I like the data you're collecting and the charts you're making.
I disagree with the whole harping on the energy bit. I'm as much for conserving as the next guy, but the cost savings due to hardware changes and usage patterns are insignificant compared to the amounts squandered by typical American activities including alcohol (after work happy hour, friday and saturday night festivities, and that stash you keep at home), services not actually used (like cable tv and unlimited data wireless plans), transportation (gas, tolls), excessive social entertainment (movies, restaurants, outdoor team sports/hobbies), and impulse shopping.
But you may have missed where your SLI 580's sit on those charts for efficiency; even compared to other setups that can produce 60+ FPS.
Unfortunately, it's easy to make charts say whatever we want. In the restricted variables you compare, the results come out as they do. (As for my cards, I use them for more than gaming. But even in gaming, they shouldn't be on your list at the paltry 1920 res.) Regardless, I was only arguing what I feel is a misplaced emphasis on power costs.
Yet you are using a 9600gt? Surely some other considerations entered into your buying that.
The poll dosn't suit the thread.
How often you upgrade has nothing to do with the topic of the thread.
Where is the poll option to choose performance over "efficiency"?
Thanks for the effort OP. I didn't go in detail as to the accuracy of your calculations.
But here is a much quicker assessment:
A mid-range videocard = 150W
A high-end videocard = 250W
Difference of 100W
100W x 365 days a year x 6 hours a day at 100% load x $0.15/kWh = $32.85 per annum or less than $3 a month.
Ok what about idle power consumption? At idle, most modern GPUs are within 15-20W of each other (as has been noted in this thread), far less than the difference between modern processors that a high-end GPU gamer would have: i.e., Core i7 920-950, Core i5 750, 2500k or a Phenom II X6 1100T at idle (likely the type of CPUs that a high-end videocard owner would pair with a 250W GPU).
While your spreadsheet is useful in theory, it calculates efficiency per FPS. This is not really relevant in the real world imo. If a person can afford a $350 HD6970, for example, it makes little difference to them how much more efficient HD5770 is since they won't buy a $100 videocard in the first place.
Also, if you don't want to deal with CF/SLI game profiles, there is no way you'll purchase an HD6850 CF setup, regardless how efficient it is.
Also, while a GTX480 is rated #15 for efficiency vs. #11 for the HD5850, your chart implies that the GTX480 isn't that bad for efficiency. But the GTX480 cost almost double what the HD5850 cost on release date. Frankly, the HD6970 is ranked worse, but the GTX480 is so much louder, its efficiency parameters are largely irrelevant in that comparison. I only provide these examples to provide an explanation why efficiency in terms of costs is generally not considered before other more important factors.
If you do care so much about electricity costs of a GPU, you really need to get your priorities straight (and I mean that in the nicest way possible). The irony here is that most of us run overclocked GPUs which by themselves easily gain 150W+ in power consumption at load. So it's somewhat hypocritical for most of us to complain about GPU power consumption while running Phenom II X4/X6 and i5/i7 processors overclocked to the moon.
For instance, if you have time to game 6 hours a day (completely unrealistic), it's about time to think about getting a real job. And if you can afford to game 6 hours a day (i.e., say you work from home or have business ventures which provide you with sustainable cash flow), then you are also probably making a decent amount of $ to not care about electricity costs. I realize, you can control your electricity costs by buying more efficient videocards. But who runs their GPU for 6 hours a day x 365 days a year to play videogames? A student maybe. In that case, your undergrad tuition and book costs are so high, the last thing in the world you care about are electricity costs.
Basically the amount of electricity your videocard consumes is peanuts compared to the sum of all the other appliances you probably use on a daily basis. And frankly if $10-20 a month makes a huge difference, I strongly suggest and encourage getting a higher paying job before sitting there and trying to count 50 cents a day saved on electricity.
I never understood the importance of" 1 penny saved is 1 penny earned". I believe that you get rich from capital gains and cash flow generating assets (whether its your human capital asset - i.e, earning a higher salary at a job, or your physical asset that generates cash flow - such as a rental property). Savings does not equal new cash flow and therefore contributes little to wealth. In other words, instead of trying to save $0.50 a day, try to get a job that pays $30-40 more a day. But that's just my thinking.
