Geekbench METAL scores

[AndreeOnline]

Let's see where we're at in terms of MacPro Metal scores! Geekbench 4.1+ required.

I'm getting the ball rolling with a single RX 480 8GB card:

EDIT: for reference my MacBook Pro 2016, Radeon 460 gets roughly 32800 points.

 

[Synchro3]

GTX Titan X in Mac Pro 4,1/5,1:

 

[H2SO4]

Single 4GB RX470 with more CUx enabled using the mods found on this forum;

My OCL result is 119,889.

 

[cpnotebook80]

GTX Titan X in Mac Pro 4,1/5,1:

View attachment 694750

can you run 4k monitors or even 4k media with that card on that machine? I'm looking to upgrade my 7950

 

[Synchro3]

can you run 4k monitors or even 4k media with that card on that machine? I'm looking to upgrade my 7950

4K monitors (with 10 Bit color): Yes. With boot screen on 4K (MVC flashed card).

4K media: Depends on the bitrate: https://forums.macrumors.com/threads/cant-play-4k-videos.2033874/

Neither the CPU nor the GTX Titan X have HEVC hardware acceleration.

If you want HEVC hardware acceleration I would recommend a GTX 950 or GTX 960.

 

[cpnotebook80]

4K monitors (with 10 Bit color): Yes. With boot screen on 4K (MVC flashed card).

4K media: Depends on the bitrate: https://forums.macrumors.com/threads/cant-play-4k-videos.2033874/

Neither the CPU nor the GTX Titan X have HEVC hardware acceleration.

If you want HEVC hardware acceleration I would recommend a GTX 950 or GTX 960.

oh ok. Thanks.

 

[h9826790]

4K monitors (with 10 Bit color): Yes. With boot screen on 4K (MVC flashed card).

4K media: Depends on the bitrate: https://forums.macrumors.com/threads/cant-play-4k-videos.2033874/

Neither the CPU nor the GTX Titan X have HEVC hardware acceleration.

If you want HEVC hardware acceleration I would recommend a GTX 950 or GTX 960.

Just want to know if anyone can confirm that the HEVC hardware acceleration work in MacOS? I know the card can do that, but does it require software support (OS, driver, apps)?

 

[splifingate]

gb4.1; 980ti; MP 5,1:

 

[flowrider]

Geekbench 4 doesn't work with Nvidia cards for the GPU Benchmark. Are you folks saying that it works for Metal and Cuda?

Lou

 

[splifingate]

https://browser.geekbench.com/user/119505

Geekbench 4.1

CPU/CUDA/Metal produce (what they produce is up for speculation).

I cannot get OpenCL Compute to produce results (same as others (http://support.primatelabs.com/discussions/geekbench/16225-compute-benchmarks-in-geekbench-4-failing (etc.))).

Regards, splifingate

 

[Yahooligan]

Geekbench 4 doesn't work with Nvidia cards for the CPU Benchmark. Are you folks saying that it works for Metal and Cuda?

Lou

I'm not sure what you're talking about here, the CPU benchmarks have nothing to do with the GPU. Compute (GPU) benchmarks in 4.1 are OpenCL, Metal, and, if you have an Nvidia GPU, CUDA.

 

[flowrider]

^^^^Mistyped. Post corrected. Meant GPU. Specifically OpenCL.

Lou

 

[Yahooligan]

Geekbench 4 doesn't work with Nvidia cards for the GPU Benchmark. Are you folks saying that it works for Metal and Cuda?

Lou

^^^^Mistyped. Post corrected. Meant GPU. Specifically OpenCL.

Lou

Gotcha, no worries. Yeah, the OpenCL benchmarks fail with Nvidia and the web driver. If you're able to use the native macOS driver then the OpenCL benchmarks complete fine. Known issue.

 

[flowrider]

^^^^Yep, I know it's a known issue, hence why I asked the question. However, still don't have an answer to my question:

Does the Geekbench Metal and Cuda test work with an Nvidia card and Nvidia drivers?

 

[Synchro3]

Does the Geekbench Metal and Cuda test work with an Nvidia card and Nvidia drivers?

Yes. Only OpenCL benchmark does not work.

 

[Synchro3]

Ok, new Metal Benchmark with GTX Titan X, this time on Sierra, and latest Nvidia web driver:

 

[SoyCapitanSoyCapitan]

Are the tests for Metal the same as the tests for OpenCL?

 

[Tucom]

What are the comparative OpenGL scores? These scores don't have much meaning without something to compare them to; if there's a way to view the OGL scores vs. these that would be awesome.

 

[SoyCapitanSoyCapitan]

What are the comparative OpenGL scores? These scores don't have much meaning without something to compare them to; if there's a way to view the OGL scores vs. these that would be awesome.

GL isn't useful for the background tests Geekbench uses. It's a compute testing app.

 

[Tucom]

GL isn't useful for the background tests Geekbench uses. It's a compute testing app.

Ah, I see. Makes sense why you were talking about the OpenCL then. My mistake.

 

[mammal]

Adding two results for Mac Pro (GTX 680) and an iMac (M295X).

 

[linuxcooldude]

Mac Pro 2015 D700

 

[H2SO4]

View attachment 694911 Mac Pro 2015 D700

Benchmark suites are odd things. My RX470 has a higher teraflop output than both of your D700s combined yet gets a lower score.

 

[Yahooligan]

Benchmark suites are odd things. My RX470 has a higher teraflop output than both of your D700s combined yet gets a lower score.

Gaming GPU vs workstation GPU, the gaming GPUs are great for gaming but don't do as well at computations. Workstation GPUs do better at computations but aren't as good for gaming.

My D700 and my wife's GTX 680 are both in the 3.1-3.5 tflop range, my D700 kills her GTX 680 in computational benchmarks like OpenCL, her GTX 680 wins when doing rendering benchmarks like Valley.

 

[linuxcooldude]

Gaming GPU vs workstation GPU, the gaming GPUs are great for gaming but don't do as well at computations. Workstation GPUs do better at computations but aren't as good for gaming.

My D700 and my wife's GTX 680 are both in the 3.1-3.5 tflop range, my D700 kills her GTX 680 in computational benchmarks like OpenCL, her GTX 680 wins when doing rendering benchmarks like Valley.

Good observation. I wish there were better benchmarks geared more toward workstations rather than gaming benchmarks with a few other ones added. I see a lot of people remark workstation graphics are just consumer versions with a higher price. But I do think there are a few slight GPU hardware tweaks that differentiates consumer and workstation cards along with drivers/support ect. Of course it helps if you own/control the platform/ecosystem/hardware/OS rather than third party hardware and drivers.

Doing a quick lookup of the Radeon RX470 I'm showing 4.9 TFLOPS rather than 7 TFLOP's for the D700's. Thats 3.5 TFLOPS for one depending if Geekbench supports multiple GPU's.