Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
I think I've actually managed to find a tool which might help out with testing isolated parts of the memory for bandwidth, which would help me to compare the speeds of the 4GB of stock memory to the 16GB kit I'm aiming for;

http://zsmith.co/bandwidth.html

We'll have to see what info we can get out of this once everything is in place, and of what value it is for the testing.

I have just built it and done some simple tests of it on my current MacBook to verify that it will be of any help in testing the performance later on. So far so good. :)
 

MJ23

macrumors newbie
Dec 29, 2012
8
0
In what daily use of your mac mini you will notice an improvement in the performance when you upgrade your RAM?
 

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
In what daily use of your mac mini you will notice an improvement in the performance when you upgrade your RAM?

In my case; quite a lot of my usage... Since my Mini (which hopefully arrives tomorrow or wednesday) comes with the stock 4GB which I'm definitely pushing above while doing general iOS development. This I know since running Xcode + iOS Simulator + a browser at the same time generates a lot of swapping on my MB white early 2009, which is currently equipped with 4GB.

As for if I would see any difference if I already had 16GB and then changing; no friggin' idea... And that is the whole issue: so far I've not seen any proper comparison/user stories of any of the different memory types that theoretically gives some kind of performance edge when being run in a 2012 Mini.

And as such, how am I to know if there are any noticeable differences between the 16GB models? Who says I'm not getting a huge performance increase (not very realistic)?

Since I'm going from 4 to 16, I can just as well buy a "good" kit of memory and see where my performance ends up.

I'm not saying that it would be a interesting move if I already had perfectly working set of 16GB, in that case I don't think I'd be nearly as interested in finding out the answers to my questions. :)
 

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
Yay! Mini just was delivered earlier today.

So, basis of my tests is now available...

Sadly I think there will be a bit of delay in before I can do any test, as the Corsair Vengeance kit is back-ordered at most of the swedish retailers with hopes of restocking it later this week, so I will see whenever it comes back in stock.

OTOH, the Kingston memories that would be my second hand choice is also not available in stock, so I'm stuck between a rock and a hard place in this choice. :)

In the meanwhile, I will try to perform some baseline tests to see if the Apple Hardware test gives info about the CAS Latency, and also if the memory bandwidth test may give any good numbers to compare to later on.
 

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
Status update is that I've successfully started the Mini and done a batch of test benchmarks with the stock memory in place.

Benchmarks I've done for comparison is XBench (minus the Threads test as it seems to hang XBench when run on anything higher than Lion), iBench, NovaBench and GeekBench. Nothing interesting with a stock Mini alone, so I'll wait till I've got the better memory to put up the numbers for comparison...

I can also give the bad news that not even the Apple Hardware Test seems to indicate memory timing, and as such we're left with only knowing what frequency they are being run on.

I'm not ruling it out entirely yet, as I only did the hardware profile, and no actual tests, but I'm not counting on doing the tests showing much more info on the CAS Latency.

So, to verify lower timings, one would have to do thorough performance tests, probably using a few different sets of memory specced at different timings, which could be quite a hassle to get a reliable result on...

And, once again, I'm pointed even more in the direction of the 1866MHz memories as the only ones I can test if they give any performance increase in a simple manner.
 

philipma1957

macrumors 603
Apr 13, 2010
6,362
248
Howell, New Jersey
Status update is that I've successfully started the Mini and done a batch of test benchmarks with the stock memory in place.

Benchmarks I've done for comparison is XBench (minus the Threads test as it seems to hang XBench when run on anything higher than Lion), iBench, NovaBench and GeekBench. Nothing interesting with a stock Mini alone, so I'll wait till I've got the better memory to put up the numbers for comparison...

I can also give the bad news that not even the Apple Hardware Test seems to indicate memory timing, and as such we're left with only knowing what frequency they are being run on.

I'm not ruling it out entirely yet, as I only did the hardware profile, and no actual tests, but I'm not counting on doing the tests showing much more info on the CAS Latency.

