Semi-poll of Core 2 Duo Ninja users

Cooling Processors quietly

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Ryan Norton
Patron of SPCR
Posts: 169
Joined: Wed Sep 21, 2005 5:13 pm
Location: South FL

Semi-poll of Core 2 Duo Ninja users

Post by Ryan Norton » Sun Mar 25, 2007 12:55 pm

Hi everyone,

I finally got around to ripping my original Ninja off my parents' old Socket 939 A64 system and installing it on my P5B Deluxe and E6400. Since it's an old Ninja, I used the mounting system where you sticky a cross-shaped backplate under the socket area, then use it to screw two rails (one per side of the socket) that each have two receptacles for the Ninja's spring clips.

Anyway, I was more than a little hesitant to do this after reading what seems like a significant number of Core 2 Duo + Ninja horror stories. I'm curious, then, to see if my temps are higher than they ought to be.

I've only had it hooked up for a few hours, but I see 31-33C at idle and haven't seen either core go over 45C in use (iTunes open, playing Supreme Commander, plus many tabs of Firefox sucking down memory) yet, according to CoreTemp. That SEEMS like an improvement over the stock heatsink, but I don't remember what those temps tended to be.

THe other thing I Just remembered is that I did NOT use the curious little spacer included for use with LGA775 sockets, that I guess is supposed to sit between the top of the CPU's IHS and the contact surface on the base of the Ninja. I mounted the Ninja on the CPU with no paste applied to see if there was solid contact, and since there was, opted not to use the spacer. I really don't like the idea of a random piece of metal floating between the base of the Ninja and the top of the CPU for so many reasons.

Anyway, let me know what you think of this experience. My tentative conclusion is that the pre Rev. B Ninja LGA775 mounting system is less troubled than the one Scythe is currently using, for what it's worth.

derekchinese
Posts: 255
Joined: Fri Mar 02, 2007 6:57 pm
Location: Nowhere.

Post by derekchinese » Sun Mar 25, 2007 1:31 pm

I believe the horror stories are from the rev. B. The original should be fine. The difference between the rev a and reb B is the backplate. I think this is the critical problem with the rev B--No backplate = no pressure = poor contact = hot CPU.


As for your temps, idle really does not matter. You should try some torture testing like orthos or TAT to see if your load temps are safe.


Derek

cmthomson
Posts: 1266
Joined: Sun Oct 09, 2005 8:35 am
Location: Pleasanton, CA

Post by cmthomson » Sun Mar 25, 2007 4:17 pm

Original Ninjas mounted with the backplate and rail system cool Core 2 Duo CPUs exceptionally well. The problems are with the newer Ninjas and their push-pin mounts.

JazzJackRabbit
Posts: 1386
Joined: Fri Jun 18, 2004 6:53 pm

Post by JazzJackRabbit » Sun Mar 25, 2007 5:19 pm

I got no problems with revision B and push-pin mount (that's with 4300 at 2.4GHz). I get 29-31 idle, haven't tested full load, but 50% load i.e. one core my temps don't go above 40-42.

shawn122
Posts: 23
Joined: Sat Feb 03, 2007 9:56 pm

Post by shawn122 » Sun Mar 25, 2007 9:07 pm

I have the Rev B....E6600 overclocked to 2.7 Ghz (mild overclock..lol) and my temps on on full load is 47 degrees Celsius using Orthos... (I actually ran this for 27 hours non-stop and my temps stayed at 47 :D )

And thats in purely passive mode...no fan on the ninja. I have the P5B Deluxe wifi ed. and the antec P180...the fans on the case do an adeqaute enough job.

I may purchase the other mounting system to do a comparison...but by all means...the Rev b. isn't bad at all.

dragmor
Posts: 301
Joined: Sun Jul 10, 2005 7:54 pm
Location: Oz

Post by dragmor » Sun Mar 25, 2007 10:29 pm

JazzJackRabbit wrote:I got no problems with revision B and push-pin mount (that's with 4300 at 2.4GHz). I get 29-31 idle, haven't tested full load, but 50% load i.e. one core my temps don't go above 40-42.
Try measuring with TAT. Basically TAT reports +15c above what every other application does.

