Overclock.net banner
61 - 70 of 70 Posts

·
Premium Member
Joined
·
3,947 Posts
Yes, it does.

I'm not sure where I'm loosing you but I'll try explain, however when you said Martin was an idiot and tried to prove him wrong referencing Jays2cents recreation of a test that was arguably the most significant discovery from Martin's and Skinnee's work, I don't think I'm talking to someone who wants to learn more than wants to just be right.

Some background on your idea that "loop order doesn't matter":
Right off the cuff, to say "loop order doesn't matter" is, at face value, completely wrong. While I understand what you are TRYING to say, LOOP ORDER MATTERS A LOT in that (thanks to Martins and Skinees work) we know loop should be ORDERED as simple as possible. Below is some reading if you want, of my recollection of Martin's and Skinee's work. I'm remembering this from over a decade ago so I could be wrong on some details but overall I think it captures the premise of what I'm trying to say:

First point
Martin and Skinee proposed over a decade ago that flow rate was the most important part of a loop design. There was a theory before this that loops HAD to have a radiator between each heat load (CPU => Rad => GPU) which DID help temps but what Martin showed was that A: the temp benefits were not significant and at times not measurable (<1*C) and B: the resulting loop tended to be more complex than it had to be which adversely affected flow rate. He also showed that as flow rate increased, the DeltaT between any point of a loop diminished (as Jays2cents shows in his testing). His resulting assumption was that, when a loop reaches a particular flow rate (I believe it was 1gal/min) the temperature across the entire loop normalizes within 1*C. This invalidated the theory that CPU's and GPU's heated the water to a significant temperature requiring radiators to be strewn throughout a loop. HOWEVER, he did admit (IIRC) there was SOME benefit to doing this, as long as loop complexity was kept at a minimum.

Second Point
His testing also showed that 1gal/min was the minimum recommended flow rate for a loop with diminishing returns as flow rate increased over 1.5 gal/min (IIRC) and showed that proper flow rate had a much bigger affect on temps than having complex tubing/radiator labyrinths. His conclusion was it was more important to make a simple loop with a maximum flow rate than to put a radiator in between every heat load in your loop. And remember, a decade ago, people were watercooling (in addition to 1-2 CPU's and up to 3-4 GPU's) VRM's, chipsets, memory, HDD's, etc... so loops got complex fast.

To summarize, having a short and simple loop that maximized flow rate was the most important part of a loop design.

Third Point:

Martin demonstrated in his "sandwich" radiator testing that running the heat load to the exhaust radiator prior to the intake radiator netted about a 1*C benefit vs the opposite.

In conclusion
When running an intake and exhaust radiator, LOOP FLOW can help radiator efficiency by exhausting a majority of the heat straight out of the case rather than straight into the case to then be recirculated through another radiator. While the joules of heat dissipated won't make a readily measurable effect on the water temperature (due to the high specific heat of water), it still results in BETTER RADIATOR EFFICIENCY when the loop is ORDERED CORRECTLY.

Now the question is, can you maximize radiator efficiency without over complicating a loop? I'm pretty sure the answer is YES. So as I said, its a small gain in efficiency but its also a small investment in effort which is a no brainer for me. LOOP ORDER MATTERS. Even if its just a small gain.
Just to correct you but it's 0.5GPM as the minimum. 1GPM is the recommended flow rate and returns diminish at 3GPM. Anything below 0.5GPM as I have already said THEN components will heat up the loop but that just begs the question why are you running that low of a flow?
 
  • Rep+
Reactions: StAndrew

·
Old and Decrepit
Joined
·
2,321 Posts
Yes, it does.

I'm not sure where I'm loosing you but I'll try explain, however when you said Martin was an idiot and tried to prove him wrong referencing Jays2cents recreation of a test that was arguably the most significant discovery from Martin's and Skinnee's work, I don't think I'm talking to someone who wants to learn more than wants to just be right.

Some background on your idea that "loop order doesn't matter":
Right off the cuff, to say "loop order doesn't matter" is, at face value, completely wrong. While I understand what you are TRYING to say, LOOP ORDER MATTERS A LOT in that (thanks to Martin's and Skinee's work) we know loops should be ORDERED as simple as possible. Below is some reading if you want, of my recollection of Martin's and Skinee's work. I'm remembering this from over a decade ago so I could be wrong on some details but overall I think it captures the premise of what I'm trying to say:

First point
Martin and Skinee proposed over a decade ago that flow rate was the most important part of a loop design. There was a theory before this that loops HAD to have a radiator between each heat load (CPU => Rad => GPU) which DID help temps but what Martin showed was that A: the temp benefits were not significant and at times not measurable (<1*C) and B: the resulting loop tended to be more complex than it had to be which adversely affected flow rate. He also showed that as flow rate increased, the DeltaT between any point of a loop diminished (as Jays2cents shows in his testing). His resulting assumption was that, when a loop reaches a particular flow rate (I believe it was 1gal/min) the temperature across the entire loop normalizes within 1*C. This invalidated the theory that CPU's and GPU's heated the water to a significant temperature requiring radiators to be strewn throughout a loop. HOWEVER, he did admit (IIRC) there was SOME benefit to doing this, as long as loop complexity was kept at a minimum.

