Techgage logo

NVIDIA TITAN RTX Workstation Performance Review


Date: July 24, 2019
Author(s): Rob Williams

NVIDIA’s TITAN RTX means business – and a lot of it. This jack-of-all-trades graphics card caters to those with serious visual computing needs, whether it be designing and rendering 3D scenes, or poring over repositories of photos or other data with deep-learning work.

Introduction & Testing References

NVIDIA’s TITAN series of graphics cards has been an interesting one since the launch of the original in 2013. That Kepler-based GTX TITAN model peaked at 4.5 TFLOPS single-precision (FP32), performance that was boosted to 5.1 TFLOPS with the release of the TITAN Black the following year.

Fast-forward to the present day, where we now have the TITAN RTX, boasting 16.3 TFLOPS of single-precision, and 32.6 TFLOPS of half-precision (FP16). Double-precision (FP64) used to be standard fare on the earlier TITANs, but today, you’ll need the Volta-based TITAN V for unlocked performance (6.1 TFLOPS), or AMD’s Radeon VII for partially unlocked performance (3.4 TFLOPS).

Lately, half-precision has garnered a lot of attention by the ProViz market, since it’s ideal for use with deep-learning and AI, things that are growing in popularity at a ridiculously quick pace. Add specifically tuned Tensor cores to the mix, and deep-learning performance on Turing becomes truly impressive.


Tensors are not the only party trick the TITAN RTX has. Like the rest of the RTX line (on both the gaming and pro side), RT cores are present in the TITAN RTX, useful for accelerating real-time ray tracing workloads. The cores need to be specifically supported by developers, using APIs such as DXR and VKRay. While support for NVIDIA’s technology started off tepid, industry support has grown a lot since the original unveiling of RTX at SIGGRAPH last year.

At E3 in June, a handful of games had ray tracing-related announcements, including Watch_Dogs: Legion, Cyberpunk 2077, Call of Duty: Modern Warfare, and of course, Quake II RTX. On the design side, some developers have already released their RTX accelerated solutions, while many more are in the works. NVIDIA has been talking a lot lately about the Adobes and Autodesks of the world helping to grow the list of RTX-infused software. We wouldn’t be surprised if more RTX goodness was revealed at SIGGRAPH this year yet again.

For deep-learning, the TITAN RTX’s strong FP16 performance is fast on its own, but there are a few perks onboard to help take things to the next level. The Tensor cores aid in much of the acceleration, but the ability to use mixed precision is another big part. With it, minimal data tracking will be stored in single-precision, while the key data will get crunched in half-precision. Everything combined, this can boost training performance by 3x over the base GPU.

NVIDIA's TITAN RTX & GeForce RTX 2080 Ti - Backs

Also notable for Turing is concurrent integer / floating-point operations, which allows games (or software) to execute INT and FP operations in parallel without tripping over each other in the pipeline. NVIDIA has noted in the past that with games like Shadow of the Tomb Raider, a sample set of 100 instructions included 62 FP and 38 INT, and that this concurrent feature directly improves performance as a result.

Another important feature of TITAN RTX is its ability to use NVLink, which essentially combines the memory pools of two cards together, resulting in a single framebuffer that can be used for the biggest possible projects. Since GPUs scale generally very well with the sorts of workloads the card targets, it’s the true memory pooling that’s going to offer the greatest benefit here. Gaming content that could also take advantage of multi-GPU would see a benefit with two cards and this connector, as well.

Because it’s a feature exclusive to these RTX GPUs right now, it’s worth mentioning that NVIDIA also bundles a VirtualLink port at the back, allowing you to plug in your HMD for VR, or in the worst case, use it as a full-powered USB-C port, either for data transfer or phone charging.

