Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Ok, my iMac is setup. Here are the results...

i7 4GHz, 512 SSD, 395X

Stock 8GB, 1866MHz CL13
GB_8GB_1833_CL13.png


4 x Crucial 16GB Kit (8GBx2) DDR3/DDR3L-1600 MHz (PC3-12800) CL11 204-Pin SODIMM Memory for Mac CT2K8G3S160BM / CT2C8G3S160BM
http://www.amazon.com/gp/product/B008LTBJFW?psc=1&redirect=true&ref_=oh_aui_detailpage_o02_s00

GB_32GB_1600_CL13.png


4 x Crucial Ballistix Sport 16GB Kit (8GBx2) DDR3 1866 (PC3-14900) 204-Pin SODIMM Memory BLS2K8G3N18AES4 / BLS2C8G3N18AES4
http://www.amazon.com/gp/product/B00DSGLMSM?psc=1&redirect=true&ref_=oh_aui_detailpage_o01_s00

GB_32GB_1833_CL10.png


4 x Kingston Technology HyperX Impact Black 8GB 2133MHz DDR3L 1.35V Laptop Memory (HX321LS11IB2/8)
http://www.amazon.com/gp/product/B00UVOVV12?psc=1&redirect=true&ref_=oh_aui_detailpage_o00_s00

GB_32GB_2133_CL11.png
 
Now we just need some kind of test for stability. Like a long export, game, YouTube at 4K, I don't know.

So 2133 MHz vs. 1866 MHz ... 14% faster RAM results into less than 5% performance improvement. For myself, not worth it.

Makes me wonder about the SO-DIMM socket design of Apple. Did they take these higher speed into account from a signal integrity perspective or not... could be SO-DIMM dependent (so one manufacturer works, the other not). So maybe you just got lucky. Usually on mobo's there's a PDF file showing which ones work. Apple of course leaves you in the dark.
 
  • Like
Reactions: smgfreak
Now we just need some kind of test for stability. Like a long export, game, YouTube at 4K, I don't know.
I don't see why it wouldn't been stable: Intel CPUs officially support 2400MHz RAM (even fastest for "k" processors). 2133 MHz is a standard RAM speed for desktop CPUs.
And if you look deep inside MacRumors threads, you will find no complaints of 2133MHz RAM instability on 2014 Retina iMacs ;)
 
  • Like
Reactions: ohsnaphappy
So 2133 MHz vs. 1866 MHz ... 14% faster RAM results into less than 5% performance improvement. For myself, not worth it.

Makes me wonder about the SO-DIMM socket design of Apple. Did they take these higher speed into account from a signal integrity perspective or not... could be SO-DIMM dependent (so one manufacturer works, the other not). So maybe you just got lucky. Usually on mobo's there's a PDF file showing which ones work. Apple of course leaves you in the dark.
One more time: the memory controller is inside Intel CPUs, nobody is in the dark and the mobo has nothing to do in the story ;)
 
So 2133 MHz vs. 1866 MHz ... 14% faster RAM results into less than 5% performance improvement. For myself, not worth it.

It was also 25% more $$$. However I don't think it is as easy as that. Some people just want the fastest for a reasonable amount of money ($50 more is no big deal, especially when paying $3K for the iMac). Now $1200 for 64GB is not reasonable.
 
I don't see why it wouldn't been stable: Intel CPUs officially support 2400MHz RAM (even fastest for "k" processors). 2133 MHz is a standard RAM speed for desktop CPUs.
And if you look deep inside MacRumors threads, you will find no complaints of 2133MHz RAM instability on 2014 Retina iMacs ;)
Desktop CPU, laptop RAM. Notice nobody is making 2400 SO-DIMMS, only DIMMS for desktop motherboards. All the 2400 stuff has giant heat spreaders. I wonder if the lack of 2400 is due to heat concerns, crosstalk between pins or something else.
 
  • Like
Reactions: jordanz
Desktop CPU, laptop RAM. Notice nobody is making 2400 SO-DIMMS, only DIMMS for desktop motherboards. All the 2400 stuff has giant heat spreaders. I wonder if the lack of 2400 is due to heat concerns, crosstalk between pins or something else.
I don't think there is the market for ultra-expensive 2400MHz laptop RAM ;)
 
I don't think there is the market for ultra-expensive 2400MHz laptop RAM ;)
Perhaps. Hasn't stopped everybody under the sun from making it for desktops though.

Kingston showed some off almost 2 years ago, and even had a part number for it, but it doesn't look like it ever went into production.

http://www.legitreviews.com/kingsto...00mhz-performance-so-dimms-at-ces-2014_133654

Makes me wonder if there is a heat/crosstalk issue in the SO-DIMM form-factor that couldn't be overcome.

Or perhaps they just focused on DDR4 instead.
 
