AMD Radeon Vega 8 vs NVIDIA GeForce GTX 675MX Comparison

The NVIDIA GeForce GTX 675MX and AMD Radeon Vega 8 represent two very different eras of mobile graphics: a discrete Kepler-based GPU from 2012 versus a modern integrated graphics processor from 2018. Benchmark data shows the AMD part holds a higher average score of 9,221 compared to 8,427 for the NVIDIA, placing it in the 45th percentile versus the 43rd. Yet the head-to-head results are split, with each winning one of the two available tests, which makes the choice heavily dependent on the workload.

Where Each One Wins

The AMD Radeon Vega 8 is the clear winner in Geekbench Metal, scoring 10,706 against 6,131 for the GTX 675MX, a margin of 42.7%. This is a massive advantage for the integrated part in Apple’s Metal API, which is the primary graphics interface for macOS applications. The Vega 8 also supports DirectX 12 (12_1) and Vulkan 1.3, giving it access to modern API features that the older NVIDIA card lacks. Its architecture, GCN 5.0 on a 14nm process, is significantly newer and includes dedicated FP16 throughput of 2.253 TFLOPS (2:1), which can accelerate certain compute workloads.

The NVIDIA GeForce GTX 675MX wins in Geekbench OpenCL, scoring 10,723 versus 8,822 for the Vega 8, a 21.5% advantage. OpenCL is a cross-platform compute API, and here the Kepler GPU’s higher raw FP32 throughput of 1,255.7 GFLOPS (compared to 1,126.4 GFLOPS for the AMD) and its dedicated 2 GB of GDDR5 memory with 115.2 GB/s bandwidth pay off. The NVIDIA also has a much higher pixel rate of 13.08 GPixel/s and texture rate of 52.32 GTexel/s, versus 8.8 GPixel/s and 35.2 GTexel/s for the AMD, which suggests it may handle traditional rasterization workloads more efficiently despite losing in Metal.

The split is even at one win each, but the nature of the wins matters. The Vega 8 wins by a huge margin in one test while the GTX 675MX wins by a smaller margin in the other. This indicates the AMD part is better for Metal-centric tasks, while the NVIDIA part is better for OpenCL compute and possibly older DirectX titles.

Architecture Differences

The GTX 675MX is built on NVIDIA’s Kepler architecture, fabricated by TSMC on a 28nm process. It uses the GK104 chip, which contains 3,540 million transistors on a 294 mm² die, yielding a transistor density of 12.0M per mm². It has 960 shading units, 80 texture mapping units, and 32 ROPs. Its memory subsystem consists of 2 GB of GDDR5 on a 256-bit bus, providing 115.2 GB/s of bandwidth. The GPU runs at a memory clock of 900 MHz (3.6 Gbps effective) and has a TDP of 100 W. It connects via an MXM-B (3.0) interface and is an end-of-life product from the GeForce 600M generation.

The Vega 8 is an integrated GPU (IGP) from AMD’s Raven Ridge generation, built on GCN 5.0 architecture. It uses the Raven chip, fabricated by GlobalFoundries on a 14nm process, with 4,940 million transistors on a 210 mm² die, giving a much higher transistor density of 23.5M per mm². It has 512 shading units, 32 TMUs, and only 8 ROPs. Its memory is system-shared, meaning it uses the host system’s RAM, with bandwidth described as system dependent. The base clock is 300 MHz with a boost clock of 1,100 MHz. Its TDP is just 25 W, and it has no dedicated power connectors, reflecting its integrated nature. It supports DirectX 12 (12_1) and Vulkan 1.3, while the NVIDIA only reaches DirectX 12 (11_0) and Vulkan 1.2.175.

The fundamental difference is discrete versus integrated. The GTX 675MX has its own VRAM and higher fill rates, but the Vega 8 benefits from newer architecture, a smaller process node, and modern API support. The transistor counts are similar in magnitude, but the Vega 8 packs them into a smaller die with a much higher density, indicating more efficient design.

Head-to-Head Benchmarks

The Geekbench Metal test is a decisive victory for the AMD Radeon Vega 8. It scores 10,706, while the GTX 675MX manages only 6,131. The delta is 42.7% in favor of the AMD part. This is not a marginal difference; it is a commanding lead that suggests the Kepler architecture is poorly optimized for Metal, or that the Vega 8’s newer GCN 5.0 design handles this API far more efficiently. For any user running Metal-based applications, the Vega 8 is the superior choice by a wide margin.

