Amperage pull of a computer?

On 27 Jan 2004 02:04:04 GMT, snipped-for-privacy@ipal.net Gave us:

Yes, and all the processing that takes place in an OS is alone to make that aggregate change balance out at ANY given time.

Like I said, My DUAL CPU machine has NO change between when seti is up-n-runnin, and when it is not.

It crunches 20 units a day, btw.

I have another box that gets the exact same CPU temperature even when it locks up. Hmmmm.... Must be ONLY a clock thing, and a clock thing ONLY. Imagine that.

Now, go figure again.

"DarkMatter"

Install lm-sensors and see for yourself. Don't even attempt to reboot and check the bios reports, CPU temps change way too fast for that.

You should investigate and find out what's wrong.

"DarkMatter"

Well it's going to be hard to "show" you, but I just installed lm-sensors on my teeny epia M10000 system. Upon running "watch -n1 sensors" on one console screen and initiating a large package compile on another, I was able to easily watch the CPU temp vary directly with the CPU load. When just sitting there at 100% idle, the CPU reports about

30C it only takes a few seconds to climb to around 60C when compiling programs that take several seconds to compile maxing out the processor. Likewise, it drops back down to around 30C within seconds of the compile string finishing. It also hovers around 45C when blasting thru a bunch of small programs that cause allot of disk i/o.

