A really surprising ITX cooler - Noctua. Sort of review.

[UglyDuckling]

Now Noctua explains this about this cooler.

Caution: The NH-L12 is a low-profile quiet cooler designed for use in small form factor cases and HTPC environments. While it provides first rate performance in its class, it is not suitable for overclocking and should be used with care on CPUs with more than 95W TDP (Thermal Design Power). Please consult our TDP guidelines to find out whether the NH-L12 is recommended for your CPU.

So ok, i run a 32nm xeon on x58 with a base TDP rating of 80 watts.. easy to cool at stock clocks.

Well what happens if i push the xeon? and not just a small amount either, a +1.6ghz and from 1.05v to 1.3v? HTT enabled.

Thermal paste used is Arctic Silver Ceramique 2.

Images of the cooler attached to the x58 board.

A note.. i took the machine down and took heatsinks off the motherboard, cleaned up the old goop and applied Ceramique 2 to the x58 chipset and southbridge (Which had none applied)

I have been using this system and cooler since June 2016.

Idle and max temps.

A small note i forgot to apply the maximum multiplier in the bios so it is at 4.1ghz right now but the voltage is the same used for 4.2ghz, i messed in bios because i had a small issue once i rebuilt with CPU fan error.

Insane right?

Not only cooling WAY above expectation but silent too.

I don't expect all CPU's to be cooled by this cooler, but since this can chill a Gulftown based 32nm Westmere-EP xeon on x58 with a huge overclock.. that's some serious cooling chops!

On top of this looking at the images of it mounted, the entire cooler covers the VRM area and MOSFETS making it a perfect cooler for the task.

Would like to see some feedback for this since yeah even Noctua said this is not meant for overclocking and not for chips higher than 95 watts which at 1.3v i can assure you the xeon is much higher than 95w TDP.

 

[BonzaiDuck]

You didn't mention which program(s) you used to stress the processor and generate those temperatures. Now I see that CPU-Z has a benchmark or stress-test and a "stop" button, so I assume that answers my question.

Take for example my Skylake system -- I'm always talking about my SKylake system . . .

AIDA 64 stress test including CPU, FPU, cache and memory, limps out at the low 50's C -- actually I'm seeing 49C steady -- with my maximum OC @ 4.7Ghz.
Intel Burn Test set to "Maximum" is also pretty limpy -- 68C to 72C
OCCT:CPU -- 67C peak package temperature
Prime95 sFFT -- 75C peak package
LinX affinitized -- 80C peak package

Note that the sustained temperatures for a core average are lower than these peaks by 5C or more.

On the other hand, OC'd as your processor is, I'll say that's pretty good.

 

[UglyDuckling]

You didn't mention which program(s) you used to stress the processor and generate those temperatures. Now I see that CPU-Z has a benchmark or stress-test and a "stop" button, so I assume that answers my question.

Take for example my Skylake system -- I'm always talking about my SKylake system . . .

AIDA 64 stress test including CPU, FPU, cache and memory, limps out at the low 50's C -- actually I'm seeing 49C steady -- with my maximum OC @ 4.7Ghz.
Intel Burn Test set to "Maximum" is also pretty limpy -- 68C to 72C
OCCT:CPU -- 67C peak package temperature
Prime95 sFFT -- 75C peak package
LinX affinitized -- 80C peak package

Note that the sustained temperatures for a core average are lower than these peaks by 5C or more.

On the other hand, OC'd as your processor is, I'll say that's pretty good.

Ok... how about i render a full movie?

Interestingly.. same CPU max temps albeit hotter by 1c...

That's the temp it peaked at, it goes no hotter for me.

If i pump the voltage up to 1.375v and run the CPU at 4.4ghz it peaks at 73c...

 

[UglyDuckling]

Linpack OCCT... i mean sure higher temps but massively unrealistic temperatures and unrealistic load on the CPU.. so pretty much worthless.
Ran for 5 minutes and it went no higher.

60-62c is where this CPU maxes out under this cooler.

"burn tests" are exactly what they say.. burn tests.. like Furmark for a GPU.

For almost a year i have never seen higher than 61-62c at 4.2ghz pushing 1.3v

 

[Crono]

What case?

 

[BonzaiDuck]

I guess it has become a habitual bias on my part. "If it doesn't run [this or that] stress program at a target speed and voltage" it isn't "reliable."

Does that explain much about the real world? Probably not. I've just seen loosely accepted voltage and speed settings proven under weaker tests which later throw a BSOD with mild gaming. Thank God for BSOD Stop Codes that have diagnostic value. . .

Personally, I have faith in the OCCT:CPU test run for 3.5 hours. If the system falls short on stability, it should find it in that time. And the temperatures are reasonable.

But it's hard for me to jettison LinX because a run of 10 iterations or so gives me a means of fine-tuning voltage after the stable range is reached. I'd be happier if it didn't attempt to toast the processor.

 

[UglyDuckling]

What case?

Xigmatek Utguard.

Full component list.

P6X58D-E with newly replaced thermal pads and applied grease (Ceramique 2)
1x RX 580 MSI Gaming
1x Reference GeForce GTX 670 SC
Silverstone ST1000-P 80 Plus Silver PSU
2x Samsung Spinpoint F3 1TB 7200RPM
1x Corsair bottom intake to feed GPU's (Came with old broken H100 CPU cooler)
1x 170mm on top blowing outwards (VRM area top of CPU)
1x Phobya G.Silent 120 rear exhausting.

Side panel has no windows and just 1 vent hole for a fan attachment.

