MSI GeForce GTX 970 GAMING 4G Video Card Review @ [H]

yeah your power consumption readings make no sense and conflict with every other 970/980 review. the 8 pin also has nothing to do with power consumption. it has the same tdp as every other 970 +/- a few watts for differing clocks.
 
Last edited:
misterbobby said:
So a 145 watt card is using 252 watts out of the box? Yes I know its clocked higher than reference but that would make Maxwell a joke which is not true. You are literally pulling nearly 100 watts than my 970 at the SAME clocks.
Click to expand...

The voltage is also higher on the MSI 970 GAMING version I believe, 1.206v at default. I have a feeling a reference 970, or stock 970 would have a lower voltage. Also, this card does have an 8-pin and 6-pin power connector, versus two 6-pin on the reference design. That's a lot of power potential.

I have triple checked my numbers, they are correct.
 
Brent_Justice said:
The voltage is also higher on the MSI 970 GAMING version I believe, 1.206v at default. I have a feeling a reference 970, or stock 970 would have a lower voltage. Also, this card does have an 8-pin and 6-pin power connector, versus two 6-pin on the reference design. That's a lot of power potential.

I have triple checked my numbers, they are correct.
Click to expand...
Well you might be reading the numbers correctly but there is something wrong somewhere causing such numbers in the first place. My voltage is even higher at default of 1.225 and when I oc its at 1.25. Even at 1525/8000 and 1.25 I am not seeing but a hair over 300 watts. Yet again my 780 at basically stock boost was using at least 70 to 90 watts more.

An 8 pin does not mean it will actually use more power. It makes zero sense to see the 145 watt 970 using more power than the 250 watt 780. Again no one on the net is coming up with such bizarre power usage. If your numbers were accurate then Maxwell would be a joke.
 
Bought the ASUS 970 because of it's good reviews in OC and temp areas, even has the same general design, the power off fans until needed, and a back plate. I figure it'll be about the same and was 10 bucks less than the MSI when I got it (339). These are bound to be some killer cards, can't wait to get mine set up! Great review, I was waiting so eagerly for the [H] 970 write up
 
Darksword said:
Thanks for the review and substantiating the throttling issue I have with my MSI Gaming 970's. I think I will try leaving the voltage at stock instead of +40. That may still allow me to run at 1300 core (1560 boost) without down throttling.

Any chance MSI will release a BIOS update to allow a higher TDP limit?
Click to expand...

In most cases, the card makers will max out the allowed power limit allowed by NVIDIA and nothing beyond that. Therefore, I suspect you'll need an unofficial BIOS for moar power.
 
Armenius said:
I disagree. At this price point it matches Ti-levels of performance in quite a few games. More importantly, minimum framerates are much improved over the 780, providing smoother gameplay and a better experience. Even better, frame pacing in SLI is subjectively perfect, while the 780s still had some issues on that front. At $700 I think that the upgrade from 780 SLI to 970 SLI was well worth it for me. I have a ~20% increase in performance while using a lot less power, and the experience is smoother to boot.

Now if these cards were $400+ a piece, then I would agree with you. At $349 a piece it's hard to resist.
Click to expand...

Good to hear about the performance improvements. A 780 is overkill for those of running at 1080p (which is a lot of us, see Steam surveys). Obviously it depends on if you run super high resolutions or multi-monitor.

Regarding power consumption, page 10 of the review says the 780 uses less power at load. Is this accurate?
 
biggles said:
Good to hear about the performance improvements. A 780 is overkill for those of running at 1080p (which is a lot of us, see Steam surveys). Obviously it depends on if you run super high resolutions or multi-monitor.

Regarding power consumption, page 10 of the review says the 780 uses less power at load. Is this accurate?
Click to expand...
A 780 is most certainly not overkill at 1080 if you want 60 fps in demanding games.
 
misterbobby said:
Well you might be reading the numbers correctly but there is something wrong somewhere causing such numbers in the first place. My voltage is even higher at default of 1.225 and when I oc its at 1.25. Even at 1525/8000 and 1.25 I am not seeing but a hair over 300 watts. Yet again my 780 at basically stock boost was using at least 70 to 90 watts more.

An 8 pin does not mean it will actually use more power. It makes zero sense to see the 145 watt 970 using more power than the 250 watt 780. Again no one on the net is coming up with such bizarre power usage. If your numbers were accurate then Maxwell would be a joke.
Click to expand...

