AMD Radeon 680M vs NVIDIA T400 Comparison

# Head-to-Head Benchmarks

The benchmark data reveals a clear performance hierarchy between the NVIDIA T400 and the AMD Radeon 680M, with the AMD part taking both available head-to-head comparisons. In Geekbench OpenCL, the Radeon 680M scores 23,468 against the T400’s 17,039, a margin of 27.4%. The Vulkan test tells a similar story: the Radeon 680M posts 21,965 while the T400 manages 15,976, a 27.3% deficit. These are substantial gaps, not marginal differences—the AMD integrated graphics solution outperforms the NVIDIA discrete workstation card by more than a quarter in both compute APIs.

The T400’s average benchmark score of 16,508 places it at the 60th percentile of all GPUs, while the Radeon 680M’s average of 15,270 sits at the 57th percentile. Interestingly, despite losing both head-to-head tests, the T400 holds a higher overall average score. This apparent contradiction stems from the T400’s benchmark profile including only OpenCL and Vulkan results, whereas the Radeon 680M's average incorporates a third test—3DMark Steel Nomad DX12—where it scores 378, dragging its average down. The T400’s nearest rivals cluster tightly: NVIDIA GeForce RTX 5090 D V2 at 16,504 (0.0% delta), AMD Radeon PRO W7500 at 16,415 (0.6%), and NVIDIA RTX PRO 6000 Blackwell at 16,408 (0.6%). The Radeon 680M’s rivals show similar clustering: NVIDIA GeForce GTX 580 at 15,283 (−0.1%), NVIDIA GeForce RTX 2060 at 15,290 (−0.1%), NVIDIA GeForce RTX 3050 OEM at 15,199 (0.5%), and AMD Radeon RX 7600 at 15,171 (0.7%). These delta values indicate both cards perform within a narrow band of their respective competitors, but the band itself is distinctly different—the T400 competes in a tier roughly 8% higher on average.

The Vulkan result is particularly telling for real-world gaming and compute workloads, as Vulkan is a low-overhead API that exposes raw hardware capabilities. The Radeon 680M’s 37.5% advantage in Vulkan (21,965 vs 15,976) suggests its architecture scales better with modern APIs that minimize driver overhead. The OpenCL gap of 27.4% reinforces this pattern, showing consistent compute superiority across different execution models.

# Where Each One Wins

The AMD Radeon 680M wins decisively in raw compute throughput. Its FP32 performance of 3.379 TFLOPS more than triples the T400’s 1,094.4 GFLOPS. Texture fill rate follows the same pattern: 105.6 GTexel/s versus 34.20 GTexel/s. Pixel rate shows a 70.40 GPixel/s versus 22.80 GPixel/s advantage. These are not incremental gains—the Radeon 680M offers roughly three times the arithmetic throughput, three times the texture processing, and three times the pixel output of the T400. For workloads that are shader-bound, texture-heavy, or pixel-fill-limited, the AMD part is categorically faster.

The T400’s wins are more subtle and structural. It is a discrete, single-slot card with dedicated 2 GB GDDR6 memory on a 64-bit bus delivering 80.00 GB/s bandwidth. The Radeon 680M, being an IGP, relies on system shared memory with bandwidth described as "System Dependent"—meaning its memory performance varies entirely with the host system’s RAM configuration. In scenarios where the host memory is slow or shared with CPU workloads, the T400’s dedicated VRAM provides predictable, consistent memory latency and bandwidth. The T400 also carries three mini-DisplayPort 1.4a outputs, enabling direct multi-monitor connectivity, whereas the Radeon 680M’s display outputs are "Portable Device Dependent," tying its display capabilities to the laptop or mobile chassis it resides in.

For professional workstation use cases—CAD, scientific visualization, or compute tasks that need stable driver certification—the T400’s Quadro lineage (generation "Quadro Turing (Tx000)") offers an established ecosystem. The Radeon 680M, as a mobile IGP from the "Navi II IGP (Rembrandt Mobile)" generation, is designed for integrated mobile performance rather than dedicated workstation duty. Neither card supports ray tracing in the T400’s case (no RT cores listed), while the Radeon 680M includes 12 ray accelerators, making it the only one of the pair capable of hardware-accelerated ray tracing.

# Architecture Differences

The two GPUs come from fundamentally different design philosophies. The NVIDIA T400 uses the TU117 chip built on Turing architecture at TSMC’s 12 nm process. It integrates 4,700 million transistors on a 200 mm² die, yielding a transistor density of 23.5 million per square millimeter. The AMD Radeon 680M uses the Rembrandt+ chip based on RDNA 2.0 architecture at TSMC’s 6 nm node. That die packs 13,100 million transistors into 208 mm²—only 4% larger physically but containing nearly three times the transistors. Transistor density jumps to 63.0 million per square millimeter, a 2.7× improvement that directly enables the Radeon 680M’s higher compute throughput.

