Geekbench 4.10 benchmarks

[SoyCapitanSoyCapitan]

Hi,

I kept noticing CPU and GPU benchmark scores were dropping with each Sierra update.

It looks like the new version of Geekbench fixes the inconsistent scores we were seeing with Macbook Pro 2016 models.

With the i7-6820HQ 455 Pro model, I now get 4215 single core, 15122 multicore.

OpenCL is back up to a more respectable 40602.

There's a new Metal test too, but only for paid users.

Fire away.

sell $btc
sell $eth
sell $ltc

 

[Charlesje]

Interesting find. I ran the benchmark for the 6920hq and obtained 4642 single core and 15370 multicore, which is higher then before. It remains to be seen if the new version affects the scores of the previous generation macs, but otherwise it can help quieten unreasonable assumptions about the previous generation mbps performing faster.
But it doesn't change the synthetic benchmark limitations of geekbench - badly predicting actual performance of mobile devices - of course.

 

[SoyCapitanSoyCapitan]

Interesting find. I ran the benchmark for the 6920hq and obtained 4642 single core and 15370 multicore, which is higher then before. It remains to be seen if the new version affects the scores of the previous generation macs, but otherwise it can help quieten unreasonable assumptions about the previous generation mbps performing faster.
But it doesn't change the synthetic benchmark limitations of geekbench - badly predicting actual performance of mobile devices - of course.

The update boosts scores across any CPU (Intel, AMD, ARM) so older machines should see numbers climb too. There's no getting around the fact that the best mobile Haswell is marginally faster in synthetics than the best mobile Skylake. That's not relevant in real world mobile computing anyway. It's just good to see the app back to behaving consistently now.

 

[Charlesje]

The update boosts scores across any CPU (Intel, AMD, ARM) so older machines should see numbers climb too. There's no getting around the fact that the best mobile Haswell is marginally faster in synthetics than the best mobile Skylake.

I understand, but apparently I don't see this mirrored in the limited recent scores on geekbench.