GTX 750ti vs GTX 960 power consumption at low to medium loads

[riesenkamell]

Dear Tom's Hardware Community,

I have a very specific question for buying a new graphics card that is supposed to be mainly focused on power consumption compared to performance. Since my PC runs many hours a day and I live in Germany (high energy cost), power consumption is my main concern after performance/price, so I have already decided on one of the new Maxwell cards.

The question I have is if the two cards I'm considering use vastly different amounts of energy or roughly the same when faced with similar tasks. Specific tasks for me are around 50% office, 20% multimedia and 30% gaming. Gaming is mostly low graphics, Dota2, Civ 5, EU 4, Total War Rome at 1080p.

All the information I could find so far was about "idle" and "blue ray" power consumption (which was roughly the same for both) and "gaming", which was obviously vastly different since they max out the cards in these tests. What I am interested in is the "real life" power consumption with the same applications running at the same settings, one card running at 50-100%, the other probably at 25-50%.

Since the initial price of the cards does not differ that dramatically (especially considering the difference in performance) and I do not really plan on buying new parts for at least a couple of years, the 960 would seem the right choice to keep options open regarding new games, a possible new monitor or the usage of that NVidia DSR technology specifically aimed for MOBA players.

On the other hand I know that the 750ti will be more than enough for all my needs at the moment and probably for a few years to come and if I'm not only paying more initially for the 960 but then it also uses considerably more power in normal usage, it would probably not be worth it just to have some headroom since I could always upgrade again in a few years if need be.

Basically it comes down to real life power consumption in my specific case. The 750ti is enough but I feel it's overpriced at the moment compared to the 960 and it doesnt give me much room to play with. If it was to use roughly the same amount of power in my specific case, I would lean towards the 960, if not I would go for the 750ti for now and upgrade in a couple of years if the need arises.

Thanks to everyone who can help with knowledge or experience, I really couldn't find anything to answer that question for me even though I have researched for a long time now.

Have a nice day

 

[Mr5oh]

I just recently added a 750 TI card to my Media PC. I tested draw from the wall, with my G3258 max draw during 3DMark11 was 50 watts. Once I added the 750 TI to the same setup, the draw was 114 watts. My desktop machine 4690k with a GTX 760, draws 249 watts for the same test.

 

[jdw_swb]

GTX 960 review from Guru3D

 

[riesenkamell]

Guys PLEASE, I appreciate you trying to help but you are not giving me any new information and are not properly answering my specific question.

Card TDP is roughly resembling maximum power draw. I know the 960 is much higher than the 750ti (I specifically mentioned this in my post). I also do know the 750ti has a TDP of 60W.

I would appreciate if you refrained from answering questions no one asked. It helps no one and just clutters space and wastes time, would be nice for you to actually READ the question and only answer if you can provide something useful.

Thank you very much

 

[Mr5oh]

Well I hate to further not be helpful, but I do like cluttering up threads... However I think you are asking some very exact information, that not many people have the capability of measuring. You are going to have to sort of infer from results that are available.

So at idle, my media machine was drawing 31 watts, with the addition of the 750 TI this jumped to 38 watts. Meaning the 750 TI added 7 watts at idle.

My 4690k with GTX760 draws 61 watts at idle. My previous machine 3570k with a GTX 570 HD drew 96 watts at idle (107 overclocked).

 

[17seconds]

There are six different charts here covering different scenarios.
http://www.techpowerup.com/reviews/Gigabyte/GTX_960_G1_Gaming/26.html

 

[riesenkamell]

Ok, so maybe to help you, Mr5oh and others help me let's recapitulate what we do and what we do not know so far:

-both cards draw around 5-10W at Idle
-the difference in power consumption at full load is around 30-50W with the 960 being much more capable

The truth I'm looking for is somewhere in the middle and the question is whether in this specific case and my specific user profile I will be looking more at 5-10W more on average or 30W more on average, which depends on how power consumption scales. Some people must know something about this and even though it might seem trivial, given my usage we are probably talking half the price of the card itself over the duration of its usage so this question will probably tip the scales in my specific case.

The question is not made easier by all the reviews online which give differing numbers based on model vs reference cards and so forth. For example in the link provided above by jdw_swb it seems like max. power consumption differs by around 30W, whereas in other reviews it looks more like 60W or more: http://www.techpowerup.com/reviews/MSI/GTX_960_Gaming/27.html

The point for me is that 10W on average in my intended use would not tip the scale whereas 50W definitely would.