Second Point
His testing also showed that 1gal/min was the minimum recommended flow rate for a loop with diminishing returns as flow rate increased over 1.5 gal/min (IIRC) and showed that proper flow rate had a much bigger affect on temps than having complex tubing/radiator labyrinths. His conclusion was it was more important to make a simple loop with a maximum flow rate than to put a radiator in between every heat load in your loop. And remember, a decade ago, people were watercooling (in addition to 1-2 CPU's and up to 3-4 GPU's) VRM's, chipsets, memory, HDD's, etc... so loops got complex fast.

To summarize, having a short and simple loop that maximized flow rate was the most important part of a loop design.

Third Point:

Martin demonstrated in his "sandwich" radiator testing that running the heat load to the exhaust radiator prior to the intake radiator netted about a 1*C benefit vs the opposite.

In conclusion
When running an intake and exhaust radiator, LOOP FLOW can help radiator efficiency by exhausting a majority of the heat straight out of the case rather than straight into the case to then be recirculated through another radiator. While the joules of heat dissipated won't make a readily measurable effect on the water temperature (due to the high specific heat of water), it still results in BETTER RADIATOR EFFICIENCY when the loop is ORDERED CORRECTLY.

Now the question is, can you maximize radiator efficiency without over complicating a loop? I'm pretty sure the answer is YES. So as I said, its a small gain in efficiency but its also a small investment in effort which is a no brainer for me. LOOP ORDER MATTERS. Even if its just a small gain.

All I read was blah blah blah and what I get is….LOOP order doesn’t matter. Posting in a condensing manner to someone who was water cooling before Martin isn’t a good start and I’m still convinced by my own observations having 30+ years at it. LOOP ORDER DOESNT MATTER.

We will just have to agree to disagree on this one and you are on a different side of the see saw from the fat kid.
 

·
Old and Decrepit
Joined
·
2,321 Posts
I've always maintained that rads work best pulling cool air in. However unless you have a case capable of then exhausting the hot air from the rads, you'll never achieve good results. Unfortunately case designs are pretty ****e when it comes to water cooling. I consider myself lucky to have a Corsair 900D but even that isn't perfect. Such a shame that Caselabs went under!
It’s does have a lot to do with what case is used. All my rads pull fresh air in an Enthoo Elite.
All the 011D builds that get crappy performance and other tiny cases just don’t cut it. I also have a 900D that’s a gem for water cooling.
 

·
Registered
Joined
·
716 Posts
All I read was blah blah blah and what I get is….LOOP order doesn’t matter. Posting in a condensing manner to someone who was water cooling before Martin isn’t a good start and I’m still convinced by my own observations having 30+ years at it. LOOP ORDER DOESNT MATTER.

We will just have to agree to disagree on this one and you are on a different side of the see saw from the fat kid.
Well I'm glad we got that all sorted.

But since loop order doesnt matter I can run radiators how ever I want and in what ever order I want I guess. Cool.
 

·
Old and Decrepit
Joined
·
2,321 Posts
Well I'm glad we got that all sorted.

But since loop order doesnt matter I can run radiators how ever I want and in what ever order I want I guess. Cool.
Loop order yes, fans on the rads is about pulling cool air in a case that isn’t the size of a wallet and having fresh air make up and another mistake people make is having fans on different rads fighting each other. I’ve read so many people posting in different places about positive pressure on the case. That’s usually decreasing air flow when you have to rads pulling air in and pushing against each other trying to force air out. Now if you have separate exhaust fans that’s pulling the air out that’s another story. Lord knows how many cases I’ve tried over the years from the old Antec cases as all the others on the market now didn’t exist up to what I run now. One thing I’ve come to respect in a case is a separate basement where components can be separated by a solid barrier.

Case labs made decent stuff. My only problem with them is too many parts to order. 900D is an awesome case, better than the 1000D and I love the Enthoo Elite I run my main rig in. The 900D is still running, just a different use.
 

