*Reference* - Modern PC Power Requirements

[Mr Fox]

I managed to beg, borrow, or creatively acquire numerous Video Card Combinations.

Two Testbed Configurations were Assembled

These testbed systems were ran for Five Iterations Of 3DMarks 06 with TAT and HD Tach Running in background.
Power Consumption was tested, and the Peak Recorded Value was Used as Max. Value.
Operating System Used : Vista Ultimate 64

Ambient : 72 Deg F

On HSPC Tech Station:

http://www.xoxide.com/hspc-tec...ation-large-black.html

I used the Following Power Quality Analyzer:
Fluke Model 434
Three-Phase Power Quality Analyzer
NIST Cal (06/25/2007)

Fluke 400 Series


X6800 2.93GHz @ 3.6 Ghz
Asus P5N32-SLI Deluxe
Mushkin 2x2 Gig HP2-6400 DDR2
2 CD/DVD-RW
2 - 74 GB Raptors
SILVERSTONE ST85F ATX12V / EPS12V 850W Power Supply

Cards, And Configurations.

N7600GT
N7900GS
N7950GT
AX1800GTO
AX1950Pro
AX1950Pro-C/F
AX1650XT
AX1900GT
N7900GT
AX1650XT-C/F
AX1900XTX-C/F
N7600GT-SLI
AX1900XT
AX1950XTX-C/F
AX1900XT-C/F
AX1950XTX
AX1900XTX
N7900GTX
N7950GX2
N7900GS-SLI
N7900GT-SLI
N7950GT-SLI
N8800GTS
N8800GTX
N7900GTX-SLI
N7950GX2-SLI
N8800GTS-SLI
N8800GTX-SLI

Open the Image to Large, and it is Clear.

Core 2 Duo Testing Results






QX6850 Kentsfield 3.0Ghz @ 3.6 Ghz

Asus P5N32-SLI Deluxe
Mushkin 2x2 Gig HP2-6400 DDR2
2 CD/DVD-RW
2 - 74 GB Raptors
SILVERSTONE ST85F ATX12V / EPS12V 850W Power Supply


Cards, And Configurations.


N7600GT
N7900GS
N7950GT
AX1800GTO
AX1950Pro
AX1950Pro-C/F
AX1650XT
AX1900GT
N7900GT
AX1650XT-C/F
AX1900XTX-C/F
N7600GT-SLI
AX1900XT
AX1950XTX-C/F
AX1900XT-C/F
AX1950XTX
AX1900XTX
N7900GTX
N7950GX2
N7900GS-SLI
N7900GT-SLI
N7950GT-SLI
N8800GTS
N8800GTX
N7900GTX-SLI
N7950GX2-SLI
N8800GTS-SLI
N8800GTX-SLI

Open the Image to Large, and it is Clear.

Core 2 Quad Testing Results


From the Max. Value I took that, and added a 33% Safety Factor (MAX*.33=Recommended Min. PSU)

The Hard Data does not Lie....

The Only Vid. Cards of significance that I was not able to have were ATI Radeon HD 2900XT 512MB but this pretty much is quite Conclusive that you don't need your own sub-station to power your PC.


Thanks to My Friends at Mushkin, ASUS, and Silverstone !

Thanks Everyone at Ingram that made some stuff possible.

Biggest thanks to Everyone here at AT Forums... You drove me to get the data..

Thanks !!

John,

Jonny Guru,

Galvanized Yankee,

Jedi Yoda,







Edited Data, and Methodology.

 

[MrOblivious]

Two questions:

What is the Best PSU column? Edit: NVM this one looks like your 133% column. Odd label though.

At what temperature was this done?

 

[Mr Fox]

Originally posted by: MrOblivious
Two questions:

What is the Best PSU column? Edit: NVM this one looks like your 133% column. Odd label though.

At what temperature was this done?

Best is Minimum Recommended. The Spreadsheet Did not convert to .JPG well


Ambient : 72 Deg F

On HSPC Tech Station:

http://www.xoxide.com/hspc-tec...ation-large-black.html

 

[MrOblivious]

Well people will still need to take into consideration derating curves for their power supply at relevant operating temperatures (as opposed to ambient outside a case) and transient loads/response when they size a unit.

 

[Mr Fox]

Originally posted by: MrOblivious
Well people will still need to take into consideration derating curves for their power supply at relevant operating temperatures (as opposed to ambient outside a case) and transient loads/response when they size a unit.

You must understand that the Max. Value represents doing things to a PC that it will not ever see in a normal in-service situation.
The Fluke 434 is capable of data logging in many different modes, This Max Value is to establish the highest usage value that you can squeeze out.

As far as temperature is concerned, we were trying to establish Maximum Usage and were not testing the PSU for efficiency or attempting to do anything outside of gathering data on Power Usage.

Common Engineering practice is to study the need, and then establish a safety factor that you are never going to exceed. If you purchase a Quality PSU of that rating, you will not have lack of power issues.

The de-rating Curves that you are referring to is going to be different with each vendors PSU based upon the quality of the capacitors in their build Bill Of Materials. Those are Variables that are out of the scope of this study. Both good capacitors, and bad capacitors have it, and that is what separates the Yayo from the Fecal Matter.

