(Yet another) P182 build

Suosaaski · Post by **Suosaaski** » Wed Nov 07, 2007 12:43 pm

Some testing today, temps are stable at 69/69/73/77 when running two instances (to utilize all four cores) of SP2004 Orthos, small ffts cpu stress test.

http://img75.imageshack.us/img75/682/q6600templx6.jpg

Again, the fans are running at the bare minimum they start with. On the other hand, the temps go down up to 7 degrees / core if I open the case, so the same problem still. Anyways, it is stable at current settings.

Need some more testing on fan speeds / configuration.

Edit: For comparison, idle temps are 37/37/34/42

Suosaaski · Post by **Suosaaski** » Thu Nov 08, 2007 1:23 am

I'm starting to think that partly this warming when case is closed is due to that x1950pro generating quite a bit of heat which stays in the case...

Justchill · Post by **Justchill** » Fri Nov 09, 2007 6:42 am

try using another tool to measure temperature and compare. Preferred is your mobo manufacturers tool..

Suosaaski · Post by **Suosaaski** » Sun Nov 11, 2007 11:50 am

Justchill wrote:try using another tool to measure temperature and compare. Preferred is your mobo manufacturers tool..

I do trust the values coretemp says. It is the other tools I don't completely trust, since they might use some other method than DTS for temperature detection.

Speedfan reports the same temperatures as coretemp, as long as I set +15C offset for the temperatures. As far as I know, Speedfan uses the same method for temperature detection as CoreTemp, but it uses the incorrect Tjunction of 85C instead of 100C for these quadcore chips. That ofcourse explains the 15 degree difference in temperature.

There is no way the default settings are anywhere near correct for Speedfan, since at idle it reports 20/20/16/23 degrees (currently) for the core temperatures. That would be around room temperature (ambient 20-21 degrees), even less for one core. Setting the 15 degree offset (if I am correct about the tjunction being wrong) will report much more reasonable values.

I will try full load with Speedfan, I'll post my results for that too after testing.

PS. Mobo manufacturers tool doesn't seem to work under current operating system (Windows Server 2003). No idea why.

Suosaaski · Post by **Suosaaski** » Sun Nov 11, 2007 1:53 pm

2 hours running 2 instances of SP2004 Orthos:

Speedfan and Coretemp both report the same values (after the +15C offset for Speedfans settings). Temperatures are down by 5 degrees compared to previous. Current fanspeeds are:

around 1250rpm for the Scythe fan on the Ninja
around 820-830rpm for the 3 Nexus fans on the P182 default fan locations.

So, I am not completely sure about the fan speeds that were before, if I remember correctly the case fans were around 50rpm less and the fan on the Ninja was around 900rpm, so it would very much seem like an airflow problem, but not for the case as much as for the Ninja. Anyways, I'll try a Scythe 120mm PWM controlled fan to allow for rpm adjustment with the Zalman fan controller in the next couple of days, remains to be seen how much the temps change with that fan at around 950rpm or so.

PS. 96W idle, 180W under load with 2 * SP2004 Orthos running small FFT, reported by the Zalman. Kind of offtopic, and kind of not

Airshark · Post by **Airshark** » Tue Nov 13, 2007 5:05 pm

Suosaaski wrote:I'm starting to think that partly this warming when case is closed is due to that x1950pro generating quite a bit of heat which stays in the case...

I found it necessary to add a slow fan (Noctua 120mm) to the front of my intake, even after removing the intake covers entirely and cutting away the metal. This was because I'm running an 8800GTX with a TR HR-03 Plus, and that card vents a lot of heat into the case, and that was warming up the CPU. Your 7950 should be at least a little cooler, but try the fan - even if you decide not to go with it (for noise reasons), it'll let you know whether airflow is your problem, or your HSF isn't seated properly, or whatever.

Also, did you do the "penny trick" with your Ninja? I know the retention clips generally don't put enough pressure on the IHS to get good thermal transfer.

Suosaaski · Post by **Suosaaski** » Tue Nov 13, 2007 9:17 pm

Airshark wrote: I found it necessary to add a slow fan (Noctua 120mm) to the front of my intake, even after removing the intake covers entirely and cutting away the metal. This was because I'm running an 8800GTX with a TR HR-03 Plus, and that card vents a lot of heat into the case, and that was warming up the CPU. Your 7950 should be at least a little cooler, but try the fan - even if you decide not to go with it (for noise reasons), it'll let you know whether airflow is your problem, or your HSF isn't seated properly, or whatever.

Also, did you do the "penny trick" with your Ninja? I know the retention clips generally don't put enough pressure on the IHS to get good thermal transfer.

