Msi claims larger heatsinks are better

[Mira Yurizaki]

1 minute ago, schwellmo92 said:

Yes but server boards get a ridiculous amount of airflow and sit in temperature controlled rooms.

So why is the heat sink designed that way then? For shits and giggles?

[Drak3]

5 minutes ago, spartaman64 said:

 i dont understand why motherboard manufacturers removed heatfins for their heatsinks and just have solid blocks of metal

Aesthetics.

[schwellmo92]

4 minutes ago, M.Yurizaki said:

So why is the heat sink designed that way then? For shits and giggles?

The motherboards are designed as to not impede airflow as much as possible as they’re setup in rack units with very powerful fans that flow all the way through the rack unit (over the motherboard/heat sinks/etc). I would also think that the components would be quite high end and overkill as cutting prices as low as possible isn’t as important in the enterprise sector, they’d much rather pay extra for some that is going to be stable. I’d be interested to see some stats of these motherboards to see how effective the heat sinks are in non-optimal settings (like a regular motherboard is exposed to).

[Mira Yurizaki]

Just now, schwellmo92 said:

The motherboards are designed as to not impede airflow as much as possible as they’re setup in rack units with very powerful fans that flow all the way through the rack unit (over the motherboard/heat sinks/etc). I’d be interested to see some stats of these motherboards to see how effective the heat sinks are in non-optimal settings (like a regular motherboard is exposed to).

I think people are zeroing in on one snippet of my post and then going "lol, that's apples to oranges." The whole point was that motherboard servers are designed that way to maximize cooling efficiency. As far as heat dissipation is concerned, certain principles don't change regardless if you have forced convection or not. This includes increasing the surface area of the part being cooled. Increasing the surface area alone increases the heat transfer potential. Why else do you think passively cooled devices have a lot of fins on it? And it's not like they're going to be sitting in front of a fan of sorts either (if it requires a fan, then the unit should be actively cooled).

 

So when it comes to so-called gaming hardware or in this case, "high-end" motherboards, it would be better to adopt these principles of maximizing cooling efficiency so that enthusiasts can push their hardware harder for longer. Except apparently living a teenage boy's wet dream of what's "cool" is more important.

[schwellmo92]

48 minutes ago, M.Yurizaki said:

I think people are zeroing in on one snippet of my post and then going "lol, that's apples to oranges." The whole point was that motherboard servers are designed that way to maximize cooling efficiency. As far as heat dissipation is concerned, certain principles don't change regardless if you have forced convection or not. This includes increasing the surface area of the part being cooled. Increasing the surface area alone increases the heat transfer potential. Why else do you think passively cooled devices have a lot of fins on it? And it's not like they're going to be sitting in front of a fan of sorts either (if it requires a fan, then the unit should be actively cooled).

 

So when it comes to so-called gaming hardware or in this case, "high-end" motherboards, it would be better to adopt these principles of maximizing cooling efficiency so that enthusiasts can push their hardware harder for longer. Except apparently living a teenage boy's wet dream of what's "cool" is more important.

I think you’ll find the componentry on server motherboards is significantly overkill as well as cutting costs to produce the board isn’t as much of a priority as consumer boards.

[Mira Yurizaki]

2 minutes ago, schwellmo92 said:

I think you’ll find the componentry on server motherboards is significantly overkill as well as cutting costs to produce the board isn’t as much of a priority as consumer boards.

I'm pretty sure adding useless features like RGB and custom heat sinks that don't do anything better isn't exactly great for cost cutting.

[Christophe Corazza]

meanwhile, I will be sticking with ASUS, thanks for the info though MSI...

[Stefan Payne]

19 hours ago, Dabombinable said:

My Asus P5Q Deluxe, joined by shitloads of LGA775 motherboards is saying: Fuck off, those aren't real heat sinks. These are real heatsinks:

  Hide contents

https://www.anandtech.com/show/2381/2

https://www.anandtech.com/show/2427/2

http://www.pcstats.com/articleview.cfm?articleid=2099&page=3

 

 

 

Yes, because those Chipsets were betwen 20 and 30W TDP, the nForce was particularly bad, especially the 590i.

Although I think that I heard that they might have been up to 40W or so...

 

The only thing I could find was the TDP for the x48 wich is 30,5W

https://ark.intel.com/products/35143/Intel-82X48-Memory-Controller-Hub

The nForce Chips were clearly way above that...

That is why they needed those heatsinks!
They are _NOT_ for VRM Cooling, they are for the Chipset!

