in regards to the placebo effect with power supplies

i dont even know why i bother, since the one i have argued with over this subject is now on my ignore list but anyway if anyone else gains info from this little experiment i guess its worth while

in a previous thread i stood my ground that a computer that was not receiving enough power could infact operate perfectly fine in a reduced state (regarding speed)

it has been documented (altho not exactly here) that new machines can and will change their frontside bus to reduce wattage consumption and retain stability, so can most modern video cards, and in theory hard drives will operate at a reduced speed with whatever wattage is left over

but what of us who have old machines, my current beater was built in 2004...

THE STORY

once apon a time osgeld biult himself a state of the art AMD athlon 850mhz, bitch slappin, mac pimping game rig and all was good, later upgraded to a 1.3 ghz cpu ...

one fustrated drunken night, years later, he decided to force some macintosh propitiatory memory into that machine, a dremel tool later he successfully fried most of the northside memory controller interface, but luckily he was building a machine for his parents, a XP 3200+ 2.2 ghz monster, tigerdirect sent him the wrong mobo (no sata), a deal was arranged where he could get the board he needed for the monster, and keep the other board for 30% off, all was good

a year later osgeld got himself a amd 2200+, a cheap but worthy upgrade to his 1.3 ghz chip and all has been good up until last year

last year i was wanting to upgrade, so i bought myself a nice dual core amd 64, i wasnt terrifically impressed with the performance, but it was an upgrade, as fate would have it, my future wifes POS celeron died about a week later, she got the dual core, im still working with my 2200+ to this day as my main game playing machine

i do have an upgrade all ready to be bought but waiting a month or two so i can sort out bills and whatnot, computers are expensive, especially when, basically after a decade, you have to start from scratch

POWER

since she got my dual core 2.6 ghz amd she also got my pimp antech 550 watt true power supply, i took her old true power 350 watt one to run my computer off of, all was good

this thing has been in service for nearly 8 years now, and after 2 years of running nearly 100 watts over spec it finally started to die to the point where the computer would not even post correctly

enter the comp usa piece of crap, i recently built a uber pimp dual core gaming rig for a co-worker, i got a nice comp usa case at the stores closing for dirt cheap, the box boasted "400 watt power supply" once you opened up the case the unit clearly was marked 250 watts (wonder why they are out of biz)

my nearly decade old 350 watt was shot, i needed a computer, so in went the 250 watt "400 watt" power supply, and it ran perfectly stable, but really slow!

THE PLACEBO EFFECT

in the meantime some resident posted a question on why her computer was running so much better with a more adequate power supply, i posted that the computer was running in a reduced state, which turned out to be 100% true, eventho i didnt nail down the exact reason, the general statement was true

some other resident claimed that it was impossible, it was all in the first residents mind, if the computer wasnt getting enough power it would be totally unstable, which is true, but only after all desperate measures were exhausted, like my old power supply which was putting out 10.98 volts on the 12 volt rail ...

the second resident, just could not understand how a computer could operate in a reduced state, eventho admitting that they do it all the farging time with power management

this all turned out to be a huge stink fest, and resulting in me adding my second name to the ignore list in the forum since 2005

MY EXPERIENCE

not counting the decades of break fix experience ive had, nor the masters degree in computer sciences and electronic engineering , lets assume im a "joe user"

my beater of a computer mentioned above, with a power supply nearly 200 watts underpowered ran the system perfectly stable, but really slow

i just got my new power supply, a ultra LS 650, not the bleeding edge but way more than enough to power my near future upgrade (485 + watts on the 12v rail)

lets examine the results tween the 2

SYSTEM

mobo = gigabyte GA7-VT600 F15

cpu = AMD Athlon XP 2200 + 1.8ghz 133 fsb + ddr (266mhz, usually i run this overclocked at 333mhz giving me 19366 mhz cpu clock speed, but the weak power supply would not allow it, so the stock 266mhz ~ 1808mhz is the basis for testing)

chipset = via VT400, 32 bit upto 200mhz (400mhz ddr) fsb

memory = micron pc2700 (333 ddr) 512mb * 3 (1.5 gb total as of testing)

video = BFG geforce 6600 GT overcloced 25% factory, agp 8x

hard drive = maxtor 160gb 5400 rpm, 133 mbs, scavenged from a busted replay tv

BENCHMARK

i used "mad onions" (now known as futuremark) pcmark 2002

why 2002, simple it gives me numbers without having to login to the futuremark website, and since were doing both the before and after numbers on the same benchmark who really cares, its the improvements were looking at not the actual numbers

