Rhino 7 vs. Rhino 6 - 30% to 50% Slower GPU performance

Hello All, and @Holo

I clearly notice an inferior viewport performance on Rhino 7 compared to Rhino 6, especially in shaded mode

just randomly benchmarking my rig with Windows 11.
Same Machine, Same OS and without restarting.

Comparision.pdf (42.7 KB)

Holomark 2 R6 v2,61

Rhino 6 - Total Score: 212753
Total Runtime: 715.63 sec

GPU scores: 198030
GPU_01 - 579.70 fps - Cube 4 tests
GPU_02 - 69.00 fps - UDT Shape
GPU_03 - 769.20 fps - Wireframe
GPU_04 - 625.00 fps - Shaded
GPU_05 - 151.50 fps - Rendered
GPU_06 - 133.30 fps - Block Rendered
GPU_07 - 655 units Nurbs @ 5 fps in Wireframe
GPU_08 - 448 units Nurbs @ 5 fps in Shaded
GPU_09 - 97 units Nurbs @ 5 fps in RenderSpeed
GPU_10 - 222.20 fps - RenderMesh Render
GPU_11 - 625.00 fps - RenderMesh RenderSpeed
GPU_12 - 526.30 fps - JoinedMesh Render
GPU_13 - 1111.10 fps - JoinedMesh RenderSpeed
GPU_14 - 115 units mesh @ 15 fps in Shaded
GPU_15 - 347 units mesh @ 15 fps in Render
GPU_16 - 750 units mesh @ 15 fps in RenderSpeed
GPU_17 - 256.40 fps - mesh in Rendered Studio
GPU_18 - 51.50 fps - Nurbs in Rendered Studio
GPU_19 - 59.20 fps - Block Illustration
GPU_20 - 121.80 fps - 2D single
GPU_21 - 78.70 fps - 2D massive (20x)

CPU scores: 14723
CPU_01 - 6.87 sec - Booleans and Contours
CPU_02 - 1.36 sec - Twist and Taper (UDT)
CPU_03 - 3.65 sec - Meshing Mini
CPU_04 - 0.03 sec - Extract Render Mesh
CPU_05 - 0.02 sec - Join Render Mesh
CPU_06 - 17.58 sec - Reduce Mesh
CPU_07 - 1.28 sec - Calculating Technical display
CPU_08 - 3.17 sec - Making Silhouettes

EVGA INTERNATIONAL CO.,LTD
To be filled by O.E.M.

NVIDIA Quadro RTX 4000 - 4095.0 MB
DriverVersion: 30.0.14.7111
NVIDIA Tesla V100-PCIE-16GB - ERROR AdapterRAM
DriverVersion: 30.0.14.7111
NVIDIA Tesla V100-PCIE-16GB - ERROR AdapterRAM
DriverVersion: 30.0.14.7111

Intel(R) Core™ i7-4790K CPU @ 4.00GHz
NumberOfCores: 4 NumberOfLogicalProcessors: 8
MaxClockSpeed: 4.0 GHz

TotalPhysicalMemory: 32.0 GB

Microsoft Windows 11 Pro

  • None - 64-bit

Rhino 6 sr 34 64 bit
AA level = 8

Holomark 2 R6 v2,61

Holomark 2 R6 v2,61

Rhino 7 Total Score: 198333

Total Runtime: 741.22 sec

GPU scores: 183080
GPU_01 - 615.40 fps - Cube 4 tests
GPU_02 - 55.60 fps - UDT Shape
GPU_03 - 1111.10 fps - Wireframe
GPU_04 - 625.00 fps - Shaded
GPU_05 - 120.50 fps - Rendered
GPU_06 - 106.40 fps - Block Rendered
GPU_07 - 764 units Nurbs @ 5 fps in Wireframe
GPU_08 - 209 units Nurbs @ 5 fps in Shaded
GPU_09 - 64 units Nurbs @ 5 fps in RenderSpeed
GPU_10 - 227.30 fps - RenderMesh Render
GPU_11 - 588.20 fps - RenderMesh RenderSpeed
GPU_12 - 769.20 fps - JoinedMesh Render
GPU_13 - 1250.00 fps - JoinedMesh RenderSpeed
GPU_14 - 100 units mesh @ 15 fps in Shaded
GPU_15 - 336 units mesh @ 15 fps in Render
GPU_16 - 584 units mesh @ 15 fps in RenderSpeed
GPU_17 - 277.80 fps - mesh in Rendered Studio
GPU_18 - 30.80 fps - Nurbs in Rendered Studio
GPU_19 - 56.50 fps - Block Illustration
GPU_20 - 55.80 fps - 2D single
GPU_21 - 8.60 fps - 2D massive (20x)

