3 ms·
M3 performance so far on Geekbench: all M3 seems to have the same single-core score around 3000, which is 10% higher than M2 Max, same as i9-13900KS. In terms
by magnio 3y ago
M3 performance so far on Geekbench: all M3 seems to have the same single-core score around 3000, which is 10% higher than M2 Max, same as i9-13900KS.
In terms of multi-core
- M3 (8-core): 11500, 10% higher than M2, in the middle of M2 and M2 Pro 10-core
- M3 Max (16-core): 20600, 50% higher than M2 Max (12-core), same as i9-13900KS
For comparison, the Snapdragon X Elite (at 23W) announced recently scores 2800 in single-core and 14000 in multi-core, same as the M2 Max.
In terms of graphics (OpenCL):
- M3: 30000, 10% higher than M2, same as AMD Radeon 780M (currently best iGPU in x86 land)
- M3 Max: 94000, 10% higher than M2 Max, a bit above the RTX 2070.
- galad87 3y agoI wouldn't use the OpenCL score, OpenCL is a second class citizen that has been deprecated 5 or more years ago. It still works, but as you can see in Geekbench, Metal get a much higher score running the same workload.
- nine_k 3y agoCan you run Metal on an NVidia GPU? If not how to compare at all?
- skavi 3y agoI gave it a go: https://news.ycombinator.com/item?id=38111256 https://news.ycombinator.com/item?id=38111256
- duskwuff 3y ago> Metal get a much higher score running the same workload. Are you sure those scores are comparable?
- galad87 3y agoIf it's doing the same work on the GPU, yes.
- solarkraft 3y agoThe comparisons are actually exciting me more than the M3 itself. Yay, more choice! However I guess you can crank any core up to any performance if you're willing to throw power at it - performance/watt seems to be the by far most interesting metric. I suppose it's certain that the i9 can't keep up at that - but maybe the Qualcomm chip can? If it's 30% or so worse it would actually still be quite good ...