Daz Studio Iray - Rendering Hardware Benchmarking
RayDAnt
Posts: 1,135
Table of Contents
- Daz Studio Iray Plugin Version History
- Iray 2017.0.1
- Iray 2018.1.3
- Iray RTX 2019.1.1
- Iray RTX 2019.1.3
- Iray RTX 2019.1.4
- Iray RTX 2019.1.5
- Iray RTX 2019.1.6
- Iray 2020.0.1
- Iray 2020.1.1
- Iray 2020.1.1 (4.14.x)
- Iray 2020.1.2
- Iray 2020.1.3
- Iray 2020.1.4
- Iray 2020.1.5
- Iray 2020.1.6
- Iray 2021.0.2
- Iray 2021.0.3
- Iray 2021.0.4
- Iray 2021.1.0 beta (build 349500.5279)
- Iray 2021.1.0
- Iray 2021.1.1
- Iray 2021.1.2
- Iray 2021.1.3
- Iray 2021.1.6
- Iray 2022.0.1
- Iray 2022.1.1 (build 363600.1657)
- Iray 2022.1.2
- Iray 2022.1.4
- Iray 2022.1.6
- Iray 2022.1.7
- Iray 2023.1.0
- Iray 2023.1.2
- Iray 2023.1.3
- Iray 2023.1.4
- Summarized Benchmark Results <<< see here first!
- Individual Benchmark Results
- Nvidia GPUs
- Maxwell Architecture (Quadro Mxxx & GeForce 900 series)
- Pascal Architecture (Quadro Pxxx & GeForce 10 series)
- Volta Architecture (Quadro GVxxx series)
- Turing Architecture (Quadro Tx00/Tx000/RTX x000 & GeForce 16/20 series)
- Ampere Architecture (RTX Axx00 & GeForce 30 series)
- Ada Lovelace Architecture (RTX xx00 & GeForce 40 series)
- Intel CPUs
- AMD CPUs
- GPU/CPU Combinations (deprecated)
- Nvidia GPUs
- Contributing Your Own Results To This Thread <<< then go here!
- About The Benchmarking Scene Used
- Design Factors
- The End Result <<< find benchmarking scene here!
- Tips & Tricks
- Additional Resources
1. Daz Studio Iray Plugin Version History
1.1 Iray 2017.0.1
Found in the following Daz Studio release(s):
- Daz Studio 4.10.0.123
1.2 Iray 2018.1.3
Found in the following Daz Studio release(s):
- Daz Studio 4.11.0.383 Beta
- Daz Studio 4.11.0.383
1.3 Iray RTX 2019.1.1
Found in the following Daz Studio release(s):
- Daz Studio 4.12.0.033 Beta
1.4 Iray RTX 2019.1.3
Found in the following Daz Studio release(s):
- Daz Studio 4.12.0.042 Beta
- Daz Studio 4.12.0.047 Beta
- Daz Studio 4.12.0.060 Beta
- Daz Studio 4.12.0.067 Beta
- Daz Studio 4.12.0.073 Beta
- Daz Studio 4.12.0.083 Beta
- Daz Studio 4.12.0.085 Beta
- Daz Studio 4.12.0.086
1.5 Iray RTX 2019.1.4
Found in the following Daz Studio release(s):
- Daz Studio 4.12.1.016 Beta
1.6 Iray RTX 2019.1.5
Found in the following Daz Studio release(s):
- Daz Studio 4.12.1.040 Beta
- Daz Studio 4.12.1.055 Beta
1.7 Iray RTX 2019.1.6
Found in the following Daz Studio release(s):
- Daz Studio 4.12.1.076 Beta
1.8 Iray 2020.0.1
Found in the following Daz Studio release(s):
- Daz Studio 4.12.1.109 Beta
- Daz Studio 4.12.1.115 Beta
- Daz Studio 4.12.1.117 Beta
- Daz Studio 4.12.1.117
- Daz Studio 4.12.2.006 Beta
1.9 Iray 2020.1.1
Found in the following Daz Studio release(s):
- Daz Studio 4.12.2.050 Beta
- Daz Studio 4.12.2.054 Beta
- Daz Studio 4.12.2.060 Beta
1.10 Iray 2020.1.1 (4.14.x)
Found in the following Daz Studio release(s):
- Daz Studio 4.14.0.008 Beta
- Daz Studio 4.14.0.008
- Daz Studio 4.14.0.010 Beta
- Daz Studio 4.14.0.010
1.11 Iray 2020.1.2
Found in the following Daz Studio release(s):
- Daz Studio 4.14.1.022 Beta
- Daz Studio 4.14.1.028 Beta
1.12 Iray 2020.1.3
Found in the following Daz Studio release(s):
- Daz Studio 4.14.1.038 Beta
- Daz Studio 4.15.0.002 Beta
- Daz Studio 4.15.0.002
1.13 Iray 2020.1.4
Found in the following Daz Studio release(s):
- Daz Studio 4.15.0.009 Beta
- Daz Studio 4.15.0.012 Beta
- Daz Studio 4.15.0.013 Beta
1.14 Iray 2020.1.5
Found in the following Daz Studio release(s):
- Daz Studio 4.15.0.014 Beta
1.15 Iray 2020.1.6
Found in the following Daz Studio release(s):
- Daz Studio 4.15.0.025 Beta
- Daz Studio 4.15.0.026 Beta
- Daz Studio 4.15.0.028 Beta
- Daz Studio 4.15.0.029 Beta
- Daz Studio 4.15.0.030 Beta
- Daz Studio 4.15.0.030
- Daz Studio 4.15.1.072 Beta
- Daz Studio 4.15.1.084 Beta
- Daz Studio 4.16.0.003 - regressed versus the beta release channel
1.16 Iray 2021.0.2
Found in the following Daz Studio release(s):
- Daz Studio 4.15.1.091 Beta
1.17 Iray 2021.0.3
Found in the following Daz Studio release(s):
- Daz Studio 4.15.1.096 Beta
- Daz Studio 4.16.1.002 Beta
1.18 Iray 2021.0.4
Found in the following Daz Studio release(s):
