MSI GeForce GTX 980 Ti LIGHTNING Video Card Review @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
MSI GeForce GTX 980 Ti LIGHTNING Video Card Review - We take the MSI GTX 980 Ti LIGHTNING video card and see how it plays the most demanding games. This video card "engineered to perfection" for extreme overclocking and has a TriFrozr cooling setup. We compare it to the overclocked ASUS GTX 980 Ti STRIX DCIII and an AMD R9 390X CrossFire configuration to find its true value.
 
I cant believe how heavily 390x crossfire got stomped!
I expected it to be ahead on average.

Good review, easy to read and everything I wanted to see addressed.
 
great review. fantastic cooler on this thing! i dig your reviews that include max overclock. thanks for the work Grady and Brent.
 
something interesting of note (though it's just an intellectual curiosity for me):

in GTA5, with the advanced options enabled, the max OC only nets a couple extra FPS. However, when disabling the advanced options the max OC gets you an additional ~5 FPS.

I wonder what is driving that? Seems like with the advanced options enabled, something other than clock speed is limiting the performance.
 
Weird...in your review of the 390 it gets 64fps in BF4 @ 1440p Ultra 4xMSAA. TWO cards get 79fps at the same settings here, isn't that like 25% scaling? That seems awful. Wonder what the heck they did to the drivers this time.

Anyway, good review, personally I would have given the LE version a gold award (which you didn't have to review, I know), perhaps this one a silver as it's so expensive compared to other cards that perform the same at less cost. But, depends on what you are looking for I suppose, to give it a gold award I would think reviewing the LN2 capabilities would be a good way to go. Especially when companies tout an [H] Gold award all over the place, and for good reason!. It feels like the component that is worth the extra cash was left unanswered. Just some feedback for future reviews, anyone else agree? Or am I in left field.
 
Weird...in your review of the 390 it gets 64fps in BF4 @ 1440p Ultra 4xMSAA. TWO cards get 79fps at the same settings here, isn't that like 25% scaling? That seems awful. Wonder what the heck they did to the drivers this time.

They're using DirectX instead of Mantle for the API, so it's going to be less optimized for AMD cards in general.
 
Weird...in your review of the 390 it gets 64fps in BF4 @ 1440p Ultra 4xMSAA. TWO cards get 79fps at the same settings here, isn't that like 25% scaling? That seems awful. Wonder what the heck they did to the drivers this time.

That was overclocked, the not manually overclocked (but slightly factory OC'd) 390 got 59.6 fps

So 59.6 to 79 is still poor scaling, but closer to 33%.

Just realized this is 390 vs 390X CFX - I have no explanation for that. Hopefully this is a driver issue.

Also regarding BF4 & Mantle, I was pretty sure that DX11 performance was better than Mantle which is why [H] swapped back to DX11. But that is a Brent/Kyle/Grady question.
 
Last edited:
Great review. Price is high but it's their halo product.

Only wish msi would have released this sooner. :p
 
What is fantastic about the 980ti and Maxwell, is the ability to OC like crazy. Really, 1,4GHZ or 1,5GHZ on a GPU is crazy.
What i havent found on the "interweb" is a clock for clock benchmark between Fiji and Maxwell. At this moment, AMD is really gimped by the low OC capabilities of Fiji (personally i was disapointed by the "overclockers dream").
The r9 390x Xfire is not viable against the 980ti, no way at all :)
 
What is fantastic about the 980ti and Maxwell, is the ability to OC like crazy. Really, 1,4GHZ or 1,5GHZ on a GPU is crazy.

Although true, a great amount of the reference board 980ti's can hit the 1400 and some stable at 1500. For the price, I would want to see more on air from this GPU.

Now when it comes to customized boards I would like to see higher OC on air, such as the classified.

Under water is a different story, [H] should get a water setup only for those high end customized boards,step up the voltages, and OC them to the max....:D
 
Weird...in your review of the 390 it gets 64fps in BF4 @ 1440p Ultra 4xMSAA. TWO cards get 79fps at the same settings here, isn't that like 25% scaling? That seems awful. Wonder what the heck they did to the drivers this time.

.

Different reviewers, different machines, none of the results can be directly compared unless used in the same machine.. however I also found the 390X results.. hmmm. Strange.. but well there are the numbers..

They're using DirectX instead of Mantle for the API, so it's going to be less optimized for AMD cards in general.

Completely fake, they use DX11 because in general offer better performance than mantle, Brent always put that "situation" in his review.. even in a review he offered mantle number to prove it have worse performance than DX in bf4..

That was overclocked, the not manually overclocked (but slightly factory OC'd) 390 got 59.6 fps

So 59.6 to 79 is still poor scaling, but closer to 33%.