 

[MrOblivious]

I am not bagging on you as the data is interesting and I know it took a lot of work but I think you might be overextending what you can do with it for various reasons.

Originally posted by: Mr Fox

Originally posted by: MrOblivious
Well people will still need to take into consideration derating curves for their power supply at relevant operating temperatures (as opposed to ambient outside a case) and transient loads/response when they size a unit.

You must understand that the Max. Value represents doing things to a PC that it will not ever see in a normal in-service situation.
The Fluke 434 is capable of data logging in many different modes, This Max Value is to establish the highest usage value that you can squeeze out.

As far as temperature is concerned, we were trying to establish Maximum Usage and were not testing the PSU for efficiency or attempting to do anything outside of gathering data on Power Usage.

Common Engineering practice is to study the need, and then establish a safety factor that you are never going to exceed. If you purchase a Quality PSU of that rating, you will not have lack of power issues.

The de-rating Curves that you are referring to is going to be different with each vendors PSU based upon the quality of the capacitors in their build Bill Of Materials. Those are Variables that are out of the scope of this study. Both good capacitors, and bad capacitors have it, and that is what separates the Yayo from the Fecal Matter.

That is all well and good but the power meters won't generally catch transient load values. The peaks you refer to are longer duration.

I mention derating curves because telling someone a 400w power supply is sufficient at RT is not the same as a power supply being sufficient at operating temperature and when you pull a value like that out of a hat such as:

From the Max. Value I took that, and added a 33% Safety Factor (MAX*.33=Recommended Min. PSU)

You really don't know if it will be sufficent and you don't know if you will exceed that safety factor as the units capacity is lower than you are predicting from RT. The point being....that 33% number is just not based on anything concrete.

And yes I know the derating curve is different for each vendor which makes a number such as 33% nebulous and yes I understand how to size a unit. This isn't my first rodeo.

 

[Mr Fox]

Originally posted by: MrOblivious
I am not bagging on you as the data is interesting and I know it took a lot of work but I think you might be overextending what you can do with it for various reasons.

Originally posted by: Mr Fox

Originally posted by: MrOblivious
Well people will still need to take into consideration derating curves for their power supply at relevant operating temperatures (as opposed to ambient outside a case) and transient loads/response when they size a unit.

You must understand that the Max. Value represents doing things to a PC that it will not ever see in a normal in-service situation.
The Fluke 434 is capable of data logging in many different modes, This Max Value is to establish the highest usage value that you can squeeze out.

As far as temperature is concerned, we were trying to establish Maximum Usage and were not testing the PSU for efficiency or attempting to do anything outside of gathering data on Power Usage.

Common Engineering practice is to study the need, and then establish a safety factor that you are never going to exceed. If you purchase a Quality PSU of that rating, you will not have lack of power issues.

The de-rating Curves that you are referring to is going to be different with each vendors PSU based upon the quality of the capacitors in their build Bill Of Materials. Those are Variables that are out of the scope of this study. Both good capacitors, and bad capacitors have it, and that is what separates the Yayo from the Fecal Matter.

That is all well and good but the power meters won't generally catch transient load values. The peaks you refer to are longer duration.

Fluke 434 is a very expensive piece of equipment ($4400.00) that is used industrially, and picks up peaks and transients and can data log, and do SPC etc. etc... etc... We use it to assure CNC and Robotic Computers are getting proper clean power, and also to study power savings incentives, preventive maintenance,etc

I mention derating curves because telling someone a 400w power supply is sufficient at RT is not the same as a power supply being sufficient at operating temperature and when you pull a value like that out of a hat such as:

Quality PSU's are rated at 50c ... I cannot control or dictate what some one does with the data, but if you buy based upon my study there is a good safety factor there.

From the Max. Value I took that, and added a 33% Safety Factor (MAX*.33=Recommended Min. PSU)

You really don't know if it will be sufficent and you don't know if you will exceed that safety factor as the units capacity is lower than you are predicting from RT. The point being....that 33% number is just not based on anything concrete.

And yes I know the derating curve is different for each vendor which makes a number such as 33% nebulous and yes I understand how to size a unit. This isn't my first rodeo.

The 33% factor takes Max. Observed Peak Value and factors in the DeRating Curve for RubyCon PSU Capacitor Technical Data that is located here : RubyCon PSU Capacitors Design Data
RubyCon makes the best Capacitors.
All of the Quality Japanese Capacitors are basically the same Per IEEE standards.
So there is nothing "Nebulous" in my Data or my Recommendations.

 

[MrOblivious]

Liek I sadi I wasn't trying to bag on you but you are overestating your postion here.


"Fluke 434 is a very expensive piece of equipment ($4400.00) that is used industrially, and picks up peaks and transients and can data log, and do SPC etc. etc... etc... We use it to assure CNC and Robotic Computers are getting proper clean power, and also to study power savings incentives, preventive maintenance,etc "

That's nice and all but seriously going from experience I'll tell you transeint loads of 10ms or less are generally missed even by really expensive brand name units.