Delta_42
-- Vendor --
Posts: 131
Joined: Mon Mar 05, 2007 8:36 am
Location: Lancs, UK
Contact:

Post by Delta_42 » Mon Mar 26, 2007 12:16 am

I just ditched a Ninja Rev B because of mounting problems, running the latest Core Temp (TAT won't run on my E6600 for some reason) I was getting 50deg idle and as high as 72 under load. Now I'm running a Noctua NH12 that has a bolt-thru mount and my temps are 32 idle 50 load (E6600 running on a 350Mhz FSB).

It seems to be pot luck whether the push-pin mount is adequate, given the size of the cooler I think using that mount is a little optimistic, I'm just sorry I didn't read all these threads before I bought the Ninja!

JazzJackRabbit
Posts: 1386
Joined: Fri Jun 18, 2004 6:53 pm

Post by JazzJackRabbit » Mon Mar 26, 2007 8:58 am

I wonder if people who had high temps with Ninja Rev. B simply didn't snap pins in place properly.

Snapping pins in place on 775 can be a little tricky, especially last one because when you snap it, it may "click" twice. It doesn't always happen, but happens often enough. If the user hears the first click and stops pushing, he may not have mounted heatsink properly.

Delta_42
-- Vendor --
Posts: 131
Joined: Mon Mar 05, 2007 8:36 am
Location: Lancs, UK
Contact:

Post by Delta_42 » Mon Mar 26, 2007 10:15 am

I mounted and re-mounted the Ninja several times and pushed on each pin as hard as I dared. Its a crappy mechanism that works OK on low-rise coolers but just isn't good enough for something the size of a Ninja, even the £300 HP Celeron PCs I buy at work use a bolt-thru HSF mount.

prostuff1

Post by prostuff1 » Mon Mar 26, 2007 10:35 am

Delta_42 wrote:I mounted and re-mounted the Ninja several times and pushed on each pin as hard as I dared. Its a crappy mechanism that works OK on low-rise coolers but just isn't good enough for something the size of a Ninja, even the £300 HP Celeron PCs I buy at work use a bolt-thru HSF mount.
Totally agree!!

The push pins are great for convenience but really are not really good enough for a tall, heavy CPU cooler.

Pooh-Bah
Posts: 41
Joined: Sat Apr 02, 2005 9:26 am
Location: Los Angeles

Post by Pooh-Bah » Mon Mar 26, 2007 6:50 pm

I've had two Rev B ninjas:

#1: Could barely keep my C2D below 50C. Performed worse than the stock intel cooler (which is surprisingly good, IMHO). Reinstalled several times with no change. Also didn't work well on a 939 platform. RMA'd it.

#2 (RMA replacement): Felt tighter on the C2D push-pin installation. Without a fan, it keeps the 2.9ghz C2D at 45 or lower - with no fan - totally passive. Heats up to maybe 50-55 under normal daily loads (non TAT or orthos type loads). Love it!

So...your milage may vary. When it works, it's great. When it doesn't, it is no better (or worse) than the stock intel cooler.

I saw a ninja-clone from OCZ. If they can put out a ninja-like setup that works consistently, they might have a winner. I haven't seen the mounting system on that one.

dentaku
Posts: 110
Joined: Tue Dec 14, 2004 7:33 am
Location: Canada

Post by dentaku » Wed Mar 28, 2007 5:40 am

shawn122 wrote:I have the Rev B....E6600 overclocked to 2.7 Ghz (mild overclock..lol) and my temps on on full load is 47 degrees Celsius using Orthos... (I actually ran this for 27 hours non-stop and my temps stayed at 47 :D )

And thats in purely passive mode...no fan on the ninja. I have the P5B Deluxe wifi ed. and the antec P180...the fans on the case do an adeqaute enough job.

I may purchase the other mounting system to do a comparison...but by all means...the Rev b. isn't bad at all.
I'm getting a P180 and an E6600 too and I was hoping to hear that the perfectly placed exhaust fans on that case would be enough to cool a Ninja.

By "other mounting system" do you mean one made by Scythe or something else like the Thermalright LGA 775 Bolt-Thru-Kit?

continuum
*Lifetime Patron*
Posts: 213
Joined: Tue Jul 19, 2005 10:23 pm
Location: California
Contact:

Post by continuum » Thu Mar 29, 2007 9:45 am

Ninja's seem to run out of capacity as the load goes up. My original rev Ninja keeps my E6400 at 45C or 50C or something very low even in Orthos or TAT load at stock 2.13ghz.