- Daz Studio 4.16.1.006 Beta
1.19 Iray 2021.1.0 beta (build 349500.5279)
Found in the following Daz Studio release(s):
- Daz Studio 4.16.1.017 Beta
- Daz Studio 4.16.1.021 Beta
- Daz Studio 4.16.1.031 Beta
1.20 Iray 2021.1.0
Found in the following Daz Studio release(s):
- Daz Studio 4.16.1.034 Beta
- Daz Studio 4.16.1.040 Beta
- Daz Studio 4.16.1.043 Beta
- Daz Studio 4.16.1.047 Beta
- Daz Studio 4.20.0.002 Beta
- Daz Studio 4.20.0.002
- Daz Studio 4.20.0.003 Beta
- Daz Studio 4.20.0.004 Beta
- Daz Studio 4.20.0.005 Beta
- Daz Studio 4.20.0.006 Beta
- Daz Studio 4.20.0.008 Beta
- Daz Studio 4.20.0.011 Beta
- Daz Studio 4.20.0.017 Beta
- Daz Studio 4.20.0.017
1.21 Iray 2021.1.1
Found in the following Daz Studio release(s):
- None - first public release of this subversion already upgraded to next Iray version
1.22 Iray 2021.1.2
Found in the following Daz Studio release(s):
- Daz Studio 4.20.1.034 Beta
1.23 Iray 2021.1.3
Found in the following Daz Studio release(s):
- Daz Studio 4.20.1.038 Beta
- Daz Studio 4.20.1.043 Beta
Iray 2021.1.4
- Internal-only release
1.24 Iray 2021.1.6
- Daz Studio 4.20.1.058 Beta
- Daz Studio 4.20.1.078 Beta
- Daz Studio 4.20.1.088 Beta
- Daz Studio 4.20.1.091 Beta
- Daz Studio 4.21.0.005 Beta
1.25 Iray 2022.0.1
- Daz Studio 4.21.1.013 Beta
1.26 Iray 2022.1.1 (build 363600.1657)
- Daz Studio 4.21.1.026 Beta
1.27 Iray 2022.1.2
- Daz Studio 4.21.1.029 Beta
1.28 Iray 2022.1.4
- Daz Studio 4.21.1.041 Beta
Iray 2022.1.5
- No public releases
1.29 Iray 2022.1.6
- Daz Studio 4.21.1.045 Beta
1.30 Iray 2022.1.7
- Daz Studio 4.21.1.048 Beta
- Daz Studio 4.21.1.080 Beta
- Daz Studio 4.21.1.104 Beta
- Daz Studio 4.21.1.108 Beta
- Daz Studio 4.22.0.001 Beta
- Daz Studio 4.22.0.007 Beta
- Daz Studio 4.22.0.009 Beta
- Daz Studio 4.22.0.010 Beta
- Daz Studio 4.22.0.012 Beta
- Daz Studio 4.22.0.015 Beta
- Daz Studio 4.22.0.016
1.31 Iray 2023.1.0
- Daz Studio 4.22.1.041 Beta
- Daz Studio 4.22.1.054 Beta
- Daz Studio 4.22.1.058 Beta
1.32 Iray 2023.1.2
- Daz Studio 4.22.1.074 Beta
1.33 Iray 2023.1.3
- Daz Studio 4.22.1.088 Beta
- Daz Studio 4.22.1.099 Beta
1.34 Iray 2023.1.4
- Daz Studio 4.22.1.110 Beta
- Daz Studio 4.22.1.123 Beta
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
2. Summarized Benchmark Results
notes:
3. Benchmark Results
3.1 Nvidia GPUs
3.1.1 Maxwell Architecture (Quadro Mxxx & GeForce 900 series)
GTX 960:
GTX 980Ti:
GTX 970:
3.1.2 Pascal Architecture (Quadro Pxxx & GeForce 10 series)
Titan Xp:
Titan X:
GTX 1080Ti:
GTX 1080:
GTX 1070Ti:
GTX 1070:
GTX 1050Ti:
GTX 1050 2GB:
3.1.3 Volta Architecture (Quadro GVxxx series)
Titan V:
3.1.4 Turing Architecture (Quadro Tx00/Tx000/RTX x000 & GeForce 16/20 series)
Quadro RTX 4000:
Titan RTX:
RTX 2080Ti:
RTX 2080 SUPER:
RTX 2080:
RTX 2070 SUPER:
RTX 2070:
RTX 2060 SUPER:
RTX 2060:
GTX 1660Ti:
GTX 1660:
GTX 1650:
3.1.5 Ampere Architecture (RTX Axx00 & GeForce 30 series)
RTX A6000:
RTX A5000:
RTX 3090:
RTX 3080:
RTX 3060Ti:
3.1.6 Ada Lovelace Architecture (RTX xx00 & GeForce 40 series)
RTX 6000:
RTX 4090:
RTX 4080:
3.2 Intel CPUs
i9-9900K:
i7-8700K:
i9-7980XE:
3.3 AMD CPUs
Threadripper 3970X:
Threadripper 3960X:
Ryzen 7 3700X:
Ryzen 7 2700X:
Ryzen 5 2600X:
Threadripper 1950X:
3.4 CPU/GPU Combinations
Titan RTX + Titan RTX + Threadripper 3970X:
Titan RTX + Titan RTX:
Titan RTX + i7-8700K:
Titan V + Titan V + Titan Xp + Titan Xp:
Titan V + Titan V:
RTX 2080Ti + RTX 2080Ti + RTX 2080Ti + RTX 2080Ti:
RTX 2080Ti + RTX 2080Ti:
RTX 2080Ti + GTX 1080Ti:
RTX 2080Ti + Quadro RTX 4000:
RTX 2080Ti + GTX 980Ti:
RTX 2080Ti + Ryzen 7 2700X:
RTX 2080 SUPER + RTX 2080 SUPER + Threadripper 3970X:
RTX 2080 SUPER + RTX 2080 SUPER:
RTX 2080 SUPER + GTX 1070:
RTX 2080 + GTX 1080:
RTX 2080 + GTX 1070:
RTX 2070 + GTX 1070:
RTX 2060 + RTX 2060 + FX-8370:
RTX 2060 + RTX 2060:
Titan Xp + Titan Xp + Titan X (Pascal):
GTX 1080Ti + GTX 1080Ti:
GTX 1080Ti + GTX 980Ti:
GTX 1080 + GTX 1080 + Ryzen 7 3700X:
GTX 1080 + GTX 1080:
GTX 1070 + GTX 1070:
GTX 1070 + GTX 960 + Ryzen 5 2600X:
GTX 1070 + GTX 960:
GTX 980Ti + GTX 980Ti:
* Function of OptiX Prime acceleration checkbox uncertain due to changes in raytracing handling by low-level APIs.