The testing scenario can be different. I am using real games to find the peak wattage. I use the same game's, the same spots in the game's, to sample the wattage. Your method of finding the wattage may be different, and different games do have different demands on wattage. Also different ways of playing the games do create different loads. So I watch the power meter across all the games we use.
 
Bout Time:D anyway as usual worth the wait good review guys.when i buy a 970 im getting this one.
 
Brent_Justice said:
Note I used a reference GTX 780 for comparison. My actual clock speed while gaming with it was 1006MHz.

I have verified the total system wattage, it is what I reported in the review between both video cards.
Click to expand...

Well then the obvious question would be: did you verify the actual clock rates during the power draw testing? The reference 780 throttles both the voltage and the clock rate when it reaches 80 degrees. If it throttled significantly, then that could perhaps be the key to making sense of those results.
 
Great review of the particular card.

But where was the valid apples to apples comparison?

Couldn't you have thrown in the MSI gaming version of the GTX780 and the R9 290?

Why even bother comparing the gaming version to the bone stock versions of the competitors? Unless you are trying to favor the card you are reviewing?

There is no question that the 970 would still have beat the R9 290 in most if not all cases. But you would at least have been giving your readers a much more honest comparison. I wouldn't have bothered throwing in the 780 numbers as that card is discontinued.

Anyway, great card review. You guys always give the best o/c details.

However, please start showing more valid apples to apples comparisons in your video card reviews. You should be above showing bias. You are already better than all of the other review sites.

Meh, my two bits. Take 'em for what they are worth.
 
Brent_Justice said:
The voltage is also higher on the MSI 970 GAMING version I believe, 1.206v at default. I have a feeling a reference 970, or stock 970 would have a lower voltage. Also, this card does have an 8-pin and 6-pin power connector, versus two 6-pin on the reference design. That's a lot of power potential.

I have triple checked my numbers, they are correct.
Click to expand...
The ASUS STRIX uses 1.200v out of the box, for reference. Since there are no reference 970s I guess you would have to get a look at NVIDIA's manufacturer specifications to see what "stock" voltage is.

biggles said:
Good to hear about the performance improvements. A 780 is overkill for those of running at 1080p (which is a lot of us, see Steam surveys). Obviously it depends on if you run super high resolutions or multi-monitor.

Regarding power consumption, page 10 of the review says the 780 uses less power at load. Is this accurate?
Click to expand...
You're right, not overkill at all for 1080p/144Hz. In my experience thus far I would say that 970 SLI is the sweet spot for this resolution and refresh rate. I've been getting 115-146 FPS in most games without needing to compromise graphical settings, while I was getting 80-110 FPS with 780 SLI.

At 60Hz, a single GTX 780 will do at 1080p. As this review shows, a single GTX 970 can be good at 1440p/60Hz. I guess most people would still be on a 60Hz monitor regardless of panel type.
 
No 4K benchmarks?

I didn't see it mentioned in the article, but does the improved cooler impede the SLI bridge in any way?
 
specops said:
Well then the obvious question would be: did you verify the actual clock rates during the power draw testing? The reference 780 throttles both the voltage and the clock rate when it reaches 80 degrees. If it throttled significantly, then that could perhaps be the key to making sense of those results.
Click to expand...

Yes

Guys, my numbers are correct, I've verified them, it is what it is. I'm moving on from this topic.
 
DeepS6X said:
Great review of the particular card.

But where was the valid apples to apples comparison?

Couldn't you have thrown in the MSI gaming version of the GTX780 and the R9 290?

Why even bother comparing the gaming version to the bone stock versions of the competitors? Unless you are trying to favor the card you are reviewing?

There is no question that the 970 would still have beat the R9 290 in most if not all cases. But you would at least have been giving your readers a much more honest comparison. I wouldn't have bothered throwing in the 780 numbers as that card is discontinued.

Anyway, great card review. You guys always give the best o/c details.

However, please start showing more valid apples to apples comparisons in your video card reviews. You should be above showing bias. You are already better than all of the other review sites.

Meh, my two bits. Take 'em for what they are worth.
Click to expand...

I had less than a week to do this review, again, without rest and sleepless nights. I squeezed in everything I could.

There will be follow-up articles that cover things we did not in the original reviews. There will also be future retail card reviews that will have comparisons by price, as usual. This is not the only 970 review we are ever going to do.
 
Brent_Justice said:
Yes

Guys, my numbers are correct, I've verified them, it is what it is. I'm moving on from this topic.
Click to expand...