"Quality PSU's are rated at 50c ... I cannot control or dictate what some one does with the data, but if you buy based upon my study there is a good safety factor there. "

Then why does Seasonic rate at 40c, Zippy/Emacs at 45c?

"The 33% factor takes Max. Observed Peak Value and factors in the DeRating Curve for RubyCon PSU Capacitor Technical Data that is located here : RubyCon PSU Capacitors Design Data
RubyCon makes the best Capacitors.
All of the Quality Japanese Capacitors are basically the same Per IEEE standards.
So there is nothing "Nebulous" in my Data or my Recommendations. "

Precisely the number comes from a shot in the dark based on on subset of data not representative of the whole.

It's not my stuff on the line so really if people want to cut things too close and I probably should have known that after jon tried to explain it nicely before that it would fall on deaf ears again.

Oh well. I tried politely pointing out the problems. Caryy on.

 

[Mr Fox]

Originally posted by: MrOblivious

Liek I sadi I wasn't trying to bag on you but you are overestating your postion here.


"Fluke 434 is a very expensive piece of equipment ($4400.00) that is used industrially, and picks up peaks and transients and can data log, and do SPC etc. etc... etc... We use it to assure CNC and Robotic Computers are getting proper clean power, and also to study power savings incentives, preventive maintenance,etc "

That's nice and all but seriously going from experience I'll tell you transeint loads of 10ms or less are generally missed even by really expensive brand name units.

"Quality PSU's are rated at 50c ... I cannot control or dictate what some one does with the data, but if you buy based upon my study there is a good safety factor there. "

Then why does Seasonic rate at 40c, Zippy/Emacs at 45c?

"The 33% factor takes Max. Observed Peak Value and factors in the DeRating Curve for RubyCon PSU Capacitor Technical Data that is located here : RubyCon PSU Capacitors Design Data
RubyCon makes the best Capacitors.
All of the Quality Japanese Capacitors are basically the same Per IEEE standards.
So there is nothing "Nebulous" in my Data or my Recommendations. "

Precisely the number comes from a shot in the dark based on on subset of data not representative of the whole.

It's not my stuff on the line so really if people want to cut things too close and I probably should have known that after jon tried to explain it nicely before that it would fall on deaf ears again.

Oh well. I tried politely pointing out the problems. Caryy on.

The thing that does not seem to be hitting home here is that the Peak Readings were coming in a condition, that no home user PC would ever encounter for anything other than nano seconds. Your Mean Distribution(Average)would be much lower. But that was not the intent of the study....

The Design of the Experiment was to establish Max. Peak Usage per each given Configuration. This is the Magic Number that you need to be able to Properly Size a PSU.

I cannot tell you why Zippy/Emac, or Seasonic have chosen the temperature that they have.
But I can tell you that if you use 40C as a standard then a Quality PSU has virtually no de-rate based upon the RubyCon technical data. I'm sure that it was based upon their Bill Of Material, and the Technical Data that guides the Engineers for whatever Caps.

If you take the time to review the ATX 2.2 Design Guidelines that are referenced in My
PSU Resource Thread you would learn that my Safety Factor is right on the money and conservative but not to the point of wasted value.....

 

[JEDIYoda]

All good information!!

 

[MrOblivious]

Originally posted by: Mr Fox

Originally posted by: MrOblivious

Liek I sadi I wasn't trying to bag on you but you are overestating your postion here.


"Fluke 434 is a very expensive piece of equipment ($4400.00) that is used industrially, and picks up peaks and transients and can data log, and do SPC etc. etc... etc... We use it to assure CNC and Robotic Computers are getting proper clean power, and also to study power savings incentives, preventive maintenance,etc "

That's nice and all but seriously going from experience I'll tell you transeint loads of 10ms or less are generally missed even by really expensive brand name units.

"Quality PSU's are rated at 50c ... I cannot control or dictate what some one does with the data, but if you buy based upon my study there is a good safety factor there. "

Then why does Seasonic rate at 40c, Zippy/Emacs at 45c?

"The 33% factor takes Max. Observed Peak Value and factors in the DeRating Curve for RubyCon PSU Capacitor Technical Data that is located here : RubyCon PSU Capacitors Design Data
RubyCon makes the best Capacitors.
All of the Quality Japanese Capacitors are basically the same Per IEEE standards.
So there is nothing "Nebulous" in my Data or my Recommendations. "

Precisely the number comes from a shot in the dark based on on subset of data not representative of the whole.

It's not my stuff on the line so really if people want to cut things too close and I probably should have known that after jon tried to explain it nicely before that it would fall on deaf ears again.

Oh well. I tried politely pointing out the problems. Caryy on.

The thing that does not seem to be hitting home here is that the Peak Readings were coming in a condition, that no home user PC would ever encounter for anything other than nano seconds.

Oh I understand what you are trying to imply...I just respectfully disagree with how you are applying it and noticed some things that are absent.

Your Mean Distribution(Average)would be much lower. But that was not the intent of the study....

The Design of the Experiment was to establish Max. Peak Usage per each given Configuration. This is the Magic Number that you need to be able to Properly Size a PSU.

No this is the magic number you have decided is correct that doesn't take into account a number of variables. Like I said had you left it at "this is what I found" it would be fine.