Overclocked to 3.2ghz and 1.35v (almost default) it hits +40~43C over ambient, which means 65C or more in TAT. With more mild overclockers, temperatures go down quite a bit, which suggests to me that the Ninjas are simply getting saturated.

I've seen almost identical performance with a few Rev B Ninja's on E6400 and E6600 chips also overclocked to 3.2ghz/1.35v. And yeah, the push pin mounting system is not ideal.

(these are all with Nexus 120mm or Scythe SFF12D fans...)

The Thermalright Ultra-120 didn't look like much of an improvement over the Ninja, but with the Ultra-120 Extreme or other heatsinks designed with more cooling capacity in their heatpipes, I'm thinking about upgrading as soon as they become available...

cmthomson
Posts: 1266
Joined: Sun Oct 09, 2005 8:35 am
Location: Pleasanton, CA

Post by cmthomson » Thu Mar 29, 2007 5:27 pm

continuum wrote:Ninja's seem to run out of capacity as the load goes up. My original rev Ninja keeps my E6400 at 45C or 50C or something very low even in Orthos or TAT load at stock 2.13ghz.

Overclocked to 3.2ghz and 1.35v (almost default) it hits +40~43C over ambient, which means 65C or more in TAT. With more mild overclockers, temperatures go down quite a bit, which suggests to me that the Ninjas are simply getting saturated.

I've seen almost identical performance with a few Rev B Ninja's on E6400 and E6600 chips also overclocked to 3.2ghz/1.35v.
Nope.

A Ninja (properly mounted) can easily handle over 150W, as in my old overclocked Pentium D 830 system.

No Core 2 Duo puts out anywhere near this much heat. My heavily overclocked E6600 spits out about 100W running TAT.

An Ultra-120 can outperform a Ninja (again both properly mounted) if it has a fast fan. With slow fans, the two have very similar performance. The Ninja, with its wide fin spacing, has a slight edge due to lower impedance. It also stays cleaner...


Bear in mind that CPU power consumption increases linearly with the clock rate, and with the square of the core voltage. So in general an aggressively overclocked and slightly overvolted CPU will burn as much as 2-3x as much power as stock.

Also, the thermal resistance of any particular IHS/HS/fan assembly is pretty constant. So as the CPU wattage goes up, so does its peak temperature, almost linearly.

JazzJackRabbit
Posts: 1386
Joined: Fri Jun 18, 2004 6:53 pm

Post by JazzJackRabbit » Sat Mar 31, 2007 9:23 am

Either you all guys have been extremely unlucky or you had concave cpu heatspreader or didn't mount heatsink properly by locking pins in place. I've got one Ninja B on C2D and my dad has two, all three work perfectly with temps around 30 in idle.

I have to reiterate it is almost impossible to push pins in place with motherboard mounted in the case. The motherboard just bends and the last pin which is hardest to push in place does not lock. Couple that with the double click sound which is typical when mounting 775 and which leads people to mistakenly think they've locked all pins in place and you have a problem. All three of the Ninja B's were mounted outside the case, verified that they are fully locked and they just work as good as the first revision.

Mikael
Posts: 206
Joined: Mon Dec 06, 2004 3:12 am
Location: Gothenburg, Sweden

Post by Mikael » Mon Apr 02, 2007 10:26 am

I have an E6600 running at 3GHz on 1.17V and a Ninja rev. B cooling it. I also use the supplied Scythe S-Flex on reduced voltage (~800RPM).

Idle temp is around 33C and full load on both cores using Orthos takes the temp up to 48C. Not that bad, I think.

miks2u
Patron of SPCR
Posts: 26
Joined: Sat Dec 25, 2004 10:48 am
Location: Berkeley, CA

Post by miks2u » Mon Apr 02, 2007 1:18 pm

If you want the old mounting system you can buy the Scythe Universal Retention kit SCURK1. http://www.newegg.com/Product/Product.a ... 6835185022. Newegg shows out of stock at the moment, but I think I saw it there the other day.

rpsgc
Friend of SPCR
Posts: 1630
Joined: Tue Oct 05, 2004 1:59 am
Location: Portugal

Post by rpsgc » Thu Apr 05, 2007 10:10 am

I have a E6400 @ 2,36GHz with a Ninja Plus (not Rev.B) running fanless and I have an idle temp of 30ºC (core). It's around 25/26ºC (core) running at stock speed (2,13GHz).