(UPDATED) 08/03/19
System Configuration
Operating System: Windows 10 64bits Pro version 1903 build
Nvidia Drivers Version: 431.36 GRD
System/Motherboard: MSI Z270 SLI Plus
CPU: Intel i7-6700K @ 4.2 Ghz
GPU: GTX 1070 x2 @ 1980Mhz
System Memory: G-Skill 32GB DDR4 @ 2200 Mhz
Benchmark Results - GTX 1070 x2 Only
Daz Studio Version: 4.12.0.033 Beta 64-bit
Optix Prime Acceleration: No
Total Rendering Time: 6 minutes 14.55 seconds
CUDA device 1 (GeForce GTX 1070): 881 iterations, 4.438s init, 367.075s render
CUDA device 0 (GeForce GTX 1070): 919 iterations, 3.736s init, 367.699s render
Iteration Rate: (1800 / 367.387) = 4.899 iterations per second
System Overhead: ((0 + 373 + 14.55) - 367.387) = (373.1455 - 367.387) = 5.7585 seconds
4. Contributing Your Own Results To This Thread:
4.1 Before You Begin
Make sure the following Daz Studio settings are set to their defaults:
4.2 Running The Benchmark
Step 1: Close all running programs (including any open instances of Daz Studio) before continuing (failing to do so may artificially decrease your measured rendering performance.)
Step 2: Download this benchmarking scene and open it in Daz Studio.
Step 3: Go to Render Settings [pane] > Advanced > Hardware > Photoreal Devices and select only the device(s) you currently wish to test.
Step 4: Daz Studio 4.12.0.086 or earlier only In the same settings window make sure to select/deselect OptiX Prime Acceleration depending on your preference for this specific test run.
Step 5: Press Render and wait until the scene fully finishes rendering.
Step 6: optional Save (with a unique file name) the final rendered image for later reference.
Step 7: Close Daz Studio completely before continuing (failing to do so may make it impossible for you to fully report your benchmarking results.)
4.3 Reporting Your Results
Step 1: Start a new post in this thread and paste the following template into it:
System/Motherboard: BRAND MODEL
CPU: BRAND MODEL @ SPEED/stock (if left at defaults)
GPU: BRAND MODEL @ SPEED/stock (if left at defaults)
System Memory: BRAND MODEL CAPACITY TYPE @ SPEED
OS Drive: BRAND MODEL CAPACITY
Asset Drive: BRAND MODEL CAPACITY/Same (if same as OS)
Power Supply: BRAND MODEL WATTAGE
Operating System: EDITION VERSION BUILD
Nvidia Drivers Version: VERSION MODE
Daz Studio Version: VERSION BITS
Optix Prime Acceleration: STATE (Daz Studio 4.12.1.086 or earlier only)
Benchmark Results
DAZ_STATS
IRAY_STATS
Iteration Rate: (DEVICE_ITERATION_COUNT / DEVICE_RENDER_TIME) iterations per second
Loading Time: ((TRT_HOURS * 3600 + TRT_MINUTES * 60 + TRT_SECONDS) - DEVICE_RENDER_TIME) seconds
Step 2: Fill in System Configuration with your info. If specific branding information is unavailable for a particular component (eg. unbranded SSDs/RAM sticks in a pre-built system) simply omit it.
Step 3: Navigate to the Daz Studio log file (found in "AppData\Roaming\DAZ 3D\Studio4\log.txt" or "AppData\Roaming\DAZ 3D\Studio4 Public Build\log.txt" for the public beta), scroll to the very end of the file, and look for lines of text resembling the following:
DAZ_STATS:
IRAY_STATS:
Step 4: Copy these lines and replace DAZ_STATS and IRAY_STATS under Benchmark Results in the template with them. Then take special care in calculating the correct values for Iteration Rate and Loading Time. These two statistics are essential for understanding how your system's rendering performance compares to others' - not Total Rendering Time itself.
Total Rendering Time, because of the way it is computed by Daz Studio, is an invalid statistic for measuring rendering performance. Especially on systems with high-performance rendering hardware.
Step 5: optional Attach the rendered image from your benchmark run to your post.
Step 6: Submit your post and check back (by seeing which post this note under Table of Contents currently leads to) for it to be incorporated into the Performance Tables at the beginning of this thread (If you really want to help speed the update process along, copy the relevant tables from the beginning of this thread, incorporate your results, and then post the updated tables in the same post as your original benchmark results.)
5. About The Benchmarking Scene Used
5.1 Design Factors
In order to present the most robust benchmarking scene possible, the following key factors were first identified (as part of an extensive study on the relative advantages/pitfalls of a variety of existing DS/Iray benchmarking scenes freely available at the time) and then used as design paramaters for the creation of the scene used to generate all of this thread's relative performance data. These factors are:
5.2 The End Result
>>> RayDAnt_DS_Iray_Benchmark_2019A_r4.duf <<< download here!
