Production FormD T1 Classic (READ FIRST POST)

[JRey]

First look at the RX 6000 heat sink, confirmed by Radeon on Twitter.
Radeon 6000 Heatsink

It looks to be a large card, very close to the Radeon 7990 which is ~12 inches (304mm). Based on the where the 2x 8-pins are located, water cooling with the DC-LT 2600 + 40 res may not be possible in in the location @fabio popularized.

Edit.
AMD added the card to some map in Fortnite. Here’s some screenshots showing the card along with the PCI-E connector for a size comparison.
https://t.co/pdysKfK7rc

Someone released a photo next to a 2080 Ti (just pasted below it).

https://pbs.twimg.com/media/Eh592ORWkAAC6zN?format=jpg&name=medium

Judging by the power connector, it looks like it’s super close to the size of a 2080 Ti. I hope this is true as it would give more options to watercool it in the T1.

 

[AP2]

@Goatee - if you need a 2080 Ti block message me ?

 

[Luke NK]

I've UPDATED THE TESTS WITH 2 MORE + 1 EASTER EGG @Luke NK

Here the pictures for the easter egg!

To have the A12x25 mounted, I need the mounting brackets from the Noctua L12, without the spring, and the AM4 Kit from Noctua! (the mounting brackets are different from the one included in the L12S! But are the same of the L12, if you have one. You can swap the mounting systems between the two cooler without any problem)

Plus, I had to revert to the normal PSU cabling, because the J-Hack M2427 is too tall for the 25mm fan.

I am using VLP 2666MHz RAM Innodisk 2x16gb, Samsung B-Die, Running at 3200MHz 17,18,18,18,38 1,35V

But apart from these small details, it FITS without any modification my Gigabyte X570!

Thanks! I was a little bit disapointed when you said you would not try it!

it seems to me around 77° is about the limits of what L12S is capable to do regarding fans or configurations with your CPU. How did you find the noise levels?

If I were to take just an LPX ram and remove its heat spreader, do you think it could fit under A12X25 without having to use VLP ram? what do you think?

 

[ellroy80]

Almost correct! BUT:

• The L12S intake, no fan on top combination produces a CPU temperature of 75C - only 1 degree warmer than the lowest temperature of 74C. However, it results in the highest GPU, SSD (front and back), and VRM temperatures. Furthermore, the chipset temperature is the second hottest measured. These results suggest that the cool air being sucked in from outside the case is doing a great job at keeping the CPU cool, but the warm air is then trapped within the case and heating everything else up. Rotating the cooler 90 degrees makes things even worse, consistent with Noctua's recommendation that the heatpipes should be facing up.
  Actually Noctua suggests also the 90 degrees orientation.
  
  
• The L12S 90 degrees intake, top A12x25 exhaust produces the lowest CPU and GPU temperatures (74C and 39C, respectively). The SSDs are the second coolest at 45C and 42C, while the cooling of the VRM (55C, third lowest) and chipset (58C, third lowest) is adequate. This is consistent with the above results, in that the top fan is removing the stagnant warm air from the case.
• The L12S exhaust, no fan on top results are quite remarkable. The CPU is the third coolest at 76C - only 2 degrees warmer than the coolest; and the GPU at 40C is only 1 degree warmer than the coolest (the coolest for both is in the setup immediately above). However, the front and rear SSDs are significantly cooler than in the setup immediately above (by 5 and 1 C, respectively), and the chipset is 4C cooler than in that immediately above. Clearly, the fan in exhaust works to pull air in from everywhere it can (which is probably literally from everywhere, given the amount of mesh in the case) and therefore helps to cool all components. The one exception to that rule are the VRMs, which are the second hottest at 59C.
  
  
• Having the top fan as intake produces the lowest VRM temperatures. This is not surprising given that the fan is blowing directly on the VRMs. However, the temperatures of most of the other components are worse than having the fan set to exhaust. I assume this is because the fan is pushing warm air back into the case.
• Surprisingly, having a bottom intake makes things worse when the L12S fan is set to exhaust. Perhaps this is because the bottom fan used is more powerful than that on the HSF, and therefore more air is being pushed in than exhausted out? If so, the additional air pushed in might be trapped and be stagnating in the case?
  This is true with the S12A Fan. I am testing now the same setup with the A12x25.