Well, running the case fans and the cpu fan at higher rpm results in notably lower temps, so I would say that airflow is the reason.

The Ninja also heats up considerably, so I would guess that the contact is fine now. Although I don't know what "penny trick" is?

I should make a vent that helps the gfx card exhaust the hot air directly out of the case and see if that makes any difference. The system is stable so I am not too concerned, but I'd like it better if it ran cooler.

Suosaaski · Post by **Suosaaski** » Sun Nov 18, 2007 11:15 am

Some new pics!

As you might see, I got a bit lazy with the cables, here's a sample:

I also opted for a Arctic Cooling PWM controlled fan, just because the Scythe one was out of stock. I have no idea which one would have been better, or if there even is any notable difference between them. At least the AC is with fluid bearing.

A lot of colourful leds this new motherboard has...

Here's a pic of the Zalman fan controller. Also this pic shows (thanks to those leds, I might add

) that the front door after modification at least seems better conserning the airflow.

And finally the basic overview pic.

Suosaaski · Post by **Suosaaski** » Sun Dec 09, 2007 2:03 pm

Today I decided to check why the temps on the cores are so uneven, so first to check the contact for the cooler:

Doesn't look too good. I decided to lap the CPU.

After lapping under full load cores 0 and 1 are pretty much at about the same temperature, but core 2 is now 9 degrees cooler and core 3 a whopping 12 degrees cooler than before. So the temps are a lot more even than they used to be. I would say it is a HUGE difference compared to what it was.

Starting to be a little more satisfied with this Q6600.

Vahan · Post by **Vahan** » Thu Dec 13, 2007 7:06 am

very nice. waiting for my p182 to arrive so i dont have to be jealous any more.
just out of curiosity, how fast can you run that cpu fan before it becomes audible at 2ft with case closed?

Suosaaski · Post by **Suosaaski** » Thu Dec 13, 2007 11:19 pm

Vahan wrote:very nice. waiting for my p182 to arrive so i dont have to be jealous any more.
just out of curiosity, how fast can you run that cpu fan before it becomes audible at 2ft with case closed?

Around 800rpm, though it is pretty hard to tell since my HDD is not as quiet as I'd like, it is noisier than the Nexus fans. But I'll live with it.

As a note, I changed that AC fan to the lower chamber and I run it at below 700rpm, and the CPU now has that Nexus at 800rpm.

Suosaaski · Post by **Suosaaski** » Mon Dec 17, 2007 8:40 am

Got some new toys today:

Another 2 * 1GB set of OCZ Platinum DDR2-800

And a Radeon HD 3850

The gfx card is fairly quiet, so at least for now I am not considering changing the cooler on that.

Suosaaski · Post by **Suosaaski** » Sun Jan 06, 2008 9:58 am

Rebuild, some new parts, some old parts:

I like this setup more than the Q6600 setup. It's cooler, quieter... and... umm... "cooler"

AMD Overdrive seems a bit unstable for overclocking, sometimes freezes the system when changing settings, sometimes doesn't. When it doesn't, things keep running stable with the new settings. Or I might be missing something, I'll keep trying things out.

thejamppa · Post by **thejamppa** » Sun Jan 06, 2008 3:10 pm

oh, you became fellow crossfire HD 3850 user. Heh, those stock coolers are damn quiet, although it makes them run bit hot. But I am sure you won't regret getting them. Also, I am sure that is one of the first unless the first Phenom build in SPCR.

Suosaaski · Post by **Suosaaski** » Mon Jan 07, 2008 3:12 am

thejamppa wrote:oh, you became fellow crossfire HD 3850 user. Heh, those stock coolers are damn quiet, although it makes them run bit hot. But I am sure you won't regret getting them. Also, I am sure that is one of the first unless the first Phenom build in SPCR.

Yeah, I like the whole setup now

The temps as reported by AMD Overdrive are around 28 idle (needs to cool down a while to get there after running 3D applications though, thanks to those radeons

) and a bit over 50 degrees under full load with Prime after a couple of hours. The Ninja does not seem to get too hot either, so I do believe it really is a lot cooler than the Q6600 was. This makes me really consider removing the fan on the Ninja. On the other hand it really does not make that much of a difference to the overall noise level and I have been considering overclocking the Phenom, remains to be seen what is the final config I end up with.

Nevertheless, I am very satisfied with this setup.

Suosaaski · Post by **Suosaaski** » Mon Jan 07, 2008 1:06 pm

Test run:

Phenom @ 2,6GHz
Radeons @ 750 (GPU) / 999 (MEM)

http://img258.imageshack.us/img258/4941 ... nomup0.jpg

Needs rather high voltage to run stable. at least with AMD Overdrive. On the other hand AOD seems a bit unstable, but it is a work in progress, will hopefully get better over time. The potential is there.

