AMD Instinct MI350P vs Lisuan Tech LX ULTRA Comparison

Head-to-Head Benchmarks

The database records no head-to-head benchmark entries for the AMD Instinct MI350P and the Lisuan Tech LX ULTRA. Both cards hold a 50th percentile position among all GPUs, and neither has an average benchmark score recorded. This means direct performance comparisons must be derived from the architectural and specification data available in the database.

The MI350P delivers 36.04 TFLOPS of FP32 compute, which is 46.6% higher than the LX ULTRA's 24.58 TFLOPS. In FP16, the MI350P again lists 36.04 TFLOPS with a 1:1 ratio, while the LX ULTRA reaches 49.15 TFLOPS via a 2:1 ratio. The MI350P's texture rate of 1,126.4 GTexel/s is nearly triple the LX ULTRA's 384.0 GTexel/s, indicating a substantial advantage in tasks that rely on texture filtering. Conversely, the LX ULTRA has a pixel rate of 192.0 GPixel/s, while the MI350P's pixel rate is recorded as 0 MPixel/s, reflecting its lack of traditional raster output units.

Memory capacity and bandwidth diverge sharply. The MI350P uses 144 GB of HBM3e on an 8192-bit bus, yielding 8.19 TB/s of bandwidth. The LX ULTRA uses 24 GB of GDDR6 on a 192-bit bus, producing 432.0 GB/s. The MI350P's bandwidth advantage is roughly 19x, a decisive factor for data-intensive workloads. The LX ULTRA's memory clock runs at 2250 MHz (18 Gbps effective), while the MI350P's memory clock is 2000 MHz (8 Gbps effective), though the latter's much wider bus entirely overshadows the clock speed difference.

The MI350P's FP32 throughput is 1.47x the LX ULTRA's, but the LX ULTRA's FP16 throughput is 1.36x the MI350P's. This inverted relationship matters for mixed-precision tasks. The MI350P's shading units total 8192, compared to 6144 on the LX ULTRA, a 33% advantage. Texture mapping units stand at 512 versus 192, a 167% advantage for the MI350P. The LX ULTRA has 96 ROPs; the MI350P has none listed.

Power draw differs, with the MI350P at 600 W and the LX ULTRA at 225 W. The suggested power supply for the MI350P is 1000 W, while the LX ULTRA lists 550 W. Both use a single 16-pin power connector and a dual-slot form factor. The MI350P's dimensions are 267 mm by 111 mm by 40 mm; the LX ULTRA is 268 mm by 112 mm by 40 mm, making them nearly identical in physical footprint.

The Verdict

The data indicates two distinct usage profiles. The AMD Instinct MI350P is oriented toward memory-bound and compute-heavy workloads, given its 144 GB HBM3e pool, 8.19 TB/s bandwidth, and higher FP32 and texture rates. The Lisuan Tech LX ULTRA is positioned for graphics output and rasterization, as it is the only one of the two with display outputs (4x DisplayPort 1.4a), a pixel rate of 192.0 GPixel/s, and 96 ROPs.

Benchmark results are absent, so no direct score comparison is possible. However, the recorded specifications show the MI350P wins on FP32 compute, texture rate, memory capacity, memory bandwidth, shading units, and TMUs. The LX ULTRA wins on FP16 compute, pixel rate, ROP count, lower power draw, and API support (DirectX 12 Ultimate, OpenGL 4.6, Vulkan 1.3). The MI350P lists no API support, no display outputs, and a 0 MPixel/s pixel rate.

The MI350P is the choice for server-side inference, scientific simulation, or any workload where memory bandwidth and capacity dominate. The LX ULTRA is the choice for workstation graphics, rendering pipelines that require raster output, or any scenario needing display connectivity. Neither card has a recorded benchmark score, so performance rankings remain hypothetical based on the specification sheet alone.

FAQ

Q: Which card has higher FP32 performance?

A: The AMD Instinct MI350P records 36.04 TFLOPS FP32, while the Lisuan Tech LX ULTRA records 24.58 TFLOPS. The MI350P is 46.6% higher.

Q: Which card has higher FP16 performance?

A: The Lisuan Tech LX ULTRA records 49.15 TFLOPS FP16 (2:1 ratio), while the AMD Instinct MI350P records 36.04 TFLOPS FP16 (1:1 ratio). The LX ULTRA is 36.4% higher.

Q: How do memory capacities compare?

A: The AMD Instinct MI350P has 144 GB of HBM3e, while the Lisuan Tech LX ULTRA has 24 GB of GDDR6. The MI350P's capacity is 6x larger.

Q: What is the memory bandwidth difference?

A: The MI350P delivers 8.19 TB/s on an 8192-bit bus. The LX ULTRA delivers 432.0 GB/s on a 192-bit bus. The MI350P's bandwidth is approximately 19x higher.

Q: Which card supports display outputs?

A: Only the Lisuan Tech LX ULTRA has display outputs, specifically 4x DisplayPort 1.4a. The AMD Instinct MI350P lists no outputs.

Q: What are the power requirements?

A: The AMD Instinct MI350P has a TDP of 600 W and suggests a 1000 W power supply. The Lisuan Tech LX ULTRA has a TDP of 225 W and suggests a 550 W power supply.

Specification Differences

The two cards differ across nearly every measured field. The MI350P uses a 3 nm process node, while the LX ULTRA uses 6 nm. Both are fabricated by TSMC. The MI350P has 73,000 million transistors on a 1190 mm² die, while the LX ULTRA's transistor count and die size are unknown. Transistor density for the MI350P is 61.3M per mm²; the LX ULTRA's density is not recorded.