If all of my assumptions are correct, then:

• The coolest results would be observed with the heatpipes facing up, HSF fan set to intake, and an exhaust fan above the cooler. This doesn't look possible with your setup though, because the ends of the heatpipes are in the way of mounting a fan?
  WELL, it is partially correct! I can mount a slim 92mm fan to help the exhaust... But the problem will still be the chipset and SSD temp IMO.
  
  
• The quietest setup of those tested is the L12S set to exhaust. Surprisingly, this produces great temperatures too (heatpipes need to be up, though). I wonder if the cooling potential of this setup would be improved by using a bottom intake fan (i.e. case flipped) that is the same model as that used on the L12S? i.e. an A12x15. This means that the air pulled in should be exhausted by the L12S fan, resulting in potentially less stagnant air in the case. Do you have another A12x15 to test this hypothesis?
  I can try it with the A12x25 (why the 15mm?), but I still think the best orientation of the case is with the fans on TOP! Especially with almost every NOT BLOWER GPU, that exhaust inside the case, top and bottom. (I am using a blower style now because I don't have an open style cooler ATM available).

Super interesting stuff!!! It has certainly got me thinking about how I might cool the CPU in my case. Heatpipe orientation (and CPU socket position as an extension of that), fan size, fan position relative to the case, and HSF fin density are all things to take into account. Thanks for doing all of this testing!!!

EDIT: disregard my request above for an A12x15 bottom intake fan + exhaust setup. I think it all makes sense now, after a second look at your photos (as long as I'm interpreting them correctly). With the L12S fan set to exhaust and having no fan on top, cool air will be pulled from the bottom of the case, past the chipset and SSD, then out through the HSF. This is why the VRM cooling isn't great in this setup - air coming from the bottom of the case will be exhausted before it reaches them. If you flip the case, orient the HSF so the heatpipes face up, set the HSF fan to exhaust, and add a bottom intake fan, the VRM cooling should be better but the chipset/SSD temperatures should be worse because the cool air will be exhausted out by the HSF before it reaches them. I think.

In general, the hot air has to be forced out. The top and bottom panel, and even more the mesh, don't allow the air to exhaust properly otherwise. I am finding them a bit compelling without any fan!

@fabio, thanks for your response. Thank you for correcting me on Noctua's recommended orientations for the L12S. Still, your results show that there is a significant benefit to having the heatpipes up. It would be interesting to see that orientation plus the 92mm exhaust up top IMO, for science if no other reason. Either way, I think you are correct - this case needs the hot air actively exhausted for best system cooling results.

I've included the data from your latest tests in my analysis and the results are below. These data are normalised to ambient temperature, so the displayed values are delta temps (Component temp - Ambient temp). Note that the gradient is from coolest (green) to warmest (red).


The data shows that the coolest system overall is the Easter Egg one. Not surprising really - there's more air blowing over the components to cool them than in any other combination.

The interesting one is the L12S 90 degrees exhaust, top A12x25 exhaust. This produces the coolest system components EXCEPT for the CPU, which is the second hottest of all combinations. My hypothesis on this is that the more powerful A12x25 is sucking air past the A12x15, starving it of air to exhaust through the L12S. This is why I thought a L12x15 might be better as the intake or exhaust fan, to provide a balance of air flow.

It really does beg the question of what is the best setup. L12S Intake + case exhaust = best CPU temps (and really good GPU temps), but worse temps for other system components. L12S exhaust + case exhaust = poor CPU temps, but all other temps are the best. Perhaps a 120mm AIO is the best? This might provide the best of both worlds, though it may pose problems once the GPU is loaded and warm air is getting sucked through the rad. (and the results in general might change with the GPU also loaded)

Just for interest sake, do you mind running the L12S without a fan at all and with the top exhaust? I wonder if that top exhaust will be enough to pull air through the L12S. I don't want you to damage your CPU, so it's entirely up to you of course! From reading your posts you know the limits of hardware much better than I, so I'll leave it to you to decide whether or not that is a good idea. I still do wonder if you could get away with a 65W CPU and a fanless L9x65 as an example (or a larger copper heatsink) and use the top exhaust fan to pull air through it........

