NVIDIA H100 CNX vs Lisuan Tech LX ULTRA Comparison

Head-to-Head Benchmarks

The recorded data shows no direct benchmark comparisons between the NVIDIA H100 CNX and the Lisuan Tech LX ULTRA. Both entries carry an average benchmark score of zero, and the head-to-head benchmark list is empty. The percentile versus all GPUs is identical at 50 for both parts, indicating that neither has measurable performance data recorded in the database relative to other GPUs. Without benchmark results, the only meaningful comparisons come from the raw specification sheets, which differ substantially across compute, memory, and interface characteristics.

The NVIDIA H100 CNX delivers FP32 performance of 53.84 TFLOPS, which is more than double the Lisuan Tech LX ULTRA's 24.58 TFLOPS. In FP16 workloads, the H100 CNX reaches 215.4 TFLOPS using a 4:1 ratio, while the LX ULTRA achieves 49.15 TFLOPS with a 2:1 ratio. The H100 CNX therefore leads by 29.26 TFLOPS in FP32 and 166.25 TFLOPS in FP16, representing a 119% advantage in FP32 and a 338% advantage in FP16. These are decisive margins for compute-heavy tasks.

The LX ULTRA counters in pixel throughput. Its pixel rate of 192.0 GPixel/s far exceeds the H100 CNX's 44.28 GPixel/s, a 147.72 GPixel/s gap. Texture rate tells a different story: the H100 CNX delivers 841.3 GTexel/s versus the LX ULTRA's 384.0 GTexel/s, a 457.3 GTexel/s lead for the NVIDIA part. These figures indicate divergent design priorities, with the H100 CNX optimized for texture-heavy workloads and the LX ULTRA targeting pixel-intensive rendering.

Memory bandwidth favors the H100 CNX overwhelmingly. The NVIDIA card offers 2.04 TB/s from HBM2e across a 5120-bit bus, while the LX ULTRA provides 432.0 GB/s from GDDR6 across a 192-bit bus. That is a 1.608 TB/s difference, a 4.7x advantage for the H100 CNX. However, the LX ULTRA's memory clock runs at 2250 MHz with 18 Gbps effective, compared to the H100 CNX's 1593 MHz with 3.2 Gbps effective. The HBM2e implementation compensates with far wider bus width, which is the dominant factor in bandwidth calculations.

One notable win for the LX ULTRA is in shading units per memory capacity. The LX ULTRA has 6144 shading units with 24 GB of memory, yielding 256 shading units per GB. The H100 CNX has 14592 shading units with 80 GB, yielding 182.4 shading units per GB. The Lisuan part packs its compute density more tightly relative to memory. The LX ULTRA also leads in ROPs, with 96 versus the H100 CNX's 24, a 4x difference that explains its pixel rate advantage.

FAQ

Q: Which GPU has higher FP32 compute performance?

A: The NVIDIA H100 CNX delivers 53.84 TFLOPS, more than double the Lisuan Tech LX ULTRA's 24.58 TFLOPS, a 29.26 TFLOPS gap.

Q: How does memory bandwidth compare between the two cards?

A: The H100 CNX offers 2.04 TB/s from HBM2e on a 5120-bit bus, versus the LX ULTRA's 432.0 GB/s from GDDR6 on a 192-bit bus. The H100 CNX holds a 1.608 TB/s lead.

Q: Which card has faster pixel fill rate?

A: The Lisuan Tech LX ULTRA achieves 192.0 GPixel/s, while the NVIDIA H100 CNX manages 44.28 GPixel/s. The LX ULTRA leads by 147.72 GPixel/s.

Q: What process nodes and foundries are used?

A: Both use TSMC as the foundry. The H100 CNX uses a 5 nm process, while the LX ULTRA uses a 6 nm process.

Q: What are the power requirements?

A: The H100 CNX has a TDP of 350 W with a suggested PSU of 750 W, using an 8-pin EPS connector. The LX ULTRA has a TDP of 225 W with a suggested PSU of 550 W, using a 1x 16-pin connector.

Q: Do both cards support display outputs?

