NVIDIA GeForce GTX 750 vs NVIDIA GRID K2 Comparison

Head-to-Head Benchmarks

The recorded database results place the NVIDIA GRID K2 clearly ahead of the GeForce GTX 750 in both available benchmark tests. In Geekbench Metal, the GRID K2 scores 5,557 against 4,274 for the GTX 750, a 30% advantage. That is the largest single-test gap between these two parts. The OpenCL result is closer but still decisive: 10,602 for the GRID K2 versus 9,315 for the GTX 750, a 13.8% margin. Across the two tests, the GRID K2 wins both, giving it a 2-0 head-to-head record.

The average benchmark score reinforces this ordering. The GRID K2 sits at 8,080, while the GTX 750 lands at 7,222. That 858-point gap places the two cards in different performance tiers despite their shared 42nd and 40th percentiles among all GPUs respectively. The percentile difference is small, but the raw score delta is not trivial.

Looking at the nearest rivals for context, the GRID K2's average of 8,080 puts it within 0.2% of the GeForce GTX 650 Ti Boost (8,067) and 0.3% of the GTX 650 Ti (8,053). It also trails the GeForce 945M (8,099) by 0.2% and leads the GTX 880M (8,040) by 0.5%. This places the GRID K2 in a tightly clustered group of mid-range Kepler-era parts. The GTX 750, meanwhile, sits at 7,222, which is 0.3% ahead of the AMD Radeon Vega 8 Mobile (7,203), 0.7% ahead of both the GeForce GTX 560 SE (7,171) and the Intel Iris Pro Graphics P580 (7,170), and 0.9% ahead of the GeForce GTX 970 (7,157). The GTX 750's nearest rivals are a mixed bag spanning integrated graphics, older desktop parts, and a higher-tier Maxwell card, which suggests its score is more of an outlier relative to its own product class.

The Metal test shows the GRID K2's advantage is most pronounced in compute workloads that leverage the full shader array. The 30% delta there is nearly double the OpenCL margin. That pattern indicates the GRID K2's larger execution resource pool matters more in Metal's specific dispatch pattern, whereas OpenCL exercises both cards in a way that narrows the gap. Still, the GTX 750 never overtakes the GRID K2 in any recorded test.

Architecture Differences

The two cards come from different NVIDIA architectures. The GRID K2 uses the GK104 chip on the Kepler architecture, while the GTX 750 uses the GM107 chip on Maxwell. Both are fabricated at TSMC on a 28 nm process, but the transistor counts diverge sharply. The GRID K2 packs 3,540 million transistors on a 294 mm² die, giving a transistor density of 12.0 million transistors per mm². The GTX 750 uses 1,870 million transistors on a 148 mm² die, with a slightly higher density of 12.6 million per mm². The GRID K2's die is nearly double the size, and its transistor budget is 89% larger, yet the density figures are close because the larger chip scales area roughly proportionally.

The execution resources differ by a factor of three. The GRID K2 has 1,536 shading units, 128 texture mapping units, and 32 raster output units. The GTX 750 has 512 shading units, 32 TMUs, and 16 ROPs. This means the GRID K2 can process three times as many shader operations per clock and four times as many texture fetches per clock. The ROP count is double on the GRID K2. Pixel throughput reflects this: the GRID K2 achieves 23.84 GPixel/s versus 17.36 GPixel/s for the GTX 750. Texture rate shows a larger gap: 95.36 GTexel/s versus 34.72 GTexel/s, a 2.75x advantage for the GRID K2.

Clock behavior also differs. The GTX 750 has a base clock of 1020 MHz and a boost clock of 1085 MHz. The GRID K2 has no recorded base or boost clock in the database, so its performance is driven entirely by its wider architecture rather than higher frequency. Memory clocks are effectively identical: both run GDDR5 at 1250 MHz base, 5 Gbps effective. The GRID K2's memory subsystem is wider, with a 256-bit bus versus 128-bit, and it carries 4 GB of VRAM against 1 GB on the GTX 750. Bandwidth scales accordingly: 160.0 GB/s for the GRID K2, 80.19 GB/s for the GTX 750, exactly double.

The floating-point throughput figures reflect the same story. The GRID K2 delivers 2.289 TFLOPS of FP32 compute, while the GTX 750 delivers 1,111.0 GFLOPS, which is 1.111 TFLOPS. The GRID K2 is 2.06x faster in raw FP32. Neither card has recorded FP16, ray tracing, or tensor core capabilities.

The GRID K2 also draws significantly more power. Its TDP is 225 W versus 55 W for the GTX 750, a 4x difference. The GTX 750 requires no power connectors and a suggested 250 W PSU, while the GRID K2 needs one 6-pin and one 8-pin connector with a suggested 550 W PSU. The GTX 750 is single-slot and 145 mm long; the GRID K2 is dual-slot and 267 mm long. The GRID K2 has no display outputs, confirming its role as a compute or virtualization part, while the GTX 750 offers 2x DVI and 1x mini-HDMI 1.4a.