The shading engine configuration differs substantially. The T400 has 384 shading units, 24 texture mapping units, and 16 render output units. The Radeon 680M doubles these counts: 768 shading units, 48 TMUs, and 32 ROPs. This 2:1 ratio in every major fixed-function block explains the roughly 3× performance advantage in raw throughput metrics. The Radeon 680M also includes 12 ray tracing cores, a feature entirely absent from the T400 (no RT cores listed). Clock speeds favor the AMD part as well: the Radeon 680M bases at 2000 MHz and boosts to 2200 MHz, while the T400 bases at 420 MHz and boosts to 1425 MHz. The T400’s low base clock suggests aggressive power management, but even at boost it trails the AMD part’s baseline frequency by 40%.

Memory architecture represents the starkest philosophical divide. The T400 uses dedicated 2 GB GDDR6 at 1250 MHz (10 Gbps effective) over a 64-bit bus, achieving 80.00 GB/s. The Radeon 680M has no dedicated VRAM—its memory size, type, and bus width are all "System Shared," with bandwidth rated as "System Dependent." This means the T400 offers fixed, guaranteed memory performance, while the Radeon 680M’s memory throughput is entirely contingent on the laptop’s RAM configuration and whether the CPU is simultaneously accessing memory.

Power and physical design differ predictably. The T400 is a 30 W single-slot card with no power connectors and a 200 W suggested PSU. The Radeon 680M is a 50 W IGP with no slot width (integrated directly onto the processor package) and no discrete power connectors. The T400 uses PCIe 3.0 x16, while the Radeon 680M uses PCIe 4.0 x8—the newer standard but half the lanes, a tradeoff suited to integrated designs. API support gives the Radeon 680M a modern edge: DirectX 12 Ultimate (12_2) versus the T400’s DirectX 12 (12_1), though both support OpenGL 4.6 and Vulkan 1.4.

# FAQ

Q: Which GPU is faster in Geekbench OpenCL?

A: The AMD Radeon 680M scores 23,468 versus the NVIDIA T400’s 17,039, a 27.4% advantage for the AMD part.

Q: Does the NVIDIA T400 have ray tracing hardware?

A: No. The T400 lists no RT cores in its specification, while the AMD Radeon 680M includes 12 ray accelerators.

Q: What is the memory bandwidth of each GPU?

A: The T400 has dedicated 80.00 GB/s bandwidth from 2 GB GDDR6 on a 64-bit bus. The Radeon 680M uses system shared memory with bandwidth rated as "System Dependent."

Q: Which card has a higher average benchmark score?

A: The NVIDIA T400 averages 16,508 across its benchmarks, placing it at the 60th percentile, while the Radeon 680M averages 15,270 at the 57th percentile. However, the Radeon 680M wins both direct head-to-head tests.

Q: What are the power requirements for each?

A: The T400 has a 30 W TDP and suggests a 200 W PSU. The Radeon 680M has a 50 W TDP with no suggested PSU listed, consistent with its integrated design.

Q: How do the transistor counts compare?

A: The Radeon 680M’s Rembrandt+ chip contains 13,100 million transistors, nearly three times the T400’s 4,700 million, despite a similar die size (208 mm² vs 200 mm²).

# Specification Differences

| Specification | NVIDIA T400 | AMD Radeon 680M |

|---|---|---|

| Chip | TU117 | Rembrandt+ |

| Architecture | Turing | RDNA 2.0 |

| Generation | Quadro Turing (Tx000) | Navi II IGP (Rembrandt Mobile) |

| Process Node | 12 nm | 6 nm |

| Transistors | 4,700 million | 13,100 million |

| Die Size | 200 mm² | 208 mm² |

| Transistor Density | 23.5M / mm² | 63.0M / mm² |

| Base Clock | 420 MHz | 2000 MHz |

| Boost Clock | 1425 MHz | 2200 MHz |

| Memory Size | 2 GB | System Shared |

| Memory Type | GDDR6 | System Shared |

| Memory Bus Width | 64 bit | System Shared |

| Memory Bandwidth | 80.00 GB/s | System Dependent |

| Shading Units | 384 | 768 |

| TMUs | 24 | 48 |

| ROPs | 16 | 32 |

| RT Cores | None | 12 |

| Pixel Rate | 22.80 GPixel/s | 70.40 GPixel/s |

| Texture Rate | 34.20 GTexel/s | 105.6 GTexel/s |

| FP32 | 1,094.4 GFLOPS | 3.379 TFLOPS |

| FP16 | 2.189 TFLOPS (2:1) | 6.758 TFLOPS (2:1) |

| TDP | 30 W | 50 W |

| Slot Width | Single-slot | IGP |

| Suggested PSU | 200 W | None |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x8 |

| Display Outputs | 3x mini-DisplayPort 1.4a | Portable Device Dependent |

| DirectX | 12 (12_1) | 12 Ultimate (12_2) |

| Production Status | End-of-life | Active |

| Release Date | 2021-05-05 | 2023-01-02 |

| Predecessor | Quadro Volta | Vega II IGP |

| Successor | Workstation Ampere | Navi III IGP |

| Percentile vs All GPUs | 60 | 57 |

| Avg Benchmark Score | 16,508 | 15,270 |