So, to verify lower timings, one would have to do thorough performance tests, probably using a few different sets of memory specced at different timings, which could be quite a hassle to get a reliable result on...

And, once again, I'm pointed even more in the direction of the 1866MHz memories as the only ones I can test if they give any performance increase in a simple manner.

you can get timings you need to burn memtest 86+ to a dvd stick it in a dvd player and restart with your finger on options pick the dvd player and you get this. those are a pair of crucial 8gb sticks timings are

11- 11- 11 -28

I ran that test for 21 hours. It was a 2.3 quad 2012.. if you can get the 1866 to run better then 1600 during 10.8.x ML I would be stunned.


They may unlock with windows running in bootcamp. I did some testing and decided a dead end. May opinion on the fastest you can do for a new mini is a pair of the kingston 8gb plug n play sticks


http://www.newegg.com/Product/Product.aspx?Item=N82E16820104317 these may time at 9-9-9-26 or so.


run the mini (quad 2.6) with a 500 gb samsung ssd and the stock 1tb hdd as a diy fusion call it a day. yeah a raid0 pair of samsung ssd's would be a little faster. but a lot more money
 

Attachments

  • IMG_2308.JPG
    IMG_2308.JPG
    2.9 MB · Views: 100
Last edited:

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
you can get timings you need to burn memtest 86+ to a dvd stick it in a dvd player and restart with your finger on options pick the dvd player and you get this. those are a pair of crucial 8gb sticks timings are

11- 11- 11 -28

I ran that test for 21 hours. It was a 2.3 quad 2012.. if you can get the 1866 to run better then 1600 during 10.8.x ML I would be stunned.


They may unlock with windows running in bootcamp. I did some testing and decided a dead end. May opinion on the fastest you can do for a new mini is a pair of the kingston 8gb plug n play sticks


http://www.newegg.com/Product/Product.aspx?Item=N82E16820104317 these may time at 9-9-9-26 or so.


run the mini (quad 2.6) with a 500 gb samsung ssd and the stock 1tb hdd as a diy fusion call it a day. yeah a raid0 pair of samsung ssd's would be a little faster. but a lot more money

Yes, that seems quite in line with my thoughts and ideas so far, with some exceptions.

What I'm wondering a bit however is if different EFI settings is used when running OS X and when running other boot alternatives, such as booting Memtest86+ from disc or running BootCamp.

The discussion on VRAM used by the HD4000 seems to indicate that the settings are not the same, as under Windows the HD4000 has almost double the VRAM available to it, which I'm not sure if it can be entirely attributed to drivers. However, this is no definite answer, just my suspicion.

Great with the full timing info of the stock RAM, have only seen the CAS before.

My wishes for the 1866MHz RAM is not all out of line, seeing as the 2011 Mini has manged to run at that speed with some RAM modules. Will have to see where it lands, but my suspicion is also that it will clock down to 1600MHz.

My thought on it though is that as the 1866MHz 16GB Kit has quite good timings (CL10) @1866, it will hopefully be able to match the Kingston timings when running @1600.

The 1866MHz vengeance memory is actually one SEK (about $0.15) cheaper here in sweden than the Kingston kit at the moment, so price wise they're roughly the same.

If I were to go with the Kingston memory in the end though, I would aim for the Low Voltage (not carried by NewEgg, so alternate retailer link http://www.ncix.ca/products/?sku=79091) to keep temperatures down a bit. All the same specs, other than the minimum voltage required (so, top of the class in quality control, I'm assuming).

My Mini was 2,6GHz i7, with the factory Fusion Drive. Only part I thought I'd improve myself was the RAM.