So 2133 MHz vs. 1866 MHz ... 14% faster RAM results into less than 5% performance improvement. For myself, not worth it.

.

There isn't that much difference in price is you shop around.
 
It's also worth pointing out that the iMac comes with a PCIe SSD which is faster and more expensive than its SATA counterparts.

It sounds like you really love your Hackintosh. It could probably even run DDR4 now that there's Skylake support. Congratulations. Not sure why you feel compelled to sh*t on the iMac party since you can talk Hackintosh anywhere, but whatevs.
 
[MOD NOTE]
A number of posts were removed for rules violations. This thread is about 2133mhz Ram Test Results not hackintoshes. Please stay on topic.
 
[MOD NOTE]
A number of posts were removed for rules violations. This thread is about 2133mhz Ram Test Results not hackintoshes. Please stay on topic.

Interestingly, according to Intel data, the 6XXX series (skylake) only supports 2133MHz and 1867MHz RAM for DDR4. the maximum DDR3 speed supported is 1600MHz : http://ark.intel.com/products/88200/Intel-Core-i7-6700T-Processor-8M-Cache-up-to-3_60-GHz

Anandtech has an article about the DDR4 vs DDR3 debate:
http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/7

IMHO, what the benchmarks above are measuring isn't about RAM frequency (seeing that the i5/i7 doesn't support 2133MHz DDR3, RAM wouldn't run at that speed anyway), but just showing that the various RAM have different latencies, and some are much better than others.
 
Interestingly, according to Intel data, the 6XXX series (skylake) only supports 2133MHz and 1867MHz RAM for DDR4. the maximum DDR3 speed supported is 1600MHz : http://ark.intel.com/products/88200/Intel-Core-i7-6700T-Processor-8M-Cache-up-to-3_60-GHz

Anandtech has an article about the DDR4 vs DDR3 debate:
http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/7

IMHO, what the benchmarks above are measuring isn't about RAM frequency (seeing that the i5/i7 doesn't support 2133MHz DDR3, RAM wouldn't run at that speed anyway), but just showing that the various RAM have different latencies, and some are much better than others.
No, the RAM is at the correct frequency, you can check with CPUz if you have Windows. It is not in Intel specs (I don't know why!), but i5/i7 CPUs sold today can handle at least 2400MHz memory.
Even Apple writes that the last Retina iMac runs its DDR3 RAM at 1867MHz ;)
 
Last edited:
IMHO, what the benchmarks above are measuring isn't about RAM frequency (seeing that the i5/i7 doesn't support 2133MHz DDR3, RAM wouldn't run at that speed anyway), but just showing that the various RAM have different latencies, and some are much better than others.

The memory I posted above that got the best score had the highest frequency, but not the lowest latency. Frequency has to have something to do with it.

(CL10) Crucial Ballistix Sport 16GB Kit (8GBx2) DDR3 1866 (PC3-14900) 204-Pin SODIMM Memory BLS2K8G3N18AES4 / BLS2C8G3N18AES4

(CL11) Kingston Technology HyperX Impact Black 8GB 2133MHz DDR3L 1.35V Laptop Memory (HX321LS11IB2/8)
 
The memory I posted above that got the best score had the highest frequency, but not the lowest latency. Frequency has to have something to do with it.

(CL10) Crucial Ballistix Sport 16GB Kit (8GBx2) DDR3 1866 (PC3-14900) 204-Pin SODIMM Memory BLS2K8G3N18AES4 / BLS2C8G3N18AES4

(CL11) Kingston Technology HyperX Impact Black 8GB 2133MHz DDR3L 1.35V Laptop Memory (HX321LS11IB2/8)
I think you get slightly different results everytime you run it.. so you need to run it probably like 10 times on each configuration and see the average and then say if there is visible difference. But just based on these it looks like HyperX was the correct choice. For those worried 2133MHz might be too much.. according to Kingston HyperX has been tested to run on 2400MHz too, so this 2133MHz shouldn't be a problem. It should automatically overclock itself to whatever is the highest your system can support. Still wondering though how on last year's 5K iMac results someone got theirs running with 2400MHz memory. Maybe they booted to windows and overclocked manually though I don't know if it would show on mac results after that.
 
I think you get slightly different results everytime you run it.. so you need to run it probably like 10 times on each configuration...

I'll give you that, however I started out running each one 5 times, I got more half way through the memory about 14 runs as I had 4 sets (4 sets x 5 runs) and each time the single and multi core values for the memory performance always fell within a range of 20-30 points and the overall geekbench scores always fell between 80-100.