FWIW, I'm seeing exactly the same thing. For my entire system measured at the wall, 100W idle, 345W on Furmark, 430W playing Metro:2033(Redux). My card is overclocked at 1400 GPU, 8000 VRAM, 110% power, no voltage bump.

The MSI Gaming model is specifically designed for overclocking. If you want the most power-efficient 970, I think the Asus Strix is your best bet.
 
Quartz-1 said:
No 4K benchmarks?

I didn't see it mentioned in the article, but does the improved cooler impede the SLI bridge in any way?
Click to expand...

If you are going to run 4K I suggest the 980 or 980 SLI. 4K is very demanding, reserved for the flagship single GPU from AMD and NVIDIA, or SLI and CrossFire. No single card can yet max out 4K.

In 2x SLI there's room for a hard connector or flex connector just fine, there is an indention for one of the SLI fingers.

For 3x SLI you are going to need two flex connectors, a hard connector doesn't look like it would work with the raised heatsink shroud covering one of the fingers.
 
BTW, about the power numbers, last thing I will say. We have an ASUS 970 STRIX to evaluate, I will make sure we have both the MSI power numbers and ASUS power numbers in that review to compare side-by-side the cards and see which is more power efficient, along with the 780.
 
Brent_Justice said:
If you are going to run 4K I suggest the 980 or 980 SLI. 4K is very demanding, reserved for the flagship single GPU from AMD and NVIDIA, or SLI and CrossFire. No single card can yet max out 4K.

In 2x SLI there's room for a hard connector or flex connector just fine, there is an indention for one of the SLI fingers.

For 3x SLI you are going to need two flex connectors, a hard connector doesn't look like it would work with the raised heatsink shroud covering one of the fingers.
Click to expand...


I have 3 MSI 970's coming in tomorrow, I'll be able to confirm this.
 
Morphish said:
FWIW, I'm seeing exactly the same thing. For my entire system measured at the wall, 100W idle, 345W on Furmark, 430W playing Metro:2033(Redux). My card is overclocked at 1400 GPU, 8000 VRAM, 110% power, no voltage bump.

The MSI Gaming model is specifically designed for overclocking. If you want the most power-efficient 970, I think the Asus Strix is your best bet.
Click to expand...
Then that is jacked up. I just checked and running the same clocks you are with my non Gaming MSI 970, I am showing 280 watts peak in Metro LL and 2033. How on earth could you be pulling 150 watts more than me? Thats also about 50 watts more than I pulled with an MSI 780 Gaming card. If what you are pulling is accurate then the 970 Gaming card is the most inefficient pos out there.

EDIT: I see you have an 8350 at 4.4 which is a hog compared to my cpu so that would explain probably about 100 watts.
 
Last edited:
misterbobby said:
Then that is jacked up. I just checked and running the same clocks you are with my non Gaming MSI 970, I am showing 280 watts peak in Metro LL and 2033. How on earth could you be pulling 150 watts more than me? Thats also about 50 watts more than I pulled with an MSI 780 Gaming card. If what you are pulling is accurate then the 970 Gaming card is the most inefficient pos out there.
Click to expand...

He's also running a completely different system than you are.

Comparisons of wattage between different CPUs, power supplies (as efficiency factors into wall draw) and even area tested within a game can yield significantly different power results.
 
David_Schroth said:
He's also running a completely different system than you are.

Comparisons of wattage between different CPUs, power supplies (as efficiency factors into wall draw) and even area tested within a game can yield significantly different power results.
Click to expand...
Yeah I noticed that afterwards and edited my post. :eek:
 
misterbobby said:
Then that is jacked up. I just checked and running the same clocks you are with my non Gaming MSI 970, I am showing 280 watts peak in Metro LL and 2033. How on earth could you be pulling 150 watts more than me? Thats also about 50 watts more than I pulled with an MSI 780 Gaming card. If what you are pulling is accurate then the 970 Gaming card is the most inefficient pos out there.

EDIT: I see you have an 8350 at 4.4 which is a hog compared to my cpu so that would explain probably about 100 watts.
Click to expand...

Yeah.... it's normally at 4.4Ghz with a 4.8Turbo, but I turned off the CPU overclock to get the power numbers. It's, well, not the most power efficient CPU out there. :)

However, the Furmark numbers are interesting to me since it represents maximum GPU load with the CPU nearly idle. What is your Furmark delta? Mine is 245W, which I don't mind at all given the ease of OC and the low temps. I might feel differently if I wanted to run more than one.
 
