Compute & Synthetics

Shifting gears, we'll look at the compute and synthetic aspects of the GTX 1660 Ti.

Beginning with CompuBench 2.0, the latest iteration of Kishonti's GPU compute benchmark suite offers a wide array of different practical compute workloads, and we’ve decided to focus on level set segmentation, optical flow modeling, and N-Body physics simulations.

Compute: CompuBench 2.0 - Level Set Segmentation 256

Compute: CompuBench 2.0 - N-Body Simulation 1024K

Compute: CompuBench 2.0 - Optical Flow

On paper, the GTX 1660 Ti looks to provide around 85% of the RTX 2060's compute and shading throughput; for Compubench, we see it achieving around 82% of the latter's performance.

Moving on, we'll also look at single precision floating point performance with FAHBench, the official Folding @ Home benchmark. Folding @ Home is the popular Stanford-backed research and distributed computing initiative that has work distributed to millions of volunteer computers over the internet, each of which is responsible for a tiny slice of a protein folding simulation. FAHBench can test both single precision and double precision floating point performance, with single precision being the most useful metric for most consumer cards due to their low double precision performance.

Compute: Folding @ Home Single Precision

Next is Geekbench 4's GPU compute suite. A multi-faceted test suite, Geekbench 4 runs seven different GPU sub-tests, ranging from face detection to FFTs, and then averages out their scores via their geometric mean. As a result Geekbench 4 isn't testing any one workload, but rather is an average of many different basic workloads.

Compute: Geekbench 4 - GPU Compute - Total Score

In lieu of Blender, which has yet to officially release a stable version with CUDA 10 support, we have the LuxRender-based LuxMark (OpenCL) and V-Ray (OpenCL and CUDA).

Compute/ProViz: LuxMark 3.1 - LuxBall and Hotel

Compute/ProViz: V-Ray Benchmark 1.0.8

We'll also take a quick look at tessellation performance.

Synthetic: TessMark, Image Set 4, 64x Tessellation

Finally, for looking at texel and pixel fillrate, we have the Beyond3D Test Suite. This test offers a slew of additional tests – many of which we use behind the scenes or in our earlier architectural analysis – but for now we’ll stick to simple pixel and texel fillrates.

Synthetic: Beyond3D Suite - Pixel Fillrate

Synthetic: Beyond3D Suite - Integer Texture Fillrate (INT8)

Synthetic: Beyond3D Suite - Floating Point Texture Fillrate (FP32)

The practically identical pixel fill rates for the GTX 1660 Ti and RTX 2060 might seem odd at first blush, but it is an entirely expected result as both GPUs have the same number of ROPs, similar clockspeeds, same GPC/TPC setup, and similar memory configurations. And being the same generation/architecture, there aren't any changes or improvements to DCC. In the same vein, the RTX 2060 puts up a 25% higher texture fillrate over the GTX 1660 Ti as a consequence of having 25% more TMUs (96 vs 120).

 

Total War: Warhammer II Power, Temperature, and Noise
POST A COMMENT

157 Comments

View All Comments

  • Yojimbo - Saturday, February 23, 2019 - link

    My guess is that in the next (7 nm) generation, NVIDIA will create the RTX 3050 to have a very similar number of "RTX-ops" (and, more importantly, actual RTX performance) as the RTX 2060, thereby setting the capabilities of the RTX 2060 as the minimum targetable hardware for developers to apply RTX enhancements for years to come. Reply
  • Yojimbo - Saturday, February 23, 2019 - link

    I wish there were an edit button. I just want to say that this makes sense, even if it eats into their margins somewhat in the short term. Right now people are upset over the price of the new cards. But that will pass assuming RTX actually proves to be successful in the future. However, if RTX does become successful but the people who paid money to be early adopters for lower-end RTX hardware end up getting squeezed out of the ray-tracing picture that is something that people will be upset about which NVIDIA wouldn't overcome so easily. To protect their brand image, NVIDIA need a plan to try to make present RTX purchases useful in the future being that they aren't all that useful in the present. They can't betray the faith of their customers. So with that in mind, disabling perfectly capable RTX hardware on lower end hardware makes sense. Reply
  • u.of.ipod - Friday, February 22, 2019 - link

    As a SFFPC (mITX) user, I'm enjoying the thicker, but shorter, card as it makes for easier packaging.
    Additionally, I'm enjoying the performance of a 1070 at reduced power consumption (20-30w) and therefore noise and heat!
    Reply
  • eastcoast_pete - Friday, February 22, 2019 - link

    Thanks! Also a bit disappointed by NVIDIA's continued refusal to "allow" a full 8 GB VRAM in these middle-class cards. As to the card makers omitting the VR required USB3 C port, I hope that some others will offer it. Yes, it will add $20-30 to the price, but I don't believe I am the only one who's like the option to try some VR gaming out on a more affordable card before deciding to start saving money for a full premium card. However, how is VR on Nvidia with 6 GB VRAM? Is it doable/bearable/okay/great? Reply
  • eastcoast_pete - Friday, February 22, 2019 - link

    "who'd like the option". Google keyboard, your autocorrect needs work and maybe some grammar lessons. Reply
  • Yojimbo - Friday, February 22, 2019 - link

    Wow, is a USB3C port really that expensive? Reply
  • GreenReaper - Friday, February 22, 2019 - link

    It might start to get closer once you throw in the circuitry needed for delivering 27W of power at different levels, and any bridge chips required. Reply
  • OolonCaluphid - Friday, February 22, 2019 - link

    >However, how is VR on Nvidia with 6 GB VRAM? Is it doable/bearable/okay/great?

    It's 'fine' - the GTX 1050ti is VR capable with only 4gb VRAM, although it's not really advisable (see Craft computings 1050ti VR assessment on youtube - it's perfectly useable and a fun experience). The RTX 2060 is a very capable VR GPu, with 6gb VRAm. It's not really VRAM that is critical in VR GPU performance anyway - more the raw compute performance in rendering the same scene from 2 viewpoints simultaneously. So, I'd assess that the 1660ti is a perfectly viable entry level VR GPU. Just don't expect miracles.
    Reply
  • eastcoast_pete - Saturday, February 23, 2019 - link

    Thanks for the info! About the miracles: Learned a long time ago not to expect those from either Nvidia or AMD - fewer disappointments this way. Reply
  • cfenton - Friday, February 22, 2019 - link

    You don't need a USB C port for VR, at least not with the two major headsets on the market today. Reply

Log in

Don't have an account? Sign up now