In all honesty and brevity I don't really ever want my power supply working above 75% too often becasue in general the DC output in many models really starts to suffer after this point. Nothing like trying to stretch a platform a bit too far.

I cannot tell you why Zippy/Emac, or Seasonic have chosen the temperature that they have.

But they are quality manufacturers..........which would be my point. Quality manufacturers don't necessarily rate at 50c as you suppose/are stating.

But I can tell you that if you use 40C as a standard then a Quality PSU has virtually no de-rate based upon the RubyCon technical data. I'm sure that it was based upon their Bill Of Material, and the Technical Data that guides the Engineers for whatever Caps.

And I can tell you then that you are wrong. They do indeed still derate above 40c which would mean that the data from Rubycon (who is as you state the best) being used is not representative of the average (which would not be the best) but of that of an example of an outlier. IMO it would make more sense to take the worst example you can find not the best to start with because the worst case scenario is the problem not the best case scenario.

If you take the time to review the ATX 2.2 Design Guidelines that are referenced in My
PSU Resource Thread you would learn that my Safety Factor is right on the money and conservative but not to the point of wasted value.....

I did, and none of it says anything about your "safety factor".

 

[ryboto]

I can't take the poll, my power supply is only 120W....

 

[Mr Fox]

Originally posted by: MrOblivious

Originally posted by: Mr Fox

Originally posted by: MrOblivious

Liek I sadi I wasn't trying to bag on you but you are overestating your postion here.


"Fluke 434 is a very expensive piece of equipment ($4400.00) that is used industrially, and picks up peaks and transients and can data log, and do SPC etc. etc... etc... We use it to assure CNC and Robotic Computers are getting proper clean power, and also to study power savings incentives, preventive maintenance,etc "

That's nice and all but seriously going from experience I'll tell you transeint loads of 10ms or less are generally missed even by really expensive brand name units.

"Quality PSU's are rated at 50c ... I cannot control or dictate what some one does with the data, but if you buy based upon my study there is a good safety factor there. "

Then why does Seasonic rate at 40c, Zippy/Emacs at 45c?

"The 33% factor takes Max. Observed Peak Value and factors in the DeRating Curve for RubyCon PSU Capacitor Technical Data that is located here : RubyCon PSU Capacitors Design Data
RubyCon makes the best Capacitors.
All of the Quality Japanese Capacitors are basically the same Per IEEE standards.
So there is nothing "Nebulous" in my Data or my Recommendations. "

Precisely the number comes from a shot in the dark based on on subset of data not representative of the whole.

It's not my stuff on the line so really if people want to cut things too close and I probably should have known that after jon tried to explain it nicely before that it would fall on deaf ears again.

Oh well. I tried politely pointing out the problems. Caryy on.

The thing that does not seem to be hitting home here is that the Peak Readings were coming in a condition, that no home user PC would ever encounter for anything other than nano seconds.

Oh I understand what you are trying to imply...I just respectfully disagree with how you are applying it and noticed some things that are absent.

Your Mean Distribution(Average)would be much lower. But that was not the intent of the study....

The Design of the Experiment was to establish Max. Peak Usage per each given Configuration. This is the Magic Number that you need to be able to Properly Size a PSU.

No this is the magic number you have decided is correct that doesn't take into account a number of variables. Like I said had you left it at "this is what I found" it would be fine.

In all honesty and brevity I don't really ever want my power supply working above 75% too often becasue in general the DC output in many models really starts to suffer after this point. Nothing like trying to stretch a platform a bit too far.

I cannot tell you why Zippy/Emac, or Seasonic have chosen the temperature that they have.

But they are quality manufacturers..........which would be my point. Quality manufacturers don't necessarily rate at 50c as you suppose/are stating.

But I can tell you that if you use 40C as a standard then a Quality PSU has virtually no de-rate based upon the RubyCon technical data. I'm sure that it was based upon their Bill Of Material, and the Technical Data that guides the Engineers for whatever Caps.

And I can tell you then that you are wrong. They do indeed still derate above 40c which would mean that the data from Rubycon (who is as you state the best) being used is not representative of the average (which would not be the best) but of that of an example of an outlier. IMO it would make more sense to take the worst example you can find not the best to start with because the worst case scenario is the problem not the best case scenario.

If you take the time to review the ATX 2.2 Design Guidelines that are referenced in My
PSU Resource Thread you would learn that my Safety Factor is right on the money and conservative but not to the point of wasted value.....

I did, and none of it says anything about your "safety factor".

EDIT : By the way the 50 Deg. C Spec is Dictated Here: ATX 12V Ver. 2.2 Specification.
Section 5.1 Page 40. You really need to review the documents that you use to write your Methodology.... I will be following up with Kyle Bennett about this, as you are clearly not researching what you write. www.Formfactors.org are the people that write the guidance standards for PSU's... It would seem that you dropped the ball here. Our friend Jonny Guru kinda missed this also....

I'm sorry it is time for you to either say what you are going to say, and back it up with Empirical Data, and References, such as I have done with every word that I have said.