CPU scores: 15253
CPU_01 - 5.42 sec - Booleans and Contours
CPU_02 - 1.72 sec - Twist and Taper (UDT)
CPU_03 - 3.72 sec - Meshing Mini
CPU_04 - 0.08 sec - Extract Render Mesh
CPU_05 - 0.03 sec - Join Render Mesh
CPU_06 - 17.32 sec - Reduce Mesh
CPU_07 - 1.29 sec - Calculating Technical display
CPU_08 - 3.20 sec - Making Silhouettes

EVGA INTERNATIONAL CO.,LTD

To be filled by O.E.M.

NVIDIA Quadro RTX 4000 - 4095.0 MB
DriverVersion: 30.0.14.7111
NVIDIA Tesla V100-PCIE-16GB - ERROR AdapterRAM
DriverVersion: 30.0.14.7111
NVIDIA Tesla V100-PCIE-16GB - ERROR AdapterRAM
DriverVersion: 30.0.14.7111

Intel(R) Core™ i7-4790K CPU @ 4.00GHz
NumberOfCores: 4 NumberOfLogicalProcessors: 8
MaxClockSpeed: 4.0 GHz

TotalPhysicalMemory: 32.0 GB

Microsoft Windows 11 Pro

  • None - 64-bit

Rhino 7 sr 8 64 bit
AA level = 8
Total Runtime: 741.22 sec

1 Like

What does SystemInfo say? 'Cause that’s not reporting the specs properly. Is this something you’re actually noticing?

I’m generally noticing that shaded view and rendered view is 30-50% slower to Rhino 6 .

I took the mini Cooper modern. Array it to be 10x10 (totalling 100 instances) then run TestMaxSpeed.

On wireframe mode. RH 7 is faster
On Shaded and Rendered view RH 6 is faster.

1 Like

Rhino 7 SR8 2021-6-30 (Rhino 7, 7.8.21181.05001, Git hash:master @ dd41d5edfa591ef81b3f34039b4e3ae5cfe59fd1)
License type: Commercial, build 2021-06-30
License details: Cloud Zoo

Windows 10.0.22000 SR0.0 or greater (Physical RAM: 32Gb)

Computer platform: DESKTOP

Standard graphics configuration.
Primary display and OpenGL: NVIDIA Quadro RTX 4000 (NVidia) Memory: 8GB, Driver date: 6-21-2021 (M-D-Y). OpenGL Ver: 4.6.0 NVIDIA 471.11
> Accelerated graphics device with 4 adapter port(s)
- Windows Main Display attached to adapter port #0
- Secondary monitor attached to adapter port #1

Secondary graphics devices.
NVIDIA Tesla V100-PCIE-16GB (NVidia) Memory: 0MB, Driver date: 6-21-2021 (M-D-Y).
> Accelerated graphics device with 0 adapter port(s)
- There are no monitors attached to this device!
NVIDIA Tesla V100-PCIE-16GB (NVidia) Memory: 0MB, Driver date: 6-21-2021 (M-D-Y).
> Accelerated graphics device with 0 adapter port(s)
- There are no monitors attached to this device!

OpenGL Settings
Safe mode: Off
Use accelerated hardware modes: On
Redraw scene when viewports are exposed: On
Graphics level being used: OpenGL 4.6 (primary GPU’s maximum)

Anti-alias mode: 8x
Mip Map Filtering: Linear
Anisotropic Filtering Mode: High

Vendor Name: NVIDIA Corporation
Render version: 4.6
Shading Language: 4.60 NVIDIA
Driver Date: 6-21-2021
Driver Version: 30.0.14.7111
Maximum Texture size: 32768 x 32768
Z-Buffer depth: 24 bits
Maximum Viewport size: 32768 x 32768
Total Video Memory: 8 GB