With all of that covered, let’s take a quick look at the overall current NVIDIA workstation stack:

NVIDIA’s Quadro & TITAN Workstation GPU Lineup
CoresBase MHzPeak FP32MemoryBandwidthTDPPrice
GV1005120120014.9 TFLOPS32 GB 8870 GB/s185W$8,999
RTX 80004608144016.3 TFLOPS48 GB 5624 GB/s???W$5,500
RTX 60004608144016.3 TFLOPS24 GB 5624 GB/s295W$4,000
RTX 50003072135011.2 TFLOPS16 GB 5448 GB/s265W$2,300
RTX 4000230410057.1 TFLOPS8 GB 1416 GB/s160W$900
TITAN RTX4608135016.3 TFLOPS24 GB 1672 GB/s280W$2,499
TITAN V5120120014.9 TFLOPS12 GB 4653 GB/s250W$2,999
P60003840141711.8 TFLOPS24 GB 6432 GB/s250W$4,999
P5000256016078.9 TFLOPS16 GB 6288 GB/s180W$1,999
P4000179212275.3 TFLOPS8 GB 3243 GB/s105W$799
P2000102413703.0 TFLOPS5 GB 3140 GB/s75W$399
P100064013541.9 TFLOPS4 GB 380 GB/s47W$299
P62051213541.4 TFLOPS2 GB 380 GB/s40W$199
P60038413541.2 TFLOPS2 GB 364 GB/s40W$179
P40025610700.6 TFLOPS2 GB 332 GB/s30W$139
Notes1 GDDR6; 2 GDDR5X; 3 GDDR5; 4 HBM2
5 GDDR6 (ECC); 6 GDDR5X (ECC); 7 GDDR5 (ECC); 8 HBM2 (ECC)
Architecture: P = Pascal; V = Volta; RTX = Turing

The TITAN RTX matches the Quadro RTX 6000 and 8000 for having the highest number of cores in the Turing lineup. NVIDIA says the TITAN RTX is about 3 TFLOPS faster in FP32 over the RTX 2080 Ti, and fortunately, we have results for both cards covering a wide-range of tests to see how they compare.

What’s not seen in the specs table above is the actual performance of the ray tracing and deep-learning components. This next table helps clear some of that up:

NVIDIA’s Quadro & TITAN – RTX Performance
RT CoresRTX-OPSRays Cast 1FP16 2INT8 3Deep-learning 2
TITAN RTX7284 T1132.6206.1130.5
RTX 80007284 T1032.6206.1130.5
RTX 60007284 T1032.6206.1130.5
RTX 50004862 T822.3178.489.2
RTX 40003643 T614.228.557
Notes1 Giga Rays/s; 2 TFLOPS; 3 TOPS

You’ll notice that the TITAN RTX has a higher “rays cast” spec than the top Quadros, which might owe its thanks to higher clocks. The other specs are identical across the top three GPUs, with obvious downgrading taking place as we move downward. Currently, the Quadro RTX 4000 (roughly a GeForce RTX 2070 equivalent) is the lowest-end current-gen Quadro from NVIDIA. Again, SIGGRAPH is almost upon us, so it could be that NVIDIA will have a hardware surprise in store; perhaps an RTX 2060 Quadro equivalent.

When the RTX 2080 Ti already offers so much performance, who exactly is the TITAN RTX for? NVIDIA is targeting it largely at researchers, but it secondarily acts as one of the fastest ProViz cards on the market. It could be opted for by those who want the fastest GPU solution going, and not to mention a huge 24GB framebuffer. 24GB might be a bit much for a lot of current visualization work, but with deep-learning, 24GB provides a lot of breathing room.

Despite all it offers, TITAN RTX can’t be called an “ultimate” solution for ProViz since it lacks some Quadro optimizations that the namesake GPUs have. That means in certain high-end design suites like Siemens NX, a true Quadro might prove better. But if you don’t use any workloads that experience specific enhancements, the TITAN RTX is going to be quite attractive given its feature-set (and that framebuffer!) If you’re ever confused about optimizations in your software of choice, please leave a comment!

A couple of years ago, NVIDIA decided to give some love to the TITAN series with driver enhancements that brings some parity between TITAN and Quadro. We can now say that TITAN RTX enjoys the same kind of performance boosts that the TITAN Xp did two years ago, something that will be reflected in some of the graphs ahead.

