Production FormD T1 Classic (READ FIRST POST)

[Wahaha360]

I have nothing to do with the testing, obviously, BUT how in hell is going to be reflective of real world if you remove the side panels?
The heat generated inside the case can effect cooling in various ways OR all the component would be effected similarly?
Thanks.

chyll2 hit the nail on the head. I just want to compare 240mm vs 120mm radiators, get the relative performance to each other. I also want to see how good the HWL LS120 is relative to my Koolance HX-CU420V.

When I get prototype 2, I will test these radiators inside the case.

You can't perform that test on the prototype since IRC, it doesn't accept 240 rad

Prototype 2 will support the Alphacool LT240 with a little hack.

 

[Wahaha360]

It's perfect for me!

I confirm:

- A4 V3 no panels
- 9900K (normal sample) + ASUS Z390 ITX, GSkill 3200 C14 + Samsung 970 Pro (512GB)
- Eisbaer LT DC-LT Pump + Noctua Sterrox + TX240 + 10mm Tube.
- Corsair SF600 Platinum.

I think we need to have the same tubing length maybe? How do you want to setup the DAN? Everything inside, only the radiator Outside?
You can also setup the BIOS profile, save the file and send to me! I am using the last bios.

Anyway, let me know how do you want to procede!

BTW, I cannot use the Titan RTX for now, I have an RTX 2070 or Quadro P4000 free. Is it ok?

Even without, using the iGPU

Hardware matched.

Let's do SFX outside the case, iGPU. I want to get as close to an open bench as possible.

We should benchmark stock 9900K, 4.7ghz on all cores, and 5.0ghz on all cores. We should run Prime95 and AIDA64 and Cinebench.

I don't overclock (paranoid of crashing), so I'm practically a newbie on that front. If you are comfortable, I can use your bios settings for the Asus Z390 ITX.

Btw, what kind of thermal compound do you want to use?

 

[Wahaha360]

If Koolance 120x30mm HX-CU420V performs close to Alphacool LT240, typical 240mm AIO support (Corsair H100i Pro, EVGA CLC 240, NZXT Kraken X52) is unnecessary.

Without OC, the Corsair H60 2018 on the 9900K (USD 70) might be the best easy-solution for most.

With OC, the Alpahcool LT240 > most 240 AIOs. The pump has a 3 pin power connector (instead of SATA power connector on other AIOs), that is way better for cable mgmt. For SFF, it's worth it for looks, airflow obstruction / temperature, and your sanity.

 

[dondan]

@Wahaha360 If you buy 10x 9900k every CPU will generates different heat so you can't compare coolers that are not tested on the exact same CPU.

 

[Allhopeforhumanity]

@Wahaha360 If you buy 10x 9900k every CPU will generates different heat so you can't compare coolers that are not tested on the exact same CPU.

If you lock them to the same voltage, frequency (ideally stock) and turbo stepping profile, a pair of soldered CPUs should give pretty comparable performance. Sure they wont be totally identical, but within a degree or 2 unless you have a wildly defective or golden chip.

 

[fabio]

@Wahaha360 you can start to check the bios; I propose:

1: Everything in AUTO, XMP I, FAN STANDARD PROFILE (Maybe we can even avoid this, but a lot of people is not spending the time to overclock their system)
2: MCE Disabled, ALL CORE 47, 1.250V Core, Cache 43, AVX offset 0, XMP I, FAN STANDARD PROFILE
3: MCE Disabled, ALL CORE 50, 1.280V Core, Cache 43, AVX offset 3, XMP I, FAN STANDARD PROFILE

Under Prime95 sFFT, for a 120mm 4.7GHz with the sFFT test is the MAX we can have. a 280mm custom loop can achieve 4.9, but is the limit for it.
A more realistic and good frequency will be 4.6GHz for that test, but we are testing the limit also, so is good to stay at 4.7GHz for this purpose.

I suggest adding also IntelBurnTest, 10 loops, Very High settings.

BTW, here also the test on radiators for the slim A12x15! You should really think to this as well for the minimum thickness of the setup. They are very good on slim radiators!