P.S.: Coincidence you would link to that very same article at the very same website, 17seconds. I'm afraid they are just using reference numbers for the 750ti here though since the max Wattage of 65 doesn't match the real life cards in the link further above which go more to around 90W. The devil is really in the detail here for me, especially considering it isn't even really max power consumption I'm particularily interested in.

Thanks to everyone for trying though

 

[Mr5oh]

The point for me is that 10W on average in my intended use would not tip the scale whereas 50W definitely would.

I don't have figures handy, but I could take power draw measurements while playing a DVD for both my media machine (750 TI) and my desktop (760). However I don't think this will directly help either, as the machines are very different, and I really don't expect power draw to be much over idle. Much less those measurements would still be the complete machine, and not just the video card.

 

[riesenkamell]

Thanks for the offer but that wouldn't really help much since all the parts in both machines are different and we have no comparison to a 960 card.

If you already have some means of measuring actual power draw and have steam and Dota 2 installed and have some free time or are curious yourself, you could measure the actual power draw of the 750ti when playing that game at 1080 with all details maxed out since this is probably the point where the two cards might differ most in my specific case.

Even then we would be missing the same figures for a 960 and this method is by no means scientific. It might clear up the difference between the two articles linked above as to the actual power draw during gaming (even though the 750ti shouldnt be maxed out in this configuration if you chose VSync to 60 Hz).

Anyway, to me it has over the last two weeks become as much a question of curiosity and principle as it is actually important for my decision which card to buy. If it turns out to be a bit like that feature on Top Gear where they drove a Toyota Prius flat out over their track and had a BMW M3 follow at the same speed, I don't wanna be the sucker driving that Prius

P.S.: By the way, Mr5oh, I would like to apologize for hating on you earlier, those numbers you provided might prove useful after all. Seems like your specific 750ti is drawing around 64W during that 3DMark11 benchmark, which I am assuming is at least close to maxing out the card. This would suggest it is actually closer to the 60W in the techpowerup article than to the 90W in the Guru3D article. Seems weird how they can arrive at such different conclusions. May I ask which model you are using specifically?

 

[renz496]

since you care power consumption so much how about just get that 750Ti.

 

[riesenkamell]

since you care power consumption so much how about just get that 750Ti.

Because first of all I feel like the difference in price (at least in Germany) doesn't come close to being justified by difference in performance. I also don't want to buy a new card in a year or two. The 960 would have me covered for at least some years to come. The question here is whether it is worth the extra and how much I will be paying extra every month on top of it just to have that luxury.

I know this question might not seem relevant to most gamers who actually dont care about power consumption at all but since I went to the trouble of researching and putting together a tailor made machine for myself I want to get that last part right as well.

So the question remains: How big a difference are we talking about in my specific case? This might actually be an interesting question to everyone remotely interested in energy efficiency of their machines, which will only become more important in the future. Researching the whole thing a bit I was actually quite surprised and disappointed about the lack of dependeable data availeable considering they spend hours and many pages on telling you which card produces half a decibel less at 75 cm distance.

I know which question is more important to me...

 

[renz496]

the problem here is you want an answer that is tailored exactly to your need hence regular information from reviewer sites become useless to you. it is something i believe none of us here can provide because none of us can provide something better that the reviewer already give. the only solution i can think of is test both card yourself for a month and then get which one will provide the best bang for buck for your situation. that's assume if you somehow can 'borrow' the card before actually buy it or have friend with exact cards to test it for you.

 

[Mr5oh]

Seems like your specific 750ti is drawing around 64W during that 3DMark11 benchmark, which I am assuming is at least close to maxing out the card. This would suggest it is actually closer to the 60W in the techpowerup article than to the 90W in the Guru3D article. Seems weird how they can arrive to such different conclusions. May I ask which model you are using specifically?

I wasn't worried about power draw, more I just needed a card that would fit in my HTPC case, which very few cards will.

GIGABYTE GV-N75TOC-2GL (Newegg Link)

Since there is no power cable to my video card, I don't have any way to measure the draw of the card directly.

I haven't played PC games in a while, so I'm afraid I don't have any of the modern PC games to test.