Test PC & What We Test

On the following pages, the results of our workstation GPU test gauntlet will be seen. The tests chosen cover a wide range of scenarios, from rendering to compute, and includes the use of both synthetic benchmarks and tests with real-world applications from the likes of Adobe and Autodesk.

Nineteen graphics cards have been tested for this article, with the list dominated by Quadro and Radeon Pro workstation cards. There’s a healthy sprinkling of gaming cards in there as well, however, to show you any possible optimization that may be taking place on one or the other.

Please note that the testing for this article was conducted a couple of months ago, before an onslaught of travel and product launches. Graphics card drivers released since our testing might improve performance in certain cases, but we wouldn’t expect any notable changes, having sanity checked a bunch of our usual tested software on both AMD and NVIDIA GPUs. Likewise, the previous version of Windows was used for this particular testing, but that also didn’t reveal any disadvantages when we sanity checked in 1903.

In recent months, we’ve spent a lot of time polishing our test suites, and also our internal testing scripts. We’re currently in the process of rebenchmarking a number of GPUs for an upcoming look at ProViz performance with cards from both AMD’s Radeon RX 5700 and NVIDIA’s GeForce SUPER series. Fortunately, results from those cards don’t really eat into a top-end card like the TITAN RTX, so tardiness hasn’t foiled us this time.

The specs of our test rig are seen below:

Techgage Workstation Test System
ProcessorIntel Core i9-9980XE (18-core; 3.0GHz)
MemoryHyperX FURY (4x16GB; DDR4-2666 16-18-18)
GraphicsAMD Radeon VII (16GB)
AMD Radeon RX Vega 64 (8GB)
AMD Radeon RX 590 (8GB)
AMD Radeon Pro WX 8200 (8GB)
AMD Radeon Pro WX 7100 (8GB)
AMD Radeon Pro WX 5100 (8GB)
AMD Radeon Pro WX 4100 (4GB)
AMD Radeon Pro WX 3100 (4GB)
NVIDIA GeForce RTX 2080 Ti (11GB)
NVIDIA GeForce RTX 2060 (6GB)
NVIDIA GeForce GTX 1080 Ti (11GB)
NVIDIA GeForce GTX 1660 Ti (6GB)
NVIDIA Quadro RTX 4000 (8GB)
NVIDIA Quadro P6000 (24GB)
NVIDIA Quadro P5000 (12GB)
NVIDIA Quadro P4000 (8GB)
NVIDIA Quadro P2000 (5GB)
StorageKingston KC1000 960GB M.2 SSD
Power SupplyCorsair 80 Plus Gold AX1200
ChassisCorsair Carbide 600C Inverted Full-Tower
CoolingNZXT Kraken X62 AIO Liquid Cooler
Et ceteraWindows 10 Pro build 17763 (1809)
DriversAMD Radeon: Adrenaline 19.4.1
AMD Radeon Pro: Enterprise 19.Q1.2
NVIDIA GeForce & TITAN: Creative Ready 419.67
NVIDIA Quadro: Quadro 419.67

Our benchmark results are categorized and spread across the following six pages. On page 2, we’re looking at some CUDA-based renderers, including V-Ray, Redshift, OctaneRender, and Arnold GPU. Some of these will add support for non-CUDA GPUs in time, and when that happens, we’ll introduce the tests to our Radeon benchmarking suite. Page 3 includes a number of more neutral renderers, like Blender, Radeon ProRender, and LuxMark.

On page 4, we’re tackling encoding with the help of Adobe’s Premiere Pro and MAGIX’s Vegas Pro, while page 5 is home to viewport performance, largely covered with the help of SPECviewperf. In total, 8 test results are featured here, covering important design suites like CATIA, SolidWorks, Siemens NX, Creo, as well as Autodesk’s 3ds Max and Maya. Our own Blender viewport test wraps the page up. And speaking of wrapping things up, page 6 covers mathematical performance with the help of SiSoftware’s Sandra.