A: No. The NVIDIA H100 CNX has no display outputs, while the Lisuan Tech LX ULTRA provides 4x DisplayPort 1.4a outputs.

Architecture Differences

The NVIDIA H100 CNX is built on the Hopper architecture using the GH100 chip, belonging to the Server Hopper (Hxx) generation. The Lisuan Tech LX ULTRA uses the TrueGPU architecture with the 7G105 chip, part of the 7G100 generation. These are fundamentally different design lineages: Hopper targets data center compute acceleration, while TrueGPU appears oriented toward general graphics with a full API stack.

Process technology differs by one node step. The H100 CNX uses a 5 nm process at TSMC, while the LX ULTRA uses a 6 nm process at TSMC. The H100 CNX integrates 80,000 million transistors on an 814 mm² die, achieving a transistor density of 98.3M per mm². The LX ULTRA's transistor count and die size are unknown, so density cannot be calculated. The H100 CNX's die is substantially larger, consistent with its higher shading unit count and tensor core presence.

The H100 CNX includes 456 tensor cores, a feature absent from the LX ULTRA's specification sheet. The LX ULTRA has no listed tensor cores or RT cores, while the H100 CNX also lacks RT cores. The H100 CNX's tensor cores enable its FP16 4:1 ratio, which doubles the throughput relative to a 2:1 ratio when compared to the LX ULTRA's FP16 implementation. This architectural difference explains the massive FP16 performance gap.

The H100 CNX uses HBM2e memory, which requires a 5120-bit bus to reach 2.04 TB/s bandwidth. The LX ULTRA uses GDDR6 with a 192-bit bus and 432.0 GB/s bandwidth. Memory type dictates the physical design: HBM stacks sit close to the processor, while GDDR6 modules spread across the board. The H100 CNX's 80 GB capacity versus the LX ULTRA's 24 GB further separates these products.

API support differs completely. The LX ULTRA supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.3. The H100 CNX has no listed API support for DirectX, OpenGL, or Vulkan, reinforcing its compute-only positioning. The LX ULTRA also has display outputs, 4x DisplayPort 1.4a, while the H100 CNX has none, confirming the NVIDIA part is designed for headless server deployment.

Specification Differences

The two cards differ across nearly every major specification field. Transistor count stands at 80,000 million for the H100 CNX, while the LX ULTRA's count is unknown. Die size is 814 mm² for the H100 CNX, unknown for the LX ULTRA. Process node is 5 nm for NVIDIA versus 6 nm for Lisuan, both at TSMC.

Clock speeds show a split. The H100 CNX has a base clock of 690 MHz and a boost clock of 1845 MHz. The LX ULTRA has no base or boost clock listed. Memory clocks differ: the H100 CNX runs at 1593 MHz with 3.2 Gbps effective, while the LX ULTRA runs at 2250 MHz with 18 Gbps effective.

Memory capacity, type, bus width, and bandwidth all differ. The H100 CNX has 80 GB HBM2e on a 5120-bit bus with 2.04 TB/s bandwidth. The LX ULTRA has 24 GB GDDR6 on a 192-bit bus with 432.0 GB/s bandwidth.

Compute units differ substantially. The H100 CNX has 14592 shading units, 456 TMUs, 24 ROPs, and 456 tensor cores. The LX ULTRA has 6144 shading units, 192 TMUs, 96 ROPs, and no tensor cores. Pixel rate is 44.28 GPixel/s for NVIDIA versus 192.0 GPixel/s for Lisuan. Texture rate is 841.3 GTexel/s versus 384.0 GTexel/s.

Power and connectivity also diverge. The H100 CNX has a 350 W TDP, 8-pin EPS connector, and suggests a 750 W PSU. The LX ULTRA has a 225 W TDP, 1x 16-pin connector, and suggests a 550 W PSU. Bus interface is PCIe 5.0 x16 for the H100 CNX versus PCIe 4.0 x16 for the LX ULTRA.

