NVIDIA GeForce GTX 1070 vs NVIDIA GeForce GTX 675MX Comparison

Head-to-Head Benchmarks

The benchmark data is unambiguous: the NVIDIA GeForce GTX 1070 dominates the GTX 675MX in every recorded head-to-head comparison. In the two shared tests, the GTX 1070 delivers results that are not merely incremental improvements, but generational leaps.

In Geekbench Metal, the GTX 1070 scores 18,801 points against the GTX 675MX's 6,131. That represents a 206.7% advantage, meaning the GTX 1070 more than triples the older card's output. This is a decisive margin that places the two GPUs in entirely different performance tiers.

The gap widens further in Geekbench OpenCL. Here, the GTX 1070 posts 44,700 points, while the GTX 675MX manages 10,723. The delta reaches 316.9%, a fourfold-plus improvement. For compute-oriented workloads that scale well with parallel throughput, the GTX 1070 is in a different league altogether.

The aggregate benchmark scores reinforce this picture. The GTX 1070 carries an average benchmark score of 9,780 across all recorded tests, while the GTX 675MX averages 8,427. Notably, the GTX 1070's average is pulled down by several legacy DirectX tests where it scores unusually low (for example, 48 in Passmark DirectX 12 and 82 in Passmark DirectX 10), yet it still holds a 16.1% overall lead.

Looking at the rival context in the database, the GTX 1070 sits at the 47th percentile of all GPUs, with its closest competitors being the AMD FirePro W5000 (average score 9,803, just 0.2% ahead), the NVIDIA Quadro M2000M (9,832, 0.5% ahead), the NVIDIA Tesla M10 (9,724, 0.6% behind), and the NVIDIA Tesla C2070 (9,716, 0.7% behind). This places the GTX 1070 in a tightly contested mid-pack region where small percentage swings separate cards.

The GTX 675MX, by contrast, holds the 43rd percentile, with its nearest rivals being the AMD Radeon 880M (8,436, 0.1% ahead), the NVIDIA GeForce MX330 (8,458, 0.4% ahead), the AMD Radeon HD 8870M (8,462, 0.4% ahead), and the AMD Radeon R9 M375X (8,325, 1.2% behind). While the percentile gap (47 vs. 43) seems modest, the absolute score difference of 1,353 points is substantial.

The head-to-head record stands at 2 wins for the GTX 1070 and 0 for the GTX 675MX. There are no benchmark categories where the older Kepler-based card manages to outperform its Pascal successor.

Architecture Differences

The two GPUs represent fundamentally different eras of NVIDIA design. The GTX 1070 uses the GP104 chip built on the Pascal architecture, fabricated on a 16 nm process at TSMC. It packs 7,200 million transistors into a 314 mm² die, yielding a transistor density of 22.9 million per square millimeter.

The GTX 675MX, in contrast, employs the GK104 chip based on the older Kepler architecture, also from TSMC but on a much larger 28 nm node. It contains 3,540 million transistors across a 294 mm² die, resulting in a density of just 12.0 million per square millimeter. The Pascal chip crams more than twice the transistors into a slightly larger physical area.

The compute resources differ dramatically. The GTX 1070 fields 1,920 shading units, 120 texture mapping units, and 64 ROPs. The GTX 675MX has 960 shading units, 80 TMUs, and 32 ROPs. That is exactly half the shaders, two-thirds the texture units, and half the ROPs. These are not minor tweaks; the GTX 1070 doubles the parallel execution hardware in key areas.

The clock behavior also separates them. The GTX 1070 has a base clock of 1506 MHz and a boost clock of 1683 MHz. The GTX 675MX has no recorded base or boost clocks in the database, only its memory clock of 900 MHz (3.6 Gbps effective). The GTX 1070's memory runs at 2002 MHz (8 Gbps effective), more than double the effective rate.

Memory capacity and bandwidth follow the same pattern. The GTX 1070 offers 8 GB of GDDR5 on a 256-bit bus, delivering 256.3 GB/s of bandwidth. The GTX 675MX has 2 GB of GDDR5 also on a 256-bit bus, but bandwidth drops to 115.2 GB/s. The bus width is identical, but the faster memory on the GTX 1070 nearly doubles the throughput.

Theoretical pixel and texture rates tell the same story. The GTX 1070 achieves 107.7 GPixel/s and 202.0 GTexel/s. The GTX 675MX manages 13.08 GPixel/s and 52.32 GTexel/s. The pixel rate gap is over 8x, while the texture rate gap is nearly 4x.

Floating-point performance is equally lopsided. The GTX 1070 delivers 6.463 TFLOPS of FP32 compute. The GTX 675MX produces 1,255.7 GFLOPS (approximately 1.26 TFLOPS). The GTX 1070 also has a recorded FP16 figure of 101.0 GFLOPS at a 1:64 ratio, while the GTX 675MX has no FP16 data recorded.

Power and physical specifications diverge as well. The GTX 1070 has a TDP of 150 W, uses a dual-slot cooler, requires a single 8-pin power connector, and suggests a 450 W power supply. It connects via PCIe 3.0 x16. The GTX 675MX has a TDP of 100 W, comes as an MXM module, requires no external power connectors, and uses an MXM-B (3.0) interface. It is clearly designed for laptop integration rather than desktop builds.

API support shows generational progress. The GTX 1070 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The GTX 675MX supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. Both support the same OpenGL version, but the GTX 1070 has a higher DirectX feature level and a newer Vulkan revision.