[yian88]

Custom water cooling for cpu/gpu/ram/mobo is better than passive/air cooling, by 10 bar lenghts.

                                                                                                               - MSI 2018

 

What a shock i had....

[Dabombinable]

4 hours ago, Stefan Payne said:

Yes, because those Chipsets were betwen 20 and 30W TDP, the nForce was particularly bad, especially the 590i.

Although I think that I heard that they might have been up to 40W or so...

 

The only thing I could find was the TDP for the x48 wich is 30,5W

https://ark.intel.com/products/35143/Intel-82X48-Memory-Controller-Hub

The nForce Chips were clearly way above that...

That is why they needed those heatsinks!
They are _NOT_ for VRM Cooling, they are for the Chipset!

The heatsinks are on the VRM as well-and I never had the VRM get more than warm on my P5Q Deluxe, even with my Xeon X5450 at 4.4GHz and drawing well over 150W.
Point is, those are real heatsinks. Not the solid aluminium shit that mobo manufacturers use now.

[Mira Yurizaki]

4 hours ago, Stefan Payne said:

They are _NOT_ for VRM Cooling, they are for the Chipset!

The fact something needs cooling should not preclude manufacturers to ignore basic thermal dynamic transfer properties of materials for the sake of "looking cool", regardless of what's being cooled. Heat is heat.

[Dabombinable]

@Stefan Payne

How are these real heatsinks not for the VRM? And you didn't follow the 3 links to see what I was talking about did you?

Link 1:

https://www.anandtech.com/show/2381/2

 

Link 2: https://www.anandtech.com/show/2427/2

 

Link 3: http://www.pcstats.com/articleview.cfm?articleid=2099&page=3

 

A P5Q Deluxe:

[ScratchCat]

29 minutes ago, M.Yurizaki said:

The fact something needs cooling should not preclude manufacturers to ignore basic thermal dynamic transfer properties of materials for the sake of "looking cool", regardless of what's being cooled. Heat is heat.

If they want to stay in business they will make what the customer wants to buy even if the customer is wrong.

Most customers will simply buy it because of the 'cool' (not literally) design. When was the last time you heard someone exclaim that the design on a car was 'awesome'? Now compare that to when someone last mentioned that a car was 'amazing' due to it's high performance alternator (or similar component).

 

Most users will not even reach the cooling limits of these aluminum blocks and those who would normally have adequate airflow to reduce the temperatures. Investing into a niche wouldn't be a wise strategy.

[Drak3]

14 minutes ago, ScratchCat said:

Now compare that to when someone last mentioned that a car was 'amazing' due to it's high performance alternator

I know people that bitch about their shitty alternators. Does that count?

[Mira Yurizaki]

6 minutes ago, ScratchCat said:

If they want to stay in business they will make what the customer wants to buy even if the customer is wrong.

Most customers will simply buy it because of the 'cool' (not literally) design. When was the last time you heard someone exclaim that the design on a car was 'awesome'? Now compare that to when someone last mentioned that a car was 'amazing' due to it's high performance alternator (or similar component).

Most times I hear people liking their car, it's because of some feature that's not part of its exterior design like how much gas millage they get, how much "oomph" they have in acceleration, etc.

 

Likewise I think most people who aren't super into computers only considered what the chassis looked like for a moment and focused on whether or not the thing had an "intel i7" or "a terabyte of memory" or if it can do "10 JPEG"

6 minutes ago, ScratchCat said:

Most users will not even reach the cooling limit of these aluminum blocks and those who would normally have adequate airflow to reduce the temperatures. Investing into a niche wouldn't be a wise strategy.

I guess wanting wanting efficiency in all aspects of hardware as much as possible is a niche market. Oh well.

[NumLock21]

1 hour ago, Dabombinable said:

@Stefan Payne

Spoiler

 

How are these real heatsinks not for the VRM? And you didn't follow the 3 links to see what I was talking about did you?

Link 1:

https://www.anandtech.com/show/2381/2

 

Link 2: https://www.anandtech.com/show/2427/2

 

Link 3: http://www.pcstats.com/articleview.cfm?articleid=2099&page=3

 

A P5Q Deluxe:

 

 

I still have that board in the 2nd pic. It's either a Asus Maximus Formula or Rampage Formula where both uses the exact same heatsink. Maximus uses Intel X38 and has the old Asus ROG logo. Rampage uses Intel X48 and has the new ROG logo that's still being used today. The P5E also has the same style heatsink, with some minor changes. That P5Q Deluxe also has a flash drive built in for linux base OS, the called ExpressGate. Lower tier P5Q series relies on HDD for ExpressGate. Some other heatsink design back in the days

[ScratchCat]

2 minutes ago, Drak3 said:

I know people that bitch about their shitty alternators. Does that count?