From what I have seen you attempt to be disruptive by playing your coy little word games...... And doing nothing to back it up with Cold Hard Data...

You also are clearly a republican, as you read black, and white, and come up with Grey...

 

[MrOblivious]

There is a bunch stuff I could go over here and try and help you out but you have already shown you really aren't interested so just a couple quick points.

Yes I research all of it.

As for being disruptive, not really or I wouldn't have continued to try and get through that you are possibly underestimating a wee bit. I was simply trying to be helpful and point out somethings you missed. There are no coy word games I plainly stated what I meant. You don't want any help or advice fine.

On references, I spoke only of items I work on/produce personally. So I am one of the people producing the empirical data on the facets I was talking about...........The only exception being that using the best case scenario/parts to establish a worst case "safety factor" may not be the best way to do it. That I'll freely admit was simply a logical assumption on how I would take those two tyoes of data and combine them for a recommendation.

Finally, I must add though that this little logical fallacy/personal attack/really what was this? is just awesome:

"You also are clearly a republican, as you read black, and white, and come up with Grey... "

If you try real hard you might be able to still squeeze in Godwin's law.

 

[Mr Fox]

Originally posted by: MrOblivious
There is a bunch stuff I could go over here and try and help you out but you have already shown you really aren't interested so just a couple quick points.

Yes I research all of it.

As for being disruptive, not really or I wouldn't have continued to try and get through that you are possibly underestimating a wee bit. I was simply trying to be helpful and point out somethings you missed. There are no coy word games I plainly stated what I meant. You don't want any help or advice fine.

On references, I spoke only of items I work on/produce personally. So I am one of the people producing the empirical data on the facets I was talking about...........The only exception being that using the best case scenario/parts to establish a worst case "safety factor" may not be the best way to do it. That I'll freely admit was simply a logical assumption on how I would take those two tyoes of data and combine them for a recommendation.

Finally, I must add though that this little logical fallacy/personal attack/really what was this? is just awesome:

"You also are clearly a republican, as you read black, and white, and come up with Grey... "

If you try real hard you might be able to still squeeze in Godwin's law.

I'm sorry... I just mistook you for Karl Rove when you began twisting what was said.. Well the cat is finally out of the bag...........

When did being a republican make you a nazi ?

I was not the one that was playing the wolf in sheep's clothing.. If you would like to continue to Joust.... I have a few more Ace's up my sleeve.

I'm sorry, but I have no conflicts to serve, and am dead frank....
I guess you can call it a curse, or a gift depending upon how you look at it.

Nothing was missed ... I have been designing experiments for years...

We have been trying to put a value to peak loads on this forum for a while so I pulled together the equipment, and the hardware to make that happen.

Your testing methodology while it is reasonably sound tests the PSU, and only the PSU.
your experience as a PSU tester really does not qualify you as an engineer, or make you knowledgeable when it comes to design.

My experiments were designed to give real world maximum loads, and then use statistical tools to establish a bottom line minimum, that would see less than 3.45 data points per million that would be out of tolerance.

I am a very firm believer of Six Sigma Methodology as it applies to many situations everyday. By using the methods of Genichi Taguchi, I used Orthogonal Data Arrays to use predictive statistics to establish a low end baseline.

This is some very boring stuff when it comes down to it.... But I do have a method to my madness...

You can put your Groucho Glasses back on now !!

 

[Yellowbeard]

Cool testing IMO. I wish you had the resources on hand, the time, and the inclination to check a few more variables as far as system setup such as:

4 x 1gb of memory ( IIRC, a DDR DIMM will consume about 12.5w or so)
More HDs
Case Lighting
Water cooling
A plethora of fans
A P35 chipset vs a 680i

But, I am certainly not complaining. This is by far more scientific than most of the generic PSU calculators I see. And, in keeping the test system relatively simple, it is a good baseline for users that plan to build more exotic systems. IMO, it is easier to calculate the effect of adding more gear than it is to use an exotic system for this test and then have to subtract a bunch of extraneous stuff.

One other note, on both the dual and quad CPUs, the 8800GTX single card results vs SLI results differ by 177 watts (14.75a) under load. IIRC, this is a very close measurement to what those cards are supposed to consume under load.

 

[PurdueRy]

Originally posted by: Mr Fox

Originally posted by: MrOblivious
There is a bunch stuff I could go over here and try and help you out but you have already shown you really aren't interested so just a couple quick points.

Yes I research all of it.

As for being disruptive, not really or I wouldn't have continued to try and get through that you are possibly underestimating a wee bit. I was simply trying to be helpful and point out somethings you missed. There are no coy word games I plainly stated what I meant. You don't want any help or advice fine.

On references, I spoke only of items I work on/produce personally. So I am one of the people producing the empirical data on the facets I was talking about...........The only exception being that using the best case scenario/parts to establish a worst case "safety factor" may not be the best way to do it. That I'll freely admit was simply a logical assumption on how I would take those two tyoes of data and combine them for a recommendation.

Finally, I must add though that this little logical fallacy/personal attack/really what was this? is just awesome:

"You also are clearly a republican, as you read black, and white, and come up with Grey... "

If you try real hard you might be able to still squeeze in Godwin's law.