 

[Luke NK]

Just for interest sake, do you mind running the L12S without a fan at all and with the top exhaust? I wonder if that top exhaust will be enough to pull air through the L12S. I don't want you to damage your CPU, so it's entirely up to you of course! From reading your posts you know the limits of hardware much better than I, so I'll leave it to you to decide whether or not that is a good idea. I still do wonder if you could get away with a 65W CPU and a fanless L9x65 as an example (or a larger copper heatsink) and use the top exhaust fan to pull air through it........

I think no fan on L12S and fan on top will be worse than a fan on it and no fan on top, which was the setup he already ran and had to abort

L12S 90 degrees INTAKE, NO FAN ON TOP
CPU ABOVE 90C, SO TEST, ABORTED

 

[ghostfish]

No fan on the L12S could be workable, but probably not with full mesh side panels. I suspect you'd need to put tape/plastic/whatever over the mesh near the motherboard except for directly above the heatsink.

I think that clear adhesive backed plastic in strategic locations could improve the temperatures of both air and water cooled builds by controlling where the intake/exhaust air comes from and goes to. In most SFF cases including Ghost S1, NCASE (early revisions) and DAN, the limited grille area forces air to take certain paths. With a full mesh side panel the air can come from anywhere, so any given airstream cross section has less flow.

For a given exhaust volume from an exhaust fan, if there's twice the intake area along the side panels you're getting half the air speed. There's also the concern of most of the intake air coming from the part of the mesh closest to the fan and having very little airflow through the mesh far from the far. A small grille area may equalize flow over its whole area better.

 

[PKAWA]

Big Navi

https://pbs.twimg.com/media/Eh53l3-XsAADDTj?format=jpg&name=large

 

[fabio]

@fabio, thanks for your response. Thank you for correcting me on Noctua's recommended orientations for the L12S. Still, your results show that there is a significant benefit to having the heatpipes up. It would be interesting to see that orientation plus the 92mm exhaust up top IMO, for science if no other reason. Either way, I think you are correct - this case needs the hot air actively exhausted for best system cooling results.

I've included the data from your latest tests in my analysis and the results are below. These data are normalised to ambient temperature, so the displayed values are delta temps (Component temp - Ambient temp). Note that the gradient is from coolest (green) to warmest (red).


The data shows that the coolest system overall is the Easter Egg one. Not surprising really - there's more air blowing over the components to cool them than in any other combination.

The interesting one is the L12S 90 degrees exhaust, top A12x25 exhaust. This produces the coolest system components EXCEPT for the CPU, which is the second hottest of all combinations. My hypothesis on this is that the more powerful A12x25 is sucking air past the A12x15, starving it of air to exhaust through the L12S. This is why I thought a L12x15 might be better as the intake or exhaust fan, to provide a balance of air flow.

It really does beg the question of what is the best setup. L12S Intake + case exhaust = best CPU temps (and really good GPU temps), but worse temps for other system components. L12S exhaust + case exhaust = poor CPU temps, but all other temps are the best. Perhaps a 120mm AIO is the best? This might provide the best of both worlds, though it may pose problems once the GPU is loaded and warm air is getting sucked through the rad. (and the results in general might change with the GPU also loaded)

Just for interest sake, do you mind running the L12S without a fan at all and with the top exhaust? I wonder if that top exhaust will be enough to pull air through the L12S. I don't want you to damage your CPU, so it's entirely up to you of course! From reading your posts you know the limits of hardware much better than I, so I'll leave it to you to decide whether or not that is a good idea. I still do wonder if you could get away with a 65W CPU and a fanless L9x65 as an example (or a larger copper heatsink) and use the top exhaust fan to pull air through it........