6. Tips & Tricks
6.1 Submitting multiple benchmark results at the same time
If you want to contribute multiple sets of results from the same computer at the same time (eg. two back-to-back runs illustrating the performance difference between OptiX Prime On and OptiX Prime Off) format your post like this:
System/Motherboard: BRAND MODEL
CPU: BRAND MODEL @ SPEED/stock (if left at defaults)
GPU: BRAND MODEL @ SPEED/stock (if left at defaults)
System Memory: BRAND MODEL CAPACITY TYPE @ SPEED
OS Drive: BRAND MODEL CAPACITY
Asset Drive: BRAND MODEL CAPACITY/Same (if same as OS Drive)
Power Supply: BRAND MODEL WATTAGE
Operating System: EDITION VERSION BUILD
Nvidia Drivers Version: VERSION MODE
Daz Studio Version: VERSION BITS
Benchmark Results
Optix Prime Acceleration: On
DAZ_STATS
IRAY_STATS
Iteration Rate: (DEVICE_ITERATION_COUNT / DEVICE_RENDER_TIME) iterations per second
Loading Time: ((TRT_HOURS * 3600 + TRT_MINUTES * 60 + TRT_SECONDS) - DEVICE_RENDER_TIME) seconds
Optix Prime Acceleration: Off
DAZ_STATS
IRAY_STATS
Iteration Rate: (DEVICE_ITERATION_COUNT / DEVICE_RENDER_TIME) iterations per second
Loading Time: ((TRT_HOURS * 3600 + TRT_MINUTES * 60 + TRT_SECONDS) - DEVICE_RENDER_TIME) seconds
6.2 Submitting benchmark results for multi-GPU and/or CPU combinations
Use the following template when submitting results for CPU + GPU rendering or GPU1 + GPU2 + etc... in multi-GPU systems:
System/Motherboard: BRAND MODEL
CPU: BRAND MODEL @ SPEED/Stock (if left at defaults)
GPU: GPU1 BRAND MODEL @ SPEED/stock (if left at defaults), GPU2 BRAND MODEL @ SPEED/stock (if left at defaults), ...
System Memory: BRAND MODEL CAPACITY TYPE @ SPEED
OS Drive: BRAND MODEL CAPACITY
Asset Drive: BRAND MODEL CAPACITY/Same (if same as OS)
Power Supply: BRAND MODEL WATTAGE
Operating System: EDITION VERSION BUILD
Nvidia Drivers Version: VERSION MODE
Daz Studio Version: VERSION BITS
Optix Prime Acceleration: STATE (Daz Studio 4.12.1.086 or earlier only)
Benchmark Results
DAZ_STATS
IRAY_STATS
Rendering Performance: [DEVICE_ITERATION_COUNT(sum of all values) / DEVICE_RENDER_TIME_(largest value)] iterations per second
Loading Time: [(TRT_HOURS * 3600 + TRT_MINUTES * 60 + TRT_SECONDS) - DEVICE_RENDER_TIME_(largest value)] seconds
7. Additional Resources
7.1 Other Daz Studio user-generated benchmarking efforts
7.2 Official Daz Studio development information
7.3 Official Nvidia Iray development Information
OK Ill play
System Configuration
System/Motherboard: GIGABYTE B450 AORUS PRO WIFI-CF
CPU: AMD Ryzen 5 2600X @ 4.0 ghz
GPU: MSI GTX 1070 Armor OC/Aero OC @ 1648mhz clock 2230mhz memory, GIGABYTE GeForce GTX 960 OC 4GB @ 1276mhz clock 1810mhz memory MSI RTX 2080 Super Gaming X Trio
System Memory: Corsairw DDR4-3200 8gb x2 gskill DDR4-3200 8gb x2
OS Drive: Samsung 960 EVO 250gb
Asset Drive: Seagate 2TB 7200 rpm
Operating System: Microsoft Windows 7 Ultimate (x64) Build 7601 (Service Pack 1)
Nvidia Drivers Version: 431.36
Daz Studio Version: 4.12.0.33 x64
Both GPU and CPU
Optix Prime Acceleration: on
TRT_STATS 9 minutes 12.87 seconds
DEVICE_STATS (GeForce GTX 1070): 1131 iterations, 6.028s init, 544.168s render
(GeForce GTX 960): 445 iterations, 7.191s init, 543.245s render
CPU: 224 iterations, 3.912s init, 546.102s render
Iteration Rate: 3.307 iterations per second
System Overhead: 6.768 seconds
Optix Prime Acceleration: Off
TRT_STATS 8 minutes 27.25 seconds
DEVICE_STATS (GeForce GTX 1070): 1161 iterations, 7.172s init, 496.600s
(GeForce GTX 960): 444 iterations, 6.061s init, 497.486s
CPU: 195 iterations, 3.975s init, 499.952s
Iteration Rate: 3.619 iterations per second
System Overhead: 10.65 seconds
Both GPU no CPU
Optix Prime Acceleration: on
TRT_STATS 8 minutes 32.69 seconds
DEVICE_STATS (GeForce GTX 1070): 1294 iterations, 4.460s init, 505.825s render
(GeForce GTX 960): 506 iterations, 5.339s init, 504.821s render
Iteration Rate: 3.560 iterations per second
System Overhead: 6.865 seconds
Optix Prime Acceleration: Off
TRT_STATS 8 minutes 32.89 seconds
DEVICE_STATS (GeForce GTX 1070): 1294 iterations, 4.733s init, 504.448s
(GeForce GTX 960): 506 iterations, 4.953s init, 504.705s
Iteration Rate: 3.567 iterations per second
System Overhead: 8.442 seconds
CPU only
Optix Prime Acceleration: on
TRT_STATS 1 hours 1 minutes 53.71 seconds
DEVICE_STATS CPU: 1800 iterations, 3.933s init, 3706.529s render
Iteration Rate: 0.485 iterations per second
System Overhead: 3660 seconds
1070 only
Optix Prime Acceleration: on
TRT_STATS 11 minutes 33.74 seconds
DEVICE_STATS (GeForce GTX 1070): 1800 iterations, 4.181s init, 686.323s render
Iteration Rate: 2.622 iterations per second
System Overhead: 7.417 seconds
Optix Prime Acceleration: Off
TRT_STATS 11 minutes 32.81 seconds
DEVICE_STATS (GeForce GTX 1070): 1800 iterations, 4.182s init, 685.427s
Iteration Rate: 2.626 iterations per second
System Overhead: 7.383 seconds
960 only
Optix Prime Acceleration: On
TRT_STATS 28 minutes 53.53 seconds
DEVICE_STATS (GeForce GTX 960): 1800 iterations, 4.320s init, 1725.972s render
Iteration Rate: 1.042 iterations per second
System Overhead: 7.028 seconds
RTX 2080 Super Gaming X Trio
2080 only
Optix Prime Acceleration: On
2019-09-17 16:59:06.053 Finished Rendering
2019-09-17 16:59:06.077 Total Rendering Time: 4 minutes 56.54 seconds
2019-09-17 17:01:35.930 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Device statistics:
2019-09-17 17:01:35.930 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : CUDA device 0 (GeForce RTX 2080 SUPER): 1800 iterations, 9.843s init, 283.125s render
Iteration Rate: 6.36 iterations per second
Loading Time: 13.42 seconds
2080 +1070
Optix Prime Acceleration: On
2019-09-17 17:49:26.765 Finished Rendering
2019-09-17 17:49:26.781 Total Rendering Time: 3 minutes 41.87 seconds
2019-09-17 18:01:03.860 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : CUDA device 0 (GeForce RTX 2080 SUPER): 1300 iterations, 4.035s init, 214.542s render
2019-09-17 18:01:03.860 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : CUDA device 1 (GeForce GTX 1070): 500 iterations, 5.179s init, 213.341s render
PS: I hate the forum formatting quirks.