Also regarding BF4 & Mantle, I was pretty sure that DX11 performance was better than Mantle which is why [H] swapped back to DX11. But that is a Brent/Kyle/Grady question.

You are correct..
 
I cant believe how heavily 390x crossfire got stomped!
I expected it to be ahead on average.

Good review, easy to read and everything I wanted to see addressed.

I wasn't sure what to think of the people in the thread for the dual 390 card that said, "meh, I'll just get a 980Ti." I thought, really? Great card, but can it hang with dual 390/390X's? I don't know...

Answer: Yes, yes it can. Except replace "hang" with "crush." Wow.
 
They're using DirectX instead of Mantle for the API, so it's going to be less optimized for AMD cards in general.


This statement is so VERY wrong. We use DirectX because it has better performance in BF4 than Mantle, and has for a good while now. Yes, we test for this.
 
Weird...in your review of the 390 it gets 64fps in BF4 @ 1440p Ultra 4xMSAA. TWO cards get 79fps at the same settings here, isn't that like 25% scaling? That seems awful. Wonder what the heck they did to the drivers this time.

BF4 is the most difficult game in our suite to get consistent performance as we play it in a full 64 player server. Yes, we have specific things that we tend to do to try to bring some consistency to it, but you really can't control the other 63 players in the match....

Also, which review are you referring to with a 390 at 64FPS?



I wasn't sure what to think of the people in the thread for the dual 390 card that said, "meh, I'll just get a 980Ti." I thought, really? Great card, but can it hang with dual 390/390X's? I don't know...

Answer: Yes, yes it can. Except replace "hang" with "crush." Wow.

The interesting thing to consider here is that this evaluation focused on 1440p gameplay while the dual 390 card focused on 4k game play. We've previously found that a single card is not sufficient for 4k, period, thus, the conclusion that I would draw from this evaluation is that the 980 Ti is the way to go for the 1440p experience and seek out additional data to draw an independent conclusion about 4k (or Eyefinity/Surround) resolutions.
Ok I'm super confused

This review
Project Cars:
390X cross fire gets 51 FPS @ 1440
http://hardocp.com/article/2015/10/...ti_lightning_video_card_review/4#.VifRFSu5Lzw

Yet on the Devil 390 review from days ago @ 2160 it gets 52.9 FPS (with a higher min FPS)
http://hardocp.com/article/2015/10/..._core_r9_390_video_card_review/4#.VifQtiu5Lzw


Did I miss something here, in the one review you said 5.10 wouldn't affect anything so you didn't use it...looks to MAJORLY have affected it

1440p and 4k are two totally different resolutions and cards behave differently between those two resolutions....
Interesting. I've heard it was the other way around. I have a pair of R9 290's so maybe I should give it a go.

I've found Mantle to be unstable with my pair of 290X's and Eyefinity screens for quite a while now - switching back to DX11 boosted performance and stopped 90% of the crashes for me.
 
Ok I'm super confused

This review
Project Cars:
390X cross fire gets 51 FPS @ 1440
http://hardocp.com/article/2015/10/...ti_lightning_video_card_review/4#.VifRFSu5Lzw

Yet on the Devil 390 review from days ago @ 2160 it gets 52.9 FPS (with a higher min FPS)
http://hardocp.com/article/2015/10/..._core_r9_390_video_card_review/4#.VifQtiu5Lzw


Did I miss something here, in the one review you said 5.10 wouldn't affect anything so you didn't use it...looks to MAJORLY have affected it
One is at 4k, the other at 1440p - don't understand your confusion :confused:, but 4k (3840x2160) is a much higher resolution than 1440p (2560x1440). Moar pixels to render, hence the difference.

Great review, thank you. It's ashame this card wasn't out a bit earlier - I purchased the ASUS STRIX card two weeks ago and would definitely have considered this if it was available.
 
Last edited:
One is at 4k, the other at 1440p - don't understand your confusion :confused:, but 4k (3840x2160) is a much higher resolution than 1440p (2560x1440). Moar pixels to render, hence the difference.

Great review, thank you. It's ashame this card wasn't out a bit earlier - I purchased the ASUS STRIX card two weeks ago and would definitely have considered this if it was available.



Yeah

With Identical settings it runs with a better avg frame rate and min frame rate @ 4k so why wouldn't they run it at 4k in this review rather than 1440, thought the whole point was max playable setting and since they couldn't turn on features, up the resolution it's like free AA in this instance.

Please feel free to explain why there's a performance gain upping the res to 4k


1440p and 4k are two totally different resolutions and cards behave differently between those two resolutions....
Yup and 4k runs faster according to [H], so why wasn't the game run on that res?


2 different resolutions and the higher resolution outperforms the lower at identical settings....da heck.
 
