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

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
MSI GeForce GTX 980 Ti SEA HAWK Video Card Review - Today we have the liquid cooled MSI GTX 980 Ti SEA HAWK video card, with features that have it screaming to be overclocked. We will compare it to the overclocked MSI GTX 980 Ti LIGHTNING with air cooling in a battle of premium GTX 980 Ti video cards. We will also compare it to a reference GTX 980 Ti to determine the full range of GPU performance.
 
Wish you guys tossed in the Titan-X for comparison. Maybe it's just me, but I find it fun seeing a much cheaper card completely obliterate a $1,000+ card. I imagine the 980-Ti SeaHawk at max OC will still obliterate a Titan-X at max OC.
 
That extra $100 is the water cooler. I hate that it adds that much, but obviously it cools so much more efficiently than the traditional method.
 
Noticed something weird Grady, why the card it's running at PCI-E 2.0?...
 
:pthiiis is how we do itttt:p

love me an AIO GPU cooler. nice to see it yielded a top max oc.

this is how i want my pascal card.
 
Noticed something weird Grady, why the card it's running at PCI-E 2.0?...

In GPUz you have to run a 3D program actively to update and show the correct PCIe speeds in GPUz. The question mark beside it, if you press it, enables a 3D app you can full-screen and then it shows the correct speeds. Just ignore that, it's a GPUz thing.
 
Wish you guys tossed in the Titan-X for comparison. Maybe it's just me, but I find it fun seeing a much cheaper card completely obliterate a $1,000+ card. I imagine the 980-Ti SeaHawk at max OC will still obliterate a Titan-X at max OC.

Noted, that would be an interesting comparison in the future.
 
In what instances does the on-card fan run? Is that fan silent even under heavy load? I hate that whine that those fans will produce at high RPM. One thing I've always liked about the NZXT and other video card water cooler kits is that the on-card fan is larger than a reference card provides.

I wish [H]ard would do sound pressure and decibel tests as well as use a better fan than the stock 120mm provided by MSI, perhaps a Noctua?
 
Awesome review guys.

Maybe I am just going blind, but I read the review and then went back and re-read it to see if I could find ambient temps, but was unsuccessful. I vaguely remember the H using standard controlled 20C for all reviews. Is that accurate?

I'm really impressed with this thing. I'm doing the same myself but manually using Corsairs HG10 N980 brackets on my video cards. (they are having some issues with those right now, but I hope that will be resolved shortly)

if they can achieve these temps with a H50 based cooler, I should be able to get fantastic temps with my H90's.

Did you happen to check the ASIC level of the card for reference?
 
In GPUz you have to run a 3D program actively to update and show the correct PCIe speeds in GPUz. The question mark beside it, if you press it, enables a 3D app you can full-screen and then it shows the correct speeds. Just ignore that, it's a GPUz thing.

this time I think you are wrong brent (I could be also ;) ) but..PCI-E don't change the PCI-E Generation availability. The card drop the lines and gen to save power, so they generally stay for example PCI-E 3.0 x16 @ x16 1.1 that mean the card dropped form Gen 3.0 to Gen 1.1 and save power, however the first sentence PCI-E 2.0 x16 (see image below with red square) show the capacity or max availabity of the motherboard, so that mean the card is only able to run at PCI-E 2.0 with 16 lines. so the card is indeed unable to run at full PCI-E 3.0 x16. that could be a issue with the card or just with an option in the motherboard that allow to switch between PCI-E Gen2 and Gen3.

fx9et2.jpg


Just as example I took the same Grady Machine from the 980TI lightning review

and see what it say: PCI-E 3.0x16 @x8 1.1 that mean the card is able to run at full PCI-E 3.0 x16.
1445206223qVCPPl1iL2_2_3.png


Just for the sake and/or sanity of the mind you can check with another program to see if a GPU-Z Bug or another failure. PCI-E 2.0 vs 3.0 can be a significant drop in performance in some games.
 
Just for the sake and/or sanity of the mind you can check with another program to see if a GPU-Z Bug or another failure. PCI-E 2.0 vs 3.0 can be a significant drop in performance in some games.

I am sure Brent and Grady will look into this, but I would have to fully disagree with you on the point that there can be a "significant drop in performance" from PCIe 2 to PCIe 3 with 16 lanes available with a single GPU.
 
this time I think you are wrong brent (I could be also ;) ) but..PCI-E don't change the PCI-E Generation availability.

I have found that my primary card tends to stay at the same gen, but my secondary card often drops down in gen as well.

Not sure if this is commonplace or not.
 
Zarathustra[H];1041975254 said:
Awesome review guys.

Maybe I am just going blind, but I read the review and then went back and re-read it to see if I could find ambient temps, but was unsuccessful. I vaguely remember the H using standard controlled 20C for all reviews. Is that accurate?

I'm really impressed with this thing. I'm doing the same myself but manually using Corsairs HG10 N980 brackets on my video cards. (they are having some issues with those right now, but I hope that will be resolved shortly)

if they can achieve these temps with a H50 based cooler, I should be able to get fantastic temps with my H90's.

Did you happen to check the ASIC level of the card for reference?

All video cards should have a water cooling option. I slapped one of the Corsair HG10 brackets and a H80i GT on my R9 290. Sitting pretty at 41c idle and 53c with +200mv sent to it.

Nvidia and AMD should make a hybrid lineup of these type of cards from now on.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
I'm confused: why does this thing get a gold award when it's $80 more expensive than the Lightning LE, and provides 2% better performance overclocked, a few degrees C lower temperatures, and really unnoticeable sound levels difference?