I'm sorry... I just mistook you for Karl Rove when you began twisting what was said.. Well the cat is finally out of the bag...........

When did being a republican make you a nazi ?

I was not the one that was playing the wolf in sheep's clothing.. If you would like to continue to Joust.... I have a few more Ace's up my sleeve.

I'm sorry, but I have no conflicts to serve, and am dead frank....
I guess you can call it a curse, or a gift depending upon how you look at it.

Nothing was missed ... I have been designing experiments for years...

We have been trying to put a value to peak loads on this forum for a while so I pulled together the equipment, and the hardware to make that happen.

Your testing methodology while it is reasonably sound tests the PSU, and only the PSU.
your experience as a PSU tester really does not qualify you as an engineer, or make you knowledgeable when it comes to design.

My experiments were designed to give real world maximum loads, and then use statistical tools to establish a bottom line minimum, that would see less than 3.45 data points per million that would be out of tolerance.

I am a very firm believer of Six Sigma Methodology as it applies to many situations everyday. By using the methods of Genichi Taguchi, I used Orthogonal Data Arrays to use predictive statistics to establish a low end baseline.

This is some very boring stuff when it comes down to it.... But I do have a method to my madness...

You can put your Groucho Glasses back on now !

While your test is quite informative and nice to see...your name calling here is rather out of line. He disagreed with your test methodology but he at least handled himself in a professional manner. You on the other hand have resorted to childish name calling.

Represent the profession you are in with a bit more class.

 

[Bosconian]

Originally posted by: Mr Fox
You must understand that the Max. Value represents doing things to a PC that it will not ever see in a normal in-service situation.

Originally posted by: Mr Fox
The thing that does not seem to be hitting home here is that the Peak Readings were coming in a condition, that no home user PC would ever encounter for anything other than nano seconds. Your Mean Distribution(Average)would be much lower. But that was not the intent of the study....

A lot of hard work done here, no doubt. My question is, though, how was it determined that peak numbers calculated after running "five Iterations Of 3DMarks 06 with TAT and HD Tach Running in background" was actually more than any home user would ever encounter?

It might have been brought up after the OP, but with all of the back and forth I might have missed it.

Originally posted by: Mr Fox
EDIT : By the way the 50 Deg. C Spec is Dictated Here: ATX 12V Ver. 2.2 Specification.
Section 5.1 Page 40.

Actually, this part of the ATX12V specification states that the OPERATING TEMPERATURE RANGE be within +10C and +50C. It doesn't say anything about a PSU being rated at 50C.

Full load is subjective since the ATX12V doesn't imply any parameters for what a PSU company can rate a PSU at. It's only when someone has the luxury of knowing if a unit is rated at continuous vs. peak, if peak how long can it hold peak and the de-rating curve can one actually say, "this PSU can do 'full load' in a 50C environment."

I can't find it in this doc anywhere, but I seem to remember reading somewhere that the ATX12V examples are samples tested at a room temperature of 25C.

Yes, a higher temp rating is better, but given a choice of two equally built units, one 600W rated at 50C and one 750W rated at 40C, I would still take the 750W @ 40C because 1. odds are that ambients in the PC aren't going to exceed 40C and 2. The de-rating curve from 40C to 50C is still not going to account for 250W.

EDIT: Formatting.

 

[JEDIYoda]

PurdueRy--While your test is quite informative and nice to see...your name calling here is rather out of line. He disagreed with your test methodology but he at least handled himself in a professional manner. You on the other hand have resorted to childish name calling.

Actually PurdueRy --Mr Oblivious better known as Spectre who is a Moderator over at http://hardforum.com/ needs to quit stirring the pot !! As a represetatice of another forum should have shown more class in his dialogue with Mr Fox!!
Starting with the use of facts and references.
Then there is the arrogance factor--- using this statement--
There is a bunch stuff I could go over here and try and help you out but you have already shown you really aren't interested so just a couple quick points.When somebody says something like this it basically is an acknowledgement that they are in way over there head!!

While Mr fox has backed up everything he has said with facts.

What I see here is somebody who knows what he is talking about and uses facts to back up what he says-- Mr Fox!

Then we have somebody who reviews PSU`s and thinks they know but cannot back up words with references or facts!

So I will pull up a lawn chair and fire up the old BarbeQue and pop open a nice cold can of Diet pepsi!!

Peace!!

 

[PurdueRy]

Originally posted by: JEDIYoda

PurdueRy--While your test is quite informative and nice to see...your name calling here is rather out of line. He disagreed with your test methodology but he at least handled himself in a professional manner. You on the other hand have resorted to childish name calling.

Actually PurdueRy --Mr Oblivious better known as Spectre who is a Moderator over at http://hardforum.com/ needs to quit stirring the pot !! As a represetatice of another forum should have shown more class in his dialogue with Mr Fox!!
Starting with the use of facts and references.
Then there is the arrogance factor--- using this statement--
There is a bunch stuff I could go over here and try and help you out but you have already shown you really aren't interested so just a couple quick points.When somebody says something like this it basically is an acknowledgement that they are in way over there head!!

While Mr fox has backed up everything he has said with facts.