What say ye to this?

~~~~~~~~~~~~~~~~~~~~~~~~

Why'd you plug your paper shredder into the UPS? Everything else I can understand, but a paper shredder? Is it really that important to be able to keep shredding your documents during a power failure? }:)

I just thought it was kinda silly, thas'all...

-Z

On Tue, 27 Jan 2004 04:21:52 GMT, "Ross Mac" Gave us:

You forgot about the shunt. Also digital accuracy only tells one what flows through the shunt, and makes no consideration for the leads.

There is already a shunt in the meter, and that resistance plus the leads makes for erroneous readings. In Any METER's Amp setting. That is why current probes are so popular, and using a DMM for amperage reading is not all that popular in the REAL industrial circles.

The Amps function of most DMMs is for low level stuff. The ten or twenty amp range on many should be avoided for measurements which actually approach those values as that is where the most error will be seen at those higher amperages with standard meter leads. A set of very short, heavy gauge leads will reduce it a lot. Said error being the difference between what the circuit really uses, and what it uses with your newly installed "less than negligible" increased line drop due to the leads.

It isn't JUST the shunt resistance. in a line that has been "opened" for monitoring, all of the elements in the chain must be considered.

The resistor has the shortest lead length, and therefore would change the circuit the least, and be the most accurate "ideal" monitor.

Exactly 0.1 Ohms precision resistors are required so direct voltmeter reading can be easily converted. 0.01 makes for an order of magnitude change in said conversion (good for higher currents with high power precision resistors). One ohm changes it an order of magnitude the other way (good for low level stuff, but the one ohm must be figured in as lead resistance from the circuit's POV.. Direct volt reading across such a device presents very little change to a circuit.

Yours would be 0.21 Ohms.. Quite an error there. Particularly since the calc that gets done is based on the 0.1 Ohm value... even inside the meters on Amps mode, it is not considered.

Not knowing how accurate your meter is at telling you the lead resistance also means that your DMM will be the voltage reading it takes internally divided by your 0.11 plus the shunt value ohm reading will very likely be even farther off the mark.

The installed shunt resistor can't be beat. No way around it.

On Tue, 27 Jan 2004 05:37:22 GMT, Charles Perrin Gave us:

Which is all about shutting a non used laptop down to a crawl, not normal operation of a fukin desktop!

On Tue, 27 Jan 2004 05:37:22 GMT, Charles Perrin Gave us:

I was the one that posted that remark, and my machine doesn't change.

Perhaps you can post a link to this looping applet for a benchmark?

On Tue, 27 Jan 2004 05:38:02 GMT, "Ross Mac" Gave us:

With a 0.1 Ohm resistor, you'd have to be a total MacTard (I owe you several thousand of those... bwuahahahahaha) to have a problem with the "conversion". You can do it in your head. It's called "shift decimal, record reading".... Done! DIRECT conversion.

1V = 10A 0.1V = 1A

Any monkey can do it. Move decimal ONE place.

On Tue, 27 Jan 2004 08:17:04 GMT, "John" Gave us:

And you subsequently introduce an error in the system, due to the meter lead resistance. If you are opening the line, the precision resistor is only a $2.00 step away from all but perfect accuracy, and keeps the meter safe as well.

Read the followup. It was done to maximize load for test purposes in order to respond to the OP about power draw. End result...... a complete office with everything running doesn't pull even 15A

The last time I saw dark matter was just before I flushed the toilet.

On Tue, 27 Jan 2004 23:15:10 GMT, "Anthony Fremont" Gave us:

Wow. Must be a lame CPU. Mine run at 62... all the time when running. Both do. Seti up.... same. From DOS... same... From linux... Same.... From directly after screen saver... same. From a benchmark.... same.

What gives? Not bits. bits are bits, and they are ALWAYS there.

You're *SO* wrong your nose stinks, and I'm the "retard". DimBUlb, you really must understand your (severe) limitations.

Since *YOU* (yes, *YOU* DimBulb) are using: X-Newsreader: Forte Agent 1.91/32.564, you too are using WinBlows. Nevertheless, OS/2 used the HLT instruction before WinBlows, as does Linux.

You're the clueless moron here (again, as usual).

You haven't a clue (again, as usual).

You retard! When the *OS* enters the idle loop (as in nothing left to do) it executes the HLT instruction. When an interrupt comes along the processor wakes up and looks for something to do, then goes back to sleep.

You're amazingly stupid!

You're totally clueless. For one thing it's not the "high bits" vs. the "low bits that matter". It's the bits that are

*changing* that matter, sorta. Upwards of 40% of the power is dissipated by the clocks, so sleep modes are critical. ...again you're clueless!

SO do desktop processors you fool!

No, that would be Apples and PCs. Both use the same techniques.

You're an idiot (again, nothing new). Power saving is critical on modern processors. ...not something you have a clue about, obviously.

Wrong!

You're an idiot! ...and couldn't be more wrong. Your're at least five years out of date.

...like you have a clue about tomorrow's sunrise.

YOu're absolutely wrong. Modern PCs, yes desktops, are put to sleep whenever there are no tasks in the idle loop.

You have a retarded system (go figure). If you were running a modern OS and processor the processor will dissipate little when nothing is happening. Running yoru retarded SETI as a background job will indeed use more power.

Yes, DimBulb. I develop these things. I full well know how they work. ...and the competitions. You? ...go back to assembling your power supplies and collect your $6BILLION. LAFF!

YOu're a retard! All current processors have power-saving modes.

*ALL* have the HLT instruction (or similar), and others have various other coarse or fine-grained power-saving modes.

You're a fool (again, no surprises)!

What do *you* know? You haven't the foggiest about CPU power consumption? What the hell would you know about the power consumption of the graphics card? These are the to *largest* (and variable) contributors to the system power. ...friggin' idiot.

Of course it is. ...as are the other sleep/nap modes. However its use is not all that complicated. The OS (obviously priv) doesn't have anything queued, so it powers the processor down. An interrupt (including the timer) wakes it up. This has been going on under DimBulb's nose for a decade.

Correct! It's the old CMOS P~FCV**2 formula. If F is zero, so is P. ...of course with later processes this doesn't hold.

Nope. It's *FAR* more. A 1GHz PII is ancient technology, so I'd expect the leakage to be zip. The difference should be an order of magnitude. It can be 50-60W, or more in a P4.

Clock transitions, approximately. The clock tree is about 40% of the power of a PIII. It's somewhat less for later technologies only because the things are such power hogs otherwise.

DimBulb is out of his league. Ha hasn't a clue what he's talking about, so your comment is meaningless. The FPU clocks are

*STOPPED* if there are no FPU instructions in the pipe.

Sure, though DimBulb hasn't a clue here either. Even under power saving modes there are parts of the processor that must run. Any dirty write-back caches must be snooped and the PLL must run, in all but the most severe power-down modes. Teh execution units do

*not* run unless there is work to be done.

Good grief! The *processor* can swing 80W! Disk drives are a piece of cake, comparatively.

No, *meaningless*

DimBulb misses his nose. He hasn't a clue what he's talking about!

On Tue, 27 Jan 2004 19:17:51 -0500, Zorin the Lynx Gave us:

If you read the thread, you'll find that he put many things on the line as a test, not as the way he normally operates his stand by power system. He wanted to determine his office's entire usage.

On Tue, 27 Jan 2004 20:47:06 -0600, "Brian" Gave us:

Stop looking in the mirror then, s*****ad. You Top Posting, retarded Usenet s*****ad!

Once again, you haven't a friggin' clue, DimBulb. Execution units that aren't used *are* shut down. Processors that aren't used *are* shut down. Power dissipation is a huge problem these days. All modern CPUs have power saving, including HALT!

...moron!

-- Keith