the same 81.72 TFLOPS of FP16 performance on a 1:1 ratio. Its texture rate is 2,553.6 GTexel/s, a figure that dwarfs the 120.0 GTexel/s of the Intel part. The MI325X carries 19,456 shading units and 1,216 texture mapping units. The Intel Arc B390 counters with 1,536 shading units, 48 TMUs, and 24 ROPs. Every compute metric in the database favors the AMD accelerator by a wide margin. The MI325X also uses 256 GB of HBM3e memory on an 8,192-bit bus, delivering 6.14 TB/s of bandwidth. The Arc B390 uses system shared memory with system dependent bandwidth. There is no contest in memory throughput.

The Intel Arc B390 wins in integration and graphics feature support. It is an IGP with a 300 MHz base clock and a 2,500 MHz boost clock. It supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The MI325X lists no API support at all: DirectX is N/A, OpenGL is N/A, and Vulkan is N/A. The Arc B390 also has 12 ray tracing cores and a pixel rate of 60.00 GPixel/s. The MI325X has zero ROPs and a pixel rate of 0 MPixel/s, which confirms it is not designed for rasterized display output. The Arc B390 has display outputs described as portable device dependent, while the MI325X has no outputs.

The benchmark percentile data reinforces the split. The Arc B390 sits at the 9th percentile among all GPUs in the database, with an average benchmark score of 1482. Its nearest recorded rivals are all older NVIDIA parts: the GeForce GT 520MX (1463, 1.3% faster), the GeForce 800M (1460, 1.5% faster), the GeForce GT 625 OEM (1446, 2.5% faster), and the GeForce GT 710 (1443, 2.7% faster). The MI325X sits at the 50th percentile with an average benchmark score of 0, because no benchmark entries exist for it. That percentile value does not reflect compute potential; it reflects the absence of standardized gaming or graphics benchmarks for a server accelerator.

The use-case split is clear. The MI325X wins in server-side FP32 and FP16 throughput, memory capacity, and memory bandwidth. The Arc B390 wins in client-side graphics API compatibility, ray tracing hardware, and pixel output. Neither part is a substitute for the other.

The Verdict

The data indicates that the AMD Instinct MI325X should be selected for compute workloads that demand massive memory and FP32 or FP16 throughput. It provides 256 GB of HBM3e, 6.14 TB/s of bandwidth, and 81.72 TFLOPS in both FP32 and FP16. It has no display outputs and no graphics API support, so it is not a graphics card in the conventional sense. Its 1000 W TDP and suggested PSU of 1400 W place it in a server chassis with OAM module slot width. The Intel Arc B390 should be selected for portable devices that need integrated graphics with modern API coverage. It supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, and it includes 12 ray tracing cores. Its 80 W TDP fits an integrated processor package.

The benchmark data does not allow a direct performance comparison. The Arc B390 has one recorded 3DMark Steel Nomad DX12 score of 1482, and the MI325X has no recorded scores. The empty head-to-head table means no delta percentage exists between the two. The verdict must therefore rest on architectural roles. The MI325X occupies the 50th percentile slot in the database, which reflects its zero-score record, not its theoretical capability. The Arc B390 occupies the 9th percentile, which places it near the GeForce GT 710 and GT 625 OEM in recorded performance.

For a buyer choosing between these two, the deciding factor is the workload. The MI325X is the only choice for large memory footprints and multi-TB/s bandwidth. The Arc B390 is the only choice for a graphics output path with ray tracing and modern API support. There is no overlap in their intended operating environments.

Head-to-Head Benchmarks

The head-to-head benchmark array is empty. There are no recorded tests where both the AMD Instinct MI325X and the Intel Arc B390 ran the same workload. The only benchmark score in the entire fact set belongs to the Arc B390: 1482 in 3DMark Steel Nomad DX12. The MI325X has an average benchmark score of 0 with no individual test entries.

Without shared benchmarks, the largest wins must be read from specification deltas. In FP32 throughput, the MI325X delivers 81.72 TFLOPS against 7.680 TFLOPS for the Arc B390. That is a ratio of roughly 10.6 to 1. In FP16, the MI325X delivers 81.72 TFLOPS at a 1:1 ratio, while the Arc B390 delivers 15.36 TFLOPS at a 2:1 ratio, a 5.3 to 1 gap. Texture rate favors the MI325X at 2,553.6 GTexel/s versus 120.0 GTexel/s, a 21.3 to 1 difference. Pixel rate is the reverse: the Arc B390 produces 60.00 GPixel/s, while the MI325X produces 0 MPixel/s.

Memory bandwidth is also one-sided. The MI325X has 6.14 TB/s from HBM3e on an 8,192-bit bus. The Arc B390 uses system shared memory with system dependent bandwidth, so no fixed number exists in the database. The MI325X has 256 GB of dedicated memory; the Arc B390 has no dedicated memory size listed.

The closest recorded competition for the Arc B390 comes from older NVIDIA parts. The GeForce GT 520MX scores 1463, which is 1.3% lower than the Arc B390's 1482. The GeForce 800M scores 1460, 1.5% lower. The GeForce GT 625 OEM scores 1446, 2.5% lower. The GeForce GT 710 scores 1443, 2.7% lower. These deltas show that the Arc B390 leads that small cluster by single-digit percentages. The MI325X has no nearest rivals listed, which means the database has no comparable entries for it.