 

[UglyDuckling]

I guess it has become a habitual bias on my part. "If it doesn't run [this or that] stress program at a target speed and voltage" it isn't "reliable."

Does that explain much about the real world? Probably not. I've just seen loosely accepted voltage and speed settings proven under weaker tests which later throw a BSOD with mild gaming. Thank God for BSOD Stop Codes that have diagnostic value. . .

Personally, I have faith in the OCCT:CPU test run for 3.5 hours. If the system falls short on stability, it should find it in that time. And the temperatures are reasonable.

But it's hard for me to jettison LinX because a run of 10 iterations or so gives me a means of fine-tuning voltage after the stable range is reached. I'd be happier if it didn't attempt to toast the processor.

A year of fully stable use is quite telling that it's fully stable...
I render video, play some demanding games too even, look this is before i purchased the RX 580...

 

[BonzaiDuck]

A year of fully stable use is quite telling that it's fully stable...
I render video, play some demanding games too even, look this is before i purchased the RX 580...

That's also true -- but ceteris paribus as they say -- "all things being equal." I've had Blue screens arise from driver conflicts. For any BSODs I got on this system related to hardware settings, I can count about four during the first few days when I started to tweak it and get familiar with the chipset, board and processor. You know you're close when the stress-testers trap the error without crashing Windows.

Some may also remember a panic I had around 2013 -- for a system running 24/7 and rock-stable -- EXCEPT! for an idle instability that would occur about once every 14 days. It turned out to be a matter of C-state settings in BIOS. Those types of things are hard to diagnose (took me a couple months). They have nothing to do with temperatures or voltage.

Anyway! I can be happy for your ecstasy for the X58 and Xeon system. That's nice. So when do you plan to upgrade it -- eventually?

 

[dmoney1980]

Did you bend the cooler? Because the side shot of it mounted on the board looks like the heatsink is very crooked

 

[BonzaiDuck]

Did you bend the cooler? Because the side shot of it mounted on the board looks like the heatsink is very crooked

It may not have been perfectly level as it was shipped.

The OP seems to be a careful enthusiast who wouldn't exhibit that kind of clumsiness.

And in other respects, after taking pictures of my own system, things could appear misaligned just for the camera angle.

 

[UglyDuckling]

Did you bend the cooler? Because the side shot of it mounted on the board looks like the heatsink is very crooked

It is ever so slightly bent, the cooler was bought second hand, there is no real issue however.

Good eye!

 

[UglyDuckling]

That's also true -- but ceteris paribus as they say -- "all things being equal." I've had Blue screens arise from driver conflicts. For any BSODs I got on this system related to hardware settings, I can count about four during the first few days when I started to tweak it and get familiar with the chipset, board and processor. You know you're close when the stress-testers trap the error without crashing Windows.

Some may also remember a panic I had around 2013 -- for a system running 24/7 and rock-stable -- EXCEPT! for an idle instability that would occur about once every 14 days. It turned out to be a matter of C-state settings in BIOS. Those types of things are hard to diagnose (took me a couple months). They have nothing to do with temperatures or voltage.

Anyway! I can be happy for your ecstasy for the X58 and Xeon system. That's nice. So when do you plan to upgrade it -- eventually?

Well AMD has re-entered the arena and they look awesome, not so awesome for overclocking but for the money...

I don't need an upgrade right now, nothing slow this CPU down enough, even rendering is a breeze on it.

I am intrigued if AMD will release a overclocking friendly ZEN, if so that would catch my eye.

 

[BonzaiDuck]

How does the Xeon compare to the Gulftown -- speed and OC-a-bility? I've had "plans" that I never pursued, and I remember now that I really had it in mind for several months to buy the $1000+ Gulftown and a socket-1366 ASUS Rampage, but I didn't follow through.

I was going to put it in the case shown in "Case expense" thread and post I made just minutes ago. And I was obviously going to water-cool it. But I didn't do that, didn't keep the case (when I should have), and waited at least another year or so . . .

Or am I wrong? Was there a RAmpage or a Maximus socket-1366? And wasn't the Gulftown for socket 1366? I can't surely remember . . . .

 

[UglyDuckling]

How does the Xeon compare to the Gulftown -- speed and OC-a-bility? I've had "plans" that I never pursued, and I remember now that I really had it in mind for several months to buy the $1000+ Gulftown and a socket-1366 ASUS Rampage, but I didn't follow through.

I was going to put it in the case shown in "Case expense" thread and post I made just minutes ago. And I was obviously going to water-cool it. But I didn't do that, didn't keep the case (when I should have), and waited at least another year or so . . .

Or am I wrong? Was there a RAmpage or a Maximus socket-1366? And wasn't the Gulftown for socket 1366? I can't surely remember . . . .

My CPU is a quad core hyperthreaded part so like an i7 but has 4MB more L3 cache and has more extensions and is binned obviously.

If i used something like a D15 or H100i cooler i could easily see this hit 4.5ghz minimum, my board can push passed 200mhz BCLK easily before anything starts to kink.

The P6X58D-E is a 32NM ready board with USB 3.0 and was launched in 2010 so a much later release for 1366.

Westmere-EP XEON's all share the same tech that was used for the Gulftown Extreme Edition CPU's known as the i7 970/ 980XEE and 990XEE

You will be incredibly surprised just how fast these chips are still these days.

Here is the E5640 vs a 2600K in Crysis 3, you would never think it would beat Sandy Bridge but it does.

 

[aigomorla]

problem with gulftown and westmere is you will bottleneck a modern gpu like a 1070 or 1080.
You will even bottleneck a 970 and 980.