System Configuration
System/Motherboard: Gigabyte Z370 Aorus Gaming 7
CPU: Intel i7-8700K @ stock (MCE enabled)
GPU: Nvidia Titan RTX @ stock (watercooled)
System Memory: Corsair Vengeance LPX 32GB DDR4 @ 3000Mhz
OS Drive: Samsung Pro 970 512GB NVME SSD
Asset Drive: Sandisk Extreme Portable SSD 1TB
Operating System: Windows 10 Pro version 1903 build 18362.239
Nvidia Drivers Version: 431.36 GRD WDDM
Benchmark Results - Titan RTX Only
Daz Studio Version: 4.12.0.033 Beta 64-bit
Optix Prime Acceleration: On
Total Rendering Time: 4 minutes 4.9 seconds
CUDA device 0 (TITAN RTX): 1800 iterations, 2.355s init, 239.117s render
Iteration Rate: (1800 / 239.117) = 7.528 iterations per second
Loading Time: ((0 + 240 + 4.9) - 239.117) = (244.9 - 239.117) = 5.783 seconds
Benchmark Results - Titan RTX Only
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: On
Total Rendering Time: 4 minutes 27.50 seconds
CUDA device 0 (TITAN RTX): 1800 iterations, 2.297s init, 262.665s render
Iteration Rate: (1800 / 262.665) = 6.853 iterations per second
Loading Time: ((0 + 240 + 27.50) - 262.665) = (267.50 - 262.665) = 4.835 seconds
Benchmark Results - Titan RTX Only
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: Off
Total Rendering Time: 4 minutes 56.36 seconds
CUDA device 0 (TITAN RTX): 1800 iterations, 3.180s init, 290.852s render
Iteration Rate: (1800 / 290.852) = 6.189 iterations per second
Loading Time: ((0 + 240 + 56.36) - 290.852) = (296.36 - 290.852) = 5.508 seconds
Benchmark Results - Intel i7-8700K Only
Daz Studio Version: 4.12.0.033 64-bit
Optix Prime Acceleration: On
Total Rendering Time: 1 hours 2 minutes 25.52 seconds
CPU: 1800 iterations, 2.358s init, 3740.472s render
Iteration Rate: (1800 / 3740.472) = 0.481 iterations per second
Loading Time: ((3600 + 120 + 25.52) - 3740.472) = (3745.52 - 3740.472) = 5.048 seconds
Benchmark Results - Intel i7-8700K Only
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: On
Total Rendering Time: 1 hours 12 minutes 40.67 seconds
CPU: 1800 iterations, 6.147s init, 4352.048s render
Iteration Rate: (1800 / 4352.048) = 0.414 iterations per second
Loading Time: ((3600 + 720 + 40.67) - 4352.048) = (4360.67 - 4352.048) = 8.622 seconds
Benchmark Results - Intel i7-8700K Only
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: Off
Total Rendering Time: 1 hours 14 minutes 0.55 seconds
CPU: 1800 iterations, 3.306s init, 4434.854s render
Iteration Rate: (1800 / 4434.084) = 0.406 iterations per second
Loading Time: ((3600 + 840 + 0.55) - 4434.854) = (4440.55 - 4434.854) = 5.696 seconds
Benchmark Results - Titan RTX + Intel i7-8700K
Daz Studio Version: 4.12.0.033 64-bit
Optix Prime Acceleration: On
Total Rendering Time: 4 minutes 17.10 seconds
CUDA device 0 (TITAN RTX): 1691 iterations, 2.390s init, 250.967s render
CPU: 109 iterations, 2.162s init, 252.273s render
Iteration Rate: (1800 / 252.273) = 7.135 iterations per second
Loading Time: ((0 + 240 + 17.10) - 252.273) = (257.10 - 252.273) = 4.827 seconds
Benchmark Results - Titan RTX + Intel i7-8700K
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: On
Total Rendering Time: 4 minutes 44.23 seconds
CUDA device 0 (TITAN RTX): 1696 iterations, 2.388s init, 277.534s render
CPU: 104 iterations, 2.022s init, 279.586s render
Iteration Rate: (1800 / 279.586) = 6.438 iterations per second
Loading Time: ((0 + 240 + 44.23) - 279.586) = (284.23 - 279.586) = 4.644 seconds
Benchmark Results - Titan RTX + Intel i7-8700K
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: Off
Total Rendering Time: 4 minutes 56.52 seconds
CUDA device 0 (TITAN RTX): 1695 iterations, 3.257s init, 290.945s render
CPU: 105 iterations, 2.969s init, 290.508s render
Iteration Rate: (1800 / 290.508) = 6.196 iterations per second
Loading Time: ((0 + 240 + 56.52) - 290.508) = (296.52 - 290.508) = 6.012 seconds
LCJD, can you add the rest of the info (check the post before yours) to your post? Just Total Rendering Time isn't enough to caculate a final result for this benchmark. See here for where to find everything.
TheKD, could you add runs for each GPU rendering the scene separately as well? That way we have the whole set.
