MSI GeForce GTX 960 GAMING Video Card Review

Yeah, but it was tested against a similarly priced R9 285 running at higher clock speeds from the same manufacturer. Given the 285's 256-bit vs the 960's 128-bit memory bus, I don't think the comparison was unfair.

Well .. it would depend on if the 285 was the target and only thing that would be 100% fair would be for Nvidia to release the 960 without knowing the 285 performance ..as they have took there time tweaking the card and driver to be faster then the 285..

but I feel the 285 was just bate for Nvidia as the R9-370 has yet to be seen.. and the unfair part starts over as they now know what the 370 has to beat.
 
I did not say it was unfair and price had nothing to do with what I said either. I was simply saying that the large factory oc on the card makes it look faster than what it really is against the stock clocked competition.
Not sure what you're getting at. [H] tested an overclocked 285 and an overclocked 960. My comments were only based on the [H] review.
 
I have always thought 960 would be a more gimped version of 970 than a 970 is to a 980, but I have never expected it to be exactly half of a 970, 2/3rds of a 970 would have been my expectations.

I think those who were waiting for this card got burned bad...

Agree completely, moreover, based on "apples to apples" comparisons (which can be stupidly [H]ard to come by (and by apples to apples I mean, "instead of taking the 970 & 980 and testing them SOLELY to other high end cards running 1440P (4K), test them ALSO against other cards running at the friggin' sweet spot of 1080P") I came to the conclusion that a 970 vs my EVGA SSC 760 would be about 20% faster, but an OC 970 would be 30-40%.

A little math and some loose assumptions showed a 960, with OC might offer 10-20% improvement...and that appears to be exactly the case (if that much of an improvement).

In the future it would be helpful to have a comparison (a single chart fer pete's sake) at 1080P with whatever game it is set to Ultra-video-card-crusher-setting. Because, you know, not all of us are running SLI 980's on dual 4K monitors.
 
Thats why I always go to techpowerup. At the end they take the average of every game and compile them on a list per resolution-

http://www.techpowerup.com/reviews/ASUS/GTX_960_STRIX_OC/29.html

On another note of the 960's complete failure, even SLI 960's fails to outperform a single 970-

http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_960_SLI/23.html

Thats a first for Nvidia. SLI of the x60 cards usually easily beats the flagship. 760 SLI beats a 780, 660 SLI spanks a 680, 560 SLI beats a 580......yet 960 SLI can't even beat a 970 let alone a 980. Just sad.
 
Thats why I always go to techpowerup. At the end they take the average of every game and compile them on a list per resolution-

http://www.techpowerup.com/reviews/ASUS/GTX_960_STRIX_OC/29.html

On another note of the 960's complete failure, even SLI 960's fails to outperform a single 970-

http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_960_SLI/23.html

Thats a first for Nvidia. SLI of the x60 cards usually easily beats the flagship. 760 SLI beats a 780, 660 SLI spanks a 680, 560 SLI beats a 580......yet 960 SLI can't even beat a 970 let alone a 980. Just sad.
You need to actually go read the review a little closer instead of just looking at the summary chart. 960 SLI is most certainly faster than the 970 from a pure graphics horsepower standpoint. The reason it's behind in the overall summary is because of games that don't perform well with SLI.
 
And the 780 Ti outperforms the 980 in a few games too, whats the point? The point of an average is just that, across all the games on average 960 SLI performs worse than a single 970.

Actual GPU power, no doubt it's stronger just based on 1024 cores X2 versus 1664 cores. So in synthetics, or certain specific games that don't require much memory bandwidth, yes that 960 SLI will win. But for most games, especially newer titles, the 970 beats it out.
 
You need to actually go read the review a little closer instead of just looking at the summary chart. 960 SLI is most certainly faster than the 970 from a pure graphics horsepower standpoint. The reason it's behind in the overall summary is because of games that don't perform well with SLI.

I did read the review, and there were a few games where it did better than the 970 and even the 980, but its NOTHING like the past were a x60 card sli setup was a performance beast and there were 15 threads on every forum asking "sli for x60 or x70/80 single gpu".

For years its been an option to have two lesser cards that cost less total than the flagship card, performed better, but had the disadvantages of an SLI setup. With the difference between two gtx 960s and a single gtx 970, that is no longer the case. There is no compelling reason to spend more money on two gtx 960s for faster performance in a few games. The bang for the buck in an SLI setup is horrible. If these cards were marketed at $150 each this would be a different discussion, but for the current price, I can see no reason to SLI these cards.

In fact, take a look on that chart at these cards vs the 690. The 960 in SLI can't even beat the 690 (which is of course 670 in SLI).
 
Last edited:
Actually a 690 is two underclocked 680's.

