Welp, I was using wikipedia for research. Still, the performance-per-watt must be terrible when compared to BlueGene.peteamer wrote: Pete
OMG, a 2000W power supply
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar
-
- Posts: 41
- Joined: Fri May 20, 2005 9:01 am
They have no money. How are they going to buy a supercomputer with no money? Instead they get supercomputer power at no cost to the research organization, the cost is shared among people who purchased computers for home or business use anyway and are donating CPU cycles and electrical consumption.mantralord wrote:Welp, I was using wikipedia for research. Still, the performance-per-watt must be terrible when compared to BlueGene.
-
- SPCR Reviewer
- Posts: 1850
- Joined: Sun Sep 21, 2003 11:23 am
- Location: Vancouver, BC, Canada
Ahh, sorry for being North America centred.
On another note, how much does peak power consumption actually matter? It takes the same amount of electricity to boil a given amount of water whether it's delivered as 2000W over five minutes or 1000W over ten minutes. And, since kettles and most other appliances do not run constantly, is there anything to be gained by looking up the maximum power rating?
On another note, how much does peak power consumption actually matter? It takes the same amount of electricity to boil a given amount of water whether it's delivered as 2000W over five minutes or 1000W over ten minutes. And, since kettles and most other appliances do not run constantly, is there anything to be gained by looking up the maximum power rating?
-
- Posts: 41
- Joined: Fri May 20, 2005 9:01 am
It's possible it will take less energy for the 2KW kettle due to the heat having less time to escape through the walls of the kettle itself. Anything that holds heat, leaks it, so the quicker you can raise it to boiling, the lesser the time for heat to radiate outside the kettle.Devonavar wrote:On another note, how much does peak power consumption actually matter? It takes the same amount of electricity to boil a given amount of water whether it's delivered as 2000W over five minutes or 1000W over ten minutes. And, since kettles and most other appliances do not run constantly, is there anything to be gained by looking up the maximum power rating?
Think of a CPU heatsink; with a really small heatsink, it can dissipate, say, 20 watts and never go over 75C at all, it would always stay below 100C no matter how much energy you used. At 30 watts, it would take a lot of energy but eventually reach and exceed 100C. At 50 watts however, it should very quickly overwhelm the heatsink's ability to lose heat and blow right through 100C right away. Since no thermos or kettle can insulate heat perfectly, in theory, the higher the heat rating, the more efficiently it should reach boiling.
-
- Posts: 175
- Joined: Tue Oct 04, 2005 4:58 pm
One consideration here is that computers rarely draw their full power all the time. Instead, a PSU capable of supplying 2000W would supply that power on a peak basis instead of a root-mean-square basis. This way, even if the computer did happen to draw 2000W, it would only be for a fraction of a second, and the power supply wouldn't flinch.
Well, I'm pretty sure you have already seen this article, but I'd like to link it anyway, just for the record & more people to enjoy.
A truly beefy machine on a 500W PSU. Can it be?
CPU 3.2 GHz Core 2 QX6700
GPU 2 x 614 MHz GF8800GTX
Mainboard nForce 680i
Peak FP DP CPU 51.2 GFLOPs
Peak FP SP GPU 1.1 TFLOPs
CPU FSB Bandwidth 8.5 GB/s
Main Memory Bandwidth 17 GB/s
Peak 2xGPU Memory Bandwidth 174 GB/s
http://theinquirer.net/default.aspx?article=36066
A truly beefy machine on a 500W PSU. Can it be?
CPU 3.2 GHz Core 2 QX6700
GPU 2 x 614 MHz GF8800GTX
Mainboard nForce 680i
Peak FP DP CPU 51.2 GFLOPs
Peak FP SP GPU 1.1 TFLOPs
CPU FSB Bandwidth 8.5 GB/s
Main Memory Bandwidth 17 GB/s
Peak 2xGPU Memory Bandwidth 174 GB/s
http://theinquirer.net/default.aspx?article=36066
-
- Posts: 41
- Joined: Fri May 20, 2005 9:01 am
Which is exactly why every gamer/overclocker out there is a genetic mistake.kater wrote:Well, I'm pretty sure you have already seen this article, but I'd like to link it anyway, just for the record & more people to enjoy.
A truly beefy machine on a 500W PSU. Can it be?
CPU 3.2 GHz Core 2 QX6700
GPU 2 x 614 MHz GF8800GTX
Mainboard nForce 680i
Peak FP DP CPU 51.2 GFLOPs
Peak FP SP GPU 1.1 TFLOPs
CPU FSB Bandwidth 8.5 GB/s
Main Memory Bandwidth 17 GB/s
Peak 2xGPU Memory Bandwidth 174 GB/s
http://theinquirer.net/default.aspx?article=36066
-
- Patron of SPCR
- Posts: 749
- Joined: Sat Nov 11, 2006 6:02 pm
To be fair I was pretty surprised by that result too. Just going off the stock numbers, the GPUs+CPU draw 490W at max load, and that's not factoring in the overclock, HDD, motherboard, Peltier cooler, etc etc. All I can say is the power supply must be a genuine 500W unit; what's even more startling is the max power draw was 429W, with all other results being in the low 400s or even 300s.
That PSU is a passive(ish) unit with two 12V lines at 15A each, so it should only be able to provide 360W on the 12V rail. Even the NeoHE 430 has a slightly higher combined output of 384W on the 12V rails. So that MGE psu has to be able to provide more juice than advertised, or something is really wrong here. Here is some kind of reveiw of it. Manufacturer's site. The manual says interesting things like "Low Gravity construction, expecially created for heat wave airflow design" and "Magnetically shielded cable with UV plastic cable". They also say something about 68% efficiency and that it delivers 70% of its rated power at 50c. They also offer a lifetime warranty in th US and Canada. yay.
-
- SPCR Reviewer
- Posts: 1850
- Joined: Sun Sep 21, 2003 11:23 am
- Location: Vancouver, BC, Canada
You'll all want to note that the measurement tool used to calculate power usage is built into the PSU itself. I have little confidence in the accuracy of this tool; similar "power meters" built into other PSU's I've seen (Coolermaster's comes to mind) have been substantially inaccurate ... by as much as 50~75W IIRC. Check our CM review for details.
I did not pay much attention to the power meter to start with. What it says is irrelevant, my point was that the PSU is rated for a max of 360W load on the 12V rail at 25c, and 70% of that at 50c. An overclocked QX6700 and 8800GTX SLi setup should consume more than that particular PSU can handle. Im just surprised, and lsightly confused