I had X5677's in my 4,1 and I agree, It was a great middle ground between Single/Multithreaded performance. Following this thread to see what I should upgrade my 5,1 to next, currently has the stock 2.4GHz Quads 
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Easy 4,1 dual CPU upgrade?
- Thread starter handheldgames
- Start date
- Sort by reaction score
I know I don’t have to ask, but please post your results.
I’m more interested in the top performer, regardless of the cost between the two.
Not saying money be damned, just interested in which chip is the champ, and by how much
Fun stuffs...
If money is no object...
I could pick up a 96gb of RAM to test with the new CPU's. I've heard that PC3L-12800R 2Rx4 down clocks without issue. I'm wondering if leveraging a RDIMM, with potentially better timings, would deliver better performance than PC3L-10600R?
If money is no object...
The higher the density the ram is, the slower the timings are, 8g sticks offer the fastest timings CL9 some 16G sticks are CL14 I would think the logic board would detect the JDEC value from the ram. inserting faster ram and down clocking it doesn't always offer faster timings. I like the Hynix CL9 8g sticks myself, I have no need for more that 48g of ram. but anyone who needs more should check the CL rating of the ram they are purchasing. micron sticks offer good value but timings are slower.
DDR3-1333 offers the best over all timings on that chart. CAS latency being 9 instead of 11. if the Mac Pro reads the JEDEC of ram as I would imagine it would, installing DDR3 1600 and running it at DDR3 1333 might even be slower than dedicated DDR3 1333. if you look at the NS at the end that's nano seconds, there is nothing in it between 1333 and 1600 but CL there is 9 to 11 I have even seen CL14 on some sticks.
The Mac Pro offers no way of changing ram timings they are all read from the JEDEC of the ram. the same way it offers no way of changing the voltage of the ram or CPU. they were built to be reliable work horses. I would stick to the fastest JEDEC DDR3 1333 you can find, lower the voltage of the ram the less heat it will produce. also it won't throttle with heat either as ram does do that which affects band width.
The Mac Pro offers no way of changing ram timings they are all read from the JEDEC of the ram. the same way it offers no way of changing the voltage of the ram or CPU. they were built to be reliable work horses. I would stick to the fastest JEDEC DDR3 1333 you can find, lower the voltage of the ram the less heat it will produce. also it won't throttle with heat either as ram does do that which affects band width.
on the whole timing thing
you can check what timings your Mac Pro is running at by firing up CPU Z and going to the memory tab.
you can also see the timings in a CPU-Z validation: http://valid.x86.fr/cwdkvs
as for the timings of 1600Mhz RAM vs 1333Mhz RAM
the if you have a 1600Mhz DIMM that has CL11 timings, keep in mind thats for 1600Mhz only, at 1333Mhz the timings will be CL9 or such (the lower the clock speed the tighter you can tighten up the timings) for the the stick to be able to down-clock to 1333Mhz in the first place it must contain SPD info for that speed for the MP5,1 to work with.
so theres generally no speed penalty for running 1600Mhz RAM at 1333 vs native 1333Mhz RAM
if you do find RAM with nice tight timings in SPD a Mac generally will use them where applicable for example my MBP9,1 has some slightly fancy SODIMMs which are 1600Mhz but have CL9 timings http://valid.x86.fr/avq4rp
(SPD data is Data on the Memory stick that tell the computer what the RAM stick is exactly and what speeds and timings it supports, so if the computer wants to RAM at a certain speed it can look at this SPD data and see which timings to use for that speed)
you can check what timings your Mac Pro is running at by firing up CPU Z and going to the memory tab.
you can also see the timings in a CPU-Z validation: http://valid.x86.fr/cwdkvs
as for the timings of 1600Mhz RAM vs 1333Mhz RAM
the if you have a 1600Mhz DIMM that has CL11 timings, keep in mind thats for 1600Mhz only, at 1333Mhz the timings will be CL9 or such (the lower the clock speed the tighter you can tighten up the timings) for the the stick to be able to down-clock to 1333Mhz in the first place it must contain SPD info for that speed for the MP5,1 to work with.
so theres generally no speed penalty for running 1600Mhz RAM at 1333 vs native 1333Mhz RAM
if you do find RAM with nice tight timings in SPD a Mac generally will use them where applicable for example my MBP9,1 has some slightly fancy SODIMMs which are 1600Mhz but have CL9 timings http://valid.x86.fr/avq4rp
(SPD data is Data on the Memory stick that tell the computer what the RAM stick is exactly and what speeds and timings it supports, so if the computer wants to RAM at a certain speed it can look at this SPD data and see which timings to use for that speed)
on the whole timing thing
you can check what timings your Mac Pro is running at by firing up CPU Z and going to the memory tab.
you can also see the timings in a CPU-Z validation: http://valid.x86.fr/cwdkvs
as for the timings of 1600Mhz RAM vs 1333Mhz RAM
the if you have a 1600Mhz DIMM that has CL11 timings, keep in mind thats for 1600Mhz only, at 1333Mhz the timings will be CL9 or such (the lower the clock speed the tighter you can tighten up the timings) for the the stick to be able to down-clock to 1333Mhz in the first place it must contain SPD info for that speed for the MP5,1 to work with.
so theres generally no speed penalty for running 1600Mhz RAM at 1333 vs native 1333Mhz RAM
if you do find RAM with nice tight timings in SPD a Mac generally will use them where applicable for example my MBP9,1 has some slightly fancy SODIMMs which are 1600Mhz but have CL9 timings http://valid.x86.fr/avq4rp
(SPD data is Data on the Memory stick that tell the computer what the RAM stick is exactly and what speeds and timings it supports, so if the computer wants to RAM at a certain speed it can look at this SPD data and see which timings to use for that speed)
It’s a shame that ram timings are hidden from Mac OS. While I’ve been avoiding the need to run windows on the Mac, your link reinforces the need for me to bite the bullet and install a copy of Redmond’s finest to get a better understanding of what’s going on under the hood.
