Intel Core i9 Processor And Intel Xeon W Processor

EDLIU

macrumors regular
Original poster
Oct 20, 2015
169
1
What are the differences between the "8-core 9th-generation Intel Core i9 processor(iMac)" and "8-core Intel Xeon W processor(iMac Pro)"?

What are their advantages respectively?

Thanks.
 

joema2

macrumors 68000
Sep 3, 2013
1,580
789
What are the differences between the "8-core 9th-generation Intel Core i9 processor(iMac)" and "8-core Intel Xeon W processor(iMac Pro)"?

What are their advantages respectively?...
The 8-core Xeon W-2145B is used in the iMac Pro and the i9-9900k in the 2019 iMac. They both use 14 nm lithography.

In general it appears the i9-9900k in the 2019 iMac provides roughly similar performance to the 8-core iMac Pro for less money. The iMP still has better GPU performance than the Vega48 in the 2019 iMac. It appears the 2019 i9 iMac is quieter than the 2017 i7 model but not as quiet as the iMP.

The iMP has more ports, two separate Thunderbolt controllers, and 10-gig Ethernet.

I think the Xeon W-2145 is Skylake or 7th generation, the i9-9900k is "Canon Lake Refresh" (CLR) or 9th generation. CLR has some hardware mitigations for the Spectre/Meltdown vulnerabilities, whereas older CPUs have to use firmware or OS patches. This might make a performance difference on some workloads but I don't think anyone has specifically tested that yet.

For video editing or transcoding the i9-9900K has the latest generation of Intel's Quick Sync. If software uses this it can accelerate encode/decode to/from H264 or H265/HEVC significantly. Xeon does not have this. On the iMac Pro FCPX uses AMD's similar (but less effective) technology called UVD/VCE.

The Xeon supports ECC RAM (used by the iMac Pro), and the i9 does not. In theory this can improve reliability. I don't know reliable numbers but I've seen estimates that without ECC RAM, a single-bit error every week is possible.

The i9-9900K has somewhat better single-thread performance.

Apple's power consumption numbers state for similar configurations the 2019 i9 iMac 27 consumes 262W whereas the 2017 i7 iMac 27 consumed 217W. However limited testing to date has shown the i9 iMac apparently runs cooler with less fan noise than the 2017 iMac. The reason for this has yet to be determined.

https://support.apple.com/en-us/HT201918

Apple doesn't list power consumption for the 8-core iMac Pro but for the 18-core it is 370W: https://support.apple.com/en-us/HT208378 Of course the iMP has a greatly improved cooling system vs the iMac.

https://ark.intel.com/content/www/us/en/ark/compare.html?productIds=126707,186605
 
Last edited:

carlos700

macrumors 6502
Dec 17, 2004
263
35
Omaha, NE
That's a really good reply. The only thing I would add is the Xeon W has 48 PCI-Express lanes while the Core i9 has 16. You can see that play out in the number of Thunderbolt 3 ports each iMac has.
 

curmudgeonette

macrumors 6502a
Jan 28, 2016
507
325
California
The on chip bus is different between Xeon and i-series. The Xeon has some sort of crossbar matrix. This allows any core to quickly get to memory. I think in the i-series it is a ring bus that loops past each core, the GPU, and the memory port. This ring bus arrangement is fine for a quad core, but may not be so good for an octacore.
 
  • Like
Reactions: AlexJoda

AlexJoda

macrumors 6502a
Apr 8, 2015
758
579
The 8-core Xeon W-2145B is used in the iMac Pro and the i9-9900k in the 2019 iMac. They both use 14 nm lithography.

In general it appears the i9-9900k in the 2019 iMac provides roughly similar performance to the 8-core iMac Pro for less money. The iMP still has better GPU performance than the Vega48 in the 2019 iMac. It appears the 2019 i9 iMac is quieter than the 2017 i7 model but not as quiet as the iMP.

The iMP has more ports, two separate Thunderbolt controllers, and 10-gig Ethernet.

I think the Xeon W-2145 is Skylake or 7th generation, the i9-9900k is "Canon Lake Refresh" (CLR) or 9th generation. CLR has some hardware mitigations for the Spectre/Meltdown vulnerabilities, whereas older CPUs have to use firmware or OS patches. This might make a performance difference on some workloads but I don't think anyone has specifically tested that yet.

For video editing or transcoding the i9-9900K has the latest generation of Intel's Quick Sync. If software uses this it can accelerate encode/decode to/from H264 or H265/HEVC significantly. Xeon does not have this. On the iMac Pro FCPX uses AMD's similar (but less effective) technology called UVD/VCE.

The Xeon supports ECC RAM (used by the iMac Pro), and the i9 does not. In theory this can improve reliability. I don't know reliable numbers but I've seen estimates that without ECC RAM, a single-bit error every week is possible.

The i9-9900K has somewhat better single-thread performance.

Apple's power consumption numbers state for similar configurations the 2019 i9 iMac 27 consumes 262W whereas the 2017 i7 iMac 27 consumed 217W. However limited testing to date has shown the i9 iMac apparently runs cooler with less fan noise than the 2017 iMac. The reason for this has yet to be determined.

https://support.apple.com/en-us/HT201918

Apple doesn't list power consumption for the 8-core iMac Pro but for the 18-core it is 370W: https://support.apple.com/en-us/HT208378 Of course the iMP has a greatly improved cooling system vs the iMac.

https://ark.intel.com/content/www/us/en/ark/compare.html?productIds=126707,186605
Didn't I read somewhere that the T2 chip can also do Video en/decoding? The HVEC performance of the iMac Pro is not so bad as one would think without Quick Sync...(seems a little bit similar to the new MacBook Air, which also uses its T2 chip for en/decoding)
 
  • Like
Reactions: rurza

adamk77

macrumors 6502a
Jan 6, 2008
513
149
The Xeon supports ECC RAM (used by the iMac Pro), and the i9 does not. In theory this can improve reliability. I don't know reliable numbers but I've seen estimates that without ECC RAM, a single-bit error every week is possible.
That was a wealth of information I didn't know about. Thanks for the great info.

I've never used ECC RAM, so never looked into it. It's definitely interesting though.

Full text in https://www.intelligentmemory.com/support/faq/ecc-dram/how-often-do-ecc-correctable-single-bit-errors-occur-and-how-about-double-multi-bit-errors.php

This study monitored the DRAM errors in the thousands of systems of the famous Google server-farm for a period of 2 1/2 years. All those servers were surely perfectly air-conditioned, dust-free and protected from radiations of all kinds. Still they came to the result of 25000 to 70000 FIT (failures per billion device hours) of 'ECC correctable errors' per Megabit of DRAM. This converts into an average of one single-bit-error every 14 to 40 hours per Gigabit of DRAM.

The field study also explains that the error-rate increases by the age of the memory. Brandnew DRAMs might not show any errors for weeks and months, but then the error-rate suddenly goes up.

Uncorrectable errors could be double- or multi-bit errors or complete functional fails of the DRAM. These can all not be corrected, but are extremely rare.

A 1 Gigabit ECC DRAM contains 16 Million blocks of 64 bit datawords. Per each of these 64 bit words, one error is correctable. In other words: Statistically one out of 16 million hits might be a double-bit error. If one error hits per day, this would mean that it takes hypothetically 16 Million days or 48000 years for a double-bit error to hit. But this is just the maths. Finally the real numbers depend on the stress and the environment the application is running in.