Vapochill LS and Mach ST/GT test part1

Personally I would like to be able to take the side panels off with warranty intact, but I don't think Nventiv's intentions are much different. The following is a quote from the Lian Li Alu-kit instructions,

"Caution! This instruction guide is for authorised personal only, inorder to guide through the installation process of the Lian-Li kit.
Private individuals may under no circumstances take off the side panels; the cooling unit contains main circuitries which are accessible when the side panels are removed"



may under no circumstances take off the side panels sounds to me like any evidence could support a voided warranty if you don't have a Lian Li kit installed, and it was done by an Authorised person :2cents: ;)

Just cut the Vapo LS case open with one of these. Asetek guys just said that warranty is voided if case is opened, nothing about cutting it all open :D

Quote:

---

Originally posted by The Stilt
Just cut the Vapo LS case open with one of these. Asetek guys just said that warranty is voided if case is opened, nothing about cutting it all open :D

---

:ROTF: Own3d :D

http://www.knighton-tools.co.uk/acat...00_SQ_MILW.JPG

Looks like it could work :wierd:

Quote:

---

Originally posted by Ragnarok
what about maintance of the unit overtime? cleaning out dust which can collect on the condensor and degrade performance?..

---

Good point - and that is why you can remove the front panel and get access to the condensor for cleaning purposes WITHOUT loosing warranty :D

Quote:

---

Originally posted by The Stilt
Just cut the Vapo LS case open with one of these. Asetek guys just said that warranty is voided if case is opened, nothing about cutting it all open :D

---

Yeah.. i say we make those holes a "little" bigger. :D


Asetek: What is your response to this?

http://www.xtremesystems.org/forums/...threadid=33797

-11c @ 100w is worse than Vapochill PE .. or classic for that matter. Wrong numbers i hope..

Quote:

---

Originally posted by Formann


-11c @ 100w is worse than Vapochill PE .. or classic for that matter. Wrong numbers i hope..

---

A Vapochill PE would never get -11C on the CPU, it would be positive core temps at 100W.
From what I can see the "dummy CPU temp" is -11, the evap surface is much colder.

Quote:

---

Originally posted by JCviggen
A Vapochill PE would never get -11C on the CPU, it would be positive core temps at 100W.
From what I can see the "dummy CPU temp" is -11, the evap surface is much colder.

---

Of course.. somehow i tought the "Test Vehicle" was the measured evap surface temp.


LS with 190w heatload. Evap -26c .. cpu +5.3c ... Difference is 31c ..

Mk2 GT 190w heatload. Evap -36c .. cpu -14.7c ... Difference is 21c


Poor contact between evap and dummy on the LS, or just bad evap?

Quote:

---

Originally posted by Formann
Yeah.. i say we make those holes a "little" bigger. :D


Asetek: What is your response to this?

http://www.xtremesystems.org/forums/...threadid=33797

-11c @ 100w is worse than Vapochill PE .. or classic for that matter. Wrong numbers i hope..

---

This guy has to drop the crack pipe. We have no review action going on with them, why we of course not would ever try to comment on it either. If he states that we have, he should be able to post the reply from us I guess?!?

They made a WaterChill review, where they used results "borrowed" from another site. We told them to drop that.

without saying too much, these results seems a "bit" strange.... With regard to the scheme they made, it does not seem to include figures from any of our products ?!?

There have never been any reason to doubt the performance graphs and temperatures we claim. This is no different with the LS.

FYI peeps,

I just sent the reviewer an email asking him to explain in further detail his testing methodology. I'll keep you posted on the response.

From what I can tell my VapoLS follows the performance chart (+-2c because of room temp) if my estimations are correct:
prescott 3,4ghz 1,4V -> 103W
prescott 4,0ghz 1,62V -> 155W
The first days I had problems with the unit but now it works very decently.

I've got a P4E 3ghz, P4C800-E and things like superPi and Prime here + a vdroop mod.
if you want to see some specific results from my rig tell me and I'll try. maybe a machII+P4E user shows up so we can compare (my room is 23-26c warm)
P.S. my max Voltage is ~1,65V and from what I feel the Mobo won't last to long with this voltage. So choose short tests (10min)

I wouldn't put much stock in that review. The numbers are pretty far off for the LS for the entire test. I suspect that this guy has an ax to grind.

One thing that I have noticed that interesting is that the evap temp decreases by 6-8 degrees if you turn on the usb connection. It also seems to affect cpu temp. I ran small FFT's on prime at 290 fsb on a 2.8mo at 1.85v in the bios, 1.88v in mbm, and with the CC on the temp was -15C. With it off it was -18C. Sandra estimates this as 177 watts. I'm not sure how accurate that is. How do you calculate the wattage.