I actually wish the 750 Ti could SLI, those cards are actually little beasts. I ran one overclocked to 1350mhz core and +600mhz memory daily for like 6 months, was right around gtx 660 performance which is not bad for consuming 70w at the wall max loaded. Played most games at 1080p at high to maxxed out settings, could even play metro 2033 on high with 35-50 fps no lag, very high would get 25-35 fps and get some lag, but high isn't bad for a card no longer than the x16 slot it was in lol.
 
Last edited:
And the 780 Ti outperforms the 980 in a few games too, whats the point? The point of an average is just that, across all the games on average 960 SLI performs worse than a single 970.

Actual GPU power, no doubt it's stronger just based on 1024 cores X2 versus 1664 cores. So in synthetics, or certain specific games that don't require much memory bandwidth, yes that 960 SLI will win. But for most games, especially newer titles, the 970 beats it out.
Gee no kidding. His comment was about the graphics power of 960 SLI not SLI itself so that is what I was addressing.
 
Any game that uses moderate or more memory bandwidth, the gtx 960 will struggle with and lose out. This is most games, especially newer titles. Having more raw GPU power is pointless without the proper memory bandwidth to back it up, as proven by this giant flop of a card called the gtx 960.

The gtx 980 is actually bottlenecked by the 256 bit bus, which is why the 780 Ti beats it out in some games. In raw GPU power the 980 easily beats the 780 Ti.
 
Any game that uses moderate or more memory bandwidth, the gtx 960 will struggle with and lose out. This is most games, especially newer titles. Having more raw GPU power is pointless without the proper memory bandwidth to back it up, as proven by this giant flop of a card called the gtx 960.

The gtx 980 is actually bottlenecked by the 256 bit bus, which is why the 780 Ti beats it out in some games.
And that has NOTHING to do with anything I said. AGAIN the ONLY reason the 960 SLI was behind the 970 was because of games with SLI issues. Those SLI issues impact ANY SLI setup not just 960 SLI. It had nothing to do with bandwidth or anything else. I am not defending the 960 as I think its a joke of a card for 200 bucks but I am simply addressing why 960 SLI was behind the 970 in the overall summary.
 
And that has NOTHING to do with anything I said. AGAIN the ONLY reason the 960 SLI was behind the 970 was because of games with SLI issues. Those SLI issues impact ANY SLI setup not just 960 SLI. It had nothing to do with bandwidth or anything else. I am not defending the 960 as I think its a joke of a card for 200 bucks but I am simply addressing why 960 SLI was behind the 970 in the overall summary.

If that was the case the gtx 690 would be performing badly as well. It's not. So, you are wrong,
 
If that was the case the gtx 690 would be performing badly as well. It's not. So, you are wrong,
Are you really that blind and lazy? Look at the review closer. The 690 has SLI issues too in some games and does not really end up all that much faster than 960 SLI.
 
Last edited:
lol thats the point, gtx 960 SLI is more GPU power than a 690, synthetic benches prove that....that means the 690 should never outperform it....so that means, the games that a 690 outperforms 960 SLI, the bandwidth lack was the cause. That was 11 of the 20 games that the 690 outperformed the 960 SLI.
 
lol thats the point, gtx 960 SLI is more GPU power than a 690, synthetic benches prove that....that means the 690 should never outperform it....so that means, the games that a 690 outperforms 960 SLI, the bandwidth lack was the cause. From what I remember that was atleast 5 games that the 690 outperformed the 960 SLI.
Um what? LOOK CLOSER as the single 960 is basically even with a 670 overall. And we know a 690 is just a little faster than 670 SLI. So in other words the 690 should be a few percent faster overall than 960 SLI and thats exactly what it is.
 
Looking at the improvement that this has over the R9 285 I think that this might be the solution for upgrading my HTPC/ living room gaming rig. Though I think that a 970 would be money well spent for like $100 too. Too many choices...
 
Um what? LOOK CLOSER as the single 960 is basically even with a 670 overall. And we know a 690 is just a little faster than 670 SLI. So in other words the 690 should be a few percent faster overall than 960 SLI and thats exactly what it is.

Yet in synthetic testing it shows a single 960 beats out a single gtx 680 by nearly 10% in raw GPU power.

http://www.guru3d.com/articles_pages/msi_geforce_gtx_960_gaming_oc_review,21.html

Therefore 960 SLI should quite easily best a 690 by quite a large margin, but it's actually worse. Why? Because games actually use memory bandwidth, synthetics don't. 960 SLI is simply handicapped from the memory bandwidth, and thats why it performs so poorly....not just because some games don't scale SLI well.
 
