NVIDIA CMP 30HX vs NVIDIA Tesla P100 PCIe 12 GB Comparison

NVIDIA
GEFORCE

NVIDIA CMP 30HX

CORE STATE TU116
VRAM 6 GB
CLOCK SPEED 1785 MHz
TDP 125 W
BUS WIDTH 192 bit
ARCHITECTURE Turing
nm
PROCESS 12 nm
LAUNCH DATE 2021
VS
NVIDIA
GEFORCE

Tesla P100 PCIe 12 GB

CORE STATE GP100
VRAM 12 GB
CLOCK SPEED 1329 MHz
TDP 250 W
BUS WIDTH 3072 bit
ARCHITECTURE Pascal
nm
PROCESS 16 nm
LAUNCH DATE 2016

PERFORMANCE BENCHMARKS

geekbench_opencl
65,199
79,396
geekbench_vulkan
62,484
N/A

Analysis: NVIDIA CMP 30HX vs NVIDIA Tesla P100 PCIe 12 GB

NVIDIA Tesla P100 PCIe 12 GB and NVIDIA CMP 30HX are both end-of-life NVIDIA boards with no display outputs, but they target entirely different workloads. The P100 is a Pascal-era compute accelerator with 12 GB of HBM2, while the CMP 30HX is a Turing-based mining card with 6 GB of GDDR6. Benchmark data shows the P100 leads in the only shared test, but the CMP 30HX counters with a lower power draw and a more recent API version.

Head-to-Head Benchmarks

In the single common benchmark test — Geekbench OpenCL — the NVIDIA Tesla P100 PCIe 12 GB scores 79,396 points, while the NVIDIA CMP 30HX scores 65,199 points. That puts the P100 ahead by 21.8%, a decisive margin that reflects the raw compute gap between the two cards. The P100’s FP32 throughput of 9.526 TFLOPS is nearly double the CMP 30HX’s 5.027 TFLOPS, and its texture rate of 297.7 GTexel/s far exceeds the 157.1 GTexel/s of the CMP 30HX. Pixel rate also favors the P100 at 127.6 GPixel/s versus 85.68 GPixel/s.

The CMP 30HX does have a second benchmark result — Geekbench Vulkan at 62,484 — which the P100 lacks, but that does not constitute a head-to-head win since no comparable score exists for the P100. In the official head-to-head comparison, the P100 wins the only matched test, securing a 1-0 advantage in wins. The delta of 21.8% is substantial enough that no amount of driver tuning or workload selection on the CMP 30HX is likely to close the gap in OpenCL compute tasks.

Looking at the nearest rivals for context, the P100’s 79,396 average score sits just 0.3% below the Tesla P100 PCIe 16 GB (79,605) and 0.6% below the GeForce RTX 5090 (79,842). It is 0.6% above the AMD Radeon RX 6850M XT (78,940) and 1.9% below the AMD Radeon Pro Vega 64X (80,959). The CMP 30HX, with an average score of 63,842, is effectively tied with the AMD Radeon RX 9060 XT LP (63,830), sits 0.1% above the RX 7600M (63,775), and 0.2% above the Radeon Pro Vega 56 (63,693). It trails the Radeon Pro WX 9100 (64,212) by 0.6%. In percentile terms, the P100 ranks in the 92nd percentile of all GPUs, while the CMP 30HX ranks in the 89th percentile — a three-point gap that aligns with the raw score difference.

Architecture Differences

The two cards come from different NVIDIA architectural generations. The Tesla P100 uses the GP100 chip on the Pascal architecture, built on a 16 nm process at TSMC with 15,300 million transistors on a 610 mm² die. The CMP 30HX uses the TU116 chip on the Turing architecture, built on a 12 nm process at TSMC with 6,600 million transistors on a 284 mm² die. Transistor density is close — 25.1M per mm² for the P100 versus 23.2M per mm² for the CMP 30HX — but the P100’s die is more than twice the physical size.