In conclusion, I would say the greatest importance of GPU efficiency to me are reduced heat and noise levels, as has been mentioned by many posters here. :thumbsup:
Agreed.Unfortunately, it's easy to make charts say whatever we want. In the restricted variables you compare, the results come out as they do. (As for my cards, I use them for more than gaming. But even in gaming, they shouldn't be on your list at the paltry 1920 res.) Regardless, I was only arguing what I feel is a misplaced emphasis on power costs.
Six hours is not unrealistic.
Some people game 12 hours a day.
I could've just as easily chosen 5 minutes or 24 hours a day for these numbers and you'd still see the differences. The numbers would just be that much more extreme in what they represent.
Finally, I would like to point out that if you find value in performance reviews, you'll also probably find value in baseline statistics; if not now, then eventually. These charts aren't for everyone, but it is my hope that more people will realize the value of efficiency and how it can affect your life - even in your gaming.
Thanks for the effort OP. I didn't go in detail as to the accuracy of your calculations.
But here is a much quicker assessment:
A mid-range videocard = 150W
A high-end videocard = 250W
Difference of 100W
100W x 365 days a year x 6 hours a day at 100% load x $0.15/kWh = $32.85 per annum or less than $3 a month.
Ok what about idle power consumption? At idle, most modern GPUs are within 15-20W of each other (as has been noted in this thread), far less than the difference between modern processors that a high-end GPU gamer would have: i.e., Core i7 920-950, Core i5 750, 2500k or a Phenom II X6 1100T at idle (likely the type of CPUs that a high-end videocard owner would pair with a 250W GPU).
While your spreadsheet is useful in theory, it calculates efficiency per FPS. This is not really relevant in the real world imo. If a person can afford a $350 HD6970, for example, it makes little difference to them how much more efficient HD5770 is since they won't buy a $100 videocard in the first place. Also, if you don't want to deal with CF/SLI game profiles, there is no way you'll purchase an HD6850 CF setup, regardless how efficient it is. Also, while a GTX480 is rated #15 for efficiency vs. #11 for the HD5850, your chart implies that the GTX480 isn't that bad for efficiency. But the GTX480 cost almost double what the HD5850 cost on release date. Frankly, the HD6970 is ranked worse, but the GTX480 is so much louder, its efficiency parameters are largely irrelevant in that comparison. I only provide these examples to provide an explanation why efficiency in terms of costs is generally not considered before other more important factors.
Similarly, someone who can afford a $500 GTX580 or a $500 HD6950 CF setup shouldn't be trying to save $5 on electricity costs a month (after all gaming is a very cheap hobby compared to other hobbies in life). If you do care so much about electricity costs of a GPU, you really need to get your priorities straight (and I mean that in the nicest way possible). The irony here is that most of us run overclocked GPUs which by themselves easily gain 150W+ in power consumption at load. So it's somewhat hypocritical for most of us to complain about GPU power consumption while running Phenom II X4/X6 and i5/i7 processors overclocked to the moon.
For instance, if you have time to game 6 hours a day (completely unrealistic), it's about time to think about getting a real job. And if you can afford to game 6 hours a day (i.e., say you work from home or have business ventures which provide you with sustainable cash flow), then you are also probably making a decent amount of $ to not care about electricity costs. I realize, you can control your electricity costs by buying more efficient videocards. But who runs their GPU for 6 hours a day x 365 days a year to play videogames? A student maybe. In that case, your undergrad tuition and book costs are so high, the last thing in the world you care about are electricity costs.
All in all, lets take a look at what typical household appliances use:
1) Clothes Dryer - 2,790 W (if you have a family, you are using this very often for 40 min at a time, if not more a week).