It should not really affect the CPU temperature, as there is no built in stress test :D

However the reason for the difference on the display, is that with the software / USB connection disabled the temperature is calculated by the processor on the ChillControl. With everything enabled, the temperature is calculated by the VapoChill Control Panel, which is more accurate (as described in the manual).

Pleas confirm if it really does affect the CPU temperature, and if so please return a feedback in our USB control panel forum, and we will have a SW engineer to look at it.

makes sense, but it did seem to make a difference. I'll check and run more tests

@pkrew

Use the following tool to calculate overclocked watttage

http://www.benchtest.com/calc.html

Thanks, according to that page it was 190 watts. Seems a little high, but not bad, -15 to -18, at 190w loaded with small FFT's. I've put in my 3.0 atm and have it running at about 280fsb. Looks like its time for a new cpu. I hate this 5/4 stuff.

Quote:

---

Originally posted by phobix
@pkrew

Use the following tool to calculate overclocked watttage

http://www.benchtest.com/calc.html

---

Whoohoooo. When using a barton at 2.2V and 3Gig you'll have a max heat of 200W :D

My config with 2,87Gig and 1,97V has 152Watts ;) (mach1 must have been at it's max ;) )

Quote:

---

Originally posted by pkrew
Thanks, according to that page it was 190 watts. Seems a little high, but not bad, -15 to -18, at 190w loaded with small FFT's. I've put in my 3.0 atm and have it running at about 280fsb. Looks like its time for a new cpu. I hate this 5/4 stuff.

---

Don't forget that P4's have TCC (thermal throtling control) which means you will never probably reach the full 190w. Some people have noted peculiar reaction with TCC under sub-ambient cooling situations with P4'S... May want to check the forums over at asetek for more info if you are interested..

I'll check it out. Thanks phobix

Quote:

---

Originally posted by SAE
Whoohoooo. When using a barton at 2.2V and 3Gig you'll have a max heat of 200W :D

My config with 2,87Gig and 1,97V has 152Watts ;) (mach1 must have been at it's max ;) )

---

Interesting. Dont know much about this, but if that's accurate then Asetek may have been closer than stated earlier.

Yeah. The Bartons are getting really hot :D

It's art ocing them so high and stable with high fsb clocks ;)

erm can someone explain how these figure are obtained :confused:

its breaking my normal knowledge of physics here

i.e. if my processor is 68.3w @ 1.65vcore
then i have a load resistance of about 0.04ohms

on this basis that resistance is constant when scaled up
to 2.1v we are using 110watts

surly the frequency cannot change the underlaying basic
properties of a transistor circuit?

is the power rating of a chip classed as 100% on or rated as
pulsed ? seems strange if its pulsed considering that the chip
takes serverel approches to design and power requirements.

/edit and think if it were pulse what's the diff in 1000mhz or
2000mhz if the on and off pulse are the same duruation on
ultimate time a slower speed will have a longer pulse a faster
speed a shorter pulse but if its all 01010101 then the chip is
surley using the same 50/50 consumption.

anyone care to correct me / show me the light because its confused me now.

I am far from a physics major... But thermal dissipation does not scale linearly, this is where Sandra's calc is flawed. If we were to increase voltage and stay at the same clockspeed then yes. But too my limited knowledge clockspeed plays a major role, perhaps someone here can elaborate more.

In the simplest possible terms a CMOS chips has two main components to the power dissipation
1. Static power dissipation due to leakage currents (mainly via the substrate). This is like a resistor, it scales with voltage.

2. Dynamic power dissipation due to the capacitance of lines, etc etc. This scales with frequency and voltage.

Since we dont know the breakdown the between dynamic and static power dissipation of these chips its hard to get a a true figure when scaled up. You could run the CPU with the clock disabled (so long as it doesnt use dynamic logic) and measure the power, this will approximate the static power consumption.


Anyway

P = (Ps + Pd * Fclock / Fclockref) * (Vcore / Vcoreref) ^ 2

where
Ps = static power consumption at reference core voltage Vcoreref
Pd = Dynamic power consumption at reference core voltage Vcoreref and reference CPU clock Fclockref.

Of course then there is significant power drawn for the I/O pins which will scale with teh I/O speed (FSB) and I/O voltages.


Just use the following for a rough approximation
P = (P0 * Fclock / Fclockref) * (Vcore / Vcoreref) ^ 2

where P0 = total power consumption at stocks speed and voltage