Riddle me this - Wraith Prism matching or beating scythe mugen 5

Relentless81

Relentless81

Relentless81

Soldato
Joined
18 May 2010
Posts
12,785
So I bought a Mugen 5 and installed it to my cooler master nr200p which has 2 X noctua a12x15 intakes at the bottom as intake and 2 X arctic p12 in the top as exhaust

The CPU is a 3600 stock settings

I run heaven for half an hour, CPU temp 72C that's weird, that's about the same as the Wraith but I'll play a game and compare it
Because I know the Wraith Prism maxes out peak temp 72c after an hour of my favourite game

After an hour and a half I check the temps, 72C the Same at the Wraith Prism

I reseat the cooler and repaste with MX4, 72C after half an hour of Heaven

Fast forward to today, I send the Mugen 5 back, reinstall the wraith Prism and run Heaven for half an hour, room temp is 1c warmer than yesterday, 69C after half an hour

So, is it the case because it's bottom to top air flow and that suits top down coolers better?

That's all I can think of, how can a 6 pipe 120mm tower cooler be matched or beaten by a 92mm stock AMD cooler?

I'm genuinely interested to know what's happening here

Cheers
 
Did you check how Mugen 5's base fitted to your CPU's heatspreader?
Scythe apparently still uses Intel optimized convex base, whose shape is harder to control in fast and cheap enough production.
Ryzen's heatspreader again is flattish or possibly smidge convex from center, which can cause mediocre/bad contact on most of the area and Zen2 has heat production off center.
Hence grinding heatsink's base flat could help lot with Ryzen.
https://www.igorslab.de/en/kuehlerb...amd-and-conkav-reader-test-and-picture-story/


Also did you check what kind speeds fans were running at those temperatures in case BIOS/some software was limiting fan speed basing on temperature?

And that case has insanely restricted bottom and needs high pressure fans as bottom intake instead of such slim very low pressure fans.
 
EsaT said:
Did you check how Mugen 5's base fitted to your CPU's heatspreader?
Scythe apparently still uses Intel optimized convex base, whose shape is harder to control in fast and cheap enough production.
Ryzen's heatspreader again is flattish or possibly smidge convex from center, which can cause mediocre/bad contact on most of the area and Zen2 has heat production off center.
Hence grinding heatsink's base flat could help lot with Ryzen.
https://www.igorslab.de/en/kuehlerb...amd-and-conkav-reader-test-and-picture-story/


Also did you check what kind speeds fans were running at those temperatures in case BIOS/some software was limiting fan speed basing on temperature?

And that case has insanely restricted bottom and needs high pressure fans as bottom intake instead of such slim very low pressure fans.
Click to expand...

I didn't check the base no but could explain it. I'm not interested in lapping CPUs

I did check the fan speeds the CPU fan is set to 100% at 65c so both coolers hit 100% fan speeds

It's either slim fans or none at all, there isn't enough room for 25mm fans, I wouldn't say it was restricted though, seems quite open at the base and I can feel the air moving from the side of the case
 
I don't mean lapping CPU but heatsink base, because that's what is more out of shape between Ryzen heatspreader and Intel optimized heatsink.
Just compare that thermal paste spread of that article's first page image to this:
https://www.igorslab.de/en/kuehlerb...-and-conkav-reader-test-and-picture-story/20/
Entirely different quality contact with CPU...


That "triple-slot GPU support" image shows that already that crappy stamping of bottom blocks 60+ % of the area and then also that filter mesh under it is quite restrictive.
https://www.coolermaster.com/catalog/cases/mini-itx/masterbox-nr200p
Here's some airflow speed comparison with different things blocking fan:
https://www.silverstonetek.com/techtalk_cont.php?tid=wh_chessis&area=usa
 
EsaT said:
I don't mean lapping CPU but heatsink base, because that's what is more out of shape between Ryzen heatspreader and Intel optimized heatsink.
Just compare that thermal paste spread of that article's first page image to this:
https://www.igorslab.de/en/kuehlerb...-and-conkav-reader-test-and-picture-story/20/
Entirely different quality contact with CPU...


That "triple-slot GPU support" image shows that already that crappy stamping of bottom blocks 60+ % of the area and then also that filter mesh under it is quite restrictive.
https://www.coolermaster.com/catalog/cases/mini-itx/masterbox-nr200p
Here's some airflow speed comparison with different things blocking fan:
https://www.silverstonetek.com/techtalk_cont.php?tid=wh_chessis&area=usa
Click to expand...

Thanks for link, interesting read. I didn't notice an uneven spread of thermal paste, it seemed to spread quite evenly on both applications. I guess the challenge is finding an air cooler suited to Ryzen but I wont be buying Scythe again which is unfortunate as the price and mounting system are really good
 
EsaT said:
Actually only direct contact heatpipe coolers are certainly with flat base.
https://www.overclockers.co.uk/alpenfohn-brocken-3-cpu-cooler-140mm-hs-05a-al.html

Also Noctuas etc have convex bases and might have fitting problems with Ryzen depending on indivual variations on both sides of contact.
https://www.igorslab.de/en/intel-vs...er-really-what-is-better-ryzen-9-or-core-9/2/
Click to expand...