I am under deadline until the 6th, so I can start to test from the 7th till the 11th. We should also decide how to register the temp, and for how long, like 30 minutes, 1 hour.

 

[fabio]

@Wahaha360 If you buy 10x 9900k every CPU will generates different heat so you can't compare coolers that are not tested on the exact same CPU.

I agree in general, but if the difference is around 2/3 degrees, It will be ok, to evaluate the efficiency of that specific configuration, I think.

 

[dlal1520]

I just recently stumbled my way across this project, and I gotta say im honestly impressed with all these tests and updates. Looking forward to seeing more of it coming together.

Just have a question about the possibilities of dust filters? apologies if thats been asked about already havent had a chance to look through all of the posts

 

[dondan]

If you lock them to the same voltage, frequency (ideally stock) and turbo stepping profile, a pair of soldered CPUs should give pretty comparable performance. Sure they wont be totally identical, but within a degree or 2 unless you have a wildly defective or golden chip.

Maybe this changed nowadays but in Sandy Bridge times (2x00) where I was very active in OC the differences was very high up to 10°C on a heavy water cooled setup.

 

[Allhopeforhumanity]

Maybe this changed nowadays but in Sandy Bridge times (2x00) where I was very active in OC the differences was very high up to 10°C on a heavy water cooled setup.

Oh, that isn't to say that 9900k and other soldered CPUs can't vary quite a bit under heavy overclock/overvolt scenarios, but the way they are binned with regard to their stock parameters are far more uniform than ivy bridge/sandy bridge for sure.

 

[Killinger]

Crazy idea, but could you mount a few 40mm fans on the part of the 240mm AIO that is directly under the motherboard? Could be an interesting trick to get a bit more airflow without needing to increase the size to incorporate a slim 120mm fan below the motherboard like in the early renders

 

[PBJ]

I’m sorry for asking this in between the radiator talk!

I plan to use a Noctua L12S which I assume will exhaust most of the warm air through the side vents. With two 25mm fans on the bottom, which placement is better for the L12S, the heatsink fins looking towards the bottom fans or towards the sides. Won’t the bottom fans, being so close to the heatsink, cause some turbulence with the L12S’s fan?

 

[dondan]

Oh, that isn't to say that 9900k and other soldered CPUs can't vary quite a bit under heavy overclock/overvolt scenarios, but the way they are binned with regard to their stock parameters are far more uniform than ivy bridge/sandy bridge for sure.

I verified my thoughts with some OC guys from a german forum and it looks like I am right. You can't compare different CPUs out of the same family even with same clock and vcore. If you like to understand why you have read about what VID means. Since the VID is different on every CPU it means that the resistent of the silicon is also different. If you have a different resistent and you setup same vcore the power draw and heat generation is also different.

 

[Allhopeforhumanity]

I verified my thoughts with some OC guys from a german forum and it looks like I am right. You can't compare different CPUs out of the same family even with same clock and vcore. If you like to understand why you have read about what VID means. Since the VID is different on every CPU it means that the resistent of the silicon is also different. If you have a different resistent and you setup same vcore the power draw and heat generation is also different.

By VID, are you talking about the Voltage Identifier requests the CPU makes to the motherboard? I'm totally on board with there being a different true resistance of the silicon chip to chip, but I was under the impression that 14nm+++ had a pretty small variance in process defect driven resistance.

If you happen to have numbers for the variance, I'd love to see them. If nothing else, they could be used as error bars in the tests that Fabio and Wahaha wan't to perform.

 

[fabio]

Thank you guys for all of this info!
@Wahaha360, I am ready to test with the TX240, later I sent pics of the setup and bios. I suggest to start with the 50 all core, 1.3V for the core. (Tried 1.28 but cinebench crash otherwise).
Let me know how do you want to procede!
(I am using Windows 10 1803).

I suggest:
Prime95 29.5 build 9 (Smallest sFFT for the maximum heat) for 1 hour.
Wait 10 minutes, than Intel burn test 10 loop very high.
10 minutes wait, cinebench for 5 times in a row
Aida 64 for one hour.
Blender with the benchmark scene (are around 25 minutes of rendering)

But, how many times all this?

