If you’re not familiar with the new Quadro RTX line, here’s what you need to know. The new RTX cards use the next generation of GPUs based on the Turing architecture with features that come together on the NVIDIA RTX platform.

Turing adds many important features over previous architecture, the primary of which is the new Tensor cores that add deep learning AI capabilities and the RT cores for real-time ray tracing acceleration. The RTX platform provides various RTX APIs for developers which allow them to build RTX applications that take advantage of the RTX Turing architecture and allows you to take advantage of those RTX features in applications that take advantage of NVIDIA’s line of RTX GPUs. Got that? Good.

To sum it up, the NVIDIA RTX GPUs can do what no other GPUs before it could do. To the point, it caused a shift in the gaming and 3D software industry, to the joy of some developers and chagrin of others yet, that made it possible for the GPU to handle the complexities of real-time ray tracing which had only been possible on CPU before.

In light of this, the RTX cards, with their RT cores and Tensor cores, introduced a new level of software + hardware acceleration. Together the real-time ray tracing results were like nothing ever seen before, in both speed and performance.

NVIDIA currently has five Quadro RTX options – RTX 3000, RTX 4000, RTX 5000, RTX 6000, RTX 8000, with the RTX 4000 sitting as the entry point to the whole bunch.

Quadro RTX 4000 Specs

Now, the Quadro RTX 4000 isn’t the most powerful GPU on the RTX platform but it hits a sweet spot with a price point right under $900. We’ve had a look at the Quadro RTX 4000 in a previous article detailing its specs. To provide a quick refresher here’s a comparison of the features among the Quadro RTX series.

CUDA Cores Quadro RTX GPU Comparison
  Quadro RTX 4000 Quadro RTX 5000 Quadro RTX 6000 Quadro RTX 8000
GPU Memory 8 GB GDDR6 16 GB GDDR6 24 GB GDDR6 48 GB GDDR6
Memory Bandwidth 416 Gbps 416 Gbps 624 Gbps 624 Gbps
CUDA Cores 2304 3072 4608 4608
Tensor Cores 288 384 576 576
RT Cores 36 48 72 72
RTX-OPS 43T 62T 84T 84T
Rays Cast 6 Giga Rays/Sec 8 Giga Rays/Sec 10 Giga Rays/Sec 10 Giga Rays/Sec
FP32 Performance 7.1 TFLOPS 11.2 TFLOPS 16.3 TFLOPS 16.3 TFLOPS
List Price $900 $2300 $6300 $10000

And if you’re on an older P(ascal), K(epler), or M(axwell) series Quadro card, you won’t have the Turing RT cores and Tensor cores that provide the real-time ray tracing and AI capabilities but you will get a lower priced card. Here’s a comparison of the RTX 4000 and the previous generation P-series cards:

CUDA Cores Quadro Mid-Range GPU Comparison
  Quadro RTX 4000 Quadro P4000 Quadro P2000 Quadro P1000
GPU Memory 8 GB GDDR6 8 GB GDDR5 5 GB GDDR5 4 GB GDDR5
Memory Bandwidth 416 Gbps 243 Gbps 140 Gbps 80 Gbps
CUDA Cores 2304 1792 1024 640
Tensor Cores 288
RT Cores 36
RTX-OPS 43T
Rays Cast 6 Giga Rays/Sec
FP32 Performance 7.1 TFLOPS 5.3 TFLOPS 3.0 TFLOPS 1.9 TFLOPS
List Price $900 $750 $399 $319

Quadro RTX 4000 Performance Comparison

This won’t be the most thorough RTX performance review but I did want to provide a comparison to the top of the line P-Series GPU that many may still have and, realistically, acts as a line between previous-generation NVIDIA GPUs and NVIDIA RTX GPUs.

Our test system is a Lenovo ThinkStation P520 workstation. It sits in the middle of the Lenovo P-Series workstation family. The base configuration with a Quadro RTX 4000 option comes in at $1900. See our ThinkStation P520 review for more details and how to swap a GPU if you need to upgrade an older card.

The P520 came with the very capable Quadro P4000. It’s a great higher-mid-range video card, especially for the price. Still, given the P4000 came out in February 2017, the RTX is sure to have some performance improvements, even with both running the latest Quadro GPU driver.

Like the Lenovo P520, there are many 3D CAD workstations that have the P4000 as a base customization option. As mentioned above, however, the Quadro P4000 lacks the Turing RT cores and Tensor cores of the RTX cards which, believe me, make all the difference.

PassMark Performance Test:

There are some great performance benchmarks available with PassMark Performance Test our favorite for the ability to compare GPUs among thousands of other system configurations. It’s a free benchmark that provides results for various compute operations while providing averages and percentiles. Here are the results for the PassMark Test.

Quadro P4000
GPU Compute (Ops./Sec.) : 5526.7 (75%)
3D Graphics Mark (Composite average) : 14163.2 (77%)
PassMark Rating (Composite average) : 6044.9 (77%)

Quadro RTX 4000
GPU Compute (Ops./Sec.) : 6456.9 (77%)
3D Graphics Mark (Composite average) : 17704.6 (90%)
PassMark Rating (Composite average) : 6161.3 (82%)

These results reveal the Quadro RTX 4000 capable of 930 more operations per second compared to the P4000 and a 3D Graphics Mark that puts it in the 90% percentile, 13 points over the P4000.

KeyShot Benchmark:

The KeyShot Benchmark is a free benchmark available in KeyShot Viewer. This is a slightly different benchmark that uses a baseline system to provide a single measure of increased (or decreased) performance. The manual describes a score of 1.0 (for either CPU or GPU) as matching the base system, an Intel Core i7-6900K CPU @3.20GHz, 2601 Mhz, 8 Core(s). Here are the results for the KeyShot Benchmark.

Quadro P4000
CPU result: (Intel(R) Xeon(R) W-2133 CPU @ 3.60GHz – Threads: 12) 1.38
GPU result: (Quadro P4000 with Driver: 442.50) 3.38

Quadro RTX 4000
CPU result: (Intel(R) Xeon(R) W-2133 CPU @ 3.60GHz – Threads: 12) 1.50
GPU result: (Quadro RTX 4000 with Driver: 442.50) 23.67

These results show the Quadro RTX 4000 is 7x faster than the P4000 and nearly 16x faster than using the CPU for the same operation.

For more in-depth reviews and wider comparison of cards, I highly recommend the following resources:

Conclusion

If you’re trying to decide whether to upgrade an older Quadro card or between a P-series Quadro and an RTX, the $150 price difference is worth it in my opinion, especially you need the GPU power for graphic intensive work. When customizing an OEM workstation, the cost to upgrade a GPU is almost always better priced than if you chose a less powerful option and upgraded later. So, if you’re looking for a new system, check the price difference with an RTX option. I think you’ll be surprised and, when you experience the difference, quite delighted at the results.

Disclosure: NVIDIA provided the RTX 4000 used in this review and Lenovo provided the workstation. No input or editing of this review was done by either. Josh Mings has worked for and currently provides services for Luxion, creator of KeyShot Viewer.

Author

Josh is founder and editor at SolidSmack.com, founder at Aimsift Inc., and co-founder of EvD Media. He is involved in engineering, design, visualization, the technology making it happen, and the content developed around it. He is a SolidWorks Certified Professional and excels at falling awkwardly.