Jay_S · Post by **Jay_S** » Mon Jan 07, 2008 9:53 pm

Suosaaski wrote:Some testing today, temps are stable at 69/69/73/77 ...

I know you're using a Phenom now, so this won't mean much. But I'm starting to understand Core Temp and Tjunction a little differently than I used to. I no longer think it is a measure of temperature in absolute terms. The delta information is certainly useful, but just doesn't tell us anything about actual temperature. Delta is just the "distance" away from shutdown.

Note that your Tjunction as reported by Core Temp is 100C. Core Temp reads the DTS as a delta or difference from Tjunction. At Tjunction your CPU shuts down. At Tj-5C, your CPU starts throttling. Your hottest core (in the above quote) is still 23C away from Tj. I'd consider this borderline-safe. After lapping, you dropped a 9 and 12 degrees. That's terrific by the way! But you might still get freaked by the resulting 65C. But don't think of it as 65C in absolute terms - think of it as 35C away from shutdown. That's way safe in my opinion.

Things said by the author of Core Temp makes me suspect that he's just guessing at Intel's Tjunction definition. It doesn't really matter where Intel defines Tj though - what matters is DELTA reported by the DTS inside your CPU. For example, Intel could have defined Tjunction at 45C - Core Temp might report a nice looking temp of 44C, but your delta is 1C, you're throttling and you're about to shut down!

The "temperature" reported by Core Temp is meaningless. In the options, switch to "display delta to Tjunction" mode or whatever it's called. This shows you how close you are to Tjuntion. Keep this number above 20C.

Jay

Suosaaski · Post by **Suosaaski** » Tue Jan 08, 2008 2:01 am

Jay_S wrote: I know you're using a Phenom now, so this won't mean much. But I'm starting to understand Core Temp and Tjunction a little differently than I used to. I no longer think it is a measure of temperature in absolute terms. The delta information is certainly useful, but just doesn't tell us anything about actual temperature. Delta is just the "distance" away from shutdown.

Note that your Tjunction as reported by Core Temp is 100C. Core Temp reads the DTS as a delta or difference from Tjunction. At Tjunction your CPU shuts down. At Tj-5C, your CPU starts throttling. Your hottest core (in the above quote) is still 23C away from Tj. I'd consider this borderline-safe. After lapping, you dropped a 9 and 12 degrees. That's terrific by the way! But you might still get freaked by the resulting 65C. But don't think of it as 65C in absolute terms - think of it as 35C away from shutdown. That's way safe in my opinion.

Things said by the author of Core Temp makes me suspect that he's just guessing at Intel's Tjunction definition. It doesn't really matter where Intel defines Tj though - what matters is DELTA reported by the DTS inside your CPU. For example, Intel could have defined Tjunction at 45C - Core Temp might report a nice looking temp of 44C, but your delta is 1C, you're throttling and you're about to shut down!

The "temperature" reported by Core Temp is meaningless. In the options, switch to "display delta to Tjunction" mode or whatever it's called. This shows you how close you are to Tjuntion. Keep this number above 20C.

Jay

Hi!

Thanks for your input. You should notice, though, that there are two different settings for coretemp to show the temperatures. Mine were reported as "normal" temperature, not delta to tjunction.

There you can see the same system measured with coretemp with two different settings. Note that the pics were taken a few seconds apart which explains why the temp for core #0 doesn't add up.

Also if it were 45 degrees, and Coretemp would show a nice looking temp of 44 degrees, that would mean that Coretemp would either have to know the correct tjunction of 45 degrees, or with a tjunction of 100 degrees, it would in fact show 99 degrees for the core (where delta to tjunction is -1 and is tjunction max is defined at 100 degrees).

Coretemp also is able to report the temperatures as delta to tjunction, where it will not matter what the tjunction max is defined as, delta to tjunction stays the same.

Note that in version 0.96 the measurements are defined more correctly, for example tjunction max is now a more accurate term than the old tjunction was.

Jay_S · Post by **Jay_S** » Tue Jan 08, 2008 6:59 am

Suosaaski wrote:You should notice, though, that there are two different settings for coretemp to show the temperatures.

Exactly!

Suosaaski wrote:Also if it were 45 degrees, and Coretemp would show a nice looking temp of 44 degrees, that would mean that Coretemp would either have to know the correct tjunction of 45 degrees, or with a tjunction of 100 degrees, it would in fact show 99 degrees for the core (where delta to tjunction is -1 and is tjunction max is defined at 100 degrees).

