Contribution links:
First 8 core hand's on
10 core hand's on
iMac Pro Ram Pricing & Availability
VESA Mount
Things to expect:
Noise tests
Video Tests
(I will need help from someone who has an 8 core with Vega 56 and someone with a 10 core with Vega 64)
iMac Pro Ram Upgrades:
So people have read that the iMac Pro can have the ram upgraded, just not by us. I actually talked with Apple about six month ago about this very subject & here's what I learned:
1) Teardown enthusiasts can upgrade their own ram at the cost of risking warranty violation
2) Apple will upgrade the ram at Apple RAM cost (you buy their RAM still)
3) A third party authorized apple service center will upgrade your ram by your choice of either Apple provided RAM or:
You can BRING YOUR OWN RAM & have the authorized third party service center install it. Your warranty will be retained. The RAM will only carry the RAM's manufacturer's warrant & you will have to have RAM related issues serviced by a third party after the install. Apple will replace it with the original quantity installed upon purchase unless you upgrade through apple. So if I upgrade 32GB RAM to 128 on my own & it flakes out, If I take it to Apple, they will replace it with 32GB Apple RAM. The authorized third party will pull the RAM & ask you what to do...that's where you can have it sent in to be replaced on your own & bring the replacement to be installed.
For where I live, "Core Care of Rocklin" provides this service. Apple sent me there when I needed my macbook's battery replaced but I still needed to keep access to the laptop. Core Care did their test, ordered the battery, & called me back to supply the MacBook when they were ready to install it. 1 hr turnaround. That's what third parties can do for you. The cost of replacing RAM will be probably $70-150 plus the cost of the RAM.
So how do you find an authorized third party service center? Just ask Apple support & they will provide you the list of nearest authorized service centers. To know they are good, they will always run a diagnostics report & submit it to Apple & give you a copy. This is what Apple does as well when they do their work.
So what are the risks?
1) it costs a service center service payment
2) They crack the screen open so you risk dust between the glass & the LCD
3) The entire Motherboard needs to be pulled out so you risk damage. Humans are imperfect. I once had a Dell Laptop returned from the service center with the trackpad broken. I couldn't fix the motherboard, but I could tear it apart myself to find that they forgot to plug the trackpad back in...one of over a dozen connections.
4) Compatibility issues: not all RAM is equal & you could have issues. Have them do a full memory test before giving it back. 2017 iMac owners learned this when using Kingston's low CL ram...they learned that if you mix it, the RAM down clocks.
I hope this helps & have a great day!
CPU Geekbench Benchmarks:
Initial benchmarks from Geekbench:
8 core: 4937 single thread / 30088 multi thread
10 core 2: 4971 single thread / 34006 multi thread
10 core: 5345 single thread / 35917 multi thread
i7 (2017 imac) 5715 single thread / 19675 multi thread
10 core is 8.2% faster on single threaded operations and 19.4% faster on multi threaded operations
Keep in mind that this is a synthetic total number and if you look at the single core ops, each will be better at certain things.
10 core: https://browser.geekbench.com/v4/cpu/4449687
10 core 2: http://browser.geekbench.com/v4/cpu/5469097
8 core: https://browser.geekbench.com/v4/cpu/5471001
i7: https://browser.geekbench.com/v4/cpu/5543682
Source:
http://browser.geekbench.com/v4/cpu/search?utf8=✓&q=Mac-7BA5B2D9E42DDD94
Update: I added a second core 2 that popped up. Basically, it shows the first benchmark where the single thread operations almost matches the 8 core...which is more along the line of how I was expecting the two cores to perform. Basically, I'm waiting for a series of longer benchmarks to get the cpus to perform more continuously for rigor.
Basically, the CPU needs to be measured where it is actually stressed. A split second, momentary test will capture the max clock (4.5 vs 3.0 for the 10 core). This will artificially show the 10 core performing single thread ops as higher. Also, the cache has an affect which helps to boost performance on the 10 core. I expected some performance boost due to that, but not a 9% gain. I can say more certainly when more time consuming benchmarks come out.
As many of you are seeing, the 8 & 10 cores are as follows:
8 core: 3.2 GHz Base, max up to 4.3
10 core: 3.0 GHz Base, max up to 4.5
It's a $800 upgrade ($720 if discounted through EPP or Edu)
The performance should be about the same on the single core tasks & the 10 core will beat the 8 by 17% on multi core tasks. For me, this equates to 10 min savings on a normally 1hr project. I personally couldn't justify this upgrade. I'll post the 8 core benchmark when mine finally comes in.
Thermal Tests:
Thermals of the 10 core:
(c/o: nick82)
Thermals of the 8 core:
(c/o: bplein)
Takeaway:
The 8 core runs at about 3.9 GHz constantly; 93 degrees
The 10 core runs at about 3.5 GHz average; 95 degrees, but thermally clocking down
Geekbench shows single thread as a higher performer (which affects gaming) in the 10 core because the thermals off of the left image shows that it does not heat up quickly...& geek bench does not last long enough to clock down.