Also, it looks like the CPU stats are missing from your first test. They should be in your log file right after the info you posted. Here's an example of what it should look like:
Didnt see a CPU line, so thought I missed it and ran it again. Still no CPU line. Maybe it has something to do with that new embree stuff, not sure what's up with that.
The log shows that it is using cpu, but it don't show the stats at the end. NVM, I see what happened. It now appeared like 2 minutes after I closed daz studio, apparently it just takes a few minutes to show up lol. Going to redo it, and add the other 2 renders for each gpu alone.
Awesome! Yeah, it's a little known fact that Iray functions almost completely independently of Daz Studio and oftentimes doesn't fully shut down until well after Daz Studio itself is closed.
ETA: And if you're feeling really adventurous, you could do a run with just your CPU as well (would love to see more AMD CPU benchmarks out there.)
Just to note, that's a lot of system info to ask for. A lot of people don't even know how to find their part info. I very seriously doubt the motherboard, RAM, and hard drives play any role in rendering, and as you will see, there is small problem with the bench I will get into. But here is my info.
System/Motherboard: Asus Z-97
CPU: i5-4690k @3.5ghz
GPU: MSI Gaming 1080ti @ 1999 Mhz + EVGA SC 1080ti @ 1980 Mhz (very minor overclocks)
System Memory: 32 GB DDR3 1866 Mhz
OS Drive: Samsung 860 Evo 1 TB
Asset Drive: WD 4TB Black, Samsung 860 EVO 1TB
Operating System: Win 10 1809
Nvidia Drivers Version: 431.36 Gaming
Daz Studio Version: 4.12 beta
OptiX ON
2019-07-23 00:39:01.369 Total Rendering Time: 4 minutes 1.76 seconds
Optix OFF TEXTURE SHADED
2019-07-23 00:31:57.418 Total Rendering Time: 4 minutes 2.32 seconds
2019-07-23 00:32:04.070 Iray INFO - module:category(IRAY:RENDER): 1.0 IRAY rend info : CUDA device 0 (GeForce GTX 1080 Ti): 903 iterations, 5.932s init, 233.194s render
2019-07-23 00:32:04.070 Iray INFO - module:category(IRAY:RENDER): 1.0 IRAY rend info : CUDA device 1 (GeForce GTX 1080 Ti): 897 iterations, 6.268s init, 233.163s render
So...I beat the RTX Titan with two 1080tis? If I am reading this right, that is not a good sign for RTX after all.
Anyway, here is the issue I think is going one with users. You ask users to have Texture Shaded on for the Viewport. However, this adds the loading time into the data, and loading times can vary based on other hardware. The new 4.12 beta loads much faster than any previous Daz, and this is giving people false impressions that they are getting faster, because this cuts about 30 or so seconds off the time that previous versions might take. I believe it why the person with the 2x 1070's is so happy right now in the SY bench, but their time is reflective of this faster loading more than the actual render speed. Using the Iray Preview equalizes this loading time, and thus eliminates unknown variables and better tests pure GPU rendering speed. That is why sickleyield asked people to use the Iray Viewport.
I had some spare time this morning, so I ran a few rounds of the benchmark. I am really interested to see more RTX results to see what difference enabling it in iray now makes.
I've run the results with 2 graphics cards, both separate and together, in DS 4.11.0.383 and in DS 4.12.33
System Configuration
System/Motherboard: ASUS X99-S
CPU: Intel i7 5960X @3GHz
System Memory: 32GB KINGSTON HYPER-X PREDATOR QUAD-DDR4
OS Drive: Samsung M.2 SSD 960 EVO 250GB
Asset Drive: 2TB WD CAVIAR BLACK SATA 6 Gb/s, 64MB CACHE (7200rpm)
Operating System: Windows 10.0.17 134 Build 17134
Nvidia Drivers Version: 430.64
Benchmark Results – 980Ti Only
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: Yes
Total Rendering Time: 14 minutes 26.48 seconds
CUDA device 1 (GeForce GTX 980 Ti): 1800 iterations, 3.517s init, 859.525s render
Iteration Rate: (1800/859.525) = 2.094 iterations per second
System Overhead: ((0+840+26.48)-859.525) =6.955 seconds
Benchmark Results – 1080Ti Only
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: Yes
Total Rendering Time: 8 minutes 59.70 seconds
CUDA device 0 (GeForce GTX 1080 Ti): 1800 iterations, 3.764s init, 532.429s render
Iteration Rate: (1800/532.429) = 3.381 iterations per second
System Overhead: ((0+480+59.7)-532.429) = 7.271 seconds
Benchmark Results – 980Ti + 1080Ti
Daz Studio Version: 4.11.0.383 64-bit
Optix Prime Acceleration: Yes
Total Rendering Time: 5 minutes 43.5 seconds
CUDA device 0 (GeForce GTX 1080 Ti): 1111 iterations, 3.537s init, 335.483s render
CUDA device 1 (GeForce GTX 980 Ti): 689 iterations, 3.913s init, 335.669s render
Iteration Rate: (1800/335.483) = 5.365 iterations per second
System Overhead: ((0+300+43.5)-335.483) = 8.017 seconds
Benchmark Results – 980Ti Only
Daz Studio Version: 4.12.033 64-bit
Optix Prime Acceleration: Yes
Total Rendering Time: 11 minutes 43.96 seconds
CUDA device 1 (GeForce GTX 980 Ti): 1800 iterations, 4.188s init, 695.982s render
Iteration Rate: (1800/695.982) = 2.586 iterations per second
System Overhead: ((0+660+43.96)-695.982) = 7.978 seconds
Benchmark Results – 1080Ti Only
Daz Studio Version: 4.12.033 64-bit
Optix Prime Acceleration: Yes
Total Rendering Time: 7 minutes 45.3 seconds
CUDA device 0 (GeForce GTX 1080 Ti): 1800 iterations, 3.953s init, 457.279s render
Iteration Rate: (1800/457.279) = 3.936 iterations per second
System Overhead: ((0+420+45.3)-457.279) = 8.021 seconds
Benchmark Results – 980Ti + 1080Ti
Daz Studio Version: 4.12.033 64-bit
Optix Prime Acceleration: Yes
Total Rendering Time: 4 minutes 49.52 seconds
CUDA device 0 (GeForce GTX 1080 Ti): 1084 iterations, 3.760s init, 281.346s render
CUDA device 1 (GeForce GTX 980 Ti): 716 iterations, 4.074s init, 281.616s render
Iteration Rate: (1800/281.616) = 6.392 iterations per second
System Overhead: ((0+240+49.52)-281.616) = 7.904 seconds
With the multiple tests, I found the formatting of the information for this thread a little time-consuming, so I put together a spreadsheet where I could paste the information form the log file and have it show the calculations etc. If this is useful to anyone else, please download below. This is in .xlsx format. It will calculate for two cards at a time - I've left the info for my 980Ti on the first row for reference.