And with all of that covered, let’s get on with things:

Rendering: V-Ray, Redshift, OctaneRender & Arnold GPU

Chaos Group V-Ray

Chaos Group V-Ray GPU Next in Autodesk 3ds Max 2019
Chaos Group V-Ray GPU Performance - Flowers Render (NVIDIA TITAN RTX)
Chaos Group V-Ray GPU Performance - Teaset Render (NVIDIA TITAN RTX)

To the surprise of no one, we’re sure, the TITAN RTX has proven dominant in our first rendering test. Notably, it isn’t exactly pulling far ahead of the RTX 2080 Ti, which depending on which GPU you’re gravitating towards most will be a great or bad thing. Regardless of how close the 2080 Ti keeps to the TITAN RTX in performance, though, the 2080 Ti will always fall short of that massive 24GB framebuffer.


Redshift in Autodesk 3ds Max 2019
Redshift GPU Performance (NVIDIA TITAN RTX)

In Redshift, the TITAN RTX domination continues, leading the pack in every way but the framebuffer (which is matched by the last-gen Quadro P6000). It wasn’t long ago 16GB of memory seemed completely normal for a desktop PC, whereas now, creators are craving that much for just their GPU.

OTOY Octane

OTOY OctaneRender 4 GPU Performance (NVIDIA TITAN RTX)

Because RTX offers unique ray tracing capabilities, OTOY has so far branched such tests to a separate version of OctaneBench, called OctaneBench RTX. The standard version is seen above, which still puts the TITAN RTX ahead of the pack a fair distance, though one not too wide to the 2080 Ti. The leap over last-gen cards is huge, though. It’s hard to believe we see a test that makes the TITAN Xp look weak, but Turing’s enhancements have sure helped a lot here.

What about when RTX is turned on?

OTOY OctaneRender RTX GPU Performance (NVIDIA TITAN RTX)

We covered a few months ago that Octane will be supporting more than just CUDA in the near-future, which ultimately means that GPUs from non-NVIDIA vendors can be used for rendering. Even Apple’s seriously locked-down macOS will keep the Octane love going thanks to what we believe will be Vulkan to Metal translation.

Even if non-CUDA cards are supported, OTOY is giving us the impression that all the rays are going to belong to NVIDIA. With “RTX On”, the performance uplift is simply incredible. That said, the RTX plugin is still in beta, and while it’d be assumed that any acceleration like this would be for quick iterative feedback rather than final renders, we were told differently at GTC in the spring.

We really can’t wait to see the final RTX-optimized Octane drop and hear some user experiences.

Autodesk Arnold GPU

Autodesk Arnold GPU Performance - E-Type Render (NVIDIA TITAN RTX)
Autodesk Arnold GPU Performance - Sophie Render (NVIDIA TITAN RTX)

We went in-depth with Arnold GPU a couple of months ago, and while the renderer was a lot of fun to play with, it definitely had some issues, and odd ones. The kind where a slower GPU was faster than one that’s technically superior. All of the performance from that article has found itself here again, with the RTX added to show that it happily dominates the rest.

Note that the 0s at the bottom of the graph meant that the render didn’t finish after an hour. How such specific GPUs face this bug, we’re not sure, but it does seem to impact the same ones each time.

As mentioned at least once before, we’re in the process of rebenchmarking a whack of GPUs for an upcoming look at ProViz performance with the latest gaming GPUs, and we can say that even with the latest Arnold version for Maya (, these scaling oddities continue to be seen.

Rendering: Blender, Radeon ProRender & LuxMark


Blender Classroom Render
Blender Performance - BMW Cycles Render (NVIDIA TITAN RTX)
Blender Performance - Classroom Cycles Render (NVIDIA TITAN RTX)
Blender Performance - The White Room Eevee Render (NVIDIA TITAN RTX)

As we learned during our massive benchmarking run with a Blender 2.80 beta a couple of months ago, NVIDIA’s Turing graphics cards are fast as heck in Blender. It’s no surprise, then, that the TITAN RTX takes a win here, with the 2080 Ti again falling not far behind.

