GPUs-Amazing variance in power demands

S

Sunin

[H]ard|DCer of the Month - August 2008
Joined
Dec 27, 2005
Messages
3,421
Certain work units my 4xGX2 pulls about 650 watts, some work units read the new P10101 or whatever they are called pull over 1,000watts! HOLY CRAP variancE!
 
People are going to be popping PSUs left and rite in multi GPU rigs if they land a whole batch of those P10101s.
 
It's going to be very interesting to run these in the summer...
 
Measured with a Kill-A-Watt meter I presume? Because that could also be contributing a fair amount to the variance you see.
 
Wouldn't the variance be consistant at the 600 level tho too?
I ask cuz I don't know, I only just started using mine and it's on skynets test bench right now

Just curious
Posted via [H] Mobile Device
 
This is interesting. I hate these p10101 units with a passion. I'm going to make an excel sheet or something and document my findings with my 260s. Unfortunately, that means it will be hard to isolate each one's power draw, but I'll see what good pausing does. I'll update this post after I've gathered some data.
*edit* Initial results tell me theres about 20 watts difference between a P5915/1888pt and a P10101/584pt. This is of course only going to apply to my case, and 260s in general I'll bet. I'll keep collecting data.
*edit2* I give up. I'll post what I collected real quick, and you can see why. So a 353 pulls as much as a p10101? Can't be right. What I was doing was looking at my APC ups and staring at it for 10+ seconds and trying to see where it levels out, as it isn't just sitting nicely at a number. It jumps within a variance of 20 watts on average, so I try to find the high and low draw during this time and average it out (leaving out any single extreme spike or dip). I also turned off my monitor and unplugged the laptop to rule that out.
powerl.jpg
 
Last edited:
Tobit-

It's what my ups was reporting, as such, I decided to scrap my readings. I said so in my post. I'm sorry if I didn't make it very clear, it was late and I probably shouldn't have been doing it at all :p
 
Zero82z said:
Measured with a Kill-A-Watt meter I presume? Because that could also be contributing a fair amount to the variance you see.
Click to expand...

I run Cyberpower Battery Back-up which has built in wattage meters. The ratings were from them. Two seperate machines too, exact same ratings!!!
 
APOLLO said:
It's going to be very interesting to run these in the summer...
Click to expand...

Good thing is my basement is below ground so it stays a near constant temp.. 72 with all machines running P10101

The heat upstairs never comes on when I'm churning all machiens on them!
 
Vaulter98c said:
Wouldn't the variance be consistant at the 600 level tho too?
I ask cuz I don't know, I only just started using mine and it's on skynets test bench right now

Just curious
Posted via [H] Mobile Device
Click to expand...


depends on how efficient the PSU is at higher loads.. so say its at 85% at 600w(system load).. it may only be 80% at 1000 watts(system load).. so it would be drawing more power to convert 1000w to the system..
 
But it's still a good benchmark no? I'm building a "build bench" today for my folding room and it's gonna have a spot specificly for my killawatt
Posted via [H] Mobile Device
 
You must log in or register to reply here.
Back
Top