What I see here is somebody who knows what he is talking about and uses facts to back up what he says-- Mr Fox!

Then we have somebody who reviews PSU`s and thinks they know but cannot back up words with references or facts!

So I will pull up a lawn chair and fire up the old BarbeQue and pop open a nice cold can of Diet pepsi!!

Peace!!

That comment was made after MrFox made these remarks:

"From what I have seen you attempt to be disruptive by playing your coy little word games...... And doing nothing to back it up with Cold Hard Data...

You also are clearly a republican, as you read black, and white, and come up with Grey... "

It's fine to ask for facts...but the name calling started here. A simple "Can you please show some facts to back up your claims" would have sufficed. IMO MrFox's response here does imply that he does not want to even consider other's opinions...that includes MrOblivious'

 

[Bosconian]

Originally posted by: JEDIYoda

Then there is the arrogance factor--- using this statement--
There is a bunch stuff I could go over here and try and help you out but you have already shown you really aren't interested so just a couple quick points.When somebody says something like this it basically is an acknowledgement that they are in way over there head!!

Actually, MrO didn't say that until his last post. One could say that the statement lacked tact, but looking at it from where I sit, it looks like MrO brought up some points and MrFox didn't want to discuss them.

Originally posted by: JEDIYoda
While Mr fox has backed up everything he has said with facts.

What I see here is somebody who knows what he is talking about and uses facts to back up what he says-- Mr Fox!

It sort of seems they both have "facts" but a lot of these facts come from vague data and ideas that have been interpreted by the individuals.

From what I've read in this thread: http://forums.anandtech.com/me...id=38&threadid=2097446 you already have an inherent dislike for MrO, so your "siding" with MrFox (if that's what this is reduced to, which I sure hope it hasn't since MrFox has done a lot of good work here and there is merit to this thread) is something that shouldn't surprise me.

Originally posted by: JEDIYoda
SPECTRE who goes by the name MrOblivious on these forums a complete something!
I told him where to shove his attitude and opinion on jonnyGURU`s forum......banned for life but it was worth it!!
Nothing more horrendous that somebody who applies double standards to everything!!

I'd like to see this thread that got you banned.

EDIT: Looks like Perdue beat me to it.

 

[Mr Fox]

Originally posted by: PurdueRy

Originally posted by: Mr Fox

Originally posted by: MrOblivious
There is a bunch stuff I could go over here and try and help you out but you have already shown you really aren't interested so just a couple quick points.

Yes I research all of it.

As for being disruptive, not really or I wouldn't have continued to try and get through that you are possibly underestimating a wee bit. I was simply trying to be helpful and point out somethings you missed. There are no coy word games I plainly stated what I meant. You don't want any help or advice fine.

On references, I spoke only of items I work on/produce personally. So I am one of the people producing the empirical data on the facets I was talking about...........The only exception being that using the best case scenario/parts to establish a worst case "safety factor" may not be the best way to do it. That I'll freely admit was simply a logical assumption on how I would take those two tyoes of data and combine them for a recommendation.

Finally, I must add though that this little logical fallacy/personal attack/really what was this? is just awesome:

"You also are clearly a republican, as you read black, and white, and come up with Grey... "

If you try real hard you might be able to still squeeze in Godwin's law.

I'm sorry... I just mistook you for Karl Rove when you began twisting what was said.. Well the cat is finally out of the bag...........

When did being a republican make you a nazi ?

I was not the one that was playing the wolf in sheep's clothing.. If you would like to continue to Joust.... I have a few more Ace's up my sleeve.

I'm sorry, but I have no conflicts to serve, and am dead frank....
I guess you can call it a curse, or a gift depending upon how you look at it.

Nothing was missed ... I have been designing experiments for years...

We have been trying to put a value to peak loads on this forum for a while so I pulled together the equipment, and the hardware to make that happen.

Your testing methodology while it is reasonably sound tests the PSU, and only the PSU.
your experience as a PSU tester really does not qualify you as an engineer, or make you knowledgeable when it comes to design.

My experiments were designed to give real world maximum loads, and then use statistical tools to establish a bottom line minimum, that would see less than 3.45 data points per million that would be out of tolerance.

I am a very firm believer of Six Sigma Methodology as it applies to many situations everyday. By using the methods of Genichi Taguchi, I used Orthogonal Data Arrays to use predictive statistics to establish a low end baseline.

This is some very boring stuff when it comes down to it.... But I do have a method to my madness...

You can put your Groucho Glasses back on now !

While your test is quite informative and nice to see...your name calling here is rather out of line. He disagreed with your test methodology but he at least handled himself in a professional manner. You on the other hand have resorted to childish name calling.

Represent the profession you are in with a bit more class.

This only happened after the black and white, became Grey, and wording that was very clear in nature became something other than that .

And I was not the one that was here stirring the pot as it were... Mr. Johnson only Identified himself after the cat was out of the bag.... at the very end of the thread when I placed the EDIT: so keep that in mind before you judge others after the fact.

I have no issue with someone having, a question or an issue with what was done...I love a good discussion on a professional level.

But no clear talking points were being used, only vague references to what knowledge he may, or may not have had........

