thanks. I ended up ordering the 2133 Kingston Ram even though I probably won't notice the difference. It wasn't that much more expensive than the 1866 Ram so I went for it.
Oh yes, I know about that... but that is just a dump of the actual spec... does;t mean it's performing to that speed - you were asking about it performing at the 2166 speeds
I purchased these G.SKILL Ripjaws Series 16GB (2 x 8G) 204-Pin DDR3 SO-DIMM DDR3 1866 (PC3 14900) which appear to have the same specs as your Kingston HyperX - 1866 MHz CL11. However, when paired with the Apple 8 GB, speed drops to 1600 MHz.
So there is something different between the HyperX and the Ripjaws such that yours stayed at 1867 with the Apple RAM and mine didn't.
I purchased these G.SKILL Ripjaws Series 16GB (2 x 8G) 204-Pin DDR3 SO-DIMM DDR3 1866 (PC3 14900) which appear to have the same specs as your Kingston HyperX - 1866 MHz CL11. However, when paired with the Apple 8 GB, speed drops to 1600 MHz.
Now I need to decide whether to:
- Live with 1600 MHz RAM (really, what is the real world difference anyway?)
- Live with 16 GB at 1867 MHz (removing Apple RAM)
- Or buy another 16 GB of Ripjaws for 32 GB at 1867 MHz (will use jet.com deal if so)
That seems strange. The CAS latency 10 Corsair Vengeance forces a clock down to 1600 MHz, while the CAS latency 11 Kingston HyperX allows the system to keep going at 1867 MHz?
Yes, the Ripjaw memory by itself runs at 1867 MHz. Just like the original Apple RAM.
By itself the ripjaws should be running at 2133mhz by themselves nit 1867mhz unless they do not support autooverclocking like the Kingstons do.Yes, the Ripjaw memory by itself runs at 1867 MHz. Just like the original Apple RAM.
I decided to do a little speed test to see how much performance I'm really losing by combining sets and stepping down to 1600 MHz. I tested a scenario where increased speed, if any, would have a measurable impact - encoding a video with ffmpeg. I'm sure this is not the best pure test of memory speed, but it's more relevant to my real-world usage. If there's not much difference, then do I really care?
Below are my results. I encoded the same file 4-6 times in each configuration and took the fastest run. 4 GHz i7 with 1 TB SSD. I also tested my 2012 Mac mini for reference (2.6 GHz quad-core i7, 16 GB memory at 1600 MHz, SATA SSD).
Results for each configuration varied by +- 1 second, so the first two are essentially equal and within the noise. The combined set was only about 1.8% slower. For a 2-hour encode, this translates to about 2 minutes longer.
- Ripjaw 16 GB @1867 MHz: 4:22.65
- Apple 8 GB @1867 MHz: 4:22.75
- Combined 24 GB @1600 MHz: 4:27.33
- Mac mini 16 GB @1600 MHz: 6:54.19
Compared to the Mac mini I was using previously, this is basically lost in the noise. The mini is about 58% slower - that hypothetical 2-hour encode would take an extra hour and 9 minutes on my old machine.
Based on this, I think I'm ok with keeping the combined sets running at 1600 MHz. Eventually when prices fall in a couple years, I'll jump to 64 GB, and hopefully at 1867 or 2133 MHz.
It might be interesting to test some other long encodes, such as with iMovie. I suspect this would depend more on the GPU than memory speed, but you don't really know until you try it. Even different ffmpeg options could stress memory speed differently, but in the end do I want to stress over every possible optimization or do I want to just use the machine?
I purchased these G.SKILL Ripjaws Series 16GB (2 x 8G) 204-Pin DDR3 SO-DIMM DDR3 1866 (PC3 14900) which appear to have the same specs as your Kingston HyperX - 1866 MHz CL11. However, when paired with the Apple 8 GB, speed drops to 1600 MHz.
Now I need to decide whether to:
- Live with 1600 MHz RAM (really, what is the real world difference anyway?)
- Live with 16 GB at 1867 MHz (removing Apple RAM)
- Or buy another 16 GB of Ripjaws for 32 GB at 1867 MHz (will use jet.com deal if so)
Considering that is EXACTLY what I have in my iMac now (ordered from Amazon when it was only $90, now jumped to 130), I think they are mistaken! They do have a memory that is similar with better timings. Not that it will make much difference. I've mixed it with my Apple RAM and all is fine.