Blender 2.80’s final release is coming very soon, and we’ve already retested a bunch of hardware using the currently available release candidate. We hope to deliver much more in-depth Blender performance not long after 2.80’s launch, so stay tuned for that.

Radeon ProRender

AMD Radeon ProRender - Vehicle Scene
AMD Radeon ProRender Performance - Nissan GT-R Render (NVIDIA TITAN RTX)
AMD Radeon ProRender Performance - Villa Render (NVIDIA TITAN RTX)

We’ve seen multiple times in the past that NVIDIA’s graphics cards are super-strong performers in AMD’s Radeon ProRender, and there hasn’t been an exception made in this testing. While the TITAN RTX didn’t get much of a lead in the villa render against the RTX 2080 Ti, it gained handsomely with the car render.

Unfortunately, ProRender is one application that’s going to skip out of our test suite for a little bit, since the latest version of the plugin (2.5) does not cooperate with our projects, either for CPU or GPU rendering. We’re hoping that with SIGGRAPH up ahead, a 2.6 (or beyond) version will be announced there, and miraculously fix our issues.


LuxMark Thumbnail
LuxMark Performance - Hotel Score (NVIDIA TITAN RTX)
LuxMark Performance - LuxBall Score (NVIDIA TITAN RTX)

We recently wrote that AMD may be EOL’ing its Radeon VII, and in some ways, it’s really too bad the company never marketed it towards the professional market instead. In many ways, it’s a very powerful GPU, and it really struts its stuff with the LuxBall render test here.

When the going gets tougher, such as with the more complex Hotel render, the VII falls behind a bit, giving a strong lead to the top gaming RTX and TITAN RTX cards.

Encoding: Adobe Premiere Pro & MAGIX Vegas

Adobe Premiere Pro CC 2019

Adobe Premiere Pro CC 2019 - Techgage Workstation GPU Test
Adobe Premiere Pro AVC Performance - 8K RED Encode (NVIDIA TITAN RTX)
Adobe Premiere Pro AVC Performance - YouTube Encode (NVIDIA TITAN RTX)

As we’ve mentioned before, we intend to expand our Premiere Pro testing when time allows us to tackle developing better tests, and also figure out the ins and outs of the software. Renderers are quite a bit easier to benchmark in comparison to video editors and image manipulators – we can say that with confidence.

Our results with simpler encodes shows diminishing returns after a certain point, which is ultimately great for the end-consumer who can get by with a lower-end GPU, but perhaps not those who were hoping to get a bit more of an encoding benefit out of their higher-end GPU. We’re currently retesting GPUs for an upcoming look at ProViz performance on the latest gaming cards to come out, so we look forward to seeing how AMD’s new Radeon Multi-Media Engine stacks up.

MAGIX Vegas Pro 16

MAGIX Vegas Pro 16 - Median FX Filter Test
MAGIX Vegas Pro 16 - LUT FX GPU Encode Performance (NVIDIA TITAN RTX)
MAGIX Vegas Pro 16 - Median FX GPU Encode Performance (NVIDIA TITAN RTX)

We posted a dedicated look at Vegas Pro a couple of months ago, and ultimately found that it’s not too favorable towards NVIDIA graphics cards. We discovered that while performance can really lack on certain NVIDIA GPUs out-of-the-box, merely adding a profile for the software inside of NVIDIA’s Control Panel can help quite a bit. It’s a bit of a MAGIX trick, you might say.

Ultimately, the TITAN RTX is a powerful GPU, but neither Premiere Pro or Vegas make it obvious.