EDIT: At 2,36GHz it reaches 44º (core) under full load (Orthos).
Last edited by rpsgc on Fri Apr 06, 2007 1:10 am, edited 1 time in total.

Maverick9611
Posts: 15
Joined: Thu Dec 28, 2006 5:32 pm

Hmm

Post by Maverick9611 » Thu Apr 05, 2007 11:39 am

Now Im starting to wonder about mine. Ive got a E6600 running at 3ghz and get 39c idle with the stock fan running full blast and up to the low 60's when running Orthos. This is in a P180 with both of the upper chamber exauhst fans installed. Ive reapplied the pushpins mutliple times and finally followed a bolt-on tutorial online with only a 1 or 2c drop to where I am now.

Any comments?

santacruzbob
Posts: 25
Joined: Thu Apr 05, 2007 4:23 pm

Post by santacruzbob » Thu Apr 05, 2007 4:29 pm

I've got an e6300 at 425fsb (~3ghz) running a ninja rev. b and my idle temps are about 40*C while under load they reach ~58*C (ambient about 20*C).. I'm really not concerned with the temps although I do have the scythe fan running at full voltage and I'm wondering if that is normal or if I should look into alternative mounting options?

rpsgc
Friend of SPCR
Posts: 1630
Joined: Tue Oct 05, 2004 1:59 am
Location: Portugal

Post by rpsgc » Fri Apr 06, 2007 7:45 am

Update: I've got my E6400 running now @ 3,2GHz (400x8) at 1,4V in BIOS (1,36V in CPU-Z).

Idle temp (core) with the Ninja running fanless is 37ºC. Under load (Orthos) it goes up to 53ºC.

abby_normal
Posts: 3
Joined: Sun Apr 15, 2007 9:32 am

Post by abby_normal » Sun Apr 15, 2007 10:30 am

Just got a Ninja Rev B and e6600 and thought my temps seemed a bit high. After reading this thread, I tried re-mounting the Ninja with the MB outside the case to see if that made a difference. I noticed a very slight improvement, maybe 1 degree C . . . hardly night and day.

Original mounting:
Ambient temp: 24.5c
TAT core 1 temp: 48c load, 27c idle
TAT core 2 temp: 48c load, 25c idle
Intel Desktop Utilities ("IDU") processor zone temp: 55c load, 34c idle
IDU Zone 1 temp: 37c load, 35c idle
IDU Zone 2 temp: 42c load, 39c idle

After re-mounting:
Ambient temp: 24.5c
TAT core 1 temp: 48c load, 27c idle
TAT core 2 temp: 46c load, 25c idle
IDU processor zone temp: 54c load, 34c idle
IDU Zone 1 temp: 35c load, 34c idle
IDU Zone 2 temp: 41c load, 39c idle

Setup details: I used AS5 both times, with the stock Scythe fan, processor at stock speed, MB is Intel D946GZIS, case is Ultra Grid with 2x120mm fans.

Load temps measured after running both cores at 100% for 10 mins using TAT. Idle temps measured after letting the system cool for 10 mins after load test.

Note that I also measured temps in Speedfan and they were very similar to the TAT and IDU processor zone numbers.

hexen
Posts: 94
Joined: Sat Mar 10, 2007 12:34 am

Post by hexen » Mon Apr 16, 2007 2:10 am

ok so what is the workaround with this issue?
(i mean: we can all keep re-stating that it is a problem, but what about a solution?)

is the solution to this issue, using screws/springs like in the mod that butters posted in another thread?
http://www.xtremesystems.org/forums/sho ... p?t=117054

or is the solution just scrapping the revB and replacing it with something else (thermalright ultra 120)... (i bought a revB, and havent opened the box yet, so id like to know BEFORE i open it as it will be easier to return)

ghia
Posts: 80
Joined: Fri Jan 05, 2007 9:48 am
Location: Cluj-Napoca, Romania

Post by ghia » Mon Apr 16, 2007 2:58 am

I've got the Ninja on a 2.8GHz, 89W, Prescott running completely fanless in a Antec Solo case (no case fan) with just the Seasonic fan (never rumped up - I don't do games) and temps are (SpeedFan):
33 - idle
55- load
It is mounted using the "Universal Retention Kit" and AS5.

continuum
*Lifetime Patron*
Posts: 213
Joined: Tue Jul 19, 2005 10:23 pm
Location: California
Contact:

Post by continuum » Mon Apr 16, 2007 9:23 am

cmthomson wrote:A Ninja (properly mounted) can easily handle over 150W, as in my old overclocked Pentium D 830 system.