RESULTS

with the POS underpowered power supply i scored
CPU = 5189
MEM = 3630
HDD = 503
screenshot = www.cheesefactory.us/slpics/pcmark1.jpg

with the new ultra ls 650 power supply, no changes what so ever to the system, the cpu's front side bus, clock speed, memory speed or even defraging the hard disk i scored
CPU = 5755
MEM = 3973
HDD = 553
screenshot = www.cheesefactory.us/slpics/pcmark2.jpg

DIFFERENCES

exactly 10% improvement in cpu scores
and almost 10% improvement in memory and hard disk scores

overall nearly a 30% speed gain in system performance , and all i did was switch power supplies form a inadequate one to much more than enough one
!! NOTHING ELSE !!

i just wish there was some sort of test i should show this all to you, oh well i guess its all in my mind

Sources are nice. They back up claims : )

Of course, you won't see this, and I'm predicting that you'll flame me or make a sarcastic response upon seeing that I made a post.

EDIT: I do hope that you don't flame me, though, it'll be a nice change.

oh robot im not always evil

Well, you have to forgive some people, they don't have a background in electronics, so they don't know any better. Power Supplies can make a huge difference in system stability, speed and overall health. It's not a theory, it's a fact.

Perhaps 40W, probably less


About 63W


Perhaps 10W, probably less


48W stock, perhaps 60W OC


About 12W max

185W total, maximum. A 250W PSU will work fine with this system. If you like lots of headroom, 350W is perfect.



Whatever. My arse is similarly stocked with figures.

I got new glasses and here are the benchmark improvements I saw:

With the POS old glasses I scored
CPU = 5189
MEM = 3630
HDD = 503

With the new ultravision 20/20 extreme magnocoated euberfocal glasses I scored
CPU = 5755
MEM = 3973
HDD = 553

Authenticated by my mom.

FWIW if you're really curious how much power your system uses, get one of these:

http://www.newegg.com/Product/Product.aspx?Item=N82E16882715001&Tpk=Kill%2bA%2bWatt

It measures power drawn from the wall. To estimate how much load your PSU is under, multiply that by the efficiency of your PSU. For a modern, efficient PSU that'll be about 80%, though the efficiency vs load curve isn't flat so it's just an estimate.

My system, 3GHz Intel C2D E8400, 4G RAM, 500G HD, nVidia 9600GT, plus a couple other boards, draws about 130W playing SL and about 145W running benchmarks. I have it undervolted, which saves IIRC 20W, so lets call it 150W and 165W for argument's sake. 80% of that gives 120W and 132W (SL and benchmarks) load on the PSU.

This system runs stable and fast on a quality 250W PSU.

Hahahaha..........i'm not getting into this one.

I'm the one who asked the question.

Sources?

Modern CPUs can reduce the clock multiplier to save power when the processing demands are low. They can not respond to reductions in available voltage or current. Get yourself a program called prime95 and run the torture tests. It will test the stability of your system under load and will tell your if your system is marginal even before it crashes.

Hard drives absolutely can not run at a reduced speed. They can be spun down to save energy and some (very few) will slow down in an idle mode that is short of spun-down, but they absolutely will not read/write at anything other than their design speed. That is, they can save power in response to no I/O demand, but they can not reduce performance in response to a reduction in available voltage or current.

You're moving into dangerous territory, Anya. Those are similar arguments that I brought up, and the whole thing dissolved into something interesting, and this thread is his latest attempt to show that I don't know anything, or something.