Great job!!!

Looking better at the tests results, I think I will delete the MAX temp. This as gotten for a few seconds, so is not really worth it! The temp reported instead is the one recorded after 30 minutes from the test start, and almost close to the average!

I also think that you should not taking care about the delta temp. This is a 30 minutes test, after 15 minutes the heatpipes are saturated anyway, at 24C or 22C ambient temperature! I think...

I've updated the test with 1 MORE, with the last configuration but in ECO Mode, to simulate a 65W CPU! And the results are impressive!

I think I can tweak the 3950X to have an incredibly balanced and quiet system!

I don't trust though in the capability of the L12s without the fan, it is not meant to work in that way! Noctua is working on 1 CPU cooler Passive for 95/120W CPU! If I will teardown the system again I will test it anyway!

The A12x25 on the L12 helps a lot in general, because the higher static pressure, allows me to run it at lower speed and still moving air through the mesh!

BTW, I am running the VLP at 18,19,19,19, 38 1.35V at 3600MHz! So, pretty neat!!

@Luke NK Nope, you can only have the VLP with the 25mm, look at the pictures!

 

[DrHudacris]

BTW, I am running the VLP at 18,19,19,19 38 1.35V at 3600MHz! So, pretty neat!!

Wow... Please link the VLP kit you're using, that's amazing for VLP RAM!

 

[Goatee]

@Goatee - if you need a 2080 Ti block message me ?

I have on on it’s way, thank you for the offer!

 

[Caleb]

I've used 6-32 UNC x 3/4" grub screws in my EK radiator to hold the fans in place since there is not enough clearance for screws with a head.

The top panel sits perfectly flush over the fans holding them down snugly. Definitely an upgrade from the sticky tape I have been using!

 

[fabio]

Wow... Please link the VLP kit you're using, that's amazing for VLP RAM!

They are pretty good! I want to try also my Crucial VLP ECC 2666 2x16gb As well maybe!!
But I think is also Ryzen and the X570 Chipset. On Intel Z390 and 9900K I was only able to run them stable at 3200 C17. And was impossible to run them at C16, no matter what speed was.

BTW, the Innodisk are these:

DDR4 UDIMM VLP | Unbuffered Long DIMM | Very Low Profile Memory | Embedded DRAM Modules | Products – Innodisk

DDR4 UDIMM VLP offers a very low-profile design and the industry's fastest memory speed with 2666MT/s, and it is fully compatible with the Intel® Purely platform and 1U devices. The modules comply

www.innodisk.com

I've bought them 1 year and and half ago on Enoch System, but seems they are out of business??

 

[fabio]

ANOTHER UPDATE WITH A TWEAKED PBO!! 120PPT, 80TDC, 117EDC!!

The ambient temp is a bit higher though.

@ellroy80 maybe you can update your spreadsheet?

 

[10101001010101011]

What mesh count does the sidepanel have?

 

[bynio]

@fabio how do you run tests, just blender test launcher? Does it matter which version?

Just for perspective, my results on crappy build. L12S 90 degrees intake, no extra fans, R5 3600, Aorus x570, LPX 3200, only front Samsung drive, crappy GPU and SSD next to it. Max stable values during 50 min run, not necessarily at the end. 75-80W on CPU. Clocks don't ever reach 4100MHz.

Ambient 24-25
Chipset 67
VRM 53
Front drive 59
CPU 83-87
GPU 48

Those are crappy CPU temps. I'm getting my Watercooling parts today.

 

[Alur]

@fabio how do you run tests, just blender test launcher? Does it matter which version?

Just for perspective, my results on crappy build. L12S 90 degrees intake, no extra fans, R5 3600, Aorus x570, LPX 3200, only front Samsung drive, crappy GPU and SSD next to it. Max stable values during 50 min run, not necessarily at the end. 75-80W on CPU. Clocks don't ever reach 4100MHz.

Ambient 24-25
Chipset 67
VRM 53
Front drive 59
CPU 83-87
GPU 48

Those are crappy CPU temps. I'm getting my Watercooling parts today.