Viewport: SolidWorks, CATIA, Siemens NX, Creo, Blender, 3ds Max & Maya

GIGABYTE Summer 2016 Event - Artesa Winery Wine Aging Room
SPECviewperf 13: Dassault Systèmes CATIA
GIGABYTE Summer 2016 Event - Artesa Winery Wine Aging Room
SPECviewperf 13: Dassault Systèmes SolidWorks


Dassault Systemes SolidWorks 1080p Viewport Performance (NVIDIA TITAN RTX)
Dassault Systemes SolidWorks 4K Viewport Performance (NVIDIA TITAN RTX)

SolidWorks is one of the industry’s most popular CAD/CAE tools, but it doesn’t require a top-end graphics card to run well. Even on modest gaming GPUs, spinning typical models around in the viewport should offer good performance – but you of course don’t want to skimp too much.

SW is one of the more interesting ProViz tools in the market, since it offers a feature exclusive to workstation cards: RealView. This viewport mode adds realistic lighting and shadows to objects, giving a better representation of the final result. We’re not really sure who chooses to keep RealView exclusive to pro cards, but AMD and NVIDIA don’t seem to mind.

Fortunately, TITAN users don’t need to fret that much. While we didn’t have access to the real SolidWorks for this article, we did for one a couple of months ago. There, we found that the TITAN Xp peaked at 30 FPS, which was a far cry from the Quadros and Radeon Pros. 30 FPS is suitable enough for temporary RealView work, however, so for many, the performance hit may not matter.

SW is a great example of software optimized for workstation GPUs, so if you work with the software day in and day out, you’ll probably want to just avoid potential future headache and go full-speed ahead with Quadro.


Dassault Systemes CATIA 1080p Viewport Performance (NVIDIA TITAN RTX)
Dassault Systemes CATIA 4K Viewport Performance (NVIDIA TITAN RTX)

CATIA is a bit more neutral towards gaming cards than SolidWorks is, but only a little. We can see an easy example of detriment to gaming GPUs with the RTX 2080 Ti – aka, the fastest gaming GPU on the planet – being pinned to the middle of the lineup. While the TITAN Xp isn’t touted as a gaming GPU, it falls a fair bit behind the Quadro P6000 at 1080p, and matches up at 4K.

For this generation’s TITAN, the extra performance was enough to help it match the P6000 at the top at 1080p, and leap ahead at 4K. The kind of lead the RTX gets here is interesting, as is the fact that the 1080p and 4K results don’t always follow the same themes. Ultimately, the TITAN RTX once again renders itself in the top spot.

Siemens NX

Siemens NX 1080p Viewport Performance (NVIDIA TITAN RTX)
Siemens NX 4K Viewport Performance (NVIDIA TITAN RTX)

“But can it run Siemens NX?”

If it’s a gaming GPU, we’re going to say “no”. Every single one of the gaming counterparts fell to the floor here, although it’s funny to note that AMD’s offerings somehow double the performance of those NVIDIA’s. But (Awful Performance x 2) does not suddenly make things that great. These results are scores, not frame rates, but it’s still impossible to ignore an $800 Quadro RTX 4000 beating the $1000 RTX 2080 Ti tenfold.

Suffice to say, you don’t want a gaming card for SNX, but you do want either a Quadro, Radeon Pro, or a TITAN. Even the last-gen TITAN Xp delivers incredible performance here, at both resolutions, and the newer RTX version adds a bunch of cherries on top. It simply runs away with the lead here.

PTC Creo

PTC Creo 1080p Viewport Performance (NVIDIA TITAN RTX)
PTC Creo 4K Viewport Performance (NVIDIA TITAN RTX)

It looks like the TITAN RTX has a point to prove with Creo. This is one tool we’re not intimately familiar with, but it clearly loves fast cards. Unlike some of the other suites on this page, workstation cards don’t seem to offer an inherent benefit here over the gaming counterparts, which is great for users of the software. And for those who want the ultimate in performance, the TITAN RTX is once again delivering on that promise, again giving us a 1-2 finish with the RTX 2080 Ti.