Osgeld doesn't like being wrong. He gets very..irritable. And don't bother asking for sources. He's his own source (at least in the initial debate).

ah but cheap power supplies lie about what they can actually do

250 watts = total bullshit

the power supply in question has a max 10 amp rating = 120 watts OMG 185 watts on a supply thats only putting out 120 watts ???? = not enough 250 watts marked on the supply is MARKETING BULLSHIT, just like your 300 watt computer speakers

and fuck your glasses, i tried to present a reasonable benchmark, if you dont beleive the numbers then go fuck yourself, i have already promised this system to robot, i cant make a dozen more for every wannabe fucktard that wants to debate me

you say yours runs fine on a QUALITY 250 watt psu, i wasnt using one IF YOU BOTHERED TO PULL YOUR HEAD OUT OF YOUR ASS AND READ, i made mention of it being a total POS a few times

prime 95, yea whats it going to do run "stressfull stuff" to my pentium mmx? please, and btw i passed prime 95 with the POS power supply as being perfectly stable so thats a load of crap

and yea hard drives CAN AND WILL run at different speeds, it DOES NOT MATTER as long as the speed is consistant at the point of read or write, you think that back in the 1950's when magnetic drum technology was introduced that is was perfectly consistant? and besides we are not talking on the fly here, shit listen to your hard drives, they dont make the same pitch constantly, which means the spinning mass is changing rpms all the time

btw say -----

edited out inappropriate personal attacks

-- Katt Linden

Anya - Point proven.

EDIT: Osgeld...ah...you don't count as a source.

Sources are third parties unrelated to you. We have no way of knowing what sort of changes took place in your configuration aside from the PSU. When desperate to make a point, you could easily do any number of things to skew your results just to prove that point (and if you weren't really desperate to prove it, you wouldn't have made this thread).

Shoot me for being traditional, but it's a reliable method : )

and when desperate to disprove a point the same tactics spawn

other than mailing you the machine, which at this point im not really in a position to right this second, i dunno what else i could possibly do

once again im a stupid lier, and your god

please forgive me robot, i forget my place in the presence of computer god, eventho god is a hobbiest who fixes junk machines for a semi part time hobby, er excuse me job

I never did get that whole "When you assume, you make an ass out of u and me" thing, because the person the assumption is being made about isn't being made to look like an ass.

In this case, you are assuming that I only fix computers as a hobby. I do fix computers as a hobby, but my main job is an IT Manager/Administrator (Sometimes I get people working for me, sometimes I don't, so it switches between Manager, and Manager+Administrator).

As for only fixing junk machines...I do enjoy fixing relics, but I tend to find myself fixing P4's and C2D's mostly...sometimes an Athlon X2 or Athlon XP. I do keep my fleet of antiques up and running, though, even if they don't get used very often.

And I never called you a liar, rather, you posted your own benchmarks as a source, when sources are traditionally third party.

I suppose first party sources would be semi-adequate when presented with evidence...I'd say having a winver window in the screenshot, as well as a System Properties window, as well as a dxdiag window. That'd probably cover it. Although, they can't beat third party sources to confirm yours.

EDIT: dxdiag should be opened to the "Display" tab.

gawd your gonna make me boot out of linux, reboot into XP, take some screenshots that will do no good becuase NOW would make me a lier,

just ignore what i present, cause im a lier and full of bullshit

btw the system changed, i leave for Florida tomorrow ... the wife wants to play sim city societies so i needed the extra 512 in my micro pizzabox Compaq, 1.6ghz celeron to take with us

please forgive me

but anyways ..

ver (like it really matters what exact verion of XP im running, i could have done both test on a non updated fresh install AND IT WOULD NOT MATTER since the base was the same)

http://www.cheesefactory.us/slpics/ver.jpg

heres the cpu info

http://www.cheesefactory.us/slpics/zcpu-cpu.jpg

heres the mobo info
http://www.cheesefactory.us/slpics/zcpu-mobo.jpg

heres dxdiag
http://www.cheesefactory.us/slpics/dxdiag.jpg

heres dxdiag again

http://www.cheesefactory.us/dxdiag.txt

like it does any good

Those are useless without rerunning the tests with both PSU's. That's why I mentioned them.

http://img246.imageshack.us/img246/7006/dx8vistaqg8.png

That is kind of how it should look.