Memory architecture is a major differentiator. The P100 packs 12 GB of HBM2 on a 3072-bit bus, delivering 549.1 GB/s of bandwidth. The CMP 30HX has 6 GB of GDDR6 on a 192-bit bus, yielding 336.0 GB/s. That is a 63% bandwidth advantage for the P100, which matters for memory-bound compute workloads. The P100’s memory clock is listed as 715 MHz (1430 Mbps effective), while the CMP 30HX runs at 1750 MHz (14 Gbps effective) — the GDDR6 runs faster per pin, but the P100’s massive bus width wins overall.

Shading resources diverge sharply. The P100 has 3,584 shading units, 224 TMUs, and 96 ROPs. The CMP 30HX has 1,408 shading units, 88 TMUs, and 48 ROPs. Neither card has ray tracing cores or tensor cores, so both rely purely on conventional shader throughput. Clock speeds favor the CMP 30HX: base 1530 MHz and boost 1785 MHz versus the P100’s 1190 MHz base and 1329 MHz boost. The CMP 30HX’s higher clocks partially offset its fewer cores, but not enough to overcome the P100’s 2.5x shading unit advantage.

API support is close but not identical. Both cards support DirectX 12 (12_1) and OpenGL 4.6. The P100 lists Vulkan 1.3, while the CMP 30HX lists Vulkan 1.4. The CMP 30HX also has a Geekbench Vulkan score of 62,484, indicating functional Vulkan capability, whereas the P100 has no Vulkan benchmark recorded. Power specifications differ substantially — the P100 has a 250 W TDP with a 600 W suggested PSU, while the CMP 30HX draws only 125 W with a 300 W suggested PSU.

The Verdict

From the benchmark data, the Tesla P100 PCIe 12 GB is the clear compute performance winner. Its 21.8% lead in OpenCL over the CMP 30HX, combined with roughly double the FP32 throughput, double the texture rate, and 63% more memory bandwidth, makes it the superior choice for any compute-heavy workload that fits within 12 GB of HBM2. The 92nd percentile ranking versus the CMP 30HX’s 89th percentile reinforces this.

However, the CMP 30HX is not without merit. It consumes half the power (125 W versus 250 W), requires a 300 W PSU instead of 600 W, and is physically smaller at 229 mm versus 267 mm. It also supports Vulkan 1.4 versus the P100’s Vulkan 1.3, and it has a documented Vulkan benchmark result. For a system with power constraints or a small chassis, the CMP 30HX is far easier to integrate. But the data does not show any performance scenario where the CMP 30HX beats the P100 in a shared test — it wins zero head-to-head benchmarks.

The launch MSRP for the Tesla P100 PCIe 12 GB is 4,599 USD, and for the CMP 30HX it is 799 USD. That is a massive price difference, but the performance-per-dollar analysis is not part of this database’s scope. What the numbers show is that the P100 delivers roughly 22% more OpenCL performance, and the buyer must decide whether that performance gap justifies the larger power envelope and board size. For raw compute, pick the P100. For efficiency and compactness, the CMP 30HX is the practical option.