Yet in synthetic testing it shows a single 960 beats out a single gtx 680 by nearly 10% in raw GPU power.

http://www.guru3d.com/articles_pages/msi_geforce_gtx_960_gaming_oc_review,21.html

Therefore 960 SLI should quite easily best a 690 by quite a large margin, but it's actually worse. Why? Because games actually use memory bandwidth, synthetics don't. 960 SLI is simply handicapped from the memory bandwidth, and thats why it performs so poorly....not just because some games don't scale SLI well.
Lol you just keep trying dont you? What you just said about some damn synthetic tests changes NOTHING. Learn to pay more attention as what I just told you put an end to this silly argument about 960 SLI.

AGAIN a single 960 is basically even with a 670 overall in that review. And we know a 690 is just a little faster than 670 SLI. So in other words the 690 should be a few percent faster overall than 960 SLI and thats exactly what it is in that SAME review.
 
Seriously? AGAIN, synthetic testing takes bandwidth out of the equation and only shows raw GPU power. In synthetic testing the gtx 960 is nearly 10% stronger than a 680, yet in real games it around 670 performance as you said. Therefore, in raw GPU power, the 960 SLI should EASILY beat the 690 but it doesn't BECAUSE OF THE LACK OF MEMORY BANDWIDTH in real games.

The synthetic test changes EVERYTHING, because it completely proves my point by taking memory bandwidth out of the equation and only showing raw GPU power. You just plain have no idea what you're talking about if you deny that.
 
Anyone found 960 vs 750 Ti benches yet? Just curious since they are similar (128-bit, 2GB) but different (ROPs, shader count).
 
Seriously? AGAIN, synthetic testing takes bandwidth out of the equation and only shows raw GPU power. In synthetic testing the gtx 960 is nearly 10% stronger than a 680, yet in real games it around 670 performance as you said. Therefore, in raw GPU power, the 960 SLI should EASILY beat the 690 but it doesn't BECAUSE OF THE LACK OF MEMORY BANDWIDTH in real games.

The synthetic test changes EVERYTHING, because it completely proves my point by taking memory bandwidth out of the equation and only showing raw GPU power. You just plain have no idea what you're talking about if you deny that.

The "LACK OF MEMORY BANDWIDTH" is exactly what memory compression is for.
 
The third slide saying 120W power is deceptive, since that's the TDP, not the actual power draw. Also, how the fuck does 140W -> 120W mean "2x power savings"?

Dirty trix nvidia.
 
I have a question... if this card had three or more gigs of VRAM would it even have the power to use it? In the instance of Far Cry 4 and watch dogs it seems like it's already on the edge with the high settings.

And have you noticed any coil whine or buzzing? I'm considering buying the MSI for my silent build.

And hi! Nice site... you guys always have solid reviews.
 
Interesting... Yeah, I've been seeing a lot of people complaining about this card only having 2GB.
 
Yes it should have been a 192 bit bus with 3gb, thats the correct match up for the raw GPU power it has.
 
It's a card that would have been worth $250 because it would be between 770 and 780 strength at that point. With the strangled bus the gtx 770 easily beats it out.

As I said before, I really hope a gtx 960 Ti comes out with those specs and price point (probably gonna be $270-280), maybe with 1280 SP's too which would definitely make it atleast 780 performance maybe a hair better.
 
There's been sightings of two possible versions of a 960 Ti; one with 1280 SPs, and one with 1536 SPs. I'm quite sure these will end up being the $250 960 Ti and the $275-$300 960 Ti Boost. There's plenty of performance room for two cards between the $200 960 and the $330 970.
 
Yeah and the first ''sightings'' of the 960 said it had 1280 SP's and 256 bit with 4gb of Vram. I don't trust any of that shit anymore that people say it's going to be, I'll believe it when I see it. I'm simply stating what I think and hope to be good for the price point and power scale.
 
I bought one of these from Newegg, mainly because of the review here. It does what I need it to do (play LoL and WoW) quite well, so I'm happy with the purchase.

One thing I'm not happy about is that I can't redeem the code for the 6 months of XSplit that came with the card, but I'm not gonna knock the card for this issue. Was planning on buying an XSplit license eventually though.
 
I bought one of these from Newegg, mainly because of the review here. It does what I need it to do (play LoL and WoW) quite well, so I'm happy with the purchase.

I also bought a 960, with the high resolution review sealing the deal. Knowing that I can play even the VRAM hogs maxed at 1080p30 is enough for me. I ended up getting the stock PNY card and not the MSI however (my 560 Ti was also from PNY, worked great for the 3 1/2 years I had it, so I went with them again), and to see that it boosts up to frequencies close to that of some of the custom overclocked cards despite the stock cooling just sweetens the deal.
 