Make sure you didn't degrade your CPU with a higher voltage, my 3700x sits at 4,3ghz on 1,275v (ye, golden chip I suppose)

 

[fabio]

@fabio how do you run tests, just blender test launcher? Does it matter which version?

Just for perspective, my results on crappy build. L12S 90 degrees intake, no extra fans, R5 3600, Aorus x570, LPX 3200, only front Samsung drive, crappy GPU and SSD next to it. Max stable values during 50 min run, not necessarily at the end. 75-80W on CPU. Clocks don't ever reach 4100MHz.

Ambient 24-25
Chipset 67
VRM 53
Front drive 59
CPU 83-87
GPU 48

Those are crappy CPU temps. I'm getting my Watercooling parts today.

Yes, I just run the blender benchmark scene for 30 minutes at a resolution of 4096x1920 on Blender 2.82.

BUT, if you don't fix manually the frequency and the voltage, you will never see anything above 4GHz on longer periods!

For example, my 3950X, without any tweak spikes at 4.6, BUT during rendering, it stays at 3900Mhz on all cores.

The same was with a 240 Custom loop: all cores, 3950MHz (lower temps though).

But of course in that case I could overclock to 4.3GHz all cores with 1.275V and still keep the CPU at around 80C

In 65W ECO MODE, all cores stay at 3400MHz, and with the tweaked PBO of 120W, all cores reach 3800MHz stable.

Also, if you don't have an exhaust fan, I suggest putting the CPU fan in exhaust!

And check the BIOS VOltage for the CPU. Don't keep it in AUTO but put it on NORMAL!

And let us know!!

 

[Luke NK]

Very nice work @fabio

Before you said a 120 mm AIO was better than noctua nh-l12s if CPU had a tdp above 120. Would this mean temps would be the same in l12s if running a 3700x than running it on a good 120 AIO with a noctua fan swap?

I'm always interested in noise profile of course. Curious if I could run a potential upcoming 4700x with either L12S or a 120 AIO at reason able noise levels

 

[bynio]

Yes, I just run the blender benchmark scene for 30 minutes at a resolution of 4096x1920 on Blender 2.82.

BUT, if you don't fix manually the frequency and the voltage, you will never see anything above 4GHz on longer periods!

For example, my 3950X, without any tweak spikes at 4.6, BUT during rendering, it stays at 3900Mhz on all cores.

The same was with a 240 Custom loop: all cores, 3950MHz (lower temps though).

But of course in that case I could overclock to 4.3GHz all cores with 1.275V and still keep the CPU at around 80C

In 65W ECO MODE, all cores stay at 3400MHz, and with the tweaked PBO of 120W, all cores reach 3800MHz stable.

Also, if you don't have an exhaust fan, I suggest putting the CPU fan in exhaust!

And check the BIOS VOltage for the CPU. Don't keep it in AUTO but put it on NORMAL!

And let us know!!

I've only enabled things along the line of 1usmus power plan and PBO. No specific voltage adjustment. And there is an auto drop of 35MHz from max clocks every +10deg over 50deg on CPU. So mine stabilises on 4030MHz on 87deg (offset around -140MHz) which would suggest that on 50deg it could do close to nominal 4200MHz boost on all cores.

Blender Benchmark Launcher doesn't let be choose resolution. Do you start it from the main app?

 

[fabio]

I've only enabled things along the line of 1usmus power plan and PBO. No specific voltage adjustment. And there is an auto drop of 35MHz from max clocks every +10deg over 50deg on CPU. So mine stabilises on 4030MHz on 87deg (offset around -140MHz) which would suggest that on 50deg it could do close to nominal 4200MHz boost on all cores.

Blender Benchmark Launcher doesn't let be choose resolution. Do you start it from the main app?

Sorry, is a demo scene called Cosmo Lunadromat, but the file is called benchmark

I don't have any PBO enabled, and in the first main page of the BIOS the CPU Voltage I've set on NORMAL.