Good review, as usual.
I'm really close to moving from my GTX770 to this card

Spelling Nazi comment:
On the last page, under Gaming Summary
The GTX 780 and AMD R9 290 could come anywhere near that level of performance.
Click to expand...

I believe you mean couldn't.
 
What! THIS IS FALSE ADVERTISING! The name clearly says that it had 4G connectivity yet I see no mention of LTE!
 
Nice review guys. If I was in the market for new cards the 970's would be very temping. But as with the 980's I'll pass and wait for the next generation. I'm quite happy with my rig as it sits today.
 
Brent_Justice said:
Yes

Guys, my numbers are correct, I've verified them, it is what it is. I'm moving on from this topic.
Click to expand...

So yes, you verified the clock rates for the GTX 780 but no, you don't want to talk about it or tell us what they were? Okay, well that's... interesting :)
 
The carda we got here in Finland seem to have pretty serious coil whine issues. How was the reviewed card?
 
One important thing, if you buy this card - be very, very cautious, when removing sticker from fans, because you may ruin your card.

This is Official MSI rep reply regarding fans ramping up to 100% from Official MSI forums:

"We discussed this internally and the fan issue (fan2 @100%) is most likely caused by the sticker.
Many fans can withstand the removal of the sticker without any damage but a small amount of fans seem to get damaged.
Our production has already stopped using these stickers.

So we recommend all users that suffer from this issue to return the card to the point of purchase for warranty service."
 
specops said:
So yes, you verified the clock rates for the GTX 780 but no, you don't want to talk about it or tell us what they were? Okay, well that's... interesting :)
Click to expand...

Page 2 of the review states what the real-time clock frequency of the 780 is. Look at the system table. This information was there when the article was published.
 
MorgothPl said:
One important thing, if you buy this card - be very, very cautious, when removing sticker from fans, because you may ruin your card.

This is Official MSI rep reply regarding fans ramping up to 100% from Official MSI forums:

"We discussed this internally and the fan issue (fan2 @100%) is most likely caused by the sticker.
Many fans can withstand the removal of the sticker without any damage but a small amount of fans seem to get damaged.
Our production has already stopped using these stickers.

So we recommend all users that suffer from this issue to return the card to the point of purchase for warranty service."
Click to expand...
Wow... So a piece of tape that (ripping # out my ass) probably cost them $0.02 in materials and labor to apply is going to cost them thousands in warranty costs.
 
Brent_Justice said:
Page 2 of the review states what the real-time clock frequency of the 780 is. Look at the system table. This information was there when the article was published.
Click to expand...

Brent, I appreciate your work and the brilliant articles that you make, but please: give me some credit.

I could write a book on the issue of GPU Boost 2.0 induced throttling, and the first sentence would be: whatever frequency you think your GPU is running at, it's not until confirmed by full realtime monitoring data through all the testing. If you are seriously going to say with a straight face that the max boost clock that you put in that system table is the one and only clock rate that your reference GTX 780 (which has a temp throttling point of 80 degrees and which you state is hitting that temp in your temperature testing table) runs at through all your testing, then I know you are not telling the truth.

I don't think and I don't want to believe that you don't know the scope to which the reference GK110 cards are designed to throttle their clock rate and voltage to keep within the very strict noise and temp limits set by Nvidia.

Just so that we are clear: it's not just the power draw that this throttling is affecting, it's an actual performance issue too. I have personally done thorough testing comparing a reference 780 to a reference 290X and despite the popular belief, the 780 throttles a lot more often and more severely than the 290X. That is why I find it interesting that [H] made a habit of using a custom model to represent the 290X performance but didn't give the 780 the same courtesy.
 
Great review. Am considering a 1440p monitor and would require a new vidcard to power it. This card seems to fit the bill perfectly.
 
Here is a Shadow of Mordor report for anyone curious. I have two of these cards and even fully maxed with Ultra textures it "only" uses 3500MB of VRAM. I ran the benchmark yesterday and was disappointed to be getting about 95 fps. I then changed the SLI profile to Fear 3 and am now getting 165. I went in and used the built in ability to run the game at a higher res and then downsample to test 4K speeds. I was still averaging 60FPS at 4K with the SLI working. This is a great card for the money.
 
You must log in or register to reply here.
Back
Top