Blender Viewport

Blender 1080p Viewport Performance (NVIDIA TITAN RTX)
Blender 1440p Viewport Performance (NVIDIA TITAN RTX)
Blender 4K Viewport Performance (NVIDIA TITAN RTX)

Blender is one of the most neutral design suites on earth, and we’re sure that owes to the fact that an “open” mentality permeates throughout the Blender Foundation. When the software first supported NVIDIA’s Turing architecture, we saw some fantastic gains over the previous-gen cards, and we can see that doesn’t just impact rendering, but also viewport performance. Just look at the performance deltas between the top two NVIDIA cards and the rest of the pack.

There’s not much else to say here, except that we’ll be taking an even deeper look at Blender 2.80 soon, now that the full release is imminent.

3ds Max & Maya

Autodesk 3ds Max Viewport Performance (NVIDIA TITAN RTX)
Autodesk Maya 1080p Viewport Performance (NVIDIA TITAN RTX)
Autodesk Maya 4K Viewport Performance (NVIDIA TITAN RTX)

Like Creo and Blender, 3ds Max and Maya don’t obsess over a requirement for a workstation GPU, with pure performance being the ultimate decider of your experience. Notably though, NVIDIA’s Turing architecture once again exhibits some nice strengths over the previous-gen GPUs. The RTX 4000 outpaces the Pascal-based P5000, and the TITAN RTX gained a 28% performance advantage over the TITAN Xp in Maya’s 4K test.

Medical & Energy

Medical 1080p Viewport Performance (NVIDIA TITAN RTX)
Medical 4K Viewport Performance (NVIDIA TITAN RTX)
Energy 1080p Viewport Performance (NVIDIA TITAN RTX)
Energy 4K Viewport Performance (NVIDIA TITAN RTX)

In both the medical and energy tests, the TITAN RTX dominates at the top, at some points coming well ahead of the GPU occupying the #2 slot. It’s notable that AMD’s Radeon VII performs extremely well in these tests, as well. Admittedly, neither energy or medical markets are often looking at gaming GPUs for their critical work, but this does let us peer into the world of Tesla and Instinct a little bit.

Math: Cryptography, Financial & Scientific

Cryptography, Financial & Scientific Performance

SiSoftware Sandra - Techgage Workstation GPU Test
Sandra Cryptography (High) GPU Performance (NVIDIA TITAN RTX)
Sandra Cryptography (Higher) GPU Performance (NVIDIA TITAN RTX)

AMD has a way of throwing truly impressive cryptography performance into the mix when you least expect it. With the normal 256+256-bit test, the VII manages to outpace the TITAN RTX, while NVIDIA’s top dog and the RTX 2080 Ti reverse the trend with the heavier 256+512 test.

For its part, AMD’s Vega architecture is really strong for encryption, and so is NVIDIA’s Turing architecture. It’s really impressive what a difference a single generation can make sometimes. We’ve seen the same kind of leaps from the CPU vendors in crypto as well in recent years.

Sandra Financial (FP32 Single-Precision) GPU Performance (NVIDIA TITAN RTX)
Sandra Financial (FP64 Double-Precision) GPU Performance (NVIDIA TITAN RTX)

Because FP64 really isn’t a focus of Quadros and Radeon Pros, we’re not likely going to continue testing Sandra that way in the future, unless of course that reality happens to change somehow. The reason we have FP64 performance right now is obvious once you look at the graphs: the Radeon VII simply dominates the results, thanks to its partially unlocked double-precision.

It’s felt like both AMD and NVIDIA knew that the FP64 buzz on gaming cards wouldn’t last beyond a day, so the green side hasn’t released a double-precision retort, and AMD has been recently rumored to have canned the VII, which means soon, no FP64 option will be alluring to regular folk. But again.. with FP64’s focus on things other than ProViz, its performance holds overall little importance to most reading this article.

In the single-precision test, the TITAN RTX kept comfortably ahead of the rest, save for the RTX 2080 Ti, which places not far behind. The Radeon VII offers a great FP32 result against the cards behind itself, but still manages to sit well behind the 2080 Ti. For FP64, forget about it… the VII is king.