Bench result updated
They play a direct role in scene Loading Time, which is one of the main things this benchmark is designed to measure (other than pure rendering time.)
Yes, you're reading that right. Your particular two 1080ti's slightly beat a single watercooled Titan RTX with full RTX support.
The Titan RTX costs as much as it does because of how much memory it has - not because of how fast it is.
Because it's the Daz Studio default mode.
It adds loading time to the Loading Time statistic and nothing else as long as you include Device Statistics with your results. Hence why this benchmark asks for Device Statistics in addition to Total Rendering Time. As long as both things are reported measured rendering times are unaffected one way or another. The reason why the results you've posted so far:
Seem to indicate what you're saying is because you are missing Device Statistics for your first test case (OptiX On, Iray viewport.) Total Rendering Time can change significantly under different viewport modes. Render time under Device Statistics doesn't. Hence why we need to know both.
ETA: See this post. Something seems to be wrong with your 4.12 Beta installation in general.
System Configuration
System/Motherboard: Asus ROG Strix X570-F
CPU: AMD Ryzen 7 3700x @stock
GPU: Asus GeForce RTX 2080 Ti Turbo @stock
System Memory: Corsair Vengeance LPX 32GB DDR4 @3200Mhz
OS Drive: Samsung SSD 860 QVO 1TB
Asset Drive: same
Operating System: Win 10 Pro version 1903 build 18362.239
Nvidia Drivers Version: Nvidia Studio Driver Version 430.86
Daz Studio Version: 4.12.0.33 64-bit
Optix Prime Acceleration: On
Benchmark Results
TRT_STATS: 4 minutes 30.29 seconds
DEVICE_STATS: 1800 iterations, 2.193s init, 265.272s render
Rendering Performance:(1800 / 265.272) = 6.785 iterations per second
Preload Time: ((240 + 30.29) - 265.272) = 5.018 seconds
Added some more, if I remember I am going to try cpu with optix off when I go to sleep or eat supper for completion sake.
The more the merrier!
General Question:
Should I remove those calculations entirely from the part you guys have to report? As long as Total Rendering Time and Device Statistics are there, I can just compute Rendering Performance and Loading Time for the graphs myself.
I would say leave them in for the added information. It was only fiddly for me because I ran 6 benchmarks.
I found the line
misc info : Iray RTX 2019.1.1, build 317500.2554, 08 Jun 2019, nt-x86-64.
I used the texture viewport for both of these runs. I simply forgot to grab the device stats line for the first. Looking at these times, though, if I did something wrong, then I don't see how I would beat the Titan at all. If Iray is a little faster now, but I am not seeing this performance gain because I did something wrong, it would be logical that I should not be near your RTX time to begin with. Also, if RTX is any factor, again, I shouldn't be so close. But both of my times beat the Titan here. Also Titan RTX isn't just about its 24gb VRAM though that is its big draw, it has more of all 3 core types than the 2080ti.
When I ran the tests with the Iray viewport, my total render times dropped below the 4 minute mark. But I did not post these.
I feel that trying to measure anything besides GPU rendering in the same place is a wild goose chase. There are too many variables to ever get good data on that. Being that nearly all of us are on PC, every single machine can be unique.
Could you add them in to your first run? That way I can integrate it into the tables at the top.
Assuming your DS 4.12 install is good, there's no reason to think that there's anything wrong with your results. It is entirely plausible for 2 1080ti's to come out ahead of a single Titan RTX due to sheer Cuda core counts alone (the Titan RTX has 4,608 - two 1080ti's have a combined 8,704.) And although Turing Cuda cores are significantly faster than Pascal ones, they aren't that much faster.
And RTX acceleration is unfortunately only a limited boon where this particular benchmarking scene (and others similar to it like the one that you did) are concerned because of the design constraints (see "memory requirements" under 4.1 Design Factors) necessary for a single benchmarking scene to be universally runnable on recent gen hardware. I'm still collecting data on it, but it looks like the only types of scenes which get major performance uplifts from RTX acceleration are ones which take hours to render on current gen high-end cards. Meaning that it is unfortunately a feature which is highly impractical to benchmark.
But render times in your Device Statistics stayed fundamentally the same (check if you doubt me.) Daz Studio's Total Rendering Time - despite its name - isn't a reliable measure of Iray rendering performance. However, render time under Iray's Device Statistics is. That is the essential number to know.
@RayDAnt - In the first table, one of the figures has been entered incorrectly.
Currently, it reads:
The 7.904 figure is the loading time for that combination. The correct figure is 6.392. I noticed because currently the first table shows the 1080ti/980ti being faster than 2x1080ti.
Very interesting to see some numbers for different cards. Thank you for putting in the time to enter all of this data. I'm considering replacing the 980TI in my system with a 2080Ti, so these figures are very useful as I make my way to a decision.
Oops, posted this in the wrong thread. Here's my bench for 4.12 beta:
i5 4690
32 GB RAM
Samsung 500MB SSD, Western Digital 1TB External
1080ti EVGA SC2 Black 1947 mhz
1800 iterations, 5.913s init, 457.867s render
Total Rendering Time: 7 minutes 47.61 seconds
Fixed it. Thanks!
Glad to hear it! That was one of the primary purposes I had here (provide useful information to people considering GPU purchases for DS/Iray use specifically.)