Performance is a factor of both latency and frequency. Take a look at the first two I tested. The stock at 1833MHz CL13 underperformed 1600MHz CL11 by a good margin. However I had 32GB instead of 8GB, not sure if it matters for the test.
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Does anyone have any experience with Corsair? I've been looking for 2133MHz ram everywhere thats in stock and I finally found this over at amazon. Its in stock and is shipped and fulfilled by amazon. I plan to use my new iMac for video editing and hope the faster ram will help.

http://www.amazon.com/Corsair-Venge...445567869&sr=8-5&keywords=2133mhz+ddr3+sodimm

According to the specs it looks like it will work. I haven't seen seen anyone post anything related to these specs not working.
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
The memory I posted above that got the best score had the highest frequency, but not the lowest latency. Frequency has to have something to do with it.

(CL10) Crucial Ballistix Sport 16GB Kit (8GBx2) DDR3 1866 (PC3-14900) 204-Pin SODIMM Memory BLS2K8G3N18AES4 / BLS2C8G3N18AES4

(CL11) Kingston Technology HyperX Impact Black 8GB 2133MHz DDR3L 1.35V Laptop Memory (HX321LS11IB2/8)
I've checked with CPUz, I can assure you that the RAM is running at the speed reported in Mac OS X. I don't know why Intel is stuck with 1600 MHz in their specs...
And think about it: in the new Retina iMacs, Apple himself tells us the RAM is 1866 MHz... And all these +1600MHz DDR3 modules sold around the world, most of them are for Intel CPU, not to look pretty on a shelf in the living room ;)

I will post you my CPUz screenshot, where you can see that my (old!) i7 of 2013, for which Intel (and Apple) gives us 1600MHz max for RAM, is perfectly handling my 1866MHz RAM.
CPUz reports a DRAM frequency of 933MHz (1866/2 - DDR if for Dual Data Rate), and I bet that for those who have 2133MHz RAM, CPUz will report a DRAM frequency of 1066 ;)
 
I've checked with CPUz, I can assure you that the RAM is running at the speed reported in Mac OS X. I don't know why Intel is stuck with 1600 MHz in their specs...
And think about it: in the new Retina iMacs, Apple himself tells us the RAM is 1866 MHz... And all these +1600MHz DDR3 modules sold around the world, most of them are for Intel CPU, not to look pretty on a shelf in the living room ;)

That may be simply because of what EFI/Bios Emulation is reported to the OS.

When I plug a DDR3 1333MHz in a 2011 iMac 27" and a 1666MHz SODIMM so they are both in the same memory bus ; about your mac shows me both RAM with the speed for both.
And we all know that it's impossible for RAMs to run at different speed in those conditions (nor can can the i7-2600 run 1600MHz DDR3 at that speed).

So I don't trust that report. It just prints the type of RAM it detects.
 
That may be simply because of what EFI/Bios Emulation is reported to the OS.

When I plug a DDR3 1333MHz in a 2011 iMac 27" and a 1666MHz SODIMM so they are both in the same memory bus ; about your mac shows me both RAM with the speed for both.
And we all know that it's impossible for RAMs to run at different speed in those conditions (nor can can the i7-2600 run 1600MHz DDR3 at that speed).

So I don't trust that report. It just prints the type of RAM it detects.
No, it doesn't just print that report. Please check how CPUz works before telling nonsense. CPUz reports instant frequencies of your system (CPU, bus and RAM).
And I've never said RAM modules can run at different speed in one machine: we agree, all modules run at the same speed. And my advice is the good one: if you mix with stock RAM, take 1866MHz CL13 modules for the late 2015 Retina iMac (1600MHz CL11 for the previous gen). And if you don't mix, take the fastest frequency and the shortest CL you can afford (and don't mix different modules: all must have the same speed and CL, same brand if possible).

And no doubt about it: if you don't mix modules, iMacs can handle without a problem 2133MHz modules running at 2133MHz (check the 2014 Retina iMac memory thread in MacRumors, a lot of successful testimonies there). Apple is not manufacturing its computers from dragon scales: iMacs are Haswell/Broadwell/Skylake standard platforms and they can deal with 1600, 1866 and 2133MHz RAM speed like any other Haswell/Broadwell/Skylake platform.

I will post you CPUz screenshots when I will be back home, where you can read instant frequency of your system, and I can assure you my memory is running at its nominal speed, not at the 1600MHz of Intel and Apple specs :) And those who have 2133MHz can check that too if they have bootcamp.
 
No, it doesn't just print that report. Please check how CPUz works before telling nonsense. CPUz reports instant frequencies of your system (CPU, bus and RAM).
And I've never said RAM modules can run at different speed in one machine: we agree, all modules run at the same speed.

And this is why I've mentioned: the "System Report" on mac shows them in the memory list with DIMM1 as DDR3 1600 and DIMM2 as 1333; even though they can't be running at those speeds.
Hence my distrust on what it displays there is meaningless when it comes to determining what speed the DIMMs actually run at.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.