Twitter | Search | |
Paul Haddad Feb 3
Replying to @siegel
Though I haven’t found a good solution for adding an external NVMe drive, at reasonable speeds, for even more storage.
Reply Retweet Like
Rich Siegel Feb 3
Replying to @tapbot_paul
Best I've come up with for external storage is a pocket USB-C enclosure. [1] OWC has a four-slot NVMe RAID, but it's Thunderbolt 3 and behind an adapter we've had some weird problems with it. [1]
Reply Retweet Like
Paul Haddad Feb 3
Replying to @siegel
😜 The problem with that is that the MP only has 5Gb/s ports. The OWC adapter is really weird even without the adapter, each NVMe drive only gets a single PCIe lane.
Reply Retweet Like
Rich Siegel Feb 3
Replying to @tapbot_paul
Sounds like you're pretty much in the same place I am, bumping into limitations on the machine's I/O performance. Oddly, I'm finding that my 2018 four-core i5 MacBook Pro stands up to the Mac Pro for lots of routine work, because it's got faster I/O.
Reply Retweet Like
Paul Haddad Feb 3
Replying to @siegel
For ObjC code?
Reply Retweet Like
Rich Siegel Feb 3
Replying to @tapbot_paul
Yes, ObjC and ObjC++.
Reply Retweet Like
Paul Haddad Feb 3
Replying to @siegel
That’s really odd, I know Swift has issues with > 4 CPUs when compiling but C scales really well. Now if you are only doing incremental, then the higher boost speeds probably makes it faster.
Reply Retweet Like
Rich Siegel Feb 3
Replying to @tapbot_paul
I'll generate some results now and report back in a couple of minutes. Looking over my old results, upgrading the machine from 6 cores to 10 actually made builds a little bit (about 10%) slower, I think because the additional work units were offset by the lower clock rate.
Reply Retweet Like
Paul Haddad Feb 3
Replying to @siegel
Odd, I thought the 10 was a good mix between lots of cores and good clock speed. I seriously considered getting that instead of the 12.
Reply Retweet Like
Rich Siegel Feb 3
Replying to @tapbot_paul
I think, generally, the 10 is an excellent balance between cores and clock rate. (It's like the 8-core the machine was introduced with, with two bonus cores.) I'm quite happy with mine.
Reply Retweet Like
Rich Siegel
So, here are the results for "time xcodebuild -confugration Release": 2018 MacBook Pro (4@2.3GHz i5, 16GB RAM): real7m53.587s user0m8.172s sys0m2.978s 2013 Mac Pro (10@3.0GHz Xeon E5, 64GB RAM): real7m17.600s user0m10.525s sys0m4.935s
Reply Retweet Like More
Rich Siegel Feb 3
Replying to @tapbot_paul
The laptop is clearly a bit slower in overall elapsed time than the big iron, but not a *lot* slower.
Reply Retweet Like
Paul Haddad Feb 3
Replying to @siegel
Yeah that’s really unexpected, something seems off. Mind doing a clang build on the 10 core? I always like using that for a good performance test.
Reply Retweet Like