Slashdot Mirror
PS3 8x More Power Hungry Than PS2
MonsieurCreosote writes "The Playstation 3 apparently demands eight times as much electricity as the Playstation 2, and more than twice as much as the Xbox 360. It also consumes much more power than a top-end PC gaming rig. It's not clear what's causing the massive drain, but Sony is now denying reports that the PS3 experienced overheating problems at the Tokyo Games Show last month. From the article: 'While an Intel Core 2 Duo PC with high-end graphics card chews politely on a 160 watt entré, the PlayStation 3 gorges itself on 380 watts... The extra power consumption of the PS3 over the PS2 suggests that we're not really getting much better at designing efficient systems, we're just pumping more 'fuel' into existing paradigms'. Are modern console hardware designers getting sloppy?"
Are modern console hardware designers getting sloppy?
I don't think consumers care much about power consumption. If I can design something cheaper and faster, but hotter, and the consumer doesn't care, why wouldn't I do it? Lower bottom line. Higher profits. Booyah! More Ferraris for my garage.
Max power rating of the PSU does not equal power used in normal operation
You've been trolled - most likely by someone paid by Microsoft
it's in my head
Here's the link to the original source before it went through a 4-blog telephone game:
http://www.jp.playstation.com/support/qa-591.html
- The GPU
- The Cell processor
- The highly clocked Rambus XDR DRAM
IO devices like the hard drive and the Blu Ray drive are peanuts compared to those.If it's for-profit but free, you're not the customer -- you're the product (e.g., the Slashdot Beta's "audience").
Sure.
A BluRay Rewritable drive uses at most 30 watts. That's peak, and probably only used for a few seconds when spinning a CD-ROM up to 52x.
Simple floating point power of the SIMD units is easy enough to benchmark.
Core Duo has one SIMD unit. Cell has 10 (7 SPE and 3 AltiVec). You can
Google for this stuff fairly easily. It's even the same benchmarks.
In 5 years the Core Duo will be just as fast; and everyone will have bought a
new PC (or two) to get it. Sony has to put this chip out now, so that it will
still be relevant in the MIDDLE of the console lifecycle.
However, the PS3 is probably a lot less expandable than your PC, so it doesn't need an overkill power supply. Sony knows how much power it needs, and they aren't going to waste money putting in a bigger PSU than it needs.
Of course with 8 cores, chances are it will not spend all that much time at maximum power usage very often...
-- OpenVerse Visual Chat: http://openverse.com
How were these figures calculated? By taking the 127 from 100-127V range supported by the PSU and multiplying by 3A to get 381. 3 amps is what the FCC label says. But since the PS3 runs in Japan at 100V, the PS3 must demand at most 300 watts. At most. But that's just the PSU. It doesn't mean the device actually draws that power.
By way of illustration, the XBox 360 PSU run at 5 amps. 5 times 127V = 635 watts. So why no stories about the XBox demanding 635 watts? Why no stories that say the PS3 actually uses half the power of the 360? Because the XBox 360 consumes 160 watts in normal usage. It is entirely misleading to look at what the PSU can deliver to determine what the device actually uses.
The same will be true of the PS3. Unless some reputable site such as ARSTechnica, Toms Hardware etc. sticks a probe in the thing and states what power the thing draws this story should be treated as bollocks. Bollocks swallowed whole by Zonk as usual.
This whole story stinks and it is just being bounced around by people less interested in whether it is true or not than putting the PS3 down.
The actual root of this story is a someones blog entry. True story. And now the story has been repeated and repeated and repeated and now it's apparently become a fact without context. The only fact is, is that the PS3's power adapter runs has a peak power of 380W. It doesn't require that power at all times. To compare the PS3's max power consumption to max power of a single Core Duo CPU seems disingenuous at best. Remember, the PS3 is an entire system, Cell Processor, Video Card, and HDD... So it has the components of a computer and it consumes at computer-esque amount of power. Maybe I am the only person who doesn't see this power consumption as relevant. I get that it will increase my power bill by few dollars every month, maybe even a few more dollars that Xbox 360, but that's ok.
And this an irrelevant fact, but I'd be curious to see the power consumption levels or a non-core Xbox 360 powering a HDD, and also requiring another outlet for it's HD-DVD add-on. I'd be suprised if we didn't see that 200W's for a core system creeping up into the +300W range as well.
At any rate, this story seems like a non-story to me.
Care to read what you post when you use the Google? You might realize that you are essentially comparing apples and oranges.
The first paper is about adapting LIN/LAPACK to the cell processor. If you read carefully, they are running single precision for that "amazing" ~100 GFlops a second; double precision drops off quite considerably, roughly around 10 GFlops, although the scale of their graph makes it difficult to determine accurately (page 13). It should also be noted that they are tuning all of this in low level assembly for the cell processor.
For the second paper, you might start to wonder, why are the processors still considerably slower? Could it be that they are running MULTIPLE instances of LAPACK on the processors in the second paper? I think so! If you cared to read the discussion on page 6 instead of just looking at the "pretty pictures", perhaps you would've realized this. They are also comparing high level libraries; I wonder if the Xeon/Opteron might function slightly better if they were written in assembly like the Cell case, and not running an OS? Something tells me they might.
Hmm, it seems to me that those systems are rougly equivalent if you actually try to compare them on (roughly) the same test: LINPACK, double precision, single instance produces ~10 gigaflops in the Cell, ~12 gigaflops or so on the Opteron. Perhaps next time you will care to read the papers you're posting instead of just looking for pretty pictures.