FAQ

Q: Does the AMD Instinct MI325X have any recorded benchmark scores?

A: No. The database lists an average benchmark score of 0 for the MI325X and an empty benchmarks array. Its percentile of 50 reflects that absence of recorded scores.

Q: What is the Intel Arc B390's only recorded benchmark result?

A: The Arc B390 scores 1482 in 3DMark Steel Nomad DX12. That is its only benchmark entry, and it places the part at the 9th percentile among all GPUs.

Q: How much memory does each processor use?

A: The MI325X has 256 GB of HBM3e memory on an 8,192-bit bus with 6.14 TB/s bandwidth. The Arc B390 uses system shared memory with system dependent bandwidth.

Q: Which processor supports ray tracing?

A: The Intel Arc B390 has 12 ray tracing cores. The AMD Instinct MI325X lists null for ray tracing cores and has zero ROPs.

Q: What graphics APIs does each processor support?

A: The Arc B390 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The MI325X lists N/A for DirectX, OpenGL, and Vulkan.

Q: How close is the Arc B390 to its nearest rivals?

A: The Arc B390's 1482 score is 1.3% above the GeForce GT 520MX (1463), 1.5% above the GeForce 800M (1460), 2.5% above the GeForce GT 625 OEM (1446), and 2.7% above the GeForce GT 710 (1443).

Architecture Differences

The AMD Instinct MI325X uses the Aqua Vanjaram chip built on CDNA 3.0 architecture. It is fabricated by TSMC on a 5 nm process with 153,000 million transistors on a 1017 mm² die. That yields a transistor density of 150.4 million transistors per square millimeter. The Intel Arc B390 uses the Panther Lake chip built on Xe3-LPG architecture. It is fabricated by Intel on a 3 nm process, with transistor count and die size listed as unknown in the database.

The MI325X has 19,456 shading units and 1,216 TMUs but zero ROPs. It has no ray tracing cores listed and no tensor core count. Its FP16 performance matches its FP32 performance at 81.72 TFLOPS on a 1:1 ratio. The Arc B390 has 1,536 shading units, 48 TMUs, and 24 ROPs. It includes 12 ray tracing cores. Its FP16 performance is 15.36 TFLOPS, which is double its FP32 figure of 7.680 TFLOPS on a 2:1 ratio.

The MI325X has a base clock of 1000 MHz and a boost clock of 2100 MHz, with memory running at 1500 MHz or 6 Gbps effective. The Arc B390 has a base clock of 300 MHz and a boost clock of 2500 MHz, with system shared memory. The MI325X uses an OAM Module slot and has no power connectors, with a 1000 W TDP and a suggested PSU of 1400 W. The Arc B390 is an IGP with no power connectors, an 80 W TDP, and no suggested PSU listed.

The MI325X has no display outputs and no graphics API support. The Arc B390 has display outputs described as portable device dependent and supports modern graphics APIs. The MI325X connects via PCIe 5.0 x16. The Arc B390 uses an IGP bus interface. The MI325X was released on 2024-10-09, and the Arc B390 was released on 2026-01-26. The MI325X lists Radeon Instinct as its predecessor, while the Arc B390 has no predecessor listed. The Arc B390 has an active production status, while the MI325X has no production status listed.

Specification Differences

The two parts differ across nearly every recorded specification field. The MI325X uses a 5 nm TSMC process; the Arc B390 uses a 3 nm Intel process. The MI325X has 153,000 million transistors on a 1017 mm² die; the Arc B390 has unknown transistor count and die size. The MI325X has a 1000 MHz base clock and 2100 MHz boost; the Arc B390 has a 300 MHz base clock and 2500 MHz boost.

Memory separates them completely. The MI325X has 256 GB of HBM3e on an 8,192-bit bus with 6.14 TB/s bandwidth. The Arc B390 has system shared memory, system shared type, system shared bus width, and system dependent bandwidth.

Compute resources differ by an order of magnitude. The MI325X has 19,456 shading units, 1,216 TMUs, and 0 ROPs. The Arc B390 has 1,536 shading units, 48 TMUs, and 24 ROPs. The MI325X has no ray tracing cores; the Arc B390 has 12. The MI325X delivers 81.72 TFLOPS in FP32 and FP16. The Arc B390 delivers 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16.

Pixel and texture rates also diverge. The MI325X has a pixel rate of 0 MPixel/s and a texture rate of 2,553.6 GTexel/s. The Arc B390 has a pixel rate of 60.00 GPixel/s and a texture rate of 120.0 GTexel/s.

Power and packaging differ sharply. The MI325X has a 1000 W TDP with a 1400 W suggested PSU and an OAM Module slot. The Arc B390 has an 80 W TDP, no suggested PSU, and an IGP slot. Both use no power connectors. The MI325X uses PCIe 5.0 x16; the Arc B390 uses an IGP bus. The MI325X has no display outputs; the Arc B390 has portable device dependent outputs. The MI325X lists no API support; the Arc B390 supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. Release dates are 2024-10-09 for the MI325X and 2026-01-26 for the Arc B390.