Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Comparing these to 2018 Dell XPS (GTX 1050 Ti) from notebookcheck:
Time Spy: MBP is 15% faster
Fire Strike: MBP is 23% faster

Thanks! Just a small correction - 2018 XPS 15 has 1050Ti-MaxQ, a 45W GPU thus comparable to Vega16.

I run the standard 3dmark tests using the standardised settings in order to make it comparable. Why would you want to run them on native resolution? You can't really compare the results to anything then...

I asked for native desktop Windows resolution, not to run it in it, just in case it has some influence on results, I wanted to compare to our results on Polaris. Here are the scores from our runs on 555x and 560x.

The numbers are - 555x : 560x : your score on Vega 20:
  • Time Spy Score 1374 1774 2763
  • TimeSpy Graphics Score 1247 1611 2623
  • TimeSpy CPU Score 3253 4188 ????
  • Firestrike Score 4103 4942 7573
  • Firestrike graphics score 4568 5680 9147
  • Firestrike Physics score 10819 10904 8777
  • Sky Diver Score 13365 14744 21296
  • Sky Diver Graphics Score 13438 16400 27100
  • Skydiver Physics score 12699 11125 10679
Massive improvements in 3d, but the CPU scores are all lower on Vega than on Polaris, and these were 2.2 machines. So looks like Apple has some work to do to improve the CPU/GPU load balancing.
 
Here is my Cinabench on my new i9, 32gb, 2tb, vega 20 I just got in yesterday. Also Disk speed. I ran Cinabench 3 times in a row.

Screen Shot 2018-12-07 at 1.26.31 PM.png


Screen Shot 2018-12-07 at 9.07.43 AM.png
 
Thats a little low, maybe your machine is still caching? What was it on first try?

Yeah, I got it last night, and it installed my time machine backup all night. And that was the first test this morning. Spotlight is definitely still working in the background. I will try another one later to see.
 
That seems to be recurring observation of the new i9 owners, but I don't see any changes that would explain it, CPU side is identical, power delivery and cooling. Can you hit 1100 on cinebench? I personally think that Apple just tighten up the quality control and makes sure there is proper contact between baseplate and die, so you have higher chance on getting a 'good' unit. Could you turn off TM backup, kill everything that's running the background and just let your laptop sit idle for 15 minutes? Keep an eye on the CPU watts, below 0.5W Core (which is pretty amazing on its own) my 2.2/555x both show 30C flat in such scenario (proximity sensors), and there is really no reason for any other config to be any different - due to different components, and not QC. Are the idle fans still at 2k?
Are you sure the power delivery is the same? Are you sure the VRM layout is the same, and it has exactly the same function?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.