That specific screenshot was for showing someone that Vista doesn't require a DX9 GPU to function. But the example applies here. In your case, you would include everything included in that screenshot, as well as your benchmark results, before and after. Same screenshot. The basic idea is to confirm the lack of any changes. Winver, System Properties both confirm the memory amount and CPU speed (you posted a screenshot of the ver command....very different), dxdiag confirms the same GPU (although GPU-Z and CPU-Z are better...they don't eliminate the usefulness of winver and System Properties, though). Of course something like Photoshop or Paint.NET or The Gimp could be used to forge the screenshot, so there's a level of trust involved.

CPU-Z was a nice touch, though, since it shows bus speeds and whatnot. GPU-Z is also nice. Might also be good to toss up Speedfan, let it chart out the voltages over a short period of time, and show them (not entirely accurate, but close enough for the purpose).

Of course, third party sources are more valuable (because they are something that can't be tampered with. Yes, I know you're sensitive, so no, I'm not suggesting you would, don't worry, I'm not calling you a liar), but the above mentioned is better than nothing.

and if tomorrow i dig the old psw out of the trash provide the same damn screen shots will you shut the Fk up?

i honestly think not

would it? you tell me

seriously if i provided the exact same info with before and after screen shots with nothing else changed would you finally believe me????

i honestly dont think so, even then you would want me to provide you a fucking video of me swapping out power supplies, benchmarking, swapping them out again, benchmarking

then accuse me of editing the video

if thats so forget it, but if that is what it takes just to shut you the FK up then ill do it ... im just a stubborn ass as you, if not 10x as much

since you mentioned gpu-z

http://www.cheesefactory.us/slpics/gpu-z.jpg

http://www.cheesefactory.us/slpics/gpu-z2.jpg

This is your friendly warning.

Forum Guidelines state: "Flaming, Spamming, Trolling – Flaming (posting a message that is intended to incite anger or directly attack a person or persons), Spamming (multiple posts of the same topic or discussion), and Trolling (a post with an intentionally contrary opinion written with the intent of inciting or getting argumentative opinions) are strongly discouraged. If you think your post might be over-reactive, or that it might fall into one of these definitions, please reconsider posting."

This thread will be locked and those flaming or otherwise engaging in behavior violating the Forum Guidelines will be suspended, if this continues.

Thank you.

-- Katt Linden

yes its heated, no its not trolling, maybe a bit of flaming but come on, theres much worse on the forums than 2 stubborn geeks butting heads

were both providing much more technical support in this subsection than linden labs ever has in my time

and as humans were allowed to disagree and as "science" ppl were bound to debate

Do you have any sources?

I've hung out in overclocking forums, where the objective is to overclock and overvolt components, and in silent computing forums, where the objective is to run the system on the lowest voltage, sometimes underclocked as well. Both communities use the same stress tests and the same benchmarks to make sure the system is stable at the out-of-spec settings. I have never heard anyone claim any component will step down its performance to remain stable at insufficient voltage or current.

I have read many many accounts of what happens when there is insufficient voltage or current, and in hundreds of hours of my own testing have a great deal of experience with how components fail when undervolted. They simply do not underferform, by design or otherwise. There is a very slim margin of being out-of-spec where a CPU will give incorrect results (prime95 tests for this) and a video card will draw artifacts, and anything further out of spec than that slim margin will lock up your computer.

A 10% performance difference is a big, huge stinking deal that would certainly be discussed. If you have info, please share. I'm sure the guy who runs this site...

http://www.silentpcreview.com

...would love to learn what you know. He does precisely this kind of testing for a living.

To recap, modern components often have power-saving features that are turned on *in order to save power*. The reverse is not also true. That is, these features are not currently designed to engage in the absence of sufficient power. I claim lack of experience with laptops, so if you want to find holes in my statement, look there first.