Thanks for chiming in with your experiences! I will try to keep you updated on my findings. Currently I'm still waiting for my "regular" retailers to get both (or any of) the Kingston and Corsair RAM kits back in stock.
 

philipma1957

macrumors 603
Apr 13, 2010
6,362
248
Howell, New Jersey
Last night I ran a test on a pair of 4gb kingston plug n play 1600mhz. these were in my base 2012 mini

better timings take a look. 9-9-9-27 the 16gb time at 11-11-11-28

so even thought these are only 8gb they should run faster the 16gb in

bootcamp as long as you use under 8gb for the program you are running
 

Attachments

  • IMG_2399.JPG
    IMG_2399.JPG
    2.9 MB · Views: 120
  • IMG_2400.JPG
    IMG_2400.JPG
    2.7 MB · Views: 89

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
Last night I ran a test on a pair of 4gb kingston plug n play 1600mhz. these were in my base 2012 mini

better timings take a look. 9-9-9-27 the 16gb time at 11-11-11-28

so even thought these are only 8gb they should run faster the 16gb in

bootcamp as long as you use under 8gb for the program you are running

Great findings!

Now we have documentation that the Kingston RAMs with better timings at least can run them with the better CAS Latency when booting Memtest and/or BootCamp.

We still cannot be entirely sure that the EFI settings and timings will be identical when booting into OS X, and so far I've not seen any way to display them from within OS X. I will however go by the assumption that they indeed are running at the same speeds reported by Memtest, on the basis that it would be stupid to run the RAM modules at a lower speed. :)

For my sake, I will primarily be running OS X, with a potential BootCamp in a distant future.

Do you happen to have any other 8GB (2x4GB) set of memory with "worse" timings? Would be interesting to see how the good timings of the Kingston compares by doing som benchmarks from within OS X...

Great people, now we're starting to get somewhere with actual experience being shared. :p
 

philipma1957

macrumors 603
Apr 13, 2010
6,362
248
Howell, New Jersey
Great findings!

Now we have documentation that the Kingston RAMs with better timings at least can run them with the better CAS Latency when booting Memtest and/or BootCamp.

We still cannot be entirely sure that the EFI settings and timings will be identical when booting into OS X, and so far I've not seen any way to display them from within OS X. I will however go by the assumption that they indeed are running at the same speeds reported by Memtest, on the basis that it would be stupid to run the RAM modules at a lower speed. :)

For my sake, I will primarily be running OS X, with a potential BootCamp in a distant future.

Do you happen to have any other 8GB (2x4GB) set of memory with "worse" timings? Would be interesting to see how the good timings of the Kingston compares by doing som benchmarks from within OS X...

Great people, now we're starting to get somewhere with actual experience being shared. :p

no i have the 1600 kingston plug n play in 4g sticks. i have the kingston 8gb sticks for mac. i had the crucial 8gb sticks for mac. the results above are all I can test except the stock 2gb sticks which don't help much since a pair of 2gb sticks are just too small for most users. kingston sells high end 8gb plug n play sticks that should test with the better timings but they are 129 usd for a pair. I don't want to pay that much. since i personally do not need the small speed boost they will offer over the 8gb sticks I got for 69 usd a pair.

squeezing that last bit of speed from ram sticks does not do a lot for my use of minis.
 

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
no i have the 1600 kingston plug n play in 4g sticks. i have the kingston 8gb sticks for mac. i had the crucial 8gb sticks for mac. the results above are all I can test except the stock 2gb sticks which don't help much since a pair of 2gb sticks are just too small for most users. kingston sells high end 8gb plug n play sticks that should test with the better timings but they are 129 usd for a pair. I don't want to pay that much. since i personally do not need the small speed boost they will offer over the 8gb sticks I got for 69 usd a pair.

squeezing that last bit of speed from ram sticks does not do a lot for my use of minis.

Yes, and no, I think you misunderstood me. :)
Your Kingstons are still faster, by having lower latency/better timings, than the 16GB kit available from Apple, or the standard Crucial kit which I get a feeling the majority of people get...

I haven't even heard anyone report that running the lower latency Kingstons could net them better performance, or that they've verified that it actually runs at those lower timings.

This is pretty much the experience I was hoping to get to know from the beginning. :)

My aim is probably still to go for the 1866MHz, as the price difference between the Kingston PnP (or the LoVo PnP) in the 16GB kit is actually in the Corsair @1866's favor, them being a few bucks cheaper... Now, just to see when things show up in stock.
 