Clock speeds differ: the MI350P has a base clock of 1000 MHz and a boost clock of 2200 MHz. The LX ULTRA has no base or boost clock listed. Memory clocks are 2000 MHz (8 Gbps effective) for the MI350P and 2250 MHz (18 Gbps effective) for the LX ULTRA.

Memory specifications diverge completely: 144 GB HBM3e versus 24 GB GDDR6, 8192-bit versus 192-bit bus, 8.19 TB/s versus 432.0 GB/s. Shading units are 8192 versus 6144, TMUs are 512 versus 192, and ROPs are 0 versus 96. Pixel rate is 0 MPixel/s versus 192.0 GPixel/s. Texture rate is 1,126.4 GTexel/s versus 384.0 GTexel/s. FP32 is 36.04 TFLOPS versus 24.58 TFLOPS. FP16 is 36.04 TFLOPS (1:1) versus 49.15 TFLOPS (2:1).

TDP is 600 W versus 225 W. Suggested PSU is 1000 W versus 550 W. Bus interface is PCIe 5.0 x16 versus PCIe 4.0 x16. Display outputs are none versus 4x DisplayPort 1.4a. API support is N/A for the MI350P, while the LX ULTRA lists DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.3.

Release dates differ: the MI350P launched on 2026-05-06, while the LX ULTRA launched on 2026-03-16. The LX ULTRA has an active production status; the MI350P's production status is not recorded. The MI350P's predecessor is Radeon Instinct; the LX ULTRA has no predecessor. Both use a dual-slot form factor and a single 16-pin power connector. Dimensions are nearly identical: 267 mm by 111 mm by 40 mm for the MI350P, and 268 mm by 112 mm by 40 mm for the LX ULTRA.

Architecture Differences

The MI350P is built on AMD's CDNA 4.0 architecture, part of the Instinct (MIx) generation. Its chip is labeled MI350 128CU. The LX ULTRA uses the TrueGPU architecture from the 7G100 generation, with a chip labeled 7G105. These are fundamentally different design philosophies: CDNA is optimized for compute acceleration, while TrueGPU appears to target general graphics processing given its API support and display outputs.

Process nodes differ: the MI350P uses 3 nm, the LX ULTRA uses 6 nm, both from TSMC. The MI350P integrates 73,000 million transistors on a 1190 mm² die, a density of 61.3M per mm². The LX ULTRA's transistor count and die size are unknown, making density comparisons impossible.

Cache and tensor core details are not recorded for either card. The MI350P lists no RT cores and no tensor cores; the LX ULTRA also lists neither. The MI350P has no ROPs, which aligns with its lack of display outputs. The LX ULTRA has 96 ROPs, supporting its pixel rate of 192.0 GPixel/s.

The MI350P's FP16 is 1:1 with FP32, indicating equal throughput for both precisions. The LX ULTRA's FP16 is 2:1, meaning FP16 runs at twice the rate of FP32. This architectural choice suggests the LX ULTRA is designed for workloads that benefit from reduced precision, while the MI350P treats both precisions equally.

The MI350P uses HBM3e memory, a high-bandwidth stack typically found in compute accelerators. The LX ULTRA uses GDDR6, a conventional graphics memory type. The bus widths, 8192-bit versus 192-bit, reflect the different memory architectures. The MI350P has no display outputs, no API support, and no pixel rate, confirming its role as a headless compute device. The LX ULTRA has full display outputs and modern API support, confirming its role as a graphics card.

Where Each One Wins

The AMD Instinct MI350P wins in scenarios requiring massive memory bandwidth and capacity. Its 8.19 TB/s bandwidth and 144 GB HBM3e pool suit large model inference, scientific computing, and data processing where data sets exceed 24 GB. Its FP32 throughput of 36.04 TFLOPS and texture rate of 1,126.4 GTexel/s give it a clear edge in compute-heavy tasks that do not rely on raster output. The 8192 shading units and 512 TMUs reinforce this position. The PCIe 5.0 x16 interface provides double the bus bandwidth of the LX ULTRA's PCIe 4.0 x16.

The Lisuan Tech LX ULTRA wins in scenarios requiring graphics output and rasterization. Its 4x DisplayPort 1.4a outputs allow direct display connection. Its 192.0 GPixel/s pixel rate and 96 ROPs enable traditional rendering pipelines. Its FP16 throughput of 49.15 TFLOPS exceeds the MI350P's, favoring workloads that use half-precision arithmetic. Its API support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.3 makes it usable in standard graphics applications, whereas the MI350P lists no API compatibility.

The LX ULTRA also wins on power efficiency. Its 225 W TDP is less than half the MI350P's 600 W, and its suggested 550 W PSU is lower than the MI350P's 1000 W. For systems with power constraints, the LX ULTRA is the practical selection. Its 24 GB GDDR6 memory, while smaller, is sufficient for many graphics workloads and its 432.0 GB/s bandwidth, though far lower than the MI350P, is adequate for its intended tasks.

The MI350P's 50th percentile ranking matches the LX ULTRA's, and both have an average benchmark score of 0. Without recorded benchmarks, the specification sheet is the only guide. The MI350P wins on raw compute and memory resources. The LX ULTRA wins on graphics features, precision flexibility, and power draw. The choice depends on whether the workload demands memory scale or display output.