Last edited:
I bought one of these from Newegg, mainly because of the review here. It does what I need it to do (play LoL and WoW) quite well, so I'm happy with the purchase.


OK you are happy playing games that a 750 Ti plays easily.....congrats.
 
I bought one of these from Newegg, mainly because of the review here. It does what I need it to do (play LoL and WoW) quite well, so I'm happy with the purchase.

One thing I'm not happy about is that I can't redeem the code for the 6 months of XSplit that came with the card, but I'm not gonna knock the card for this issue. Was planning on buying an XSplit license eventually though.

You should try to send an email to the XSplit customer support and tell them of your issue. Hopefully that can help save you some money :)
 
I'm thinking he could have saved money with a 750Ti as well.

I disagree. When the GTX 750 Ti was all we had last February, it was a tantalizing option for low-power gaming. But now a year has passed, and new games are running all over the 750 Ti, while the GTX 960 is about twice as fast, and has enough power to really push that 2GB ram (effectively about 2.5GB real-world memory space with the new loss-less memory optimizations) to the limit. It's also low-enough power to not be a burden on cooling or PSUs.

Based on this review, I'd pay the extra 60 bucks for twice the horsepower:

http://www.hardocp.com/article/2015/03/10/dying_light_video_card_performance_review/6

You can get a lot more years out of the GTX 960 than the 750 Ti, and eventually even casual games will up their quality settings, so it's not a "wasted" investment.
 
What we were saying, with the games he listed that the 960 did good at, is that a 750 Ti would easily play those games. Yes of course there's newer games out there than the 750 Ti would struggle with, but I was just going off what was listed.

The 960 is about 30% stronger than a 750 Ti, not double at all.

However, it's the very low stock core clock of the 750 Ti and it's hindered stock wattage range thats the problem. With an altered bios I was running 1350mhz on the core and +600mhz on the memory, I had the performance of a gtx 660 Ti which is roughly 10% under gtx 960 performance. I had no problem playing Metro 2033 on high on 1080p, never below 35 FPS and an average of 45fps. Totally smooth play.
Plus, no PCIe connector needed and it still ran super cool + silent. Even fully loaded it drew 70w from the wall with that huge overclock as well. Even without a modded bios I could run like 1200 core clock which gave a huge boost in performance already.
 
What we were saying, with the games he listed that the 960 did good at, is that a 750 Ti would easily play those games. Yes of course there's newer games out there than the 750 Ti would struggle with, but I was just going off what was listed.

The 960 is about 30% stronger than a 750 Ti, not double at all.

However, it's the very low stock core clock of the 750 Ti and it's hindered stock wattage range thats the problem. With an altered bios I was running 1350mhz on the core and +600mhz on the memory, I had the performance of a gtx 660 Ti which is roughly 10% under gtx 960 performance. I had no problem playing Metro 2033 on high on 1080p, never below 35 FPS and an average of 45fps. Totally smooth play.
Plus, no PCIe connector needed and it still ran super cool + silent. Even fully loaded it drew 70w from the wall with that huge overclock as well. Even without a modded bios I could run like 1200 core clock which gave a huge boost in performance already.
Um the 960 delivers a little over 60% more performance so you are both wrong...lol :p

650 ti boost is 10% faster than the 750 ti here

http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_750_Ti/25.html

and the reference clocked 960 is a little over 50% faster than the 650 ti boost here

http://www.techpowerup.com/reviews/Gigabyte/GTX_960_G1_Gaming/28.html
 
Last edited:
The 960 is about 30% stronger than a 750 Ti, not double at all.

I don't know what you're smoking, but the 750 Ti is not in the same ballpark.

http://www.techspot.com/review/946-nvidia-geforce-gtx-960/page3.html

Take a look at EVERY PAGE OF THAT REVIEW. You'll see a curious thing happening: GTX 960 with twice the framerate of the 750 Ti in nearly every game! GTX 960 has playable framerates at maximum, while the 750 Ti does not.

The GTX 960 has twice the ROPs/texture units, and nearly 1.6 times the shaders/TMUs. This means worst-case, it's 60% faster in shader or texture-limited cases, and up to 100% faster when you exercise those ROPs (like say, 1920x1200 Ultra with AA). This is important if you like to turn up eye candy, even in casual games.

It's not limited by the 128-bit memory bus, that much is obvious. The [H] review I linked above agrees with the one from Techspot:

http://www.hardocp.com/images/articles/1425883267pY1DNqXG9i_6_3.gif

The reason people have no idea how much more powerful the GTX 960 is over the 750 Ti is because HARDLY ANYONE INCLUDED BOTH in a review. Even my usual goto Techpowerup "forgot" to include the 750 Ti in their summary charts, for some stupid reason.
 
Last edited:
Back
Top