NVIDIA GeForce GTX 950A vs NVIDIA Tesla K20c Comparison

The NVIDIA Tesla K20c and the NVIDIA GeForce GTX 950A are both end-of-life NVIDIA parts built on the same 28 nm TSMC process, but they target completely different segments of the market. In the single available head-to-head benchmark, the Tesla K20c decisively outperforms the GTX 950A, but the data reveals a far more nuanced story about compute capability versus power efficiency and architectural generation.

Head-to-Head Benchmarks

The only direct comparison available is the Geekbench OpenCL benchmark, and the result is a clear win for the Tesla K20c. The Tesla K20c scores 11,479 points, while the GeForce GTX 950A scores 10,273 points. This translates to an 11.7% advantage for the Tesla K20c, a substantial margin that underscores its positioning as a compute-oriented workstation part.

Contextualizing this result against the broader GPU landscape reinforces the gap. The Tesla K20c sits at the 51st percentile of all GPUs, while the GTX 950A sits at the 48th percentile. The Tesla K20c’s nearest rivals in the benchmark database are the AMD Radeon Pro 5500M (11,528 points, -0.4% delta), the AMD Radeon RX 7800 XT (11,627 points, -1.3% delta), and the NVIDIA GeForce GTX 1660 (11,680 points, -1.7% delta). Notably, the Tesla K20c is only 0.4% behind the Radeon Pro 5500M, a much newer mobile workstation GPU, and remarkably just 1.7% behind the GTX 1660, a dedicated desktop gaming card from a much later generation.

The GTX 950A’s nearest rivals tell a different story. Its closest competitor is the AMD Radeon RX 6500M (10,362 points, -0.9% delta), followed by the NVIDIA Tesla C2075 (10,400 points, -1.2% delta) and the AMD Radeon RX 550X (10,481 points, -2% delta). The GTX 950A is also 2% ahead of the AMD Radeon R9 M375 (10,070 points). These comparisons show that the GTX 950A is competitive with entry-level mobile and legacy workstation parts, but it is firmly in a lower performance tier than the Tesla K20c.

The benchmark data is unambiguous: the Tesla K20c wins the only head-to-head test, and its lead is significant. There are no benchmark tests in the data where the GTX 950A emerges victorious.

Architecture Differences

The two GPUs represent two distinct architectural generations from NVIDIA, despite sharing the same 28 nm manufacturing node and foundry (TSMC). The Tesla K20c is built on the Kepler architecture, specifically the GK110 chip, while the GTX 950A uses the Maxwell architecture with the GM107 chip. This generation gap is fundamental to understanding their different characteristics.

The chip sizes and transistor counts diverge dramatically. The Tesla K20c’s GK110 die is 561 mm² and contains 7,080 million transistors. In contrast, the GTX 950A’s GM107 die is only 148 mm² and houses 1,870 million transistors. Both have an identical transistor density of 12.6M per mm², which is expected given they are fabricated on the same process node. The Tesla K20c’s massive die is over 3.7 times larger than the GTX 950A’s, reflecting its heavy compute focus.

Compute resources are heavily skewed toward the Tesla K20c. It features 2,496 shading units, 208 texture mapping units, and 40 render output units. The GTX 950A, by contrast, has 640 shading units, 40 TMUs, and 16 ROPs. This means the Tesla K20c has nearly four times the shader count and over five times the TMU count of the GTX 950A. Consequently, the theoretical peak rates are vastly different: the Tesla K20c achieves 3.524 TFLOPS FP32, 36.71 GPixel/s pixel rate, and 146.8 GTexel/s texture rate, while the GTX 950A delivers 1,438.7 GFLOPS FP32, 17.98 GPixel/s pixel rate, and 44.96 GTexel/s texture rate.

Memory architecture also differs fundamentally. The Tesla K20c is equipped with 5 GB of GDDR5 memory on a 320-bit bus, yielding a bandwidth of 208.0 GB/s. The GTX 950A uses 2 GB of DDR3 memory on a 128-bit bus, delivering just 32.03 GB/s of bandwidth. This is a 6.5x difference in memory bandwidth, a critical factor for compute workloads. The memory clocks also differ, with the Tesla K20c running at 1300 MHz (5.2 Gbps effective) and the GTX 950A at 1001 MHz (2 Gbps effective).

The GTX 950A does have a clock speed advantage where the data exists. Its base clock is 993 MHz and boost clock is 1,124 MHz, while the Tesla K20c’s base and boost clocks are not listed in the data. However, this clock advantage is insufficient to overcome the massive differences in core count and memory bandwidth.

Other architectural differences include power and interface. The Tesla K20c has a 225 W TDP, requires a dual-slot cooler with 1x 6-pin and 1x 8-pin power connectors, and suggests a 550 W power supply. It uses a PCIe 2.0 x16 bus interface. The GTX 950A, on the other hand, has a 75 W TDP, uses an MXM Module form factor with no power connectors, and connects via an MXM-B (3.0) interface. The Tesla K20c has no display outputs, while the GTX 950A’s display outputs are labeled as "Portable Device Dependent".

