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

Some extra cooling makes a difference 5.1pro

Matty_TypeR

Matty_TypeR

macrumors 6502a
Original poster
Oct 1, 2016
642
556
UK
I wanted to see what difference some extra fans would make cooling the ram and IOH and this is what I used.

one of these SATA to fan controller, with high and low speed setting.



then fitted fans using double sided tape, foam filled so no vibration. looks like this.



the fan controller is also double sided taped to side of heatsink B like this.



and before and after temps, and totally silent on low setting.



and after temps.



so after it is cooler and silent, and lost 6.5c off the IOH temp plus ram cooler.
 
  • Like
Reactions: Reindeer_Games, Synchro3, MriX and 2 others
the power I took from the SATA power feed, can be via the spare optical drive, so you need a SATA power extension cable. run that to the fan controller pic 1 and then stick that to the side of heatsink B and plug in the SATA power lead from optical drive spare. the fans are 60mm corsair vengeance fans with aluminium which again are double sided foam taped to side of CPU heatsinks. plug fans into the fan controller and that's it.
 
  • Like
Reactions: mrxak
Is there any evidence that the memory needs additional cooling? 33c doesn't seem concerning enough to warrant this.
 
  • Like
Reactions: mrxak
when you are using memory intensive tasks it heats up and throttles, which reduces band width also heats up IOH so 2 simple fans can make a difference. its just an easy mod to improve cooling really, heat is the biggest cause of failure's of any component part. if you live in a warm climate 6c will make a difference.
 
  • Like
Reactions: mrxak and thornslack
thornslack said:
Is there any evidence that the memory needs additional cooling? 33c doesn't seem concerning enough to warrant this.
Click to expand...

Yes. in another thread, I've documented a drop in ram disk performance due to ddr3 throttling on the single cpu cMP with Apple's default intake/exhaust cooling profile and provided links to documentation from Intel on the specifics of the DDR3 throttling from the x58/c5500 chipsets and how to avoid it's effects. It's a good read.

Increasing the base intake/exhaust speed to 1100ish rpms, as shown by @Matty_TypeR provides an airflow level that keeps the RAM at great temps. You can also setup a dynamic fan speed based on your warmest dimm temp. that adjust speeds as the temps warm up.

The approach taken by @Matty_TypeR is a Cooling Enthusiast approach to improving Apple's designs, leading to an increase in product life and a consistent performance envelope. In 2009, Apple probably wasn't thinking about users placing 96GB+ of DDR 3 RAM with 130W CPU(s).

The chip set runs within specs, but it's a bit warm compared to most standard pc's. I think this is the 1st attempt at providing supplemental cooling to the x58/x5500 in the cMP.
 
Last edited:
  • Like
Reactions: mrxak, Synchro3, thornslack and 1 other person
I just did this as a test using the blue USB powered fan in the photo with an ambient temp sensor.

Not very scientific but close to real world data.

Got a 7°C reduction in ambient temperature from 43°C to 36°C. Considering that it is the peak of summer here in middle Japan that's amazing . .. . I might try to connect one of those ultra slim GPU fans to the USB one and position it between CPU B & CPU A . . . as a test

cMP Fan experiment.jpg
 
Last edited:
  • Like
Reactions: Matty_TypeR and crjackson2134
MIKX said:
Got a 7°C reduction in ambient temperature from 43°C to 36°C. Considering taht it isthe peak of summer here in middle Japan that's amazing . .. . I might try to connect one of those ultra slim GPU fans tothe USB one and position it between CPU B & CPU A . . . as a test
Click to expand...

interesting test there, only thing I can see is your temp sensor sits right in front of the cooling fan offering optimum detection of temp variation. this is how I see the air flow in a Mac Pro.



Cool air enters via the intake fan at the front, CPU A gets the coolest air, Memory for CPU A also gets the coolest air from intake fan at front. Boosta fan in CPU heatsink A pushes air out the back of heatsink A ( which has warmed up cooling the CPU A then has to cool the Dimms by the side of heatsink B and exit via the fan at rear. the fan above dimes next to cpu B is directly above the centre of dimms blowing down to offer better cooling to Dimms next CPU B before exit via rear exit fan.