Drive Wiping & Reinstalling OsX High Sierra without DFU:
Be careful here. I used bootcamp & had an issue that required me to install 8 (which is not supported by the iMac Pro). I had to wipe the bootcamp partition because it was corrupted after wiping. Bootcamp assistant did not give me the option to remove the bootcamp partition on its own. Bootcamp would not allow me to reinstall after that either. If you want to do the upgrade method, provided you have upgraded once in the past, follow the link to bootcamp details:
https://forums.macrumors.com/threads/imac-pro-bootcamp-windows-10-limit-fix.2097652/
I finally booted into regular recovery, disabled boot security, & wiped the OSX partition as well...only the osx partition as there appeared to be more. I then created a new APFS partition & immediately went to install OSX. Everything went successful. I did not need a DFU boot.
Geekbench testing thermals & CPU use (8 Core):
Apple Solid State Drives
The commercial market attempts to sell "fast" drives that hold a very high read & write for sequential disk operations. The Apple drives do a good job doing this. You can see this reflected in Blackmagic Disk Speed tests. The problem is, most of our operations fall within the 4K realm or close (ref: thessdreview.com). Per Les Tokar, the 4k or 4kQD32 results are what we should be looking at. Apple also does a pretty good job here as well. We used AmorphousDiskMark which is a authorized OsX conversion of Crystal Disk Mark, but does not use the same read/write methods so they will differ from windows. You can only compare Amorphous tests to other Amorphous tests.
1TB:
From the teardown, we know that the 1TB drives consists of two 512GB modules connected in RAID.
(c/o: bplein)
2TB:
Likely, the 2TB drives are also two 1TB modules are likely connected in RAID as well, but no teardown has been performed with this configuration.
GPU Tests:
Thanks to bplein, we got an early look at the graphics tests. He used a Vega 64 card & posted Cinebench R15 results. These are good to reference against a GPU master list of benchmarks located at https://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html . To update the default list, change it to show all GPUs (desktop & mobile) & add Cenebench R15.
(c/o: bplein)
With 1255.99 fps, this holds its own against nVidia. The notebook check results are currently limited with the high end cards, so the 1080 TI SLI desktop appears lower at 119fps, but there are benchmarks of the 1080 being in the 160 range. You'd have to dig into the card details to see those results. The performance should be along the line of the nVidia 1070 desktop GPU.
Awaiting Vega 56 results.
First 8 core hand's on
10 core hand's on
iMac Pro Ram Pricing & Availability
VESA Mount
Things to expect:
Noise tests
Video Tests
(I will need help from someone who has an 8 core with Vega 56 and someone with a 10 core with Vega 64)
iMac Pro Ram Upgrades:
So people have read that the iMac Pro can have the ram upgraded, just not by us. I actually talked with Apple about six month ago about this very subject & here's what I learned:
1) Teardown enthusiasts can upgrade their own ram at the cost of risking warranty violation
2) Apple will upgrade the ram at Apple RAM cost (you buy their RAM still)
3) A third party authorized apple service center will upgrade your ram by your choice of either Apple provided RAM or:
You can BRING YOUR OWN RAM & have the authorized third party service center install it. Your warranty will be retained. The RAM will only carry the RAM's manufacturer's warrant & you will have to have RAM related issues serviced by a third party after the install. Apple will replace it with the original quantity installed upon purchase unless you upgrade through apple. So if I upgrade 32GB RAM to 128 on my own & it flakes out, If I take it to Apple, they will replace it with 32GB Apple RAM. The authorized third party will pull the RAM & ask you what to do...that's where you can have it sent in to be replaced on your own & bring the replacement to be installed.
For where I live, "Core Care of Rocklin" provides this service. Apple sent me there when I needed my macbook's battery replaced but I still needed to keep access to the laptop. Core Care did their test, ordered the battery, & called me back to supply the MacBook when they were ready to install it. 1 hr turnaround. That's what third parties can do for you. The cost of replacing RAM will be probably $70-150 plus the cost of the RAM.
So how do you find an authorized third party service center? Just ask Apple support & they will provide you the list of nearest authorized service centers. To know they are good, they will always run a diagnostics report & submit it to Apple & give you a copy. This is what Apple does as well when they do their work.
So what are the risks?
1) it costs a service center service payment
2) They crack the screen open so you risk dust between the glass & the LCD
3) The entire Motherboard needs to be pulled out so you risk damage. Humans are imperfect. I once had a Dell Laptop returned from the service center with the trackpad broken. I couldn't fix the motherboard, but I could tear it apart myself to find that they forgot to plug the trackpad back in...one of over a dozen connections.
4) Compatibility issues: not all RAM is equal & you could have issues. Have them do a full memory test before giving it back. 2017 iMac owners learned this when using Kingston's low CL ram...they learned that if you mix it, the RAM down clocks.
I hope this helps & have a great day!