Physical dimensions are close: the H100 CNX measures 267 mm in length and 111 mm in height, while the LX ULTRA measures 268 mm in length, 112 mm in height, and 40 mm in width. Both are dual-slot cards. Release dates differ: the H100 CNX launched on 2023-03-20, the LX ULTRA on 2026-03-16. The H100 CNX lists a predecessor (Server Ada) and successor (Server Blackwell), while the LX ULTRA has neither.

The Verdict

The data indicates a clear split in intended workloads. The NVIDIA H100 CNX dominates compute-heavy applications: its FP32 throughput of 53.84 TFLOPS is 119% higher than the LX ULTRA's 24.58 TFLOPS, and its FP16 output of 215.4 TFLOPS is 338% higher. Memory bandwidth of 2.04 TB/s versus 432.0 GB/s makes the H100 CNX the obvious choice for large data transfers and memory-bound algorithms. The 80 GB capacity versus 24 GB allows far larger datasets to reside on-card.

The Lisuan Tech LX ULTRA wins in graphics-oriented metrics. Its pixel rate of 192.0 GPixel/s is 4.3x the H100 CNX's 44.28 GPixel/s. It offers display outputs, API support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.3, and a lower 225 W TDP. The H100 CNX has no display outputs and no listed graphics APIs, making it unsuitable for any visual output scenario.

The H100 CNX also leads in texture rate, 841.3 GTexel/s versus 384.0 GTexel/s, and in shading units, 14592 versus 6144. The LX ULTRA has more ROPs, 96 versus 24, which drives its pixel performance. The LX ULTRA's PCIe 4.0 interface is a generation behind the H100 CNX's PCIe 5.0, which could bottleneck data transfer in PCIe-bound workloads.

The release timeline shows the LX ULTRA is a newer product by nearly three years, launching on 2026-03-16 versus the H100 CNX's 2023-03-20. Despite the newer date, the LX ULTRA's specifications do not surpass the H100 CNX in compute or memory bandwidth. The H100 CNX's 5 nm process and 80,000 million transistors provide a structural advantage that the 6 nm LX ULTRA cannot overcome in raw throughput.

Users requiring maximum FP32 or FP16 compute, large memory capacity, or high memory bandwidth should select the NVIDIA H100 CNX. Users needing display output, graphics API support, high pixel fill rate, or lower power consumption should select the Lisuan Tech LX ULTRA. The H100 CNX's 350 W TDP versus the LX ULTRA's 225 W TDP reflects its higher performance ceiling, but also demands a 750 W PSU versus 550 W.

Where Each One Wins

NVIDIA H100 CNX wins in: FP32 compute (53.84 TFLOPS vs 24.58 TFLOPS), FP16 compute (215.4 TFLOPS vs 49.15 TFLOPS), texture rate (841.3 GTexel/s vs 384.0 GTexel/s), memory bandwidth (2.04 TB/s vs 432.0 GB/s), memory capacity (80 GB vs 24 GB), shading units (14592 vs 6144), TMUs (456 vs 192), tensor cores (456 vs none), transistor count (80,000 million vs unknown), die size (814 mm² vs unknown), process node (5 nm vs 6 nm), bus interface (PCIe 5.0 x16 vs PCIe 4.0 x16), and boost clock (1845 MHz vs no listed boost clock).

Lisuan Tech LX ULTRA wins in: pixel rate (192.0 GPixel/s vs 44.28 GPixel/s), ROPs (96 vs 24), memory clock speed (2250 MHz vs 1593 MHz), effective memory speed (18 Gbps vs 3.2 Gbps), TDP efficiency (225 W vs 350 W), suggested PSU (550 W vs 750 W), display outputs (4x DisplayPort 1.4a vs none), API support (DirectX 12 Ultimate, OpenGL 4.6, Vulkan 1.3 vs none listed), and transistor density per shading unit (6144 shading units on a smaller memory footprint).

The H100 CNX is the compute specialist, excelling in every metric tied to data processing throughput. The LX ULTRA is the graphics and display specialist, winning in pixel output and API compatibility while consuming less power. Both cards hold a percentile of 50, and neither has recorded benchmark scores, so real-world performance remains unmeasured in the database. The specification sheets alone, however, draw a clear line: the H100 CNX for server compute, the LX ULTRA for workstation graphics.