No Core 2 Duo puts out anywhere near this much heat. My heavily overclocked E6600 spits out about 100W running TAT.
I fully understand this; however when I've had multiple different P5B Deluxe boards with several different Ninjas all watching temperatures increase by +20C to +25C at load from stock clock speeds up to 3.2ghz or so, it makes me wonder. This is both with original Ninjas and with the later rev B's and various BIOS rev's starting at 0804 and later.

Temperatures at stock speeds are expected, given constant ambient temperature I've seen +20C over ambient at load as I expect; watching it go up another +20C or +25C over that (+40C to +45C over ambient) has happened every time, no matter who's installing the heatsinks.

Could just be that I've got a few batch of lower-performing Ninjas here, or insufficient case airflow, as about half of these builds are Antec P180's with very low speed chassis fans such as S-flex SFF12D's.

cmthomson
Posts: 1266
Joined: Sun Oct 09, 2005 8:35 am
Location: Pleasanton, CA

Post by cmthomson » Mon Apr 16, 2007 6:28 pm

You should certainly expect CPU temperature to rise when overclocking, regardless of other aspects of the system.

The power consumption of a CPU varies linearly with its clock speed, and with the square of its core voltage. So if you (for example) take an E6600 from its stock 2.4 GHz @ 1.35V to 3.2 GHz @ 1.55V, the power consumption will go up by 76%.

The thermal resistance of a particular CPU/IHS/TIM/HSF combination is relatively constant under various power/temperature situations. So for practical purposes, CPU temperature rise will increase linearly with power consumption.

So, to use the example above, you would expect the overclocked/overvolted CPU to have about 75% more temperature rise over ambient than stock.

continuum
*Lifetime Patron*
Posts: 213
Joined: Tue Jul 19, 2005 10:23 pm
Location: California
Contact:

Post by continuum » Mon Apr 16, 2007 8:16 pm

Agreed, but these boxes-- all of them-- have only been overclocked to 3.2ghz with no or a minimal voltage increase-- 1.35v from the stock 1.25v (?) is the highest done.

Temperatures are rising at a rate well beyond what I would expect-- another +15C over the +20C ambient (to +35C) I would expect, but +20C to +25C additional is well out of the expected range (to +40C to +45C), especially given the temperatures otherwise being reported with Ninjas on similar or essentially identical setups.

MalcolmC
Posts: 67
Joined: Thu Mar 17, 2005 2:05 pm
Location: UK

Post by MalcolmC » Wed Apr 25, 2007 5:11 am

I've got 3 Ninjas - a type A & a type B, on almost identical machines, and the type a runs (say) 10 degrees lower.

I just installed a second type B on a new machine today - it's very hard to tell if it's all clipped down properly, but you have to say that the clips are just not up to the job.

Anyway temps seemed to be way too high, so I exerted a bit of manual pressure on the top of the heatsink, and watched the temp fall by 15 degrees C.

I'm not buying another Ninja until they improve the mechanism.

hexen
Posts: 94
Joined: Sat Mar 10, 2007 12:34 am

Post by hexen » Wed Apr 25, 2007 7:29 am

MalcolmC wrote:I've got 3 Ninjas - a type A & a type B, on almost identical machines, and the type a runs (say) 10 degrees lower.

I just installed a second type B on a new machine today - it's very hard to tell if it's all clipped down properly, but you have to say that the clips are just not up to the job.

Anyway temps seemed to be way too high, so I exerted a bit of manual pressure on the top of the heatsink, and watched the temp fall by 15 degrees C.

I'm not buying another Ninja until they improve the mechanism.
but if you dont use the push pins and replace them with screws/nuts then the problem is rectified no?
http://www.xtremesystems.org/forums/sho ... p?t=117054

Delta_42
-- Vendor --
Posts: 131
Joined: Mon Mar 05, 2007 8:36 am
Location: Lancs, UK
Contact:

Post by Delta_42 » Wed Apr 25, 2007 8:45 am

hexen wrote:
but if you dont use the push pins and replace them with screws/nuts then the problem is rectified no?
http://www.xtremesystems.org/forums/sho ... p?t=117054
But why should you have to pay extra and/or bugger about with mods to get an adequate mounting mechanism? Scythe should provide one out of the box (and used to do so!).

Post Reply