Discussion in 'Article Discussion' started by Gareth Halfacree, 4 Sep 2015.
Looks to 'Green the Beast.'
3 months of the year, i use no heating (Scotland).
3 months of the year, my pc is 'wasting' energy
9 months of the year, it's contributing slightly to heating my house.
I'd hazard a guess that I spend 10,000 kWh/year just for heating.
If I lived in southern California, I'd perhaps be a little more concerned about that 'wasted' energy. For now, I'll continue recycling it as heating energy.
@1400 kwh/year, that's about £150, about £50 of that is 'wasted'.
I'm not sure how the research is warranted. High capability components use more power. Power efficient versions use less power. Wow
I'm interested in the incoming Quadro M4000. Waiting for the price though xD.
It has the raw power of a GTX770 (1664 cuda cores), but comes in a single slot package, 120W TDP and 8GB of memory: https://www.pny.com/nvidia-quadro-m4000
I'm glad he didn't do this research in the days of separate PSUs for the PC and cooling system, or even five years ago, might have died of the shock.
Hahahaha.. good joke.. Try that number every quarter...
About 300AUD, cause 'straya....
Edit: to be completely fair, that does also include a dual-CPU server and most of the network, but still...
The most power hungry system in that list draws 521W under full load
1400kWh/year = an average of 3.84kWh/day or 7.4 hours running that high end system every single day.
I don't think the average gaming PC draws as much power as that and I don't think the average gamer spends over 7 hours a day 365 days a year gaming.
It'll be fine: in the 70's and 80's they said that nuclear power would bring an abundance of cheap, clean electricity. It can't be far off now.
Most gaming PCs draw at least triple a console. The lowest I've seen from the wall on 2 gpus is 430 watts.
Regardless of what we think of his research, I think we could all agree that we'd benefit from more efficient components.
Efficient is better. But solar is even better. That is all.
I'm not sure this will remain a problem much longer. Games are getting pretty close to life-like detail. Once that is fully achieved, basically all newer generations of hardware just have to find more power efficient ways to maintain that same performance. That's basically what's happening to the CPU market. If you haven't noticed, Intel hasn't really tried to improve performance for something like 4 years now, and they haven't needed to. But in terms of performance-per-watt, they've continuously improved.
I think the greatest issue for GPUs right now is 4K screens.
^^ detail is only half the battle though, once games get to lifelike detail, then the RANGE at which those textures and polygons are rendered can be increased, needing more grunt, or higher resolution, etc etc. there is always going to be a desire for 'MOAR POWER'.
And then you see countries like Iran trying, and the US comes in with an attempt to shut it down hard.. With that in mind, unless you're China (who doesn't care what the US does), how do you expect the rest of the world to get the guts to start nuclear energy production?
At the same time, most of the first world have lobbies left right and center effectively stopping nuclear energy from proliferating, despite the complete failure of Fukushima leading to a meltdown, and the ensuing failures leading to (as far as I can tell) no major or widespread health effects or havy contamination...
Most gaming PCs are single-GPU affairs, with 150-300W GPUs, and 100W for the rest of the PC. Most people should be seeing under 500W at the wall for the PC.
We do, we just overclock it more or make it bigger to get more performance!
Solar is slow and expensive to deploy, and the logistics, politics and payment models are annoying to do. Plus, it's not an option for apartment-dwellers either. Realistically, we need something to bridge the fossil-fuel to renewables transition, and right now, nuclear energy is pretty much the only viable option - except for the fact it's blocked in politics.
Oh stop that crap. Intel's been improving performance just fine, it's just that the mainstream market hasn't had any use for more than 4cores up until now (blame the software consumers use). In serverland, we've had a steady rise of core counts: from 6 (Dunnington) to 18 (Haswell-EP/EX).
So right now, as far as core design goes, Intel wants two things: low power cores to grow their mobile presence (ideally break into phones and wearables), and low power cores so they can stick lots of them into a massive server CPU they can sell for IBM-sized margins. The result is an extreme focus on performance per W efficiency that serves everything but the low-core-count consumer desktop market, a market where almost nobody can use any more performance.
How about doing some actual meaningful research and look into the whole lifecycle of these PCs, not just the part where they are being used by their users. For example, how much energy goes into manufacturing , distributing, selling, disposing or recycling these. And also why not look into how software and business models play into such high energy consumption. Pussyfooting around doing such mediocre research is not worth anyone's time.
Any decent normal consumer benchmarks out on this? As far as I know, Quadro cards cost a lot more because of very high quality driver support right?
Yes, you pay for the optimized drivers and the uncrippled FP64 performances (but still crippled compared to Tesla cards) ... and the 8GB of memory vs 4GB for the GTX970 (its consumer grade counterpart). Quadro M5000 and upward have ECC memory too.
I haven't seen those cards for sale, yet. It could be a great versatile card, depending on the price.
This is a link to the paper. (From their website)
One thing I note is that they have a table describing the TDP versus the actual power usage. About 5 minutes of googling and I found a paper Intel have on the conditions used to calculate TDP. The researchers didn't use those conditions for benchmarking. They ran cinebench.
That figure is actually almost the median of this guys "Top-10" list, the 14000Kw may well be considered an outlier (as it's near twice as much as the second place one)
The median is 534,5 (with display mind you)
That's still an estimated 7 Hours per day UNDER FULL LOAD. (how realistic)
In other words, the guy's a greenie trying to raise an alarm with high figures grabbed out of thin air.
Phew, he had a lot of time, reading his actual article. Also Americans seem to have a different way of indicating sources than we have here.
Anyway, average day's use of a gaming PC according to...
Urban, B., Shmakova, V. Lim, B. Roth, K. (2014). Energy con-sumption of consumer electronics in U.S. homes in 2013.
as quoted by this guy:
Active Gaming 4.4
Web Browsing 2
Video Streaming 1,6
Idle modes 4
Hmmm, 8 hours of spare time to blow off every day...
I think he misinterpreted the "Urban" study, then again it's >150 pages, and I can't find his figures there...
Separate names with a comma.