2) Coffee Maker - 1,200 W / or Kettle (probably using this 5-6x a week)
3) Oven - 12,000 W (if not you are eating out, which costs more than cooking at home if you are eating healthy food)
4) Microwave Oven - 1,200 W+
5) Toaster - 1,100 W
6) Water Heater - 2,500 W
7) Television - 200-500 W
Let's not even pretend your GF, mother, wife doesn't use a 1000 W hair dryer on a weekly basis! What about a hair straightener? Think about that - 15 min of a blow-dryer x 5 workdays a week @ 1000W. Go ahead and tell your wife to save on electricity by drying her hair for 1 hour on the balcony - see how that will fly.
Basically the amount of electricity your videocard consumes is peanuts compared to the sum of all the other appliances you probably use on a daily basis. And frankly if $10-20 a month makes a huge difference, I strongly suggest and encourage getting a higher paying job before sitting there and trying to count 50 cents a day saved on electricity. I never understood the importance of" 1 penny saved is 1 penny earned". I believe that you get rich from capital gains and cash flow generating assets (whether its your human capital asset - i.e, earning a higher salary at a job, or your physical asset that generates cash flow - such as a rental property). Savings does not equal new cash flow and therefore contributes little to wealth. In other words, instead of trying to save $0.50 a day, try to get a job that pays $30-40 more a day. But that's just my thinking.
Honestly you are better off drying your clothes outside/on a rack than using your dryer, or take your bike/rollerblade to a convenience store 10 min away instead of driving. All of these changes in your life will bring far greater benefits than trying to save money on energy costs from GPUs.
Should we also stop using cell phones (smartphones usually require charging every day) and switch to a landline to save costs from charging our wireless devices?
In conclusion, I would say the greatest importance of GPU efficiency to me are reduced heat and noise levels, as has been mentioned by many posters here. :thumbsup:
/sigh
Once more and then I'll concede this to contextually challenged, cognitive bias and endless internet debate.
If you don't like what's being portrayed in the spreadsheets, change your mind and accept it or disprove the calculations.
I'll provide you with a pretty easy comparison for why I created the spreadsheets, since this appears to be completely missed as of late.
HD 6850 CF is a high-end GPU setup (average of all benches = 84.31 FPS)
GTX 580 SLI is a high-end GPU setup (average of all benches = 111.56 FPS)
The cost per year to OPERATE AND PURCHASE HD 6850 CF @6 hours a day (full load) with 2 hours idle, 300/365 days/year @$0.21/kWh = $484.28
The cost per year to OPERATE AND PURCHASE GTX 580 SLI @6 hours a day (full load) with 2 hours idle, 300/365 days/year @$0.21/kWh = $1,164.19
Per year, a difference of $679.91, or per month, $56.66. All to receive an extra 27.25FPS average over all benchmarks provided (well above 60FPS).
You pick multi-GPU setups and talk about "cost"?
Funny...
Once more and then I'll concede this to contextually challenged, cognitive bias and endless internet debate.
Per year, a difference of $679.91, or per month, $56.66. All to receive an extra 27.25FPS average over all benchmarks provided (well above 60FPS).[/COLOR]
Right, but you didn't need to compute the electricity costs to conclude that a GTX580 SLI setup for $900-1000 is overkill for 1920x1200 and represents poor value. We already know that from a performance per dollar perspective a GTX580 SLI setup is worse value than an HD6850 CF setup. So by you concluding that the total cost of ownership for the GTX580 SLI is much greater vs. an HD6850 CF setup is just stating the obvious since $900 of that cost is the cost of the videocards (not incremental electricity!).
In fact, based on your linked Anandtech benches, a single $220 HD6950 1GB achieves an average of 60-61 fps. So what incentive would someone have buying a faster videocard than that from your chart? I thought the point of your chart was to single out efficiency in terms of Electricity costs (which is what we care the most about in this thread, right?). YetI can't quickly determine this because nowhere does it tell me how much extra it would cost for me to operate an HD6970 over say an HD6850. I can do this task "manually" by subtracting Column J from Column L. You should probably provide this as a standalone column.
Let's just look at 1 example from your data:
HD5770 ($147.46 Total cost - $110 purchase price = $37.46 in annual electricity costs).