Yes! Except that Core Temp doesn't "know" anything. As far as I can tell, based on reading countless forum posts, Intel has never released the Tjunction specs for their desktop processors - only their mobile processors. Because of this, we can only guess at Tjunction temps. Likewise, Core Temp's author is just guessing at Intel's Tj specs.

Suosaaski wrote:Coretemp also is able to report the temperatures as delta to tjunction, where it will not matter what the tjunction max is defined as, delta to tjunction stays the same.

This was precisely my point! Delta is the useful data reported by Core Temp - not the resulting temperature.

Suosaaski wrote:Note that in version 0.96 the measurements are defined more correctly, for example tjunction max is now a more accurate term than the old tjunction was.

I still argue that "correctness" and "accuracy" do not apply to Core Temp 0.96's reported temperatures, because of the author's arbitrary decision to pick Tj=100C vs Tj=85C. (Apparently, do to complaints he is going to switch back to Tj=85C for L2 stepping E2xx series processors in his next revision - how's that for objectivity!)

Nonetheless, Core Temp is still accurate in the sense that Delta is always accurate. Delta is the important stuff.

In my opinion (of course!),
Jay

Suosaaski · Post by **Suosaaski** » Tue Jan 08, 2008 8:41 am

Jay_S wrote: I still argue that "correctness" and "accuracy" do not apply to Core Temp 0.96's reported temperatures, because of the author's arbitrary decision to pick Tj=100C vs Tj=85C. (Apparently, do to complaints he is going to switch back to Tj=85C for L2 stepping E2xx series processors in his next revision - how's that for objectivity!)

Nonetheless, Core Temp is still accurate in the sense that Delta is always accurate. Delta is the important stuff.

In my opinion (of course!),
Jay

By accuracy I meant the terms used by coretemp, not the exact temperatures reported by it.

Yes, 100C Tj. Max is just a lucky guess, and the most usefull data is either the reported temp compared to that "guess", or with setting delta to tjunction, that particular value.

mnekic · Post by **mnekic** » Tue Jan 08, 2008 9:10 pm

i cant wait to do that filter door mod, it really adds to the great looks of the case.

Eagle156 · Post by **Eagle156** » Wed Jan 09, 2008 1:49 pm

How did you get the plastic sliding thing between the chambers to seal up so good? Mine will only only go about halfway, and looking at the top of the lower chamber there doesn't seem to be any way to slide it further?

Suosaaski · Post by **Suosaaski** » Wed Jan 09, 2008 9:16 pm

Eagle156 wrote:How did you get the plastic sliding thing between the chambers to seal up so good? Mine will only only go about halfway, and looking at the top of the lower chamber there doesn't seem to be any way to slide it further?

You mean like this? http://img77.imageshack.us/img77/1456/cablesuz6.jpg

That didn't require anything special. Did you open both the screws and start sliding? How many cables do you have going through the holes?

Eagle156 · Post by **Eagle156** » Thu Jan 10, 2008 11:59 am

Yes, I only have a single SATA cable going through it. It doesn't really matter though because it looks like the hole in the panel doesn't allow for the tabs on the bottom of the plastic assembly to slide any further. Maybe mine is different from yours?

AzNightmare · Post by **AzNightmare** » Fri Jan 11, 2008 1:10 pm

Nice build. I have the P182 as well with a Scythe.
I also do have that ZM-MFC2, which helped manage cabling because
it came with 3pin extension cables + a Y connector to plug 2 fans into
one controller.

But the actual product was so disappointing because I have my P182 on the floor, and when I look at the LCD screen, all i see is 888 888 888 etc
cause the viewing angle is so bad.

what is your opinions on the ZM-MFC2? I totally regret buying it lol

Eagle156 · Post by **Eagle156** » Fri Jan 11, 2008 3:01 pm

Eagle156 wrote:Yes, I only have a single SATA cable going through it. It doesn't really matter though because it looks like the hole in the panel doesn't allow for the tabs on the bottom of the plastic assembly to slide any further. Maybe mine is different from yours?

Well I fixed it. That little plastic ring by there, wasn't put on very thoughtfully. I moved it some and now the thing can slide all the way across.

Suosaaski · Post by **Suosaaski** » Fri Jan 11, 2008 10:46 pm

AzNightmare wrote:
what is your opinions on the ZM-MFC2? I totally regret buying it lol

The viewing angle is bad, but it still allows easy fan control from outside of the case, and it is a nicer looking solution than using a bunch of fanmates etc. Apart from the viewing angle problem I like it. Then again, now that is adjusted the way I like it, the viewing angle doesn't really matter either. What would be nice, though, is if it were able to adjust fan speed according to temperatures.