Rhino plugins that do not ship with Rhino

C:\ProgramData\McNeel\Rhinoceros\7.0\Plug-ins\Twinmotion Direct Link 2020.2 (e24ee14a-9514-47da-8fe6-f6e61b20160e)\Twinmotion Direct Link 2020.rhp “Twinmotion Direct Link 2020” 1.0.6.0
C:\Program Files\Chaos Group\V-Ray\V-Ray for Rhinoceros\V7\VRayForRhino.rhp “V-Ray for Rhino”
C:\Users\tayO\AppData\Roaming\McNeel\Rhinoceros\packages\7.0\ColorPicker\6.0.0\ColorPicker.rhp “ColorPicker”
C:\Users\tayO\AppData\Roaming\McNeel\Rhinoceros\packages\7.0\EleFront\4.2.2\ElefrontProperties.rhp “ElefrontProperties” 1.0.0.0
C:\Users\tayO\AppData\Roaming\McNeel\Rhinoceros\packages\7.0\SubstanceImporter\2.0.1\Substance.Win.rhp “SubstanceImporter” 2.0.1.0

Rhino plugins that ship with Rhino
C:\Program Files\Rhino 7\Plug-ins\Commands.rhp “Commands” 7.8.21181.5001
C:\Program Files\Rhino 7\Plug-ins\rdk.rhp “Renderer Development Kit”
C:\Program Files\Rhino 7\Plug-ins\RhinoRenderCycles.rhp “Rhino Render” 7.8.21181.5001
C:\Program Files\Rhino 7\Plug-ins\rdk_etoui.rhp “RDK_EtoUI” 7.8.21181.5001
C:\Program Files\Rhino 7\Plug-ins\rdk_ui.rhp “Renderer Development Kit UI”
C:\Program Files\Rhino 7\Plug-ins\NamedSnapshots.rhp “Snapshots”
C:\Program Files\Rhino 7\Plug-ins\RhinoCycles.rhp “RhinoCycles” 7.8.21181.5001
C:\Program Files\Rhino 7\Plug-ins\Toolbars\Toolbars.rhp “Toolbars” 7.8.21181.5001
C:\Program Files\Rhino 7\Plug-ins\3dxrhino.rhp “3Dconnexion 3D Mouse”
C:\Program Files\Rhino 7\Plug-ins\Displacement.rhp “Displacement”
C:\Users\tayO\AppData\Roaming\McNeel\Rhinoceros\packages\7.0\PanelingTools\2021.3.2.446\PanelingTools.rhp “PanelingTools”
C:\Users\tayO\AppData\Roaming\McNeel\Rhinoceros\packages\7.0\SectionTools\2021.4.30.511\SectionTools.rhp “SectionTools”

How often did you run these tests? Can you really trust them for being accurate? From the total score, its about 7% difference, but that could also be due to various background processes. Also, if you use a cracked copy of Windows 11, nobody knows what they do in the background.
And, why wouldn’t you use a cracked copy of Rhino if you crack Windows? Maybe some smart Iranian hacker is using your potentially cracked Rhino 7 for some hidden GPU bitcoin mining. :stuck_out_tongue: This is as likely as having decreased performance build in due to feature development and bug fixing…

Yeah, I don’t know why you’re using the leaked Windows 11 on a work machine, that’s rather foolish, there’s nothing really that exciting coming.

And my response to these benchmarks is, shrug? I dunno? Who knows? Who cares? V6 was a huge leap in terms of actually USING advanced OpenGL features, so probably V7 uses more and is more demanding on hardware–and the first-gen RTX cards were rather a disappointment in that regard–or maybe LESS because of people running potatoes complaining.

The benchmarks developed by actual teams that you see all over the Net for testing games or Specviewperf or memory performance, they’re all kinda bunk, and the idea of trying to compare benchmarks of different versions of software instead of the same benchmark on different hardware or OS is…simply invalid from any “scientific” basis.

1 Like

One feature seems to be really nice, and that’s the ability to have more than one desktop view. That could be super useful if you work with one monitor. Instead of switching apps you just switch views based on what you want to do… But that obviously makes only sense if its fast enough…

Nothing Cracked or Leaked, it is official dev build, and I’m not using it as a work machine, instead it is my secondary machine I’m using, mostly for rendering. my work machine is not that exciting to play with.

I do these tests every time I update Nvidia GPU drivers.
I did the same test on Windows 10 installation, still there is performance difference between RH6 and RH7

I don’t know how helpful these words are.

1 Like

So??? It’s not like there’s a team of people working full-time on how to optimize this stuff, this isn’t a AAA video game made with a bigger team than the entire CAD industry, the main concern is that it “work” on a variety of hardware. This benchmarking between versions like this is not valid. If the results are correct–which only show it slower in some tests!–it’s a big whoopdee doo who cares, the biggest performance issues with Rhino are not how fast it can spin some objects with OpenGL.

