Bitcoin Forum

Please excuse me if this has been discussed bef0re [and please excuse my 0's]. I did search and w0uld appreciate a link t0 appr0priate thread if it exists.

I'd appreciate if s0me0ne had #s sh0wing whether 0r n0t 0verv0lting is pr0fitable - ign0ring the extra wear 0n the card & p0tential pr0blems due t0 excess heat.


Ideally, if s0me0ne had p0wer draw & hash rate @ st0ck speeds, pwr draw & hash rate @ max stable cl0ck speed [w/0 0verv0lt], and pwr draw & hash rate @ max stable cl0ck speed and max stable 0perating v0ltage - that'd be great. As well, if s0me0ne has pwr draw & hash rate @ vari0us undercl0cks, that'd be all the m0re dandy.

If this data d0esn't exist, I'll d0 the testing myself in a few days when I'm back with my rigs [and fully-functi0nal keyb0ard!].

that'd be interesting data.. i'd like to see it also.

I have ever seen a chart but generally speaking power load changes by the square of voltage change.

So overvolt 10% = 1.1^2 = 21% increase in power.  That is partially offset by increasing clocks but it generally is a bad deal unless you have free power.

+1



In my own, very limited testing, I was drawing more power at higher gpu volatge at same clocks. Was also generating more heat.  I also noticed that input volatge played a significant role in the cards total amp draw.  I.e., 12.4v had lower TPD at the card level than 11.7v did..  Sadly, no reliable graphing or logs of any of my limited findings. :/

so, would be interested in seeing some more careful testing.

cheers

I`m overvolting only few GPU, 3 of them are Watercolled and only 1 os on AC. WHY?
Cos i cant see VRM temperature in linux. The only one GPU is stock 5850 and software is radeonvolt.
What i obserwed:
Incresing Vgpu STOCK ( it`s 1.125 i think) -> 1.2
VRM temp@STOCK - 62,65,63*C, VRM@1.2+40MHz more = 78,80,78
Whole rig took from the wall 25-30W more.
Fan speed, this same. mem clock - this same.
What can i say....soon or the later, overvolting will fry your GPU.
I did it with 2 of my 5970, so i think i may say "i know what i`m talking about".

Test will be very straight-forward, getting the stats listed in op.

What I intend to derive from the data is whether the increased cost of electricity justifies overvolting. I will update the spreadsheet and get to work filling it out as I learn when I'm back with my rigs in MI ~Friday. The data will be at https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0Ao892S4MOoDZdGFncFpPWlVUZUozSkduYk9XS0UzV2c&output=html and I'll bump this thread when I'm done.

If any0ne's b0red and l00king f0r data t0 rifle thr0ugh, I've g0t s0me data 0n batteries y0u may be interested in [m0stly inc0mplete and I'm n0t w0rking 0n it again f0r a c0uple m0nths while I m0ve]. https://docs.google.com/spreadsheet/pub?hl=en_US&hl=en_US&key=0Ao892S4MOoDZdGVXLXBQbmdJM21xelhPN1M4bDhYS0E&output=html I like criticism and already have plenty 0f new ideas t0 implement after learning my uncle literally used t0 test batteries f0r c0rp0rati0ns.  :)

Ideally, you want to up your clocks without uping your voltage. With good, clean 12v power and proper cooling this is easier to achieve.  I've been able to drop my 5830's to 1.167v @ up to 1020MHz @ ~85a. This is on an 80+ gold with solid 12.3v on the 12v rail. On my old supply, sag at load was down to 11.75~v on 12v rail and I was not able to drop the voltage or to be stable at over 1010. Also the amperage was higher at the card. I.e., 1.178v @ 1010MHz and ~93a.  That's almost 10w more and 10MHz less.

Overvolting(probably plus ANY overclock) = FASTER electromigration!!

Actually, it's UNDERVOLTING that some experienced miners are recommending.

It's not pure hash rate what counts (unless someone else is paying for your power) but the ratio of your hash rate to power consumption.
Lowered voltage might mean that you'll have to pull the clocks back as well but the power consumption will fall significantly faster than hash rate.

I don't mind the 0s a bit. I fixed them with greasemonkey :>

I'll check underv0lting, t00. I meant t0 menti0n it.

You seem to be confused. '0' is a number, 'o' is a letter. Your posting is unreadable as is.

That shit's 1337, dog.

This forum is a joke. Full of genius, and elite script kiddies.

Gonna go out on a limb here, and guess his o key is gone...lol, hence, the 0

A friend of mine had/has a similar problem...space bar broke; use.a.period.instead

0101100101100101011000010110100000101110001000000100100100100000011101000110100 0001100000111010101100111011010000010000001001001001000000110110101100001011001 0001100101001000000110100101110100001000000111000001110010011001010111010001110 1000111100100100000011000110110110001100101011000010111001000100000011010010110 11100010000001110100011010000110010100100000001100000101000000101110

Anyh00, I still plan 0n getting the results p0sted in a few days, in n0n-1337-5P34K. Pretty curi0us t0 see h0w it turns 0ut given I've read many c0nflicting rep0rts.

conflicting report? the only issue i see here is your brain is smaller than my balls.

Fcking kid.

When d0 we replace the f0rum with 0ne big sh0vel d0g .gif, anyway?

Dear lord, I have somehow traveled back in time to the 1990s! Damnit, Percival Dunwood!

except he has 'o's in his posts as well...so much for that theory.

lol
yeah

someone missed the title.