Are they (ex-) high performance shitty alternators?

10 minutes ago, M.Yurizaki said:

Most times I hear people liking their car, it's because of some feature that's not part of its exterior design like how much gas millage they get, how much "oomph" they have in acceleration, etc.

Are these 'car people'? If yes maybe it's a bad example.

11 minutes ago, M.Yurizaki said:

Likewise I think most people who aren't super into computers only considered what the chassis looked like for a moment and focused on whether or not the thing had an "intel i7" or "a terabyte of memory" or if it can do "10 JPEG"

While that may be partly true, even at this very moment there are people considering buying motherboards "just to say I have it" or based on looks. An average user will think of the specs a few times a month while seeing the case every day.
What is the RGB craze about if not for looks? Why do we now have stylized cases for SFF PCs rather than the old grey and black boxes?

18 minutes ago, M.Yurizaki said:

I guess wanting wanting efficiency in all aspects of hardware as much as possible is a niche market. Oh well.

That is the business market (to an extent) but mainly the server market with produces such as SuperMicro and Fujitsu. At some point there is a trade off between cost, practicability and efficiency otherwise all servers would be attached to giant heatsinks directly.

[Dabombinable]

3 hours ago, NumLock21 said:

I still have that board in the 2nd pic. It's either a Asus Maximus Formula or Rampage Formula where both uses the exact same heatsink. Maximus uses Intel X38 and has the old Asus ROG logo. Rampage uses Intel X48 and has the new ROG logo that's still being used today. The P5E also has the same style heatsink, with some minor changes. That P5Q Deluxe also has a flash drive built in for linux base OS, the called ExpressGate. Lower tier P5Q series relies on HDD for ExpressGate. Some other heatsink design back in the days

And every single one of these motherboards has something that modern ones just don't: Real heatsinks that went through at least some R&D.

[Stefan Payne]

5 hours ago, Dabombinable said:

@Stefan Payne

How are these real heatsinks not for the VRM? And you didn't follow the 3 links to see what I was talking about did you?

Yes, they are placed on the VRM but they are not mainly for cooling the VRM but the Chipset.

Just look at the specifications page for the X38 Chipset:

https://ark.intel.com/products/31919/Intel-82X38-Memory-Controller

 

TDP26.5 W

The X48 is even higher and at 30W TDP.

That is a lot! Just look at graphics cards with ~25W TDP, what serious heatsinking they need...

 

A better example for my point might be the EVGA/nVidia reference Board:

http://www.guru3d.com/articles-pages/nvidia-nforce-590-sli-intel-preview,4.html

 

Look at the Cooling of the Northbridge.

But I think you get my point now, do you?

 

Chipsets at the time were really really hot buddys but also included a proprietary bus to the CPU, Memory Controller on Intel Plattforms (AMD had the memory controller integrated since 2003 and K8) and the connection to the SB.

 

The Space around the CPU is just convenient for the Heatsinks because there is a bit of airflow from the Heatsink used (they were designed around the Intel Box Heatsink), thus have a bit of airflow to cool the Chipset.

To cool the VRM is just an additional point but with the heat from the Chipset, its probably heating it...

 

5 hours ago, M.Yurizaki said:

The fact something needs cooling should not preclude manufacturers to ignore basic thermal dynamic transfer properties of materials for the sake of "looking cool", regardless of what's being cooled. Heat is heat.

Yes BUT:

If something doesn't need cooling, you can get away with Art and you don't need "real heatsinks", especially as most customers don't want "real heatsinks"...

 

 

5 hours ago, Dabombinable said:

The heatsinks are on the VRM as well-and I never had the VRM get more than warm on my P5Q Deluxe, even with my Xeon X5450 at 4.4GHz and drawing well over 150W.
Point is, those are real heatsinks. Not the solid aluminium shit that mobo manufacturers use now.

Sadly I can't link to the "Power Consumption vs. Overclocking" Article on XBit Labs because they went down and switched off the site. They had an awesome article about Power Consumption and Overclocking...

Because your CPU might consume closer to 200W...

 

Anyway, the VRM on the old high end boards were totally overbuilt for what is needed, they used more space as well on C2D Boards because Memory Interface was in the Chipsets...

 

Anyway, the main reason for those big heatsinks is keeping the Chipset cool because at the time, they were really hot, especially for Intel Plattform.