I’m sharing these ideas because I care about the people developing Rhino. not to ask for a perfect product, but something I’m noticing and should be discussed.
on my work, usually I easily hit a 500mb model and I wish I have a 10% boost to keep my experience smoother.

2 Likes

@Holo what is this test doing? It seems to be one that may need attention

1 Like

Hi Steve, it just redraws a view with lots of 2D data.

To be more focused on the issue to track, I was able to compare the Latest Rhino 7 SR 8 with the earlier Rhino 7 SR 2 from early January. the performance difference is huge.

Below is my general viewport navigation with @Holo 's 2D Massive (20) on Rhino 7 SR 2.

And you can compare it with the sluggish performance from Rhino 7 SR 8.

We are having a long weekend here in USA, and I’m planning to test out every rhino build to determine when this performance drop started.

Ok the dev build is out for 4 days, probably because it has been leaked 2 weeks ago. You were posting before that that you are using this build. So my assumption was that you are using the leaked one. And a leaked build can be compromised with malware and its sort of cracked, because at this time you had to bypass licensing. Furthermore it is not uncommon that cracked software is patched with malware… Sorry I find it no so unlikely if people using leaked builds, that they use cracked software. And it might be stupid to you, but it certainly is a difference to make performance tests with compromised software and original builds! That was my point! The opposite could also be true, by disabling any sort of telemetry within an app, a cracked software could potentially perform better.

Are you testing different Nvidia drivers too? And Windows versions? They all impact this stuff massively too. And you’re only testing one little-used video card. Hey if you like running benchmarks for fun, fill your boots, but don’t kid yourself that it’s much more than that.

1 Like

Same issue with my work machine using stable versions of Windows, Rhino and Nvidia Drivers:

Holomark 2 R6 v2,61

Total Score: 96694
Total Runtime: 379.44 sec

GPU scores: 81520
GPU_01 - 449.40 fps - Cube 4 tests
GPU_02 - 55.90 fps - UDT Shape
GPU_03 - 588.20 fps - Wireframe
GPU_04 - 277.80 fps - Shaded
GPU_05 - 78.10 fps - Rendered
GPU_06 - 70.40 fps - Block Rendered
GPU_07 - 254 units Nurbs @ 5 fps in Wireframe
GPU_08 - 120 units Nurbs @ 5 fps in Shaded
GPU_09 - 66 units Nurbs @ 5 fps in RenderSpeed
GPU_10 - 112.40 fps - RenderMesh Render
GPU_11 - 303.00 fps - RenderMesh RenderSpeed
GPU_12 - 294.10 fps - JoinedMesh Render
GPU_13 - 588.20 fps - JoinedMesh RenderSpeed
GPU_14 - 28 units mesh @ 15 fps in Shaded
GPU_15 - 107 units mesh @ 15 fps in Render
GPU_16 - 218 units mesh @ 15 fps in RenderSpeed
GPU_17 - 86.20 fps - mesh in Rendered Studio
GPU_18 - 26.00 fps - Nurbs in Rendered Studio
GPU_19 - 50.00 fps - Block Illustration
GPU_20 - 58.40 fps - 2D single
GPU_21 - 8.70 fps - 2D massive (20x)

CPU scores: 15174
CPU_01 - 5.03 sec - Booleans and Contours
CPU_02 - 1.46 sec - Twist and Taper (UDT)
CPU_03 - 3.45 sec - Meshing Mini
CPU_04 - 0.08 sec - Extract Render Mesh
CPU_05 - 0.03 sec - Join Render Mesh
CPU_06 - 17.80 sec - Reduce Mesh
CPU_07 - 1.65 sec - Calculating Technical display
CPU_08 - 3.45 sec - Making Silhouettes

Dell Inc.
Precision 5530

Intel(R) UHD Graphics 630 - 1024.0 MB
DriverVersion: 27.20.100.8935
NVIDIA Quadro P1000 - 4095.0 MB
DriverVersion: 27.21.14.6231

Intel(R) Core™ i7-8850H CPU @ 2.60GHz
NumberOfCores: 6 NumberOfLogicalProcessors: 12
MaxClockSpeed: 2.59 GHz

TotalPhysicalMemory: 16.0 GB

Microsoft Windows 10 Pro

  • None - 64-bit

Rhino 7 sr 7 64 bit
AA level = 4strong text

Beyond benchmarking, I’ve been noticing many performance issues recently.

Come on, is not a magic to join here
https://insider.windows.com/en-us/

1 Like

I’m confused, virtual desktops are already possible in Windows 10. How is this a new feature?