@talidesade can you add your current windows, nvidia driver, and Daz Studio versions into your post with your benchmark as well? That's the bare minimum I need to to enter your results into the tables at the top of the thread. Also whether or not you have OptiX Prime enabled (that still matters in 4.12 Beta if you have a GTX card.)
Some TCC benchmark results for those interested.
System Configuration
System/Motherboard: Gigabyte Z370 Aorus Gaming 7
CPU: Intel i7-8700K @ stock (MCE enabled)
GPU: Nvidia Titan RTX @ stock (watercooled)
System Memory: Corsair Vengeance LPX 32GB DDR4 @ 3000Mhz
OS Drive: Samsung Pro 970 512GB NVME SSD
Asset Drive: Sandisk Extreme Portable SSD 1TB
Operating System: Windows 10 Pro version 1903 build 18362.239
Nvidia Drivers Version: 431.60 GRD WDDM
Benchmark Results - Titan RTX Only (TCC Mode)
Daz Studio Version: 4.12.0.033 Beta x64
Optix Prime Acceleration: On
Total Rendering Time: 3 minutes 56.45 seconds
CUDA device 0 (TITAN RTX): 1800 iterations, 2.245s init, 231.590s render
Iteration Rate: (1800 / 231.590) = 7.772 iterations per second
Loading Time: ((0 + 180 + 56.45) - 231.590) = (236.45 - 231.590) = 4.860 seconds
Benchmark Results - Titan RTX Only (TCC Mode)
Daz Studio Version: 4.11.0.383 x64
Optix Prime Acceleration: On
Total Rendering Time: 4 minutes 15.93 seconds
CUDA device 0 (TITAN RTX): 1800 iterations, 2.326s init, 251.184s render
Iteration Rate: (1800 / 251.184) = 7.166 iterations per second
Loading Time: ((0 + 240 + 15.93) - 251.184) = (255.93 - 251.184) = 4.746 seconds
Benchmark Results - Titan RTX Only (TCC Mode)
Daz Studio Version: 4.11.0.383 x64
Optix Prime Acceleration: Off
Total Rendering Time: 4 minutes 35.57 seconds
CUDA device 0 (TITAN RTX): 1800 iterations, 3.137s init, 270.057s render
Iteration Rate: (1800 / 270.057) = 6.665 iterations per second
Loading Time: ((0 + 240 + 35.57) - 270.057) = (275.57 - 270.057) = 5.513 seconds
Benchmark Results - Titan RTX (TCC Mode) + Intel i7-8700K
Daz Studio Version: 4.12.0.033 x64
Optix Prime Acceleration: On
Total Rendering Time: 4 minutes 5.52 seconds
CUDA device 0 (TITAN RTX): 1693 iterations, 2.411s init, 239.909s render
CPU: 107 iterations, 2.244s init, 240.623s render
Iteration Rate: (1800 / 240.623) = 7.481 iterations per second
Loading Time: ((0 + 240 + 5.52) - 240.623) = (245.52 - 240.623) = 4.897 seconds
Benchmark Results - Titan RTX (TCC Mode) + Intel i7-8700K
Daz Studio Version: 4.11.0.383 x64
Optix Prime Acceleration: On
Total Rendering Time: 4 minutes 23.24 seconds
CUDA device 0 (TITAN RTX): 1704 iterations, 2.323s init, 257.724s render
CPU: 96 iterations, 2.082s init, 258.780s render
Iteration Rate: (1800 / 258.780) = 6.956 iterations per second
Loading Time: ((0 + 240 + 23.24) - 258.780) = (263.24 - 258.780) = 4.460 seconds
Benchmark Results - Titan RTX (TCC Mode) + Intel i7-8700K
Daz Studio Version: 4.11.0.383 x64
Optix Prime Acceleration: Off
Total Rendering Time: 4 minutes 40.62 seconds
CUDA device 0 (TITAN RTX): 1698 iterations, 3.294s init, 274.001s render
CPU: 102 iterations, 3.022s init, 275.201s render
Iteration Rate: (1800 / 275.201) = 6.541 iterations per second
Loading Time: ((0 + 240 + 40.62) - 275.201) = (280.62 - 275.201) = 5.419 seconds
I’ve been kinda scratching my head trying to figure out what we’re trying to accomplish with this effort. Apparently it was mentioned in another thread that we’re trying to make a comparison guide for users to see how well the cards perform to help in purchasing decisions?
But the excellent migenius article basically says that Iray/RTX performance is totally dependent upon scene content (“complexity”), and the only way to benchmark RTX/Iray performace is by each user trying RTX/Iray out on scenes that represent their typical complexity (whatever that means) to get a rough idea. And they’re saying the performance range they’ve seen is a staggering 5% - 300%, depending on scene content.
So why would we gather so much detailed hardware/software info to come up with a highly accurate render time comparison, down to a fraction of a second, for a single scene (which may or may not take much advantage of the RTX technology), when users’ actual results for their personal scenes could be anywhere in that 5% - 300% range? We get GPU render time comparisons accurate to 0.01 second for this benchmark scene using all of this detailed data, but a user may find their results for their particular scenes are not even close to what these benchmarks indicate. “Hey, you said there’s 300% improvement, how come I’m only getting 5% on my scenes?”.
And it sounds like we’re trying to find the most “complex” scene in the store as a benchmark, and I doubt we even know what scene characteristics are needed to get the most RTX/Iray performance gain. And why do we even care about the complex scene and the most performance gain, when it may be irrelevant to the average user?
Personally, I think the migenius article puts the nail in the coffin of standard, single scene benchmarks like this. Like I said long ago, this stuff is incredibly complex, and as you start to add other RTX-related features into the mix (Tensor de-noising, materials, physics), it becomes virtually impossible to find a single (or two, or three) benchmarks to be of much (any) use. My biggest surprise with the migenius article was that I wasn’t conservative enough in suggesting these benchmarks are, at best, ballparks within maybe +/- 15 seconds or 10% or whatever for comparison purposes. Wow, 5% to 300% variation? Amazing.
Please post ONLY benchmark results in this thread.
For discussion of these results or issues pertaining to benchmarking in general go here.
ebergerly see this post.