Cool air from the front goes across the CPU A dimms and warms then passes over IOH heatsink at the bottom of CPU B heatsink and into the boosta fan of CPU B to exit from rear fan at the back. this is the warmest path from entry cool air over CPU A dimms then IOH and then CPU B. the fan I have put on the side of heatsink A is not directly pointing downward. its at an angle pointing backward towards heatsink B directly blowing over IOH and aiding the cooling of CPU B

the Mac Pro cooling layout isn't bad, works as a tunnel blowing cool air from the front to exit at the rear. this setup aids in cooling and if you are sat in a hot climate it can help keep your Mac cool with out making it noisy. the side cover is important to keep on as this creates the tunnel affect from front to back.
 
Last edited:
  • Like
Reactions: Synchro3
I used MacsFanControl to ramp up fan speeds to 1200 rpm across the board, as an experiment. I'm achieving similar temps (as the OP) as a result, and my Mac Pro is still quite quiet.
Keep in mind, it's cooler here now in early September, so the ambient room temperature has dropped. But having said that, my Mac Pro temps have dropped significantly.
 
  • Like
Reactions: handheldgames and h9826790
Its a bit cooler here today. and I have a nice quiet Mac Pro. and increasing fan speeds does the same thing by decreasing temps, only downside is how much noise it makes.

 
  • Like
Reactions: handheldgames
Climate change is here to stay. . here in Japan we recently experienced almost 3 weeks of 39 ` 40°C temperatures daily.
I opted to remove the side cover and position a small, externally powered ( with the fan rotation = left to right button engaged ) to . . . .

(1 ) blow hot air away from the rear of the cMP. &

.. on the next left to right sweep . .

(2 ) blow hot air away from the CPU's & RAM.

That greatly reduced ambient temperatures.
 
  • Like
Reactions: handheldgames
MIKX said:
Climate change is here to stay. . here in Japan we recently experienced almost 3 weeks of 39 ` 40°C temperatures daily.
I opted to remove the side cover and position a small, externally powered ( with the fan rotation = left to right button engaged ) to . . . .

(1 ) blow hot air away from the rear of the cMP. &

.. on the next left to right sweep . .

(2 ) blow hot air away from the CPU's & RAM.

That greatly reduced ambient temperatures.
Click to expand...

40C wow that's warm, no wonder you had to do some thing there. as they say if it works it works, so well done on that extra cooling
 
  • Like
Reactions: handheldgames
MIKX said:
Climate change is here to stay. . here in Japan we recently experienced almost 3 weeks of 39 ` 40°C temperatures daily.
I opted to remove the side cover and position a small, externally powered ( with the fan rotation = left to right button engaged ) to . . . .

(1 ) blow hot air away from the rear of the cMP. &

.. on the next left to right sweep . .

(2 ) blow hot air away from the CPU's & RAM.

That greatly reduced ambient temperatures.
Click to expand...

Along with climate change comes an increase in earthquakes and vulcanism. Anyone on the pacific should have had a good show over the past few weeks. It’s amazing how the redistribution of water weight from global warming and over pumping of aquifers in China can really screw things up.

E2819752-D185-4EE3-B2A2-19CCCC7C0785.jpeg
 
  • Like
Reactions: Matty_TypeR
handheldgames said:
With the low-voltage 1.35v Samsung Registered ECC DDR3 dimms I'm running, throttling starts to have an effect, when it reaches near the 40c area.
Click to expand...

low voltage registered dimms are a gamble. Most of them dont run, some overheat. Maybe that's the cause of you run into temperature trouble.

My registered (1.5V) registered dimms in my Single 5.1 are 50+C and there is no throttling.
 
Macschrauber said:
low voltage registered dimms are a gamble. Most of them dont run, some overheat. Maybe that's the cause of you run into temperature trouble.