In Geekbench OpenCL, the tables turn. The GTX 675MX scores 10,723, beating the Vega 8’s 8,822 by 21.5%. This is a significant but not overwhelming win. The NVIDIA part’s advantage likely stems from its higher FP32 throughput (1,255.7 GFLOPS vs 1,126.4 GFLOPS) and its dedicated memory with 115.2 GB/s bandwidth, which avoids the latency and bandwidth penalties of shared system memory. The Vega 8’s FP16 capability of 2.253 TFLOPS does not help here, as OpenCL benchmarks often favor FP32 performance.

Looking at the nearest rivals provides context. The Vega 8’s average score of 9,221 puts it just 0.1% ahead of the AMD Radeon 890M, 0.6% behind the NVIDIA GeForce GTX 960, 0.8% behind the GTX 465, and 0.9% behind the GTX 850M. This shows the Vega 8 sits in a tight cluster of mid-range GPUs, with no significant gap to any of them. The GTX 675MX’s average score of 8,427 places it 1.2% ahead of the AMD Radeon R9 M375X, 0.4% behind the AMD Radeon HD 8870M, 0.4% behind the NVIDIA GeForce MX330, and 0.1% behind the AMD Radeon 880M. Both cards are surrounded by rivals within a few percentage points, indicating that neither has a dominant position in the overall performance hierarchy.

The average benchmark score difference between the two is 9,221 versus 8,427, a gap of about 9.4%. This aligns with the percentile rankings of 45 versus 43. However, the head-to-head results show that averages can be misleading. The Vega 8 wins Metal by a huge margin, the GTX 675MX wins OpenCL by a solid margin, and the overall average favors the AMD part because its Metal score is much higher than the NVIDIA’s, pulling the mean up.

The Verdict

From the data, the AMD Radeon Vega 8 is the better overall GPU for modern API workloads. Its 42.7% lead in Metal is decisive, and its higher average score of 9,221 versus 8,427 confirms that it is the faster part in general compute. The Vega 8 also supports DirectX 12 (12_1) and Vulkan 1.3, while the GTX 675MX is limited to DirectX 12 (11_0) and Vulkan 1.2.175, making the AMD part more future-proof for new games and applications. Its 25 W TDP is also a major advantage for thin-and-light laptops, compared to the 100 W TDP of the NVIDIA card.

However, the NVIDIA GeForce GTX 675MX is not obsolete. It wins OpenCL by 21.5%, and its dedicated 2 GB GDDR5 memory with 115.2 GB/s bandwidth is a tangible benefit for workloads that are sensitive to memory latency and bandwidth. Its higher pixel rate of 13.08 GPixel/s and texture rate of 52.32 GTexel/s suggest it may handle traditional rasterization better than the Vega 8, despite the latter’s newer architecture. For users running OpenCL compute or older DirectX 11-era games where Kepler was well-optimized, the GTX 675MX remains a capable performer.

Pick the AMD Radeon Vega 8 if you prioritize Metal performance, modern API support, and low power consumption. Pick the NVIDIA GeForce GTX 675MX if you need the fastest OpenCL compute or rely on its dedicated VRAM for memory-intensive tasks. The data does not support a universal winner; it supports a workload-specific choice.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon Vega 8 has an average benchmark score of 9,221, which is higher than the NVIDIA GeForce GTX 675MX’s 8,427. The Vega 8 also holds a higher percentile rank of 45 versus 43.

Q: How much faster is the Vega 8 in Geekbench Metal?

A: The Vega 8 scores 10,706 in Geekbench Metal, while the GTX 675MX scores 6,131. This gives the AMD part a 42.7% advantage in that test.

Q: In which test does the GTX 675MX outperform the Vega 8?

A: The GTX 675MX wins in Geekbench OpenCL, scoring 10,723 against the Vega 8’s 8,822, a 21.5% lead. This is the only test where the NVIDIA card comes out ahead.

Q: What are the API support differences between the two?

A: The AMD Radeon Vega 8 supports DirectX 12 (12_1) and Vulkan 1.3, while the NVIDIA GeForce GTX 675MX supports DirectX 12 (11_0) and Vulkan 1.2.175. Both support OpenGL 4.6.

Q: How do the memory configurations differ?

A: The GTX 675MX has 2 GB of dedicated GDDR5 memory on a 256-bit bus with 115.2 GB/s bandwidth. The Vega 8 uses system-shared memory, with size, type, bus width, and bandwidth all described as system dependent.

Q: What are the TDP ratings for each GPU?

A: The NVIDIA GeForce GTX 675MX has a TDP of 100 W, while the AMD Radeon Vega 8 has a TDP of just 25 W. This makes the Vega 8 far more power-efficient for portable devices.