CPU Geekbench Benchmarks:
Initial benchmarks from Geekbench:
8 core: 4937 single thread / 30088 multi thread
10 core 2: 4971 single thread / 34006 multi thread
10 core: 5345 single thread / 35917 multi thread
i7 (2017 imac) 5715 single thread / 19675 multi thread
10 core is 8.2% faster on single threaded operations and 19.4% faster on multi threaded operations
Keep in mind that this is a synthetic total number and if you look at the single core ops, each will be better at certain things.
10 core: https://browser.geekbench.com/v4/cpu/4449687
10 core 2: http://browser.geekbench.com/v4/cpu/5469097
8 core: https://browser.geekbench.com/v4/cpu/5471001
i7: https://browser.geekbench.com/v4/cpu/5543682
Source:
http://browser.geekbench.com/v4/cpu/search?utf8=✓&q=Mac-7BA5B2D9E42DDD94
Update: I added a second core 2 that popped up. Basically, it shows the first benchmark where the single thread operations almost matches the 8 core...which is more along the line of how I was expecting the two cores to perform. Basically, I'm waiting for a series of longer benchmarks to get the cpus to perform more continuously for rigor.
Basically, the CPU needs to be measured where it is actually stressed. A split second, momentary test will capture the max clock (4.5 vs 3.0 for the 10 core). This will artificially show the 10 core performing single thread ops as higher. Also, the cache has an affect which helps to boost performance on the 10 core. I expected some performance boost due to that, but not a 9% gain. I can say more certainly when more time consuming benchmarks come out.
As many of you are seeing, the 8 & 10 cores are as follows:
8 core: 3.2 GHz Base, max up to 4.3
10 core: 3.0 GHz Base, max up to 4.5
It's a $800 upgrade ($720 if discounted through EPP or Edu)
The performance should be about the same on the single core tasks & the 10 core will beat the 8 by 17% on multi core tasks. For me, this equates to 10 min savings on a normally 1hr project. I personally couldn't justify this upgrade. I'll post the 8 core benchmark when mine finally comes in.
Thermal Tests:
Thermals of the 10 core:
(c/o: nick82)
Thermals of the 8 core:
(c/o: bplein)
Takeaway:
The 8 core runs at about 3.9 GHz constantly; 93 degrees
The 10 core runs at about 3.5 GHz average; 95 degrees, but thermally clocking down
Geekbench shows single thread as a higher performer (which affects gaming) in the 10 core because the thermals off of the left image shows that it does not heat up quickly...& geek bench does not last long enough to clock down.
Drive Wiping & Reinstalling OsX High Sierra without DFU:
Be careful here. I used bootcamp & had an issue that required me to install 8 (which is not supported by the iMac Pro). I had to wipe the bootcamp partition because it was corrupted after wiping. Bootcamp assistant did not give me the option to remove the bootcamp partition on its own. Bootcamp would not allow me to reinstall after that either. If you want to do the upgrade method, provided you have upgraded once in the past, follow the link to bootcamp details:
https://forums.macrumors.com/threads/imac-pro-bootcamp-windows-10-limit-fix.2097652/
I finally booted into regular recovery, disabled boot security, & wiped the OSX partition as well...only the osx partition as there appeared to be more. I then created a new APFS partition & immediately went to install OSX. Everything went successful. I did not need a DFU boot.
Geekbench testing thermals & CPU use (8 Core):
Apple Solid State Drives
The commercial market attempts to sell "fast" drives that hold a very high read & write for sequential disk operations. The Apple drives do a good job doing this. You can see this reflected in Blackmagic Disk Speed tests. The problem is, most of our operations fall within the 4K realm or close (ref: thessdreview.com). Per Les Tokar, the 4k or 4kQD32 results are what we should be looking at. Apple also does a pretty good job here as well. We used AmorphousDiskMark which is a authorized OsX conversion of Crystal Disk Mark, but does not use the same read/write methods so they will differ from windows. You can only compare Amorphous tests to other Amorphous tests.
1TB:
From the teardown, we know that the 1TB drives consists of two 512GB modules connected in RAID.
(c/o: bplein)
2TB:
Likely, the 2TB drives are also two 1TB modules are likely connected in RAID as well, but no teardown has been performed with this configuration.
GPU Tests:
Thanks to bplein, we got an early look at the graphics tests. He used a Vega 64 card & posted Cinebench R15 results. These are good to reference against a GPU master list of benchmarks located at https://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html . To update the default list, change it to show all GPUs (desktop & mobile) & add Cenebench R15.
(c/o: bplein)
With 1255.99 fps, this holds its own against nVidia. The notebook check results are currently limited with the high end cards, so the 1080 TI SLI desktop appears lower at 119fps, but there are benchmarks of the 1080 being in the 160 range. You'd have to dig into the card details to see those results. The performance should be along the line of the nVidia 1070 desktop GPU.
Awaiting Vega 56 results.
Attachments
Last edited:
