Retina iMac 1866 MHz RAM Working

what's the increase in real-world performance though? Just by looking at the raw numbers, the speed bump seems minimal.

It would be a better comparison running various programs or games for benchmark.

Appreciate the effort though!

I wondered about this.
Thank you for posting a comparison.

----------

dragon529 said:

the speed bump seems minimal.

Click to expand...

seems 2%

Is it safe to do that?

Just a small bump for that much more money. Thanks for the update.

Im planning on buying these.

http://www.kingston.com/datasheets/H...11IB2K2_16.pdf

4x8GB
DDR3L-2133MHZ (PC3-17000)
DDR3-2133MHZ CL11-12-13 @1.35V or 1.5V

Anyone knows if they will work?

Tod

Again, appreciate the head-to-head comparison. I happened to be choosing between the two, but reports of the machine running toasty (fans kicking in) pushed me towards the stock speed RAM.

This was also discussed in this thread:

https://forums.macrumors.com/showthread.php?p=20274244#post20274244

One difficulty here is that Geekbench, except for the Memory Performance tests, has benchmarks that are fairly insensitive to memory speed.

(there's one exception, the AES test, which is more memory-bound, does get the 25% speedup)

In memory performance tests specifically, the gain is considerable, at least 25%.

There is also some evidence that window display, which involves pushing lots of bitmaps quickly from main memory to the GPU, benefits significantly.

for an additional $60... it's debatable haha

dragon529 said:

for an additional $60... it's debatable haha

Click to expand...

All upgrades are debatable, but you're talking to a group that spends $30 just to have their iMac delivered 2 days sooner.

And it is just about the only way (besides SSD) to speed up ordinary operations that don't tax the CPU or GPU at all. Web browsing in particular, where each page can have a 100 or 200 MB memory footprint, and your i5 or i7 CPU stays equally untapped waiting for RAM.

I know people love Geekbench but this is a real blind spot, microbenchmarks have nothing like the memory access patterns of real applications.

So what does that mean? Only the new Retina iMac supports greater than DDR3-1600MHz? The difference between 1866 and 1600 is negligible in my view.

What about 2133MHz RAM or even faster?

I would go safe with these Kingston HyperX Impact SODIMM which are for Apple computers, and they run at CL9 which a a lot better the the stock CL11 RAM that comes with the iMacs

Part Number: HX316LS9IBK2/16
Specs: , DDR3L, 1600MHz, CL9, 1.35V,

http://www.kingston.com/en/memory/search?DeviceType=2&Mfr=APP&Line=iMac&Model=90781

How is this able to run faster by just sticking it in? I thought that would require some settings changes at the system board level. I would have assumed cas and ram speed are set no matter what higher spec you throw in there. I guess I'm missing something.....

xgman said:

How is this able to run faster by just sticking it in?

Click to expand...

It is possible actually.

e.g. The 5k iMac is actually capable to run the RAM at 1866MHz, but Apple use slower RAM for cost saving. In this case, the whole system will automatically adopt to the RAM and lower the clock to 1600MHz. And all you need to do is upgrade the RAM by yourself and then the system will automatically increase the clock speed to 1866MHz (a PRAM reset may required).

However, I think the iMac will only use the preset CL value, no matter what RAM you through in. If the stock RAM run at 1600 CL11, install a 1600 CL9 RAM will either run the RAM at CL11 (depends on the RAM), or not boot at all. All these RAM parameters are in the firmware, and has no option for user.

While the numbers may look better with the sport Ram, the real life performance difference will be negligible at best
It was a good idea to try but not worth the cost or performance boost

h9826790 said:

It is possible actually.

e.g. The 5k iMac is actually capable to run the RAM at 1866MHz, but Apple use slower RAM for cost saving. In this case, the whole system will automatically adopt to the RAM and lower the clock to 1600MHz. And all you need to do is upgrade the RAM by yourself and then the system will automatically increase the clock speed to 1866MHz (a PRAM reset may required).

However, I think the iMac will only use the preset CL value, no matter what RAM you through in. If the stock RAM run at 1600 CL11, install a 1600 CL9 RAM will either run the RAM at CL11 (depends on the RAM), or not boot at all. All these RAM parameters are in the firmware, and has no option for user.