Every time we relaunch HWINFO 64, and screenshot the page, 1 meanwhile, 1 at the end of the stress test.

What do you think?
I am curious about also the DAN comment to check these variances between chip!

 

[Wahaha360]

Thank you guys for all of this info!
@Wahaha360, I am ready to test with the TX240, later I sent pics of the setup and bios. I suggest to start with the 50 all core, 1.3V for the core. (Tried 1.28 but cinebench crash otherwise).
Let me know how do you want to procede!
(I am using Windows 10 1803).

I suggest:
Prime95 29.5 build 9 (Smallest sFFT for the maximum heat) for 1 hour.
Wait 10 minutes, than Intel burn test 10 loop very high.
10 minutes wait, cinebench for 5 times in a row
Aida 64 for one hour.
Blender with the benchmark scene (are around 25 minutes of rendering)

But, how many times all this?

Every time we relaunch HWINFO 64, and screenshot the page, 1 meanwhile, 1 at the end of the stress test.

What do you think?
I am curious about also the DAN comment to check these variances between chip!

Let's do it this weekend, I will be in Germany with dondan.

I am flying tomorrow, so I have pack and finish work in the US.

If we can't sync results due to CPU variation, just mail me the radiator, then we don't have to guess.

 

[fabio]

@Wahaha360 I've "tested" the TX240 today.
I did, 2 times for each test (Next post the results). Here the setup:

"Open Bench" DAN Case
Corsair SF600 Platinum
ASUS Z390 ITX BIOS 1201
Intel 9900K, CPU All Core 50GHz, AVX Offset 3, CPU Cache 43, Vcore 1.300V, MCE Disabled, VCCIO 1.05, VCCSA 1.15, Load Line 6, all turbo duration to MAX
Integrated GPU
G.Skill 3200MHz C14 14 14 14 34, XMP I, 1.350V
Thermal Grizzly Kryonaut, rice grain method
Eisbaer LT Pump+Combo on AIO_PUMP header, control disabled in bios to run it 100%
35cm 13/10mm tubes
XSPC TX240 Radiator
2x Noctua A12x25 Sterrox in PULL mode, pulling air from the radiator against the VRM heatsink to keep it "cooler", BIOS FAN PROFILE: STANDARD.

AMBIENT TEMP 22/22,5 Degrees

Following picture of the BIOS Settings to help to set up it same way, and picture of the system placed.

(the MCE is DISABLED, here the pic was taken before doing that)







And the TX240 System

 

[Wahaha360]

@Wahaha360 I've "tested" the TX240 today.
I did, 2 times for each test (Next post the results). Here the setup:

"Open Bench" DAN Case
Corsair SF600 Platinum
ASUS Z390 ITX BIOS 1201
Intel 9900K, CPU All Core 50GHz, AVX Offset 3, CPU Cache 43, Vcore 1.300V, MCE Disabled, VCCIO 1.05, VCCSA 1.15, Load Line 6, all turbo duration to MAX
Integrated GPU
G.Skill 3200MHz C14 14 14 14 34, XMP I, 1.350V
Thernal Grizzly Kryonaut, rice grain method
Eisbaer LT Pump+Combo on AIO_PUMP header, control disabled in bios to run it 100%
35cm 13/10mm tubes
XSPC TX240 Radiator
2x Noctua A12x25 Sterrox in PULL mode, pulling air from the radiator against the VRM heatsink to keep it "cooler", BIOS FAN PROFILE: STANDARD.

AMBIENT TEMP 22/22,5 Degrees

Following picture of the BIOS Settings to help to set up it same way, and picture of the system placed.

(the MCE is DISABLED, here the pics was taken before doing that)







And the TX240 System

Noctua Thermal Grease?

If I have time later in the night, I will try before i pack

 

[fabio]

Noctua Thermal Grease?

If I have time later in the night, I will try before i pack

Nope, Thermal Grizzly Kryonaut

 

[Wahaha360]

Nope, Thermal Grizzly Kryonaut

I don't have Kryonaut, let me get that then test.