Specification Differences

  • Chip and architecture: GP100 on Pascal (P100) versus TU116 on Turing (CMP 30HX)
  • Process node: 16 nm (P100) versus 12 nm (CMP 30HX)
  • Transistors: 15,300 million (P100) versus 6,600 million (CMP 30HX)
  • Die size: 610 mm² (P100) versus 284 mm² (CMP 30HX)
  • Transistor density: 25.1M / mm² (P100) versus 23.2M / mm² (CMP 30HX)
  • Base clock: 1190 MHz (P100) versus 1530 MHz (CMP 30HX)
  • Boost clock: 1329 MHz (P100) versus 1785 MHz (CMP 30HX)
  • Memory clock: 715 MHz / 1430 Mbps effective (P100) versus 1750 MHz / 14 Gbps effective (CMP 30HX)
  • Memory size: 12 GB HBM2 (P100) versus 6 GB GDDR6 (CMP 30HX)
  • Memory bus: 3072 bit (P100) versus 192 bit (CMP 30HX)
  • Memory bandwidth: 549.1 GB/s (P100) versus 336.0 GB/s (CMP 30HX)
  • Shading units: 3584 (P100) versus 1408 (CMP 30HX)
  • TMUs: 224 (P100) versus 88 (CMP 30HX)
  • ROPs: 96 (P100) versus 48 (CMP 30HX)
  • Pixel rate: 127.6 GPixel/s (P100) versus 85.68 GPixel/s (CMP 30HX)
  • Texture rate: 297.7 GTexel/s (P100) versus 157.1 GTexel/s (CMP 30HX)
  • FP32: 9.526 TFLOPS (P100) versus 5.027 TFLOPS (CMP 30HX)
  • FP16: 19.05 TFLOPS (P100) versus 10.05 TFLOPS (CMP 30HX)
  • TDP: 250 W (P100) versus 125 W (CMP 30HX)
  • Suggested PSU: 600 W (P100) versus 300 W (CMP 30HX)
  • Bus interface: PCIe 3.0 x16 (P100) versus PCIe 1.0 x4 (CMP 30HX)
  • Length: 267 mm (P100) versus 229 mm (CMP 30HX)
  • Height: not listed (P100) versus 111 mm (CMP 30HX)
  • Width: not listed (P100) versus 35 mm (CMP 30HX)
  • Vulkan version: 1.3 (P100) versus 1.4 (CMP 30HX)
  • Release date: 2016-06-19 (P100) versus 2021-02-24 (CMP 30HX)
  • Launch MSRP: 4,599 USD (P100) versus 799 USD (CMP 30HX)
  • Average benchmark score: 79,396 (P100) versus 63,842 (CMP 30HX)
  • Percentile: 92nd (P100) versus 89th (CMP 30HX)

FAQ

Q: Which card is faster in OpenCL benchmarks?

A: The NVIDIA Tesla P100 PCIe 12 GB scores 79,396 in Geekbench OpenCL, which is 21.8% higher than the NVIDIA CMP 30HX’s 65,199. The P100 wins the only shared head-to-head test.

Q: Does the CMP 30HX have any benchmark where it wins?

A: No. In the head-to-head comparison, the CMP 30HX has zero wins. It does have a separate Geekbench Vulkan score of 62,484, but the P100 has no recorded Vulkan benchmark, so no comparison exists for that test.

Q: How do the memory subsystems compare?

A: The P100 has 12 GB of HBM2 on a 3072-bit bus with 549.1 GB/s bandwidth. The CMP 30HX has 6 GB of GDDR6 on a 192-bit bus with 336.0 GB/s. The P100’s bandwidth advantage is roughly 63%.

Q: What are the power requirements for each card?

A: The P100 has a 250 W TDP and requires a 600 W suggested PSU. The CMP 30HX has a 125 W TDP and a 300 W suggested PSU. Both use a single 8-pin power connector and are dual-slot cards.

Q: Which card supports a newer Vulkan version?

A: The CMP 30HX supports Vulkan 1.4, while the P100 supports Vulkan 1.3. Both support DirectX 12 (12_1) and OpenGL 4.6.

Q: How do the cards rank against all other GPUs?

A: The P100 sits in the 92nd percentile of all GPUs with an average score of 79,396. The CMP 30HX sits in the 89th percentile with an average score of 63,842.