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
Can just chime in that I just made a fine working usb stick with Memtest86+ on as well (which I will be using to verify whatever memory I'm getting later on) and my stock RAM is giving the same numbers, 11-11-11-28 when running at 798MHz - DDR3-1596)
 

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
It is on! :)

My plan as it is looking right now;

  1. Baseline tests, with stock 4GB RAM, tomorrow - Friday.
  2. RAM Switch + Apple Hardware Test - Over night Friday to Saturday
  3. Memtest86+ to get CAS Latency - First thing Saturday if AHT fine
  4. Comparison Tests - Saturday

Any specific tests you would want me to perform (which does not require any specific licenses or similar things)?

My plan, as earlier mentioned, is to run GeekBench 32 bit, iBench, NovaBench and XBench (minus threading test which does not finish).

Will also try to run the bandwidth measurement tool, to see how much a single core can soak the bus for RAM bandwidth, with various buffer sizes and executable architectures (both 32 and 64 bit).

I will most probably also run Evan/Firechild's Multicore Logic Benchmark test (http://www.gearspace.com/board/music-computers/371545-logic-pro-multicore-benchmarktest.html). I only run Logic Express, so it won't be a "proper" test, but might still give a comparative result.

If you've missed earlier, my current setup, before switching RAM;
BTO i7 2.6GHz + Fusion Drive with 4GB Stock RAM.
 

Attachments

  • vengeance1866.jpg
    vengeance1866.jpg
    79.7 KB · Views: 86

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
Last night I ran a test on a pair of 4gb kingston plug n play 1600mhz. these were in my base 2012 mini

better timings take a look. 9-9-9-27 the 16gb time at 11-11-11-28

so even thought these are only 8gb they should run faster the 16gb in

bootcamp as long as you use under 8gb for the program you are running

It is done! :)
Just wrote my initial test report for the 1866MHz Vengeance at https://forums.macrumors.com/threads/1528139/

Extended Memtest86+ runs showed issues with the RAM though, so I'm gonna try to exchange it. Maybe I'll even switch over to Kingstons CL9 1600MHz. Then we could have a comparison, even though the difference in timings between the Kingstons and the Vengeance (when running at 1600MHz) was minimal; 9-9-9-27 (as per your report) vs. 9-10-9-26...
 

bigbird

macrumors 6502
Aug 17, 2007
381
0
Canada
I just installed Corsair Vengeance 2X8GB 1600 MHz (CMSX16GX3M2A1600C10), runs just fine, no errors in rember, what more could I ask for?

This business about running 1866 MHz RAM in a Mini, which is only capable of using 1600 MHz RAM, seems pointless.
 

53x12

macrumors 68000
Feb 16, 2009
1,544
4
This business about running 1866 MHz RAM in a Mini, which is only capable of using 1600 MHz RAM, seems pointless.

The 2011 Mini was able to run 1866 MHz RAM, so for some it is not pointless. :rolleyes:
 

magebarf

macrumors member
Original poster
Dec 29, 2012
51
2
Gothenburg - Sweden
I just installed Corsair Vengeance 2X8GB 1600 MHz (CMSX16GX3M2A1600C10), runs just fine, no errors in rember, what more could I ask for?

This business about running 1866 MHz RAM in a Mini, which is only capable of using 1600 MHz RAM, seems pointless.

You're really missing some of the basic points... Apple states the Mac Mini 2012 is only capable using 1600MHz RAM. The 2011 Mac Mini is only capable using 1333MHz ram, but works just fine with both using RAM at 1600MHz and 1866MHz, and actually running the RAM at those frequencies.

My own MacBook White Early 2009 that the specifications states should use 667MHz DDR2 RAM runs the current RAM specced for 800MHz just fine at 800MHz.

Of course, it's out of the original specs, and hence it is not very certain that it will be running RAM at those frequencies, which is also why I suspected the end result, as it actually turned out, that the 2012 Mini runs the RAM at 1600MHz.