Sandra Scientific (FP32 Single-Precision) GPU Performance (NVIDIA TITAN RTX)
Sandra Scientific (FP64 Double-Precision) GPU Performance (NVIDIA TITAN RTX)

With the scientific result, we’re seeing much of the same as with the financial one, except that the Radeon VII has managed to perform extremely well in the FP32 result here as well, essentially matching the top NVIDIA cards, and of course outperforming them in the scientific test. How many scientists use gaming cards in their labs, we’re not sure, but AMD has the right to boast about some really impressive performance here.

Final Thoughts

We had quite a bit of fun testing out NVIDIA’s TITAN RTX, and we’re presuming it’s not going to be hard to figure out why – and if it is, you probably didn’t look at the test results. As expected, the TITAN RTX delivered in droves across the board, proving to be the fastest card of the bunch overall – with a few exceptions where workstation-specific cards would take the lead.

With its $2,499 price tag, the TITAN RTX clearly isn’t targeting the world at large. It’s instead focusing on delivering the ultimate solution to those working with artificial intelligence and deep-learning, as well as those building the most complex visual projects. With a massive 24GB framebuffer on-board, memory is not likely to be a bottleneck for quite some time.

NVIDIA's TITAN RTX & GeForce RTX 2080 Ti - Back-to-Back

Memory is important in the ProViz market, but it seems even more important on the deep-learning side, since the datasets that are crunched through are simply enormous, and the more memory on tap, the more efficiently batch jobs could be cut through. In our previous testing and research, we found that NVIDIA’s Volta and Turing architectures are efficient at reducing deep-learning memory requirements, which means that 12GB on a Turing card could get you further than 12GB on a Pascal card. So with 24GB on the TITAN RTX, you’re going to have a lot of room to work with.

A big framebuffer is useful, but so too are the special RTX features found on the entire GeForce and Quadro RTX lines. On one hand, we have RT cores, which accelerates ray tracing workloads, both in gaming and in design. On the other, we have Tensor cores, which can dramatically improve deep-learning performance, as we discussed on the first page of this review. For ProViz, that can improve things like AI denoising performance in renders.

Unfortunately, we didn’t really dabble much at all into ray tracing or deep-learning in this article, but we do plan to explore its performance more, as we’re now discovering some benchmarks that would be easy enough to jump into and generate results for. A deeper look at this performance will likely come along with expanded ProViz coverage for Linux.

As mentioned before, we’re hesitant to call TITAN RTX an “ultimate” ProViz card, even though it certainly feels like it most of the time. The only reason we can’t is because it still lacks some of the Quadro optimizations. To get those, as well as an ECC bonus, you’d be opting for the $4,000 Quadro RTX 6000. For the heaviest of memory requirements, there’s also a Quadro RTX 8000, sporting an enormous 48GB framebuffer, retailing for $5,500.


For regular creators and modest research labs, gaming cards are going to prove to be an excellent value depending on the required workloads. For those heavier and more serious workloads, the TITAN RTX is a bit of a no-brainer. It carries an obvious cost premium, but it also boasts a massive framebuffer, as well as additional RTX accelerators that help bolster its performance in key workloads, like deep-learning and ray traced rendering. It’s a no-nonsense GPU for getting important work done.

In addition to future coverage of deep-learning perf, we’re also hoping to dive into RTX support in popular software suites as well, and report more on our experiences there. While not tested in this article, we are planning on tackling DaVinci Resolve performance soon, which with the current Studio 16 version supports RTX for AI touch-ups to video – and based on what we saw when we visited NVIDIA’s HQ in Taiwan during Computex, it’s really impressive technology.

If you’re still unsure about anything related to the TITAN RTX, or we failed to cover something here, please feel free to leave a comment. We will note that we will have another look at ProViz performance coming up soon, with newer drivers, although we haven’t witnessed any significant changes from the testing here. We do however have the latest SUPER and RX 5700 Navi GPUs included, so keep an eye out for that.

Copyright © 2005-2019 Techgage Networks Inc. - All Rights Reserved.