Where Each One Wins

The benchmark data clearly shows the Tesla K20c wins in raw compute performance. Its 11.7% lead in Geekbench OpenCL, combined with its vastly superior shader count, texture rate, and memory bandwidth, makes it the clear choice for any workload that leverages massive parallelism and high data throughput. The Tesla K20c’s 3.524 TFLOPS FP32 performance and 208.0 GB/s memory bandwidth are the standout metrics that drive its compute advantage. Its 5 GB of memory also provides more headroom for large datasets compared to the GTX 950A’s 2 GB.

The GTX 950A, however, wins decisively in power efficiency and physical footprint. Its 75 W TDP is exactly one-third of the Tesla K20c’s 225 W TDP. It requires no external power connectors, uses a compact MXM Module form factor, and has no suggested PSU requirement listed. This makes it suitable for portable or space-constrained devices, as indicated by its "Portable Device Dependent" display outputs. The GTX 950A also has a higher boost clock (1,124 MHz) than the Tesla K20c’s unspecified boost clock, and its newer Maxwell architecture supports Vulkan 1.4, a newer API version than the Tesla K20c’s Vulkan 1.2.175.

In terms of production timeline, the Tesla K20c was released in November 2012, while the GTX 950A came later in March 2015. The Tesla K20c’s predecessor is Tesla Fermi and its successor is Tesla Maxwell, while the GTX 950A’s predecessor is GeForce 800A and it has no listed successor.

FAQ

Q: Which GPU has a higher Geekbench OpenCL score?

A: The NVIDIA Tesla K20c scores 11,479, which is 11.7% higher than the NVIDIA GeForce GTX 950A’s score of 10,273.

Q: How do the memory bandwidths compare between the two cards?

A: The Tesla K20c has a memory bandwidth of 208.0 GB/s using 5 GB of GDDR5 on a 320-bit bus. The GTX 950A has only 32.03 GB/s bandwidth using 2 GB of DDR3 on a 128-bit bus.

Q: What are the TDP requirements for each card?

A: The Tesla K20c has a TDP of 225 W and requires a dual-slot cooler with 1x 6-pin and 1x 8-pin power connectors. The GTX 950A has a TDP of 75 W and requires no power connectors.

Q: Which GPU supports a newer version of Vulkan?

A: The GeForce GTX 950A supports Vulkan 1.4, while the Tesla K20c supports Vulkan 1.2.175.

Q: What is the transistor count of each GPU?

A: The Tesla K20c contains 7,080 million transistors on a 561 mm² die. The GTX 950A contains 1,870 million transistors on a 148 mm² die.

Q: Which card has more shading units?

A: The Tesla K20c has 2,496 shading units, while the GTX 950A has only 640 shading units.

Specification Differences

The two GPUs differ on nearly every major specification. The most significant differences are:

• Chip and Architecture: GK110 (Kepler) vs GM107 (Maxwell)
• Transistors: 7,080 million vs 1,870 million
• Die Size: 561 mm² vs 148 mm²
• Base Clock: Not listed vs 993 MHz
• Boost Clock: Not listed vs 1,124 MHz
• Memory Size: 5 GB vs 2 GB
• Memory Type: GDDR5 vs DDR3
• Memory Bus Width: 320 bit vs 128 bit
• Memory Bandwidth: 208.0 GB/s vs 32.03 GB/s
• Shading Units: 2,496 vs 640
• TMUs: 208 vs 40
• ROPs: 40 vs 16
• Pixel Rate: 36.71 GPixel/s vs 17.98 GPixel/s
• Texture Rate: 146.8 GTexel/s vs 44.96 GTexel/s
• FP32 Performance: 3.524 TFLOPS vs 1,438.7 GFLOPS
• TDP: 225 W vs 75 W
• Slot Width: Dual-slot vs MXM Module
• Power Connectors: 1x 6-pin + 1x 8-pin vs None
• Suggested PSU: 550 W vs Not listed
• Bus Interface: PCIe 2.0 x16 vs MXM-B (3.0)
• Display Outputs: No outputs vs Portable Device Dependent
• Vulkan Support: 1.2.175 vs 1.4
• Release Date: November 2012 vs March 2015
• Launch MSRP: 3,199 USD vs Not listed

The Verdict

The data points to a clear split in purpose. The NVIDIA Tesla K20c is the superior compute performer, winning the only head-to-head benchmark by 11.7% and offering vastly more shader units, texture units, memory bandwidth, and memory capacity. Its 3.524 TFLOPS FP32 capability and 208.0 GB/s bandwidth make it the obvious choice for compute-intensive workloads like scientific simulation or machine learning inference, provided the system can handle its 225 W TDP and dual-slot footprint. Its percentile ranking (51st) and proximity to much newer cards like the AMD Radeon RX 7800 XT (1.3% behind) demonstrate its enduring compute strength.

The NVIDIA GeForce GTX 950A is the efficiency-oriented choice. Its 75 W TDP, lack of power connectors, and MXM form factor make it suitable for mobile or embedded systems. It is competitive within its own performance tier, sitting just 0.9% behind the AMD Radeon RX 6500M and 1.2% behind the NVIDIA Tesla C2075. Its Maxwell architecture provides newer API support, including Vulkan 1.4, which could be relevant for specific software compatibility needs.

Who should pick which? Choose the Tesla K20c if raw compute throughput and memory bandwidth are paramount, and if the platform can accommodate its power and space requirements. Choose the GTX 950A if power efficiency, compact form factor, and a newer API feature set are more important than peak performance. The benchmark results are conclusive on performance, but the specification differences are conclusive on use case.