API support is similar but not identical. Both support DirectX 12 (11_0) and OpenGL 4.6. The Vulkan versions differ: the GRID K2 lists Vulkan 1.2.175, while the GTX 750 lists Vulkan 1.4. That newer Vulkan revision on the GTX 750 reflects its later release date, 2014-02-17 versus 2013-05-10 for the GRID K2. The GTX 750 also has a recorded predecessor (GeForce 600) and successor (GeForce 900), while the GRID K2 has neither in the database.

The Verdict

The data points to a clear compute performance winner: the NVIDIA GRID K2. It wins both benchmark tests, holds a 13.8% to 30% lead in direct comparisons, and delivers roughly double the FP32 throughput and memory bandwidth of the GTX 750. For any workload that saturates shader units, texture units, or memory bandwidth, the GRID K2 is the superior choice. Its 4 GB VRAM capacity also gives it a practical edge for datasets that exceed the GTX 750's 1 GB limit.

The GTX 750, however, is not without reason for selection. Its 55 W TDP means it can run in systems with a 250 W PSU and no auxiliary power connectors. The GRID K2 requires a 550 W PSU and both a 6-pin and 8-pin connector. The GTX 750 is also physically smaller at 145 mm versus 267 mm, and it is single-slot versus dual-slot. For a compact, low-power desktop or a system with strict power and space budgets, the GTX 750 is the practical pick. It also has display outputs, making it usable as a standard graphics card, whereas the GRID K2 cannot drive a monitor.

The GTX 750's Vulkan 1.4 support is newer than the GRID K2's 1.2.175, which may matter for applications relying on newer Vulkan extensions. The GRID K2's OpenCL score advantage (13.8%) is smaller than its Metal advantage (30%), so if the primary workload is OpenCL, the gap narrows but does not disappear.

The typical user profile matters. The GRID K2's 42nd percentile and 8,080 average score place it among mid-range Kepler parts, but its compute density and large VRAM make it a server-side compute card. The GTX 750's 40th percentile and 7,222 average score put it in a similar percentile band, but its nearest rivals include integrated GPUs and older discrete parts, reflecting a more modest absolute performance level. The GRID K2 is the choice for compute-heavy tasks where power and space are available. The GTX 750 is the choice for a low-power, compact, display-capable card with adequate but not exceptional compute.

Specification Differences

The two cards differ in nearly every measured specification. Process node is identical (28 nm, TSMC), but transistor count is 3,540 million for the GRID K2 versus 1,870 million for the GTX 750. Die size is 294 mm² versus 148 mm². Transistor density is 12.0M per mm² versus 12.6M per mm². The GRID K2 has no base or boost clock recorded; the GTX 750 has a base of 1020 MHz and boost of 1085 MHz. Memory size is 4 GB versus 1024 MB. Memory bus width is 256-bit versus 128-bit. Memory bandwidth is 160.0 GB/s versus 80.19 GB/s. Shading units are 1,536 versus 512. TMUs are 128 versus 32. ROPs are 32 versus 16. Pixel rate is 23.84 GPixel/s versus 17.36 GPixel/s. Texture rate is 95.36 GTexel/s versus 34.72 GTexel/s. FP32 is 2.289 TFLOPS versus 1,111.0 GFLOPS. TDP is 225 W versus 55 W. Slot width is dual-slot versus single-slot. Power connectors are 1x 6-pin + 1x 8-pin versus none. Suggested PSU is 550 W versus 250 W. Display outputs are none versus 2x DVI, 1x mini-HDMI 1.4a. Vulkan version is 1.2.175 versus 1.4. Length is 267 mm versus 145 mm. Release date is 2013-05-10 versus 2014-02-17. Launch MSRP is 5,199 USD for the GRID K2 and 119 USD for the GTX 750. The GRID K2 has no predecessor or successor recorded; the GTX 750 lists GeForce 600 and GeForce 900 respectively.

FAQ

Q: Which card performs better in Geekbench Metal?

A: The NVIDIA GRID K2 scores 5,557 versus 4,274 for the GTX 750, a 30% advantage for the GRID K2.

Q: Is the GTX 750 ever faster than the GRID K2 in any recorded benchmark?

A: No. The GRID K2 wins both the Metal and OpenCL tests, with a 13.8% lead in OpenCL and a 30% lead in Metal.

Q: How much memory bandwidth does each card have?

A: The GRID K2 has 160.0 GB/s from a 256-bit bus with 4 GB GDDR5. The GTX 750 has 80.19 GB/s from a 128-bit bus with 1024 MB GDDR5.

Q: What are the power requirements for each card?

A: The GRID K2 has a 225 W TDP, requires one 6-pin and one 8-pin power connector, and a suggested 550 W PSU. The GTX 750 has a 55 W TDP, requires no power connectors, and a suggested 250 W PSU.

Q: Can either card output video to a display?

A: The GTX 750 has 2x DVI and 1x mini-HDMI 1.4a outputs. The GRID K2 has no display outputs.

Q: Which card has newer Vulkan API support?

A: The GTX 750 supports Vulkan 1.4, while the GRID K2 supports Vulkan 1.2.175.

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

A: The GRID K2 averages 8,080, and the GTX 750 averages 7,222, a difference of 858 points in favor of the GRID K2.

Q: How do their transistor counts compare?

A: The GRID K2 uses 3,540 million transistors on a 294 mm² die. The GTX 750 uses 1,870 million transistors on a 148 mm² die.