Where Each One Wins

The Tesla P100 PCIe 12 GB wins in every measurable performance category from the data. It has higher FP32 (9.526 TFLOPS versus 5.027 TFLOPS), higher FP16 (19.05 TFLOPS versus 10.05 TFLOPS), higher pixel rate (127.6 GPixel/s versus 85.68 GPixel/s), higher texture rate (297.7 GTexel/s versus 157.1 GTexel/s), and more memory bandwidth (549.1 GB/s versus 336.0 GB/s). It also has more shading units, TMUs, and ROPs, and it wins the only head-to-head benchmark by 21.8%. Its 92nd percentile ranking is three points higher than the CMP 30HX’s 89th. This card is for compute workloads where raw throughput and memory bandwidth are the limiting factors — think large data sets, HPC-style math, or any OpenCL-heavy application.

The CMP 30HX wins in efficiency and physical integration. It uses half the power (125 W versus 250 W), requires a 300 W PSU instead of 600 W, and is shorter (229 mm versus 267 mm) with a smaller die (284 mm² versus 610 mm²). It also supports Vulkan 1.4 versus the P100’s 1.3, which could matter for applications built against the newer API spec. Its higher boost clock (1785 MHz versus 1329 MHz) shows it runs faster per core, but the core count deficit is too large to overcome. For a system with limited power budget, a small case, or a need for the latest Vulkan feature set, the CMP 30HX is the more practical board. For everything else, the P100 is the performance pick.

DETAILED SPECIFICATIONS

SPECIFICATION
CMP 30HX
Tesla P100 PCIe 12 GB
Core Specs
Shading Units
1,408
3,584 +154.5%
Shaders
1,408
3,584 +154.5%
TMUs
88
224 +154.5%
ROPs
48
96 +100.0%
SM Count
22
56 +154.5%
Clocks
Base Clock
1530 MHz
1190 MHz
Boost Clock
1785 MHz
1329 MHz
Memory Clock
1750 MHz 14 Gbps effective
715 MHz 1430 Mbps effective
Memory
Memory Size
6 GB
12 GB
VRAM (MB)
6,144
12,288 +100.0%
Memory Type
GDDR6
HBM2
Memory Bus
192 bit
3072 bit
Bandwidth
336.0 GB/s
549.1 GB/s
Cache
L1 Cache
64 KB (per SM)
24 KB (per SM)
L2 Cache
1536 KB
3 MB
Performance
Pixel Rate
85.68 GPixel/s
127.6 GPixel/s
Texture Rate
157.1 GTexel/s
297.7 GTexel/s
FP32 (TFLOPS)
5.027 TFLOPS
9.526 TFLOPS
FP64 (TFLOPS)
157.1 GFLOPS (1:32)
4.763 TFLOPS (1:2)
FP16 (TFLOPS)
10.05 TFLOPS (2:1)
19.05 TFLOPS (2:1)
Power
TDP
125 W
250 W
TDP (W)
125
250 +100.0%
Suggested PSU
300 W
600 W
Power Connectors
1x 8-pin
1x 8-pin
Architecture
Architecture
Turing
Pascal
GPU Name
TU116
GP100
Generation
Mining GPUs
Tesla Pascal (Pxx)
Process Size
12 nm
16 nm
Transistors
6,600 million
15,300 million
Die Size
284 mm²
610 mm²
Foundry
TSMC
TSMC
Density
23.2M / mm²
25.1M / mm²
API Support
DirectX
12 (12_1)
12 (12_1)
OpenGL
4.6
4.6
Vulkan
1.4
1.3
OpenCL
3.0
3.0
CUDA
7.5
6.0
Shader Model
6.8
6.0
Physical
Slot Width
Dual-slot
Dual-slot
Length
229 mm 9 inches
267 mm 10.5 inches
Height
111 mm 4.4 inches
—
Outputs
No outputs
No outputs
Bus Interface
PCIe 1.0 x4
PCIe 3.0 x16
Other
Launch Price
799 USD
4,599 USD
Production
End-of-life
End-of-life
Predecessor
—
Tesla Maxwell
Successor
—
Tesla Volta
View CMP 30HX Details View Tesla P100 PCIe 12 GB Details