Nvidia Quadro P6000 GPU reviewed – Outperforms Titan X Pascal

Nvidia's Pascal Quadro P6000 uses an unlocked GTX Titan X Pascal GPU

Nvidia Quadro P6000 GPU reviewed – Outperforms Titan X Pascal

 

Nvidia’s $5000 Quadro P6000 has been tested against Nvidia’s Titan X Pascal, outperforming it in a variety of gaming applications. 

While the Quadro P6000 is not intended for gaming applications, it is unsurprising that this GPU outperforms the Titan X Pascal when gaming, as it uses a fully unlocked version of Nvidia’s GP-102 GPU core, giving it more CUDA cores than the GTX Titan X. 

We can see in the graph below that in Futuremark’s 3DMARK Time Spy benchmark the Quadro P6000 easily achieved a score that is over 500 points higher than the GTX Titan X, which is news will likely disappoint a lot of Nvidia fans as it clearly shows that the $1200 GTX Titan X Pascal is not the highest performance GPU that they could have potentially released using their GP-102 silicon.

  

Nvidia Quadro P6000 GPU reviewed - Outperforms Titan X Pascal

 

Nvidia’s GTX Titan X does not utilise Nvidia’s full GP 102 GPU core, opening up the possibility of a future GTX Titan X Pascal Black Edition style GPU in the future which uses the full GP 102 GPU core.

Below is a table comparing all of Nvidia’s PCIe-based Pascal GPUs, showcasing a common trend between all of Nvidia’s GP core designs, showing that Nvidia’s full GP-104 core design has exactly 1280 more cores than that GP-106 GPU core design and that the Quadro P6000 has exactly 1280 more GPU cores than the GTX 1080. 

This leads us to believe that the Tesla P6000 uses the full GP-102 GPU core design, following the same trend that we see with Nvidia’s GP-104 and GP-106 GPU cores. 

Despite the Quadro P6000’s lower memory clock speeds, it can still outperform the GTX Titan X, though we also see on the other hand that the Quadro P5000’s lower memory clock speeds do harm the GPUs performance when compared to the GTX 1080, which offers similar specifications. 

We can see here that both core GPU performance and memory bandwidth/clock speeds are important aspects to consider for gaming applications.  This will make it very interesting to see how next generation GPUs will perform with HBM2 memory, which will offer both larger memory capacities and higher memory bandwidth than HBM1.  

 

  Nvidia Tesla P6000  GTX Titan X Quadro P5000 GTX 1080 GTX 1070 GTX 1060
GPU Architecture Pascal Pascal Pascal Pascal Pascal Pascal
Process node 16nm  16nm 16nm 16nm 16nm 16nm
GPU Core GP-102 GP-102 GP-104 GP-104 GP-104 GP-106
SM Units 60  56 40 40 30 20
Cores per SM 64  64 64 64 64 64
SP FP Performance 12TFLOPs  11 TFLOPs 9TFLOPs 9 TFLOPs 6.5 TFLOPs 4.4TFLOPS
CUDA Core Count 3840  3584 2560 2560 1920  1280
VRAM Type GDDR5X  GDDR5X GDDR5X GDDR5X  GDDR5  GDDR5
VRAM Cappacity 24GB  12GB 16GB 8GB  8GB  6GB
Memory Bus Size 384-bit  384-bit 256-bit 256-bit  256-bit  192-bit
Memory Clock Speed 9008MHz 10008MHz 9008MHz 10008MHz 8000MHz 8000MHz
Memory Bandwidth 432 GB/s  480 GB/s 296GB/s 320 GB/s  256 GB/s  192 GB/s
Base clock speed 1417MHz 1417MHz 1607MHz 1607MHz  1506Mhz  1506MHz
Boost clock speed 1531MHz  1531MHz 1733MHz 1733MHz  1683MHz  1708MHz
TDP 250W  250W 180W 180W  150W  120W
Power Connection 1×8-pin 1x 6-pin  1x 8-pin 1x 6-pin 1×8-pin 1x 8-pin  1×8-pin  6-pin
PCI Express  PCIe 3.0  PCIe 3.0 PCIe 3.0 PCIe 3.0  PCIe 3.0  PCIe 3.0

  

 

2017 is looking like an interesting year for the GPU market, with AMD’s Vega architecture expected in spring and Nvidia’s GTX 1080 Ti expected within a similar timeframe, both os which will provide some much-needed competition in the high-end GPU market. 

 

You can join the discussion on Nvidia’s Quadro P6000 outperforming the GTX Titan X Pascal in games on the OC3D Forums.