·
Registered
Joined
·
262 Posts
For me water cooling a PC has always been about the temps and not how fancy it looks, that is just a bonus as far as I'm concerned. We see plenty of cool looking water cooled PC's but I always think "yeah but what's are the temps like in that thing?" This is something people should consider before getting into water cooling, do I want it to look fancy so I can show it off to my m8's or do I want it to actually cool my PC component's? If your lucky or understand the principles behind water cooling you can achieve both.

Now don't get me wrong, my 1st venture into water cooling I had no real idea. I researched what I needed to make a system but I had no idea about the methodology behind it. I had no sensors or software to control anything, I just knew the temps were better in the bios for CPU and the GFX card than they'd been under air. Understanding air flows rad space etc all came along later with Aquacomputer gear to control everything and a case specifically designed for water cooling, the Corsair 900D.
 

·
Old and Decrepit
Joined
·
2,321 Posts
When I started Corsair was non existent as was the entire water cooling market. The only thing to turn to was chat rooms as the internet was in its infancy and lighting fast speed was a 14.4K modem. GPUs we’re so weak they were passively cooled and ran in an AGP slot and corporate networks were all coaxial and 1Mb/s. There was no watercooling industry as it exist today. My first project involved a square block of copper with the top cut off with a band saw then the other side hollowed out with a pattern of circular holes drilled to a specific depth on a milling machine. A bunch of people in an AOL chat room discussing who got what results with what patterns. Orings? Pffft! Auto part store for tubing, brass barbs, black hoses and hose clamps as well as a can of formagasket which was the sealant of choice because your ability to cut a groove for an Oring to sit in let alone find an Oring to fit it was about as scarce as the smell of a fart in a hurricane. My single rad was a heater core from a 1969 Chevy truck amd the fan was a 120v condenser fan that was robbed off an old refrigerator and an old parts cooling pump for a milling machine. All sat in a wooden box with nothing but black tubes running to the case. At first it was metal strap holding the block on until I got industrious enough to make a bracket that went over the top of the whole block and it eventually evolved into tabs being soldered onto the cold plate and I did one that the mounts were integrated into the cold plate but that took forever with a manual milling machine. The holy grail was whoever made it to 1GHz first wins which took a few years as the first thing I started with water cooling was a 100MHz pentium. It want until later that I got a 500MHz Athlon XP to break the 1GHz barrier. That was back in the late 80s and early 90s well before any commercially made water cooling components came to be. If you want to see the for runners in commercial water cooling they are listed in the Martins watercooling documentation. Swiftech, Dangerden, Dtek and a small hand full of others. An entire industry began from a small group of people in AOL chat rooms just screwing around because they could.
 

·
Premium Member
Joined
·
3,947 Posts
When I started Corsair was non existent as was the entire water cooling market. The only thing to turn to was chat rooms as the internet was in its infancy and lighting fast speed was a 14.4K modem. GPUs we’re so weak they were passively cooled and ran in an AGP slot and corporate networks were all coaxial and 1Mb/s. There was no watercooling industry as it exist today. My first project involved a square block of copper with the top cut off with a band saw then the other side hollowed out with a pattern of circular holes drilled to a specific depth on a milling machine. A bunch of people in an AOL chat room discussing who got what results with what patterns. Orings? Pffft! Auto part store for tubing, brass barbs, black hoses and hose clamps as well as a can of formagasket which was the sealant of choice because your ability to cut a groove for an Oring to sit in let alone find an Oring to fit it was about as scarce as the smell of a fart in a hurricane. My single rad was a heater core from a 1969 Chevy truck amd the fan was a 120v condenser fan that was robbed off an old refrigerator and an old parts cooling pump for a milling machine. All sat in a wooden box with nothing but black tubes running to the case. At first it was metal strap holding the block on until I got industrious enough to make a bracket that went over the top of the whole block and it eventually evolved into tabs being soldered onto the cold plate and I did one that the mounts were integrated into the cold plate but that took forever with a manual milling machine. The holy grail was whoever made it to 1GHz first wins which took a few years as the first thing I started with water cooling was a 100MHz pentium. It want until later that I got a 500MHz Athlon XP to break the 1GHz barrier. That was back in the late 80s and early 90s well before any commercially made water cooling components came to be. If you want to see the for runners in commercial water cooling they are listed in the Martins watercooling documentation. Swiftech, Dangerden, Dtek and a small hand full of others. An entire industry began from a small group of people in AOL chat rooms just screwing around because they could.
For me the golden age was Cathar and Little River water blocks. I miss those days
 
61 - 70 of 70 Posts
Top