Conclusion, it's all a shot in the dark??
Damn.Yes, it usually is for the first guy.
If I do, I will.If you get one and it works (or doesn't), let us know. Then everyone that follows won't have to take a shot in the dark.
Will know soon enough, the seller will come over and we'll both install it together to see if it works. Won't have to buy it if it doesn't work. Anything in particular I should look for or test should it work and boot?Yes, it usually is for the first guy.
If you get one and it works (or doesn't), let us know. Then everyone that follows won't have to take a shot in the dark.
Anything in particular I should look for or test should it work and boot?
Sorry noob here, Did you just upgrade your Mac Pro 5,1 to a Mac Pro 6,1 with a CPU upgrade?
I don't know what an nMP, cMP or what TDP.
nMP = the 2013 cylinder mac pro. I put in a new CPU. TDP stands for thermal design power. A lower TDP results in less heat. Since it works I guess any of the lower TDP ones will.Sorry noob here, Did you just upgrade your Mac Pro 5,1 to a Mac Pro 6,1 with a CPU upgrade?
I don't know what an nMP, cMP or what TDP.
I hope I never have to open the nMP again. The electronics are so small, I almost feared I was gonna break it.Sweet success!
I'll add it to the list a bit later. Too busy at the moment. (Anyone else, feel free to do so.)
Hi,
I have a 2009 Mac Pro (with W3520 and 10 RAM) and want to update the processors and RAM.
My questions are:
Will it be better to go with the W3690 or with the X5675?
Will 32GB of RAM work and be usefull and is there a difference when using it with a W or X processor?
Regards Patrick
The original post says X5687 is not compatible, but GeekBench shows 3 Mac Pro 5,1s with that processor (or 1 with 3 benchmarks).
http://browser.primatelabs.com/geekbench3/search?utf8=✓&q=X5687,+MacPro5,1
There are all kinds of Macs listed in Geekbench that don't match actual Mac models. It's probably a hackintosh that identifies itself as a Mac Pro.
Yeah, pretty sure those are Hackintosh. In fact, 2 out of 3 has non standard RAM clock speed, which cannot happen on a real Mac Pro.
Ah, I see. That makes sense.
Hi,
I have a 2009 Mac Pro (with W3520 and 10 RAM) and want to update the processors and RAM.
My questions are:
Will it be better to go with the W3690 or with the X5675?
Will 32GB of RAM work and be useful and is there a difference when using it with a W or X processor?
Regards Patrick
Hi Patrick, over the past month I have installed 2 different CPU's in a Mac Pro 4,1 single processor flashed to 5,1.
Geekbench 4 scores:
W3520 - single core 2521 - multi 8629 - 24Gb RAM - Supports 1066mz Dimms
W3670 - single core 2860 - multi 12809 - 24Gb RAM - Supports 1066mz Dimms - CPU only supports 24Gb RAM
W3680 - single core 3088 - multi 14764 - 24Gb RAM - Supports 1066 and 1333mz Dimms - CPU supports 288Gb RAM, not sure of Mac RAM limit suspect 48gb
Suspect W3690 will only be 5% faster than the W3680
Note memory slot issue on Mac Pro single CPU, 3 CPU channels to handle 4 DIMM slots, memory slots 1 and 2 have a CPU channel each and 3 & 4 share the same CPU channel. This causes a memory slow down if you put a stick in slot 4. RAM copy speed halves! Yes 3 x 8Gb sticks ARE FASTER than 4 x 8Gb sticks. Don't use my now spare 4th stick.
1333mz Server Ram Sticks Samsung part number M293B1K70DH0-YH9 work well.
Hi Patrick, over the past month I have installed 2 different CPU's in a Mac Pro 4,1 single processor flashed to 5,1.
Geekbench 4 scores:
W3520 - single core 2521 - multi 8629 - 24Gb RAM - Supports 1066mz Dimms
W3670 - single core 2860 - multi 12809 - 24Gb RAM - Supports 1066mz Dimms - CPU only supports 24Gb RAM
W3680 - single core 3088 - multi 14764 - 24Gb RAM - Supports 1066 and 1333mz Dimms - CPU supports 288Gb RAM, not sure of Mac RAM limit suspect 48gb
Suspect W3690 will only be 5% faster than the W3680
Note memory slot issue on Mac Pro single CPU, 3 CPU channels to handle 4 DIMM slots, memory slots 1 and 2 have a CPU channel each and 3 & 4 share the same CPU channel. This causes a memory slow down if you put a stick in slot 4. RAM COPY SPEED halves in Geekbench tests! Yes 3 x 8Gb sticks ARE FASTER than 4 x 8Gb sticks. Don't use my now spare 4th stick.
1333mz Server Ram Sticks Samsung part number M393B1K70DH0-YH9 work well.
There are 16Gb sticks too.
W3680 only supported up to 24GB officially. 288GB is the X5680's limit, not W3680.
The Max demonstrated limit on cMP with W3680 so far is 56Gb (3x16 + 8). And no one try the 32GB stick with this CPU yet, so the actual limit still unknown at his moment.
The expected speed difference between W3680 and W3690 is 3.9% (multi core), and 3.6% (single core). Most user should not able to tell the real world difference without any help from instruments / monitoring aids.
The 4th stick only slow down the Mac when you really hit the memory bandwidth limit, AND you have enough memory for all process at that moment. In most cases, the system can use the 4th stick (extra memory) as cache to speed up the system (rather than slow down). Of course, it depends on the workflows. If your workflow need to regularly access the same files. Extra memory (cache) will help, because your system now only need to load the data once, and then it can access from the memory when you need the same data again, no need to load from the hard drive anymore. Or if you reboot your Mac few times a day, then the extra memory may not able to help too much, because there is always nothing in the cache.
Of course, the best case would be you only install 3 sticks, and still have enough memory for everything (including cache). However, by considering MacOS now is so good to utilise the memory, memory size is relatively small on Mac, and the apps getting bigger and bigger. It's almost always the more the better now.
Except some process require large memory bandwidth (e.g. Some computation task). I rarely heard that anyone can really feel the 4th stick significantly slow down their cMP. In fact, I doubt if anyone can tell in a double blind test. I tried both 3 sticks s 4 sticks config. I really can't tell the difference in speed in daily
The similar situation also happened on memory speed. 1333 seems quite a bit faster than 1066. However, 1333 match with CL9, and 1066 match with CL7. The real world speed difference is only ~2%, very insignificant for most users. But the benchmarks can shows large difference once the 4th stick is there.
Of course, if you allow your machine to run few % slower on every aspect. It may end up >10% slower overall. But that very last bit of performance (on each single area) usually will cost a lot more. It's the user to decided if that's worth or not.
Thanks!!
I ended up getting a very nice deal for a X5680 and 32GB 1333 RAM (4x8).
Now still in doubt wether I should let the 4th stick in there or just run with 3 sticks of 8GB RAM..
Hello Patrick again (from our discussion at apple.com). Do not overthink it with memory bandwidth. Here is a nice case study about memory:
https://macperformanceguide.com/Mac-Upgrade-CaseStudy-MacPro-Memory.html