In other words, there is no difference in practice, at least frequency wise, between the 1600 specced Corsair and the 1866 specced one. Timing wise, it could be that the 1866MHz one has a little edge over the 1600MHz one when it is being run at 1600MHz, but knowing what I do today I would not pay extra for the 1866MHz model.

But, until someone actually had tested it, none in here could really say what the outcome would be.

If it would've been running at 1866MHz, there could be a good performance increase.

Now, as for my case, I fell back to the Kingston HyperX @1600MHz, with CL9 timing.

It is very possible that a future EFI update may unlock the option of running RAM at higher frequencies, especially since the CPU itself which does contain the memory controller is capable of it, but I wouldn't keep my fingers crossed.


A good recommendation to you, it is always a good idea to let new memory run over night using either Memtest86 or Memtest86+. I would recommend you to do the same.

If you have a Bootcamp partition or know anyone with a Windows machine, there is a easy tool to create a bootable USB drive for you at http://memtest.org/

When I used the 1866MHz Vengeance, I successfully completed all of the tests, and had no visible or otherwise detectable issues from within OS X. Running Memtest86+ over night did bring these issues to my attention however.

Running Memtest86 won't really guarantee that your RAM modules are okay, that is impossible, but it will be able to show you that they are at least not showing signs of failure over the period you run them (hence, the longer the better).

It might been possible to detect it with Apple Hardware Test in extended mode for a similar amount of time, but my weapon of choice is Memtest86+ due to the good information it provides.
 

avkdm

macrumors regular
Feb 14, 2012
159
42
You're really missing some of the basic points... Apple states the Mac Mini 2012 is only capable using 1600MHz RAM. The 2011 Mac Mini is only capable using 1333MHz ram, but works just fine with both using RAM at 1600MHz and 1866MHz, and actually running the RAM at those frequencies.

My own MacBook White Early 2009 that the specifications states should use 667MHz DDR2 RAM runs the current RAM specced for 800MHz just fine at 800MHz.

Of course, it's out of the original specs, and hence it is not very certain that it will be running RAM at those frequencies, which is also why I suspected the end result, as it actually turned out, that the 2012 Mini runs the RAM at 1600MHz.

In other words, there is no difference in practice, at least frequency wise, between the 1600 specced Corsair and the 1866 specced one. Timing wise, it could be that the 1866MHz one has a little edge over the 1600MHz one when it is being run at 1600MHz, but knowing what I do today I would not pay extra for the 1866MHz model.

But, until someone actually had tested it, none in here could really say what the outcome would be.

If it would've been running at 1866MHz, there could be a good performance increase.

Now, as for my case, I fell back to the Kingston HyperX @1600MHz, with CL9 timing.

It is very possible that a future EFI update may unlock the option of running RAM at higher frequencies, especially since the CPU itself which does contain the memory controller is capable of it, but I wouldn't keep my fingers crossed.


A good recommendation to you, it is always a good idea to let new memory run over night using either Memtest86 or Memtest86+. I would recommend you to do the same.

If you have a Bootcamp partition or know anyone with a Windows machine, there is a easy tool to create a bootable USB drive for you at http://memtest.org/

When I used the 1866MHz Vengeance, I successfully completed all of the tests, and had no visible or otherwise detectable issues from within OS X. Running Memtest86+ over night did bring these issues to my attention however.

Running Memtest86 won't really guarantee that your RAM modules are okay, that is impossible, but it will be able to show you that they are at least not showing signs of failure over the period you run them (hence, the longer the better).

It might been possible to detect it with Apple Hardware Test in extended mode for a similar amount of time, but my weapon of choice is Memtest86+ due to the good information it provides.

Sorry to drag up an old thread but no one is replying to this question on another thread I started. It's a simple question really!
It will probably annoy the heck out of performance obsessed users, will 1333 ram 1.5v (not 1.35v low) work in a 2012 Mac Mini.
Forget brand, blah blah, blah, has anyone actually got any 1333 ram to work in the 2012?

edit: 2x8gb Kingston Ram (KVR1333D3S9/8G) works fine at 1333 and 1.5V for those that are interested or even care :)
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.