HD5870 ($425.95 Total cost - $380 purchase price = $45.95 in annual electricity costs).
So you see how there is only an $8 difference per annum between owning a high-end 5870 vs. a low-end HD5770? So how are you showing that Electricity costs are so substantial in the total cost of ownership? Looks to me like that $380 purchase price for the 5870 is completely scewing your findings in favor of the HD5770 from FPS / $ perspective. Not to mention you could have picked up an HD5870 for $180-200 4-5 months ago.
I think you should create 3 charts: (1) showing ONLY the total electricity costs - this is required when trying to show how in/efficient each card is in terms of power consumption; (2) FPS / $ graph to show which card provides the best value from a performance per $ perspective; (3) Combine the electricity costs + cost / FPS. This way, we'll be able to use any one of those 3 charts and isolate for electricity costs.
Can you elaborate how are you deriving the Upgrade costs every 2 years? This would be extremely difficult to do since it doesn't take into account reselling your previous hardware (A GTX580 will sell for a higher price than a GTX460. As a result, it makes it difficult to determine the future cash outlay for that user).
To summarize, I understand why you clump together purchase price + electricity cost. But you should separate the electricity cost entirely. We want to know how in/efficient a particular card is from a power consumption perspective. My main point is that you are including the purchase price -- but this component greatly reduces the value proposition for high -end cards (which most of us understand since you pay top dollar for best performance) - thus this screws the results unfavourably for high-end cards like the HD6970/GTX570/580 and at the same time tells us little about how energy efficient they are relative to each other.
If you can alter your excel sheet to isolate specifically for electricity costs, that would get the point across much better.
The primary focus of the charts are to show how efficiently your money is being spent compared to what else is on the market.
Resale is not considered for the upgrade frequency. That would be difficult to determine IMHO. Especially after 4 years of ownership...
I could generate an extra column to include a generalized resale value on each frequency of ownership, but like I've emphasized against before, this will cause the data to be subjective and not actual.
I'm trying to keep the charts as unbiased as possible so if you have your own considerations or information to interpret, then you can use the variables on the charts as baseline.
These charts took a tremendous amount of time to create, and this is only a hobby.
Yep, that pretty much nullifies the whole electricity costs argument. If you can't afford an extra 10-15 bucks a year for electricity then you can't afford a $300 video card to begin with.But generally speaking, it appears that for people already buying $200+ videocards, the cost of electricity among mid-range and high end cards is only about $10-15, which is pretty minor considering some of these videocards cost $250-500 at launch!
Yep, that pretty much nullifies the whole electricity costs argument. If you can't afford an extra 10-15 bucks a year for electricity then you can't afford a $300 video card to begin with.
TDP only makes sense from the point of view of noise given hotter cards tend to be louder than colder cards. Youll be affected by the extra noise every time you game, so it becomes very relevant in that case.
Yep, that pretty much nullifies the whole electricity costs argument. If you can't afford an extra 10-15 bucks a year for electricity then you can't afford a $300 video card to begin with.
TDP only makes sense from the point of view of noise given hotter cards tend to be louder than colder cards. Youll be affected by the extra noise every time you game, so it becomes very relevant in that case.
Before I get into the real math and figures, I will say that in a year's time, the differences in costs are pretty substantial. Substantial like the fact that we critically argue how crucial initial costs of CPUs and GPUs per their performance marks, but when power efficiency comes into play we seem to forget operational cost per performance. If the video card you just bought for $30 cheaper performs even on par with another, but is significantly inefficient with its watt usage and heat generation in comparison and you'll spend that $30 you saved over the next 6 months - then more in the following months - wouldn't that be something you'd consider when originally purchasing it? Don't think energy costs that much? I dare you to continue reading.
Great!!
Now you can clearly see from your charts that the power costs between a low-end (HD5770) and a high-end card (HD6970) is very small: $37 vs. $49
Also the difference between the HD5870 and GTX480 is also pretty small: $46 vs. $61
In other words, most of the difference comes from Price of Card/Setup vs. FPS achieved.