If he would have first clearly identified his interests, and/or expertise, I would have treated him a little more cool... He came on the thread throwing around quasi-knowledge, and never laid out any supporting information.... just cheeky little references that he had been around the block a few times...

it is hard to argue with Data, and references are what the industry is based off of.

You really need to look at who was acting unprofessionally from the start of the thread, and then only after his interests were exposed were any jabs laid...... and they were mild at most he is a big boy, and messed with the wrong person.............

but I am not one to hold grudges.... I only laugh and move on !!

 

[Mr Fox]

Originally posted by: Yellowbeard
Cool testing IMO. I wish you had the resources on hand, the time, and the inclination to check a few more variables as far as system setup such as:

4 x 1gb of memory ( IIRC, a DDR DIMM will consume about 12.5w or so)
More HDs
Case Lighting
Water cooling
A plethora of fans
A P35 chipset vs a 680i

But, I am certainly not complaining. This is by far more scientific than most of the generic PSU calculators I see. And, in keeping the test system relatively simple, it is a good baseline for users that plan to build more exotic systems. IMO, it is easier to calculate the effect of adding more gear than it is to use an exotic system for this test and then have to subtract a bunch of extraneous stuff.

One other note, on both the dual and quad CPUs, the 8800GTX single card results vs SLI results differ by 177 watts (14.75a) under load. IIRC, this is a very close measurement to what those cards are supposed to consume under load.

Thanks YB ! I was shooting for quick and dirty.

the 4 x 1 GB look at the chipset, it would be a gamble and time consuming to tweak it in. I tried 4 x 2 GB and it puked.

Hard Drives add 10-12 watts each. per WD/Seagate Data sheets.
Fans add 3-4 watts tops maybe 5 with a Screaming Delta.

Water Cooling would have been subjective, as the various kits have different requirements according to the design.

As for the P-35 the requirements in most cases would have been less, and it would have limited Multi/GPU testing.

The 8800 GTX's were my own personal pair BFG Tech BFGR88768GTXOCE

 

[Mr Fox]

Originally posted by: PurdueRy

Originally posted by: JEDIYoda

PurdueRy--While your test is quite informative and nice to see...your name calling here is rather out of line. He disagreed with your test methodology but he at least handled himself in a professional manner. You on the other hand have resorted to childish name calling.

Actually PurdueRy --Mr Oblivious better known as Spectre who is a Moderator over at http://hardforum.com/ needs to quit stirring the pot !! As a represetatice of another forum should have shown more class in his dialogue with Mr Fox!!
Starting with the use of facts and references.
Then there is the arrogance factor--- using this statement--
There is a bunch stuff I could go over here and try and help you out but you have already shown you really aren't interested so just a couple quick points.When somebody says something like this it basically is an acknowledgement that they are in way over there head!!

While Mr fox has backed up everything he has said with facts.

What I see here is somebody who knows what he is talking about and uses facts to back up what he says-- Mr Fox!





Then we have somebody who reviews PSU`s and thinks they know but cannot back up words with references or facts!

So I will pull up a lawn chair and fire up the old BarbeQue and pop open a nice cold can of Diet pepsi!!

Peace!!

That comment was made after MrFox made these remarks:

"From what I have seen you attempt to be disruptive by playing your coy little word games...... And doing nothing to back it up with Cold Hard Data...

You also are clearly a republican, as you read black, and white, and come up with Grey... "

It's fine to ask for facts...but the name calling started here. A simple "Can you please show some facts to back up your claims" would have sufficed. IMO MrFox's response here does imply that he does not want to even consider other's opinions...that includes MrOblivious'

And also the double speak that was his opening statement of his third post was particularly cute I most admit !!

Originally posted by: MrOblivious
I am not bagging on you as the data is interesting and I know it took a lot of work but I think you might be overextending what you can do with it for various reasons.

Ry please consider this :

Opinions are just that, until you back-up your position and establish some type of credibility.

If Mr. Johnson had laid out his case, instead of beating around the bush... I would have then laid out a few more of my cards... but that credibility was never established, or laid out there.

Additionally being an un-identified representative of a commercial entity is a conflict of interest, and that becomes an issue when you have an agenda as Mr. Johnson clearly did.

The information that Mr Johnson was so interested about was the sensor polling rate of the Fluke 434 but he never just came out and said that. (My first Ace in the Hole)

The other thing that Mr. Johnson was obviously not aware of was the use of statistics based upon design of experiment to predict tolerances, and probability. (My Second Ace in The Hole)

The other issue that Mr Johnson was pressing at was 33% and where I got that.
.3 happens to be the exact de-rate rule of thumb for a Tant. cap. @ 50 Deg C.
add to that : if the designer did his work properly there is 15-20% built in safety factor in the design bill of materials.

 

[MrOblivious]

Actually no. You already knew who I was as John and I discussed it previously on the forums and typically using the I know so becuase I do xy y or z is considered bad form. So instead I made my points for you to consider.

As for the rest I had simple observations and a difference of opinion sorry you can't see past it. I act in no official capacity beyond my own curiosity here or elsewhere beyond my employ and have participated in this area for some time on a number of forums.