 

[riesenkamell]

the problem here is you want an answer that is tailored exactly to your need hence regular information from reviewer sites become useless to you. it is something i believe none of us here can provide because none of us can provide something better that the reviewer already give. the only solution i can think of is test both card yourself for a month and then get which one will provide the best bang for buck for your situation. that's assume if you somehow can 'borrow' the card before actually buy it or have friend with exact cards to test it for you.

Well, there must be people out there who either have the cards in question and use them for the same purposes and could therefore provide real life numbers or people who have deep knowledge of the Maxwell architecture and its differences in the different cards and how that would affect power consumption at different loads (i.e. tech geeks).

Unlike 95% of posters on sites like this one I wouldn't have bothered you guys with this question if it was easily answereable by researching a bit.

So yes, you might be right. This question may not be answereable by anyone on here. We can hope and try though, can't we?

And no, I will not test the cards and return one, I would feel bad about that and I would rather buy one on a hunch if everything else fails. Thanks for the idea though and have a nice day, sorry for being difficult

 

[renz496]

nah. don't worry about it. just giving my honest opinion here

 

[riesenkamell]

Seems like your specific 750ti is drawing around 64W during that 3DMark11 benchmark, which I am assuming is at least close to maxing out the card. This would suggest it is actually closer to the 60W in the techpowerup article than to the 90W in the Guru3D article. Seems weird how they can arrive to such different conclusions. May I ask which model you are using specifically?

I wasn't worried about power draw, more I just needed a card that would fit in my HTPC case, which very few cards will.

GIGABYTE GV-N75TOC-2GL (Newegg Link)

Since there is no power cable to my video card, I don't have any way to measure the draw of the card directly.

I haven't played PC games in a while, so I'm afraid I don't have any of the modern PC games to test.

Thanks and no problem. I just looked at some more numbers from other reviews on your card (or the "Windforce" version of it, not sure if they are exactly the same but they have the same clock rates) and it seems more and more like Guru3D for some reason has the numbers all horribly wrong. Most reviews seem to suggest max power consumption is around the 60W area for the card alone, so I guess I will go with that number for now.

Doesn't really make my decision much easier though since I still have no idea how much of that possible 60W difference between the two cards will actually come into play in my specific case.

I guess if I get no hints in an opposing direction I will have to assume the 960 does use at least some extra power most of the time and it might therefore be a better idea to go for the 750ti for now and upgrade later if needed but my feeling tells me this might not be true at all.

It's really a shame the computer industry doesn't seem to care much about these practical questions and just measures things that are not really compareable over different "tier levels".

 

[Mr5oh]

It's really a shame the computer industry doesn't seem to care much about these practical questions and just measures things that are not really compareable over different "tier levels".

It's because most people don't even care. Most people don't even care about power draw at idle. I ended up using a smart power strip for my speakers, as they draw 25 watts sitting there doing nothing. Not a huge amount, but that's a ton, and quite frankly not excusable (Logitech) for something in "standby" mode. If people cared about power draw and efficiency you wouldn't see thread after thread of people using 750+ watt power supplies. Typically a power supplies aren't the most efficient at a 10% load.

Of course again, not related, more of a rant.

 

[riesenkamell]

It's really a shame the computer industry doesn't seem to care much about these practical questions and just measures things that are not really compareable over different "tier levels".

It's because most people don't even care. Most people don't even care about power draw at idle. I ended up using a smart power strip for my speakers, as they draw 25 watts sitting there doing nothing. Not a huge amount, but that's a ton, and quite frankly not excusable (Logitech) for something in "standby" mode. If people cared about power draw and efficiency you wouldn't see thread after thread of people using 750+ watt power supplies. Typically a power supplies aren't the most efficient at a 10% load.

Of course again, not related, more of a rant.

Unfortunately, you are totally right but I honestly think times may be changing now. It's unbelieveable what kinds of extra power people will blow away just to save a few bucks in initial cost or to get 5% more performance. I'm assuming in my personal case my power usage will go down by about 50% even during the most heavy tasks, which translates to roughly 100W less for the whole system.

If people realized how huge that difference can be, I think they would think a bit more about these things and then maybe I could find those numbers I'm looking for. I'm expecting my new system to refinance itself within a few years while at the same time giving much better performance - the twenty Euro more or less for any particular part that Hardware Review sites like this one make so much out of when reviewing are really nothing compared to that in the long run.

My guess is that at least 50% of readers/customers don't pay their own electric bill and therefore simply don't care - sad but probably true.

/derailment of thread by random rants over?