[doublepost=1535664176][/doublepost]
I’ve been going through data sheets and it looks like most of the RDIMMS from Samsung share the same CL rating of 9 at 1333. RDIMMS from skhynix vary in performance based on the quality level of the dimm depending on the grade of ram used, going between cl 9 and 10 at 1333. Samsung seems like a safer bet.
I used megafarad's services, he did a great job worth the $75. He tests the CPus, before and after delidding process and cleans the silicone seal properly without damaging the board.
Good reference. Thanks for sharing the resource. How did the Samsung RDIMMS work out for your 2009?
[doublepost=1535674355][/doublepost]
Recapping the hardware that I've ordered so far and the associated costs.
$100 2x 5680's || $50 2x 5770's
$275 5,1 Dual CPU Tray
$110 5,1 Logic Board
$24 2x Innovation Cooling Graphite Thermal Pad – Alternative To Thermal Paste/Grease (40 X 40mm)
$264 6 X 16GB Samsung M393B2G70DB0-YK0 2Rx4 PC3L-12800R ECC REG 1.35V
Last edited:
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Just a reference point, I upgraded my 4.1 -> 5.1 DP to dual 3.33ghz 6core X5680s a few months ago. Much cheaper than the x5690s I could find. A huge boost over the 2.26 quads that it had. I did not de-lid them. I used washers to make up for the thickness of the lids.
the Funny thing is on Uninorth PowerPC macs system profiler reported both the Speed and the timings of DIMMMs installed
(speed was in PC values not Mhz however)its one of those things we lost for some reason, when Apple moved to intel
Thx for sharing. Its great to hear the x5680's are working well. The server style sockets without the traditional lock down system in the 2009 dual processor board was something I wanted to avoid.
the Funny thing is on Uninorth PowerPC macs system profiler reported both the Speed and the timings of DIMMMs installed
its one of those things we lost for some reason, when Apple moved to intel
That's a shame. Considering it's a hardware config handle by the efi firmware, I would think that the os has access to the info somewhere. If Windows can access it, you would think that Mac OS should be equally capable.
[doublepost=1535753345][/doublepost]Unpacking the 5,1 cpu tray
Whilst waiting for the new Logicboard, CPU’s and RAM, its time to unpack the 5,1 dual Cpu Tray.
The dual CPU tray has a bit of "stickey" dust residue on it that's common on used circuit boards.
In the 90's I used environmentally unfriendly tuner cleaner to remove this type of gunk when recovering partially expired vector based Arcade Games.
Any recommendations for cleaning it off beyond a paint brush?
In the 90's I used environmentally unfriendly tuner cleaner to remove this type of gunk when recovering partially expired vector based Arcade Games.
Any recommendations for cleaning it off beyond a paint brush?
I usually clean everything with compressed air, my father has a big air compressor at his home, then some isopropanol with a anti-static brush to remove the gunk.
I'm an unfriendly sort of guy. I have some tuner cleaner packed away. If I were going to clean off circuit board gunk, I wouldn't hesitate to use it. Shame on me!
Keyboard Cleaner works too!
Last edited:
This is what I use, its a very good cleaner and evaporates away, a nice soft brush will lift all the gunk. some very small components on there so gently does it.
The area on the outer side of CPU B appears to have the most build up.
I’ve moved too many times since 1995 to know where or if my supply of tuner cleaner survived.
And yes... the little components are well covered.
I’ve moved too many times since 1995 to know where or if my supply of tuner cleaner survived.
And yes... the little components are well covered.
yes it looks like a build up of dust and muck, isopropanol will clean that up like new.
That's residue from the thermal pad decay. Nasty oily gunk.
this is what I used to replace those thermal pads, silicone thermal pad will not decay like that.
https://www.ebay.co.uk/itm/100MM-x-...416270?hash=item3b19e3974e:g:ZuUAAOSwU9xUNsAQ
Thx for sharing the link. Now to find a thermal interface material that’s thick enough, sticky on one side, and long enough for one strip per heatsink.
I will need to open my dual 5,1 one more time and do all of this cleaning too.
And I think I want my IOH diode temp a little lower (now 75°C idle). It might be well within specs, but little lower would be better, wouldn't it.
Ordered the 100x100mm sheets (2 pcs).
Thanks for pics and links.
And I think I want my IOH diode temp a little lower (now 75°C idle). It might be well within specs, but little lower would be better, wouldn't it.
Ordered the 100x100mm sheets (2 pcs).
Thanks for pics and links.