Click to expand...

Your second paragraph is not correct. The system will sync the ram at the fastest allowable timings as defined in the ram's SPD, so if it's CL9 ram, it will run it at CL9.

Be aware however that you should not mix ram of different timings. If you do, the system will drop the bus speed to the fastest speed at which ALL the ram can run the most aggressive timings. I'll paraphrase that in case it's not clear: The system will first choose CL9 (say) and then decide what clock speed can ALL the memory modules manage at CL9. In the case of Apple-supplied ram, this is 1333MHz.

So the net effect if if you drop CL9 ram into a system as an upgrade, you must replace the existing ram, not add to it. If you add to the existing ram, you will end up running your whole memory bus at 1333MHz instead of 1600MHz. This is because the stock ram cannot run at 1600MHz with CL9 timing.

The system will work OK, but it will not be running as fast as if you had just added CL11 ram instead!

Chippy99 said:

I'll paraphrase that in case it's not clear: The system will first choose CL9 (say) and then decide what clock speed can ALL the memory modules manage at CL9. In the case of Apple-supplied ram, this is 1333MHz.

Click to expand...

That's interesting. May be the iMac works differently than the Mac Pro. My old Mac Pro able to run the RAM at 1066 CL7 or 1333 CL9. The system will automatically choose the higher speed (1333 CL9) if available. Only if I install the RAM in the non-optimum configuration (with exactly the same RAM sticks, but not fully utilise the triple channel architecture), then the system will downgrade to 1066 CL7 (and the benchmark confirm it is a downgrade).

Of course, the iMac can perform in the other way around. I am not sure about that.

h9826790 said:

That's interesting. May be the iMac works differently than the Mac Pro. My old Mac Pro able to run the RAM at 1066 CL7 or 1333 CL9. The system will automatically choose the higher speed (1333 CL9) if available. Only if I install the RAM in the non-optimum configuration (with exactly the same RAM sticks, but not fully utilise the triple channel architecture), then the system will downgrade to 1066 CL7 (and the benchmark confirm it is a downgrade).

Of course, the iMac can perform in the other way around. I am not sure about that.

Click to expand...

This is how it works with late 2012 and late 2013 iMacs for sure. I don't know about RiMac, but I would imagine its the same since it is basically the same architecture.

Chippy99 said:

This is how it works with late 2012 and late 2013 iMacs for sure. I don't know about RiMac, but I would imagine its the same since it is basically the same architecture.

Click to expand...

Thanks for the info. It's good to know.

So how much so-called "real world performance" are we talking about here? And I count simple responsiveness as a need, so if everything becomes somewhat snappier, do let me in on it, won't you?

Crunch said:

So how much so-called "real world performance" are we talking about here? And I count simple responsiveness as a need, so if everything becomes somewhat snappier, do let me in on it, won't you?

Click to expand...

There's no perceptible difference between RAM speeds all things else being equal. Benchmarks will show very small differences with certain applications (and no differences with most), and with extremely RAM intensive applications, there can be benefits over the long term.

We're at a point where each generation of faster technology is only going to bring a couple of percentage points here, and few percentage points there. Add them all up over a few years, it makes a difference.

ixxx69 said:

There's no perceptible difference between RAM speeds all things else being equal. Benchmarks will show very small differences with certain applications (and no differences with most), and with extremely RAM intensive applications, there can be benefits over the long term.

We're at a point where each generation of faster technology is only going to bring a couple of percentage points here, and few percentage points there. Add them all up over a few years, it makes a difference.

Click to expand...

I'd love to get your take on DDR3 vs. the upcoming DDR4. How big a change/difference are we talking about here?

Thanks!

Crunch said:

So what does that mean? Only the new Retina iMac supports greater than DDR3-1600MHz? The difference between 1866 and 1600 is negligible in my view.

What about 2133MHz RAM or even faster?

Click to expand...

The 2013 iMacs support faster than 1600MHz RAM also, mine has been running 32GB of 1866 since early last year.

Hi!

I just ordered a 5K iMac with the i7 processor and 8GB of RAM. Can I use the two 16GB modules (see attachment) that I originally bought for my "old" 2012 iMac?

I'm not sure if the timings will be compatible.

Cheers,
Chris