Last edited:
Since this was a review of a single 980 Ti we focused on 1440p gameplay. I believe we do have performance shown at 4K on a single 980 Ti out there - http://www.hardocp.com/article/2015/07/26/amd_radeon_r9_fury_x_4k_video_card_review/

Different drivers and game versions and test systems and resolution were used between the two reviews you mentioned. It is possible the AMD CrossFire profile is not as efferent at 1440p or more CPU limited in this game at a lesser resolution. It is possible that only at 4K GPU utilization and efficiency is at its best since it is more demanding. It certainly would not be out of the norm to have less effeciency in a CrossFire profile at a lesser resolution or changed because of a different driver, we've seen it before. It goes to further our stance on the immaturity and inconsistency found in AMD CrossFire profiles. If you want better consistency scaling in resolution, it is typically found in single-GPU usage.
 
Last edited:
I forgot this is Windows 10 as well. Because the performance is so bad on the 390xs, I assumed it was Windows 8.1 with pre Win 10 driver issues.

What are AMD doing?
The hardware is capable of a lot more.
I hoped things would start moving forward but it seems the reverse.
 
In regards to Project Cars and 390X, David has been working on a review of a single-390X and we have discovered some performance issues which could translate into a simple bug on 390X but specifically at 1440p resolution.

We will have more information in his review as he lays out his experiences with the 390X in that game. Simply put, performance is a lot lower than expected for some reason on 390X at 1440p, and we have tried driver versions from 15.8 Beta to the current driver. Now, he is on Windows 8.1 still, and Grady and My own testing was on Windows 10. Still, we may be encountering some kind of bug with 390X specifically in this game at 1440p. We don't have an answer for it, and it is odd that 390X CrossFire at 4K performed better for me, but do keep in mind also that was a much earlier version of the game, version 4.0. Since then there have been several patches to the game. It is an ongoing issue we are trying to figure out. In the interim, since 390X single-GPU performance is looking not right at 1440p David is dropping the game for his review so we could move on with it. He will however explain in the conclusion his experiences so you guys know what we encountered with that game. I'm not discounting the possibility the bug persists with 390X CrossFire at 1440p as well.

I just wanted to put that out there now because 390X and Project Cars may have a crippling bug at 1440p and performance is slower than it should be. This is news to us as we haven't seen this issue talked about on other forums or here. Our testing though, cannot be denied and we are dealing with this bug in real time. We have informed AMD. This seems to be the only gpu experiencing this issue in this game.
 
Last edited:
That's some crazy detective work

Glad to hear it's been noticed and there's a theory on to wtf happened with PC, as you guys normally up resolutions if the cards can handle it.
 
That's some crazy detective work

Glad to hear it's been noticed and there's a theory on to wtf happened with PC, as you guys normally up resolutions if the cards can handle it.

We normally do not raise single-GPUs to 4K from 1440p unless it were a "4K" review. In the instance of Project Cars there are other ways to improve image quality by raising AA settings by using MSAA or DSX AA modes before we'd raise it to 4K on a single-GPU.
 
Yup and 4k runs faster according to [H], so why wasn't the game run on that res?
Cause this is a 980Ti review, not a 390X Crossfire review, and no single card that's available today is really suitable for serious 4k gaming.

Please feel free to explain why there's a performance gain upping the res to 4k
I would assume that, given AMDs history of driver issues, it's a driver problem on AMDs part. This is borne out by what Brent says about the 390X having issues at 1440p in Project Cars.
 
Last edited:
BF4 is the most difficult game in our suite to get consistent performance as we play it in a full 64 player server. Yes, we have specific things that we tend to do to try to bring some consistency to it, but you really can't control the other 63 players in the match....

Also, which review are you referring to with a 390 at 64FPS?





The interesting thing to consider here is that this evaluation focused on 1440p gameplay while the dual 390 card focused on 4k game play. We've previously found that a single card is not sufficient for 4k, period, thus, the conclusion that I would draw from this evaluation is that the 980 Ti is the way to go for the 1440p experience and seek out additional data to draw an independent conclusion about 4k (or Eyefinity/Surround) resolutions.


1440p and 4k are two totally different resolutions and cards behave differently between those two resolutions....


I've found Mantle to be unstable with my pair of 290X's and Eyefinity screens for quite a while now - switching back to DX11 boosted performance and stopped 90% of the crashes for me.

http://www.hardocp.com/article/2015/09/21/xfx_r9_390_double_dissipation_8gb_video_card_review/8#.VikEPn6rSHs

http://www.hardocp.com/article/2015/10/21/msi_geforce_gtx_980_ti_lightning_video_card_review/9#.VikEQn6rSHs

Those two reviews, same system, slightly different drivers. One is an R9 390, the other CF R9 390X - overclocking the 390 should put you right at the 390X or very close. And yet, we're looking at around ~25% scaling. I've never seen that before, it's awful for XDMA in an optimized engine like FB.
 