 

The ATi Chipsets were rather cool at the time...

[vorticalbox]

On 23/07/2018 at 9:07 PM, M.Yurizaki said:

More people need to get upset with these motherboard manufacturers and their heat sink designs. This is what a heat sink on a server board looks like:

 

And these boards are working a heckuva lot harder than any consumer board I would imagine.

I don't know I work my laptop pretty hard a work using docker to simulate a small network of services.

 

Green pci with blue/black ram slots how I've missed you.

[Dabombinable]

6 hours ago, Stefan Payne said:

Yes, they are placed on the VRM but they are not mainly for cooling the VRM but the Chipset.

Just look at the specifications page for the X38 Chipset:

https://ark.intel.com/products/31919/Intel-82X38-Memory-Controller

 

TDP26.5 W

The X48 is even higher and at 30W TDP.

That is a lot! Just look at graphics cards with ~25W TDP, what serious heatsinking they need...

 

A better example for my point might be the EVGA/nVidia reference Board:

http://www.guru3d.com/articles-pages/nvidia-nforce-590-sli-intel-preview,4.html

 

Look at the Cooling of the Northbridge.

But I think you get my point now, do you?

 

Chipsets at the time were really really hot buddys but also included a proprietary bus to the CPU, Memory Controller on Intel Plattforms (AMD had the memory controller integrated since 2003 and K8) and the connection to the SB.

 

The Space around the CPU is just convenient for the Heatsinks because there is a bit of airflow from the Heatsink used (they were designed around the Intel Box Heatsink), thus have a bit of airflow to cool the Chipset.

To cool the VRM is just an additional point but with the heat from the Chipset, its probably heating it...

 

Yes BUT:

If something doesn't need cooling, you can get away with Art and you don't need "real heatsinks", especially as most customers don't want "real heatsinks"...

 

 

Sadly I can't link to the "Power Consumption vs. Overclocking" Article on XBit Labs because they went down and switched off the site. They had an awesome article about Power Consumption and Overclocking...

Because your CPU might consume closer to 200W...

 

Anyway, the VRM on the old high end boards were totally overbuilt for what is needed, they used more space as well on C2D Boards because Memory Interface was in the Chipsets...

 

Anyway, the main reason for those big heatsinks is keeping the Chipset cool because at the time, they were really hot, especially for Intel Plattform.

 

The ATi Chipsets were rather cool at the time...

The heatsinks still cooled the VRM. That wasn't my main point though, my main point was that those are real heatsinks. Not the dirt cheap+shitty aluminium blocks we have cooling VRM now.

[Stefan Payne]

29 minutes ago, Dabombinable said:

The heatsinks still cooled the VRM. That wasn't my main point though, my main point was that those are real heatsinks. Not the dirt cheap+shitty aluminium blocks we have cooling VRM now.

Yes, because they can get away with "Art-Sinks" and don't need real heatsinks most of the time.

[leadeater]

8 hours ago, Stefan Payne said:

Chipsets at the time were really really hot buddys but also included a proprietary bus to the CPU, Memory Controller on Intel Plattforms (AMD had the memory controller integrated since 2003 and K8) and the connection to the SB.

This was ages ago so I might not remember correctly which computer it was but I think it was my old E7300 system which had a terrible heatsink on the chipset and during summer that would overheat destroying performance, would be middle of gaming then it would drop to less than 1 FPS. Replaced the heatsink with a better one and problem solved .

 

Not this board but the heatsink is exactly the same (or extremely similar) other than it was green not blue, full internet points if anyone can figure out what I actually had because I can't remember.

Spoiler

 

I replaced the crappy heatsink with this

http://www.frostytech.com/articleview.cfm?articleID=2236

[Dabombinable]

9 hours ago, leadeater said:

This was ages ago so I might not remember correctly which computer it was but I think it was my old E7300 system which had a terrible heatsink on the chipset and during summer that would overheat destroying performance, would be middle of gaming then it would drop to less than 1 FPS. Replaced the heatsink with a better one and problem solved .

 

Not this board but the heatsink is exactly the same (or extremely similar) other than it was green not blue, full internet points if anyone can figure out what I actually had because I can't remember.

  Hide contents

 

I replaced the crappy heatsink with this

http://www.frostytech.com/articleview.cfm?articleID=2236

I went this far for my P5K-VM and later P5Q Turbo. 40-50^oC off.

http://www.coolermaster.com/service/support/model/RT-UCL-L4U1/