[Solved] Low OpenCL score with Akitio Node(Thunderbolt 3) + Vega56
MacBook Pro (13-inch Mid 2018)
|Geekbench 4.4.2 Tryout for Mac OS X x86 (64-bit)|
|Upload Date||March 04 2020 02:48 AM|
|Operating System||macOS 10.15.3 (Build 19D76)|
|Model||MacBook Pro (13-inch Mid 2018)|
|Motherboard||Apple Inc. Mac-827FB448E656EC26 MacBookPro15,2|
|Memory||16384 MB 2133 MHz LPDDR3|
|BIOS||Apple Inc. 1037.80.53.0.0 (iBridge: 17.16.13050.0.0,0)|
|Name||Intel Core i5-8259U|
|Topology||1 Processor, 4 Cores, 8 Threads|
|Identifier||GenuineIntel Family 6 Model 142 Stepping 10|
|Base Frequency||2.30 GHz|
|L1 Instruction Cache||32.0 KB x 4|
|L1 Data Cache||32.0 KB x 4|
|L2 Cache||256 KB x 4|
|L3 Cache||6.00 MB x 1|
|Device Name||AMD Radeon RX Vega 56 Compute Engine|
|Maximum Frequency||1.59 GHz|
|Device Memory||7.98 GB|
I've connected my 2K monitor to eGPU directly, and I ran geekbenck4 opencl test in my external display. The score shoud be more than 150k, 130k is too low. (Akitio Node + RX580 4G: 110k, win10 + built-in Vega56: 180k).
The Akitio Node3 only has a 400W PSU, but I don't think this is the problem, because I use my eGPU only for final cut pro instead of gaming. I don't know why this is happening, any advice will be appreciated, thanks!
Run the Unigine Superposition benchmark in Windows instead.
You can get good system performance info there.
Thanks for your reply! Are you suggesting the geekbench4 opencl compute benchmark is not reliable? This eGPU setup is not for gaming, so the OpenCL compute performance is more important than famerate. PS: I've run a Geekbench5 OpenCL benchmark, the score is 59K
Superpostion results are consistent over the last several months.
So you can meaningfully compare your Superposition results with other Vega 56 owners.
(Not going to comment on other benchmarks 😂)
You can also run the Luxmark render benchmark to test Mac Open CL performance.