My registered (1.5V) registered dimms in my Single 5.1 are 50+C and there is no throttling.
Click to expand...

Nice. What brand / mode / configuration of ram are you working with.

Geekbench won't reveal throttling. The only way I've documented throttling is working with a Raid 0 RAMDisk running multiple Quickbench Large File Transfer benchmarks in succession. Yes, a striped raid 0 ramdisk is faster than a regular ram disk.

Anyways.. Per Intel, DDR3 throttling is a feature in the c5500/x58 chipset that can't be disabled. The only way to avoid it is with an modified fan profile. It's likely that Your workflow may not reveal the throttling impact of your system. DDR3 throttling is a little discussed, dirty little secret of the c5500/x58 chipset.

Throwing a bone to Geekbench 4, the 1.35v 16GB Samsung rDIMMS I'm using allows my cMP to set records in single and dual cpu setups. That's putting 3x16 or 16x16 against everything else out there. It was a gamble buying the 2Rx4 16GB rdimms. that turned out to be more of a jackpot than anything else.
 
  • Like
Reactions: eksu and crjackson2134
handheldgames said:
Nice. What brand / mode / configuration of ram are you working with.

Geekbench won't reveal throttling. The only way I've documented throttling is working with a Raid 0 RAMDisk running multiple Quickbench Large File Transfer benchmarks in succession. Yes, a striped raid 0 ramdisk is faster than a regular ram disk.

Anyways.. Per Intel, DDR3 throttling is a feature in the c5500/x58 chipset that can't be disabled. The only way to avoid it is with an modified fan profile. It's likely that Your workflow may not reveal the throttling impact of your system. DDR3 throttling is a little discussed, dirty little secret of the c5500/x58 chipset.

Throwing a bone to Geekbench 4, the 1.35v 16GB Samsung rDIMMS I'm using allows my cMP to set records in single and dual cpu setups. That's putting 3x16 or 16x16 against everything else out there. It was a gamble buying the 2Rx4 16GB rdimms. that turned out to be more of a jackpot than anything else.
Click to expand...

Oh, so I dont noticed throttling, I thaught you meant benchmarks with higher ram temperatures.

I have seen that some Low Voltage Rams have dual voltage modes. Guess they produce more heat when regulating down.
 
Sharing a Technology Primer: Low Voltage Ram

"... Low voltage RAM fares very well against standard RAM. Versus RAM of the same frequency and timings, it performs just as well as standard RAM while reducing the overall power requirements by 2-3 watts per stick. This power reduction also allows to the RAM to run much cooler than standard RAM. While this did not translate into significantly lower CPU or GPU temperatures, anything that reduces heat without sacrificing performance is a good thing in our view."
 
handheldgames said:
Sharing a Technology Primer: Low Voltage Ram

"... Low voltage RAM fares very well against standard RAM. Versus RAM of the same frequency and timings, it performs just as well as standard RAM while reducing the overall power requirements by 2-3 watts per stick. This power reduction also allows to the RAM to run much cooler than standard RAM. While this did not translate into significantly lower CPU or GPU temperatures, anything that reduces heat without sacrificing performance is a good thing in our view."
Click to expand...

JEDEC Solid State Technology Association announced the publication of JEDEC DDR3L on July 26, 2010.

https://www.jedec.org/news/pressrel...anticipated-ddr3l-low-voltage-memory-standard

The Mac Pro 2009 was introduced back in March 2009. By considering the 2010 and 2012 Mac Pro has the same logic board as the 2009 model. It's very impossible that any Mac Pro 4,1 / 5,1 logic board can support 1.35V DIMM.

And if the logic board does not support DDR3L memory, the DIMM will operate at 1.5V. All DDR3L DIMM have backward compatibility on DDR3 system.

On my flashed 5,1. The PC3L memory is running at 1.523V.
Screen Shot 2018-09-04 at 23.35.46.png

I suspect your current DIMM simply has better quality then the old one, therefore, give you better result. That's most likely nothing about 1.35V vs 1.5V
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back