If I was in the market, the LE version would look tasty. Looks like a tremendous effort out of MSI.
 
Typing skill without ability to read, interesting :p
The review has more than pictures.

It was a rhetorical question. Little, if anything, was gained by including a crossfire set-up into a current generation nVidia single card review. Silly actually........
 
It was a rhetorical question. Little, if anything, was gained by including a crossfire set-up into a current generation nVidia single card review. Silly actually........

It was made very clear.
Crossfire 390x is in the same price range as the card reviewed.
Perhaps you arent used to a value based review system.
 
Why was an AMD R9 390X CrossFire configuration in this review?

A good surprise. I would say if you have a budget around $700 that would be an option to look at. The 980Ti at this time really shine on these games. Plus other 980Ti and Fury reviews had data for those configurations.

I like this review in throwing in the 390x in cfx. The outcome speaks louder then the words. :D
 
As relevant as a GTX780 SLI configuration included in a single Fury X review. Years old architecture rebranded, scaling vs non-scaling, dual card driver issues/stuttering, heat/power consumption, case fitment issues, etc. etc.

The matter of a similar price alone does not make a logical apples-to-apples comparison.
 
I find the reviews excellent and appropriate, among the very best on the web.

If you dont like the reviews, you dont have to read them.
There are a ton of websites I avoid for the same reason.
The problem is on the seat underneath you.
 
As relevant as a GTX780 SLI configuration included in a single Fury X review. Years old architecture rebranded, scaling vs non-scaling, dual card driver issues/stuttering, heat/power consumption, case fitment issues, etc. etc.

The matter of a similar price alone does not make a logical apples-to-apples comparison.

Would you rather have more data than you need in a review, or less data than you need and miss something?

I love the fact that we are able to provide both a performance comparison and a price comparison giving our readers the most information possible to make an informed buying decision.

If you really don't like something being included in a review, just, ya know, ignore that something. :confused:

Which, to be honest, kind of disrespects the work and time of the author.
 
^
The 980ti parts were great, just didn't get the inclusion of AMD crossfire. If anything, a 970sli configuration would have made a little more sense but still not apples to apples. Just my opinion, not a big deal.
 
^
The 980ti parts were great, just didn't get the inclusion of AMD crossfire. If anything, a 970sli configuration would have made a little more sense but still not apples to apples. Just my opinion, not a big deal.

390X CF is price comparative, 390 CF is not.
 
^
The 980ti parts were great, just didn't get the inclusion of AMD crossfire. If anything, a 970sli configuration would have made a little more sense but still not apples to apples. Just my opinion, not a big deal.

Its already been established exactly why 390x crossfire was included, if you dont like that, why does it matter?
970 SLI is memory limited, it doesnt make sense to include that.

You just dont like seeing AMD get trounced.
 
Something definitely wrong with CF here, every other 390/X CF bench of BF4 Ultra Preset 1440p is around 30fps higher, or more. That's really too bad, and I seriously doubt it's just BF4. That was the one that really stood out to me. I returned a 390 by XFX because I didn't like the thermals, picked up a 380 (almost grabbed another 970 but figured I would give this a try) and am actively looking at reviews of different branded 390's for CF. This definitely turned me off until I noticed the discrepancy. On to more research!
 
Its already been established exactly why 390x crossfire was included, if you dont like that, why does it matter?

If you don't like the fact that I don't like it was included, why does it matter?


970 SLI is memory limited, it doesnt make sense to include that.

Memory limited at 1440p?

You just dont like seeing AMD get trounced.

I own a 970, what did I say to lead you to that conclusion? AMD was not worthy to be in this comparison. Again, just my opinion.......
 
Last edited:
If you don't like the fact that I don't like that it was included, why does that matter?
You asked.

Memory limited at 1440p?
Yes.

I own a 970, what did I say to lead you to that conclusion? AMD was not worthy to be in this comparison. Again, just my opinion.......
An observation.
You have made 7 posts in total at [H].

The first was asking to compare 3 FuryX cards vs 2 Titans Xs. Rather biased considering you could have compared to 3 980tis at the same price or less than the Furys.

The second was saying nice job reviewing the dual 390 card. Then wanting to compare 2 of these to 2 Titans Xs, similar bias.

Then this thread where you dont like 390x crossfire being compared to a 980ti even though they can be bought for about the same money.
I agree they are not worthy but only because I read this review.

Quite amusing tbh :)
Especially when 2 AMD GPUs were pitted against one NVidia GPU and lost.
How do you expect quadfire to go?
 
Last edited:
^
Hey Mods, I find it very creepy that this guy (Nenu) is searching through my posting history and trolling me in this thread.
I finally gave up on TH and decided to give this forum a chance. With weirdos like Nenu bullying, maybe this is not the best place for new users to come?
 
Back
Top