Can't fit the Brocken 3 unfortunately, I can fit the Eco Advance though as I was looking at it previously

It is direct touch so does that automatically mean it's flat?
 
Relentless81 said:
It is direct touch so does that automatically mean it's flat?
Click to expand...
Way of manufacturing of those doesn't allow non-flat base. (or at least easily)
They're literally made by flattening and pressing those those heatpipes into that slot/space in base piece with possibly some light final grinding.
(too much would puncture heatpipe)

Grinding/lapping base of heatsink flat is really rather easy and basically risk free.
(unlike doing anything to PGA socket CPU with pins)
 
EsaT said:
Way of manufacturing of those doesn't allow non-flat base. (or at least easily)
They're literally made by flattening and pressing those those heatpipes into that slot/space in base piece with possibly some light final grinding.
(too much would puncture heatpipe)

Grinding/lapping base of heatsink flat is really rather easy and basically risk free.
(unlike doing anything to PGA socket CPU with pins)
Click to expand...

Ok good to know thanks

Still not sure if it's the issue just because I've used noctua coolers on Ryzen before with good results and apparently they also have convex base but I've never tried a cooler with direct contact pipes before and the Brocken Eco is cheap so might be worth trying
 
EsaT said:
Was that Zen/Zen+ Ryzen?
With monolithic die in center of heatspreader those results are notably invalid for chiplet/MCM CPUs.
Zen2's CCDs are actually completely off center with IO die being closest to heatspreader center:
https://www.guru3d.com/articles-pages/amd-ryzen-7-3800x-review,2.html
Click to expand...

I had a U12S AM4 on a 1600X peak gaming temps in the mid 60's and a D15S on a 2600X again peak gaming temps in the mid 60's

If you weren't interested in lapping anything and wanted something to work well from the off with a height limit of 154mm what would you get for a 3600?
 
Mugen should run circles around Prism.
Convexity of base can't explain this
Any more data points? Are idle temps also similar? Are load temps rising immediately or gradually?
 
EsaT said:
154mm is darn demanding limit.
For example extreme bang per buck Arctic Freezer 34 is 157mm.
https://www.overclockers.co.uk/arctic-freezer-34-cpu-cooler-120mm-hs-077-ar.html

Cryorig H7 would be actually only 145mm tall, but according to Tweaktown its base is slightly convex.
https://www.tweaktown.com/reviews/6958/cryorig-h7-cpu-cooler-review/index.html

Brocken Eco Advanced would at least have flat base.
BeQuiet Pure Rock 2 would be 155m high.
Click to expand...

The Cryorig H7 4 heat pipe versions are very hard to get hold of and I've read the base is no good for Ryzen some users reporting poor results

I think the Pure Rock 2 wont fit, the Mugen is 154.5mm tall and the gap between the cooler and glass was so small I wouldn't have been able to slide a piece of paper through it

The Brocken is an option, such a shame that its so unpopular there is very little data out there on it apart from a couple of German reviews

alec said:
Mugen should run circles around Prism.
Convexity of base can't explain this
Any more data points? Are idle temps also similar? Are load temps rising immediately or gradually?
Click to expand...

It should shouldn't it, it didn't in this case though.

Idle temps were very similar, once Windows booted after a few minutes both coolers sit in the low 30's with the odd burst and settle to the low 30's again

I didn't monitor how fast temps risen but fast rising temps generally go hand in hand with poor fitment, poor application of paste or fault and very high temps and throttling. My temps didn't go over 72C with either cooler so the Mugen was cooling the CPU effectively enough just not beating the Wraith Prism in terms of peak temps which I feel is unexpected. Average temps were also similar on both coolers low to mid 60's
 
CPU temp will go from CPU idle temp of about (28c) to CPU at full load temp of about (65-70c) and stabilize in less than 3 minutes .. and when CPU goes back to idle it will reach a stable idle temp in same 3 minutes in a case with good airflow. I've tested this many times running Prime and/or rendering graphics.

But if case airflow is not matched to component airflow demands it can take 20-50 minutes of constant gradual increase in temp before they stabilize.

A graph with CPU load and fan speeds would give a better view of what CPU temps and fan's speeds are doing as CPU temps change.

Below are graphs from idle to full load and back to idle from one of my builds. First is short Prime 95 run, 2nd is while running encodes. The drops/spikes every 5 minutes are from one encode to next. The CPU drops/spikes in each 5 minute run are dificutly differences in encode. Hope that makes sense.

315332-c4361109e35daa613d764d8fb58db574.jpg
315338-eef108ef0304a1d092ffc5a8279bed37.jpg

Too small ..

Here they are big enough to read. ;)

i7-980-prime-95-27-07-2014-mod-jpg.315332
ohm-temp-load-rpm-graph-jpg.315338
 
You must log in or register to reply here.
Back
Top Bottom