The Verdict

The data supports only one conclusion: the GTX 1070 is the superior GPU for virtually any workload measured in the database. It wins both head-to-head tests, holds a higher average benchmark score (9,780 vs. 8,427), and offers multiple times the raw compute, memory bandwidth, and shading resources.

The GTX 675MX is not without merit, but its merits lie in different areas. It consumes 50 W less power (100 W vs. 150 W), requires no power connectors, and comes in an MXM form factor suited for portable devices. For a laptop user who cannot upgrade and needs a functional GPU with modest power draw, it remains a serviceable part. But the performance gap is too large to recommend it on any performance basis.

The GTX 1070's 206.7% lead in Metal and 316.9% lead in OpenCL are not marginal differences. They represent the difference between playable modern workloads and legacy-limited ones. The GTX 675MX's nearest rivals in the database (AMD Radeon 880M, GeForce MX330) all score within 0.4% of it, placing it in a crowded field of low-to-mid tier mobile GPUs. The GTX 1070, by contrast, trades blows with workstation and datacenter parts like the Quadro M2000M and Tesla M10.

For anyone choosing between these two based on the recorded measurements, the GTX 1070 is the clear pick. The GTX 675MX only makes sense if the physical constraints of an MXM module or the 100 W power envelope are non-negotiable requirements.

FAQ

Q: How much faster is the GTX 1070 in Geekbench Metal?

A: The GTX 1070 scores 18,801 versus the GTX 675MX's 6,131, a 206.7% advantage.

Q: What is the average benchmark score difference between the two cards?

A: The GTX 1070 averages 9,780 across all recorded tests, while the GTX 675MX averages 8,427, a difference of 1,353 points.

Q: Which card has more shading units?

A: The GTX 1070 has 1,920 shading units, double the GTX 675MX's 960.

Q: Do both cards use the same memory bus width?

A: Yes, both use a 256-bit bus, but the GTX 1070 has 8 GB of GDDR5 versus 2 GB, and bandwidth of 256.3 GB/s versus 115.2 GB/s.

Q: What is the TDP of each card?

A: The GTX 1070 has a TDP of 150 W, while the GTX 675MX has a TDP of 100 W.

Q: Which card supports a newer Vulkan version?

A: The GTX 1070 supports Vulkan 1.4, while the GTX 675MX supports Vulkan 1.2.175.

Where Each One Wins

The GTX 1070 wins every benchmark category where both cards have recorded data. In Geekbench Metal, it leads by 206.7%. In Geekbench OpenCL, it leads by 316.9%. Its average score across all tests is 16.1% higher than the GTX 675MX.

The GTX 1070's 8 GB memory capacity and 256.3 GB/s bandwidth make it suitable for texture-heavy workloads and higher-resolution assets. Its 6.463 TFLOPS of FP32 compute and 202.0 GTexel/s texture rate indicate strong performance in fill-rate-bound and compute-bound scenarios. The 107.7 GPixel/s pixel rate suggests solid rasterization throughput.

The GTX 675MX has no recorded wins in the database. Its advantages are structural rather than performance-based. At 100 W TDP with no external power connectors, it fits into portable devices where the 150 W, 8-pin-requiring GTX 1070 cannot operate. Its MXM-B (3.0) form factor and portable-device-dependent display outputs make it a laptop-specific component.

The GTX 1070's PCIe 3.0 x16 interface and desktop-oriented display outputs (1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.4a) suit it for desktop workstations and gaming rigs. The GTX 675MX's smaller transistor count (3,540 million vs. 7,200 million) and lower density (12.0M / mm² vs. 22.9M / mm²) reflect its older process and more limited thermal envelope.

Specification Differences

• Architecture: Pascal (GTX 1070) vs. Kepler (GTX 675MX)
• Process Node: 16 nm vs. 28 nm
• Transistors: 7,200 million vs. 3,540 million
• Die Size: 314 mm² vs. 294 mm²
• Transistor Density: 22.9M / mm² vs. 12.0M / mm²
• Base Clock: 1506 MHz vs. none recorded
• Boost Clock: 1683 MHz vs. none recorded
• Memory Clock: 2002 MHz (8 Gbps effective) vs. 900 MHz (3.6 Gbps effective)
• Memory Size: 8 GB vs. 2 GB
• Memory Bandwidth: 256.3 GB/s vs. 115.2 GB/s
• Shading Units: 1920 vs. 960
• TMUs: 120 vs. 80
• ROPs: 64 vs. 32
• Pixel Rate: 107.7 GPixel/s vs. 13.08 GPixel/s
• Texture Rate: 202.0 GTexel/s vs. 52.32 GTexel/s
• FP32: 6.463 TFLOPS vs. 1,255.7 GFLOPS
• FP16: 101.0 GFLOPS (1:64) vs. none recorded
• TDP: 150 W vs. 100 W
• Slot Width: Dual-slot vs. MXM Module
• Power Connectors: 1x 8-pin vs. None
• Suggested PSU: 450 W vs. none recorded
• Bus Interface: PCIe 3.0 x16 vs. MXM-B (3.0)
• Display Outputs: 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.4a vs. Portable Device Dependent
• DirectX: 12 (12_1) vs. 12 (11_0)
• Vulkan: 1.4 vs. 1.2.175
• Release Date: 2016-06-09 vs. 2012-09-30
• Launch MSRP: 379 USD vs. none recorded