The MSI GTX 980 Ti SEA HAWK was quieter than the MSI GTX 980 Ti LIGHTNING, but not by much. The two video cards operated so quietly, even when the three fans of the LIGHTNING were at 100%. These two video cards are designed to perfection when it comes to a silent gaming experience. The AMD Radeon R9 Fury X was also quiet with its liquid cooling and single fan. The MSI GTX 980 Ti SEA HAWK however was the most quiet of all.

I mean, you did use the existence of the Lightning LE to justify your Gold for the overpriced full Lightning with the liquid cool BIOS. So you can't pretend it doesn't exist when scoring an even more overpriced monstrosity.

The Lightning LE proves that water cooling is outrageously overpriced for barely noticeable improvements. It only makes sense when you don't have the room for a full size cooler, like a MiniITX build. And that card isn't exactly competitive with the length of the Fury X.
 
Last edited:
Perhaps it wasn't mentioned because it wasn't an issue but can you please comment on pump noise? From personal experience Corsair can be hit and miss..
 
Also I bet you could build your own version of this buying just a standard 980TI on sale and getting the bracket and a used/clearance H50 cooler somewhere and be almost right there performance wise :)
 
I'm confused: why does this thing get a gold award when it's $80 more expensive than the Lightning LE, and provides 2% better performance overclocked, a few degrees C lower temperatures, and really unnoticeable sound levels difference?.

The Ultra high end is usually only separated by a few FPS and a few degrees. This is a novel product that matches & exceeds the Lightning(top SKU) for a few more $ while being cooler to boot. That's why.
 
Just for the sake and/or sanity of the mind you can check with another program to see if a GPU-Z Bug or another failure. PCI-E 2.0 vs 3.0 can be a significant drop in performance in some games.

I am leaning toward GPUz bug, note different versions are being used between those screenshots. It is possible during idle the card drops into a low power mode. With new cards like this, especially custom cards that GPUz doesn't recognize, it can throw out wrong numbers. GPUz gets its info from pre-programmed database in the software. The fact is, his motherboard can do PCIe 3.0, his CPU allows it, and the card was in the primary slot, so everything is setup, like all video cards, to be PCIe 3.0. We will try to see what is up, but if we can't find an answer it is likely just due to software issue. I also agree with Kyle, it would not cause a significant performance difference in single-card configuration even if it were PCIe 2.0, which it is not.

I know GPUz and CPUz display PCIe bus speed, PM me some other free software ideas that also report the PCIe bus speed correctly.
 
Last edited:
Why is this thing not a single slot card?

My guess would be because the H50 Quiet edition (which this is based off of, but with beefier fans) pump/block is too thick for that.

It's much thinner than the original H50, but still, probably not single PCIe slot thin.

msi-sea-hawk-2.jpg


The Corsair HG10 bracket that allows you to convert your own video card does not turn it into a single slot either.
 
I am leaning toward GPUz bug, note different versions are being used between those screenshots. It is possible during idle the card drops into a low power mode. With new cards like this, especially custom cards that GPUz doesn't recognize, it can throw out wrong numbers. GPUz gets its info from pre-programmed database in the software. The fact is, his motherboard can do PCIe 3.0, his CPU allows it, and the card was in the primary slot, so everything is setup, like all video cards, to be PCIe 3.0. We will try to see what is up, but if we can't find an answer it is likely just due to software issue. I also agree with Kyle, it would not cause a significant performance difference in single-card configuration even if it were PCIe 2.0, which it is not.

I know GPUz and CPUz display PCIe bus speed, PM me some other free software ideas that also report the PCIe bus speed correctly.

Thanks for take the time to see about that... PM Sent. :)
 
Update on GPUz screenshot, don't know why the original screenshot showed PCIe 2.0. We have re-loaded the driver, and the latest GPUz now shows it as PCIe 3.0 x16 in that spot. A software reading glitch. I am going to update the image on that page, one moment and you will be able to see it.

Ok, the image is updated, and btw it shows @ x8 1.1 because it is not running a 3D app, and basically in idle mode or low power state mode, Windows can lower the power idle state of the PCIe bus.

http://www.hardocp.com/article/2015/11/16/msi_geforce_gtx_980_ti_sea_hawk_video_card_review/2

And yes the card does read PCIe 3.0 when running the 3D app
 
Last edited:
Update on GPUz screenshot, don't know why the original screenshot showed PCIe 2.0. We have re-loaded the driver, and the latest GPUz now shows it as PCIe 3.0 x16 in that spot. A software reading glitch. I am going to update the image on that page, one moment and you will be able to see it.

Ok, the image is updated, and btw it shows @ x8 1.1 because it is not running a 3D app, and basically in idle mode or low power state mode, Windows can lower the power idle state of the PCIe bus.

http://www.hardocp.com/article/2015/11/16/msi_geforce_gtx_980_ti_sea_hawk_video_card_review/2

Any chance you can read the ASIC of the card while you have it installed? These temps are unbelievable, I'd be curious if the GPU has been heavily binned to achieve it.
 
Question:

How come in the GPUz charts accompanying the OC results, the GPU's only doing ~63% load?

What did you use for stability testing? Was vsync on?
 
Zarathustra[H];1041978087 said:
Any chance you can read the ASIC of the card while you have it installed? These temps are unbelievable, I'd be curious if the GPU has been heavily binned to achieve it.

temps are where they should be.. it's more or less the same with EVGA Hybrid.

Zarathustra[H];1041979401 said:
Question:

How come in the GPUz charts accompanying the OC results, the GPU's only doing ~63% load?

What did you use for stability testing? Was vsync on?

IDK, probably the same bug that affected the PCI-E readings to 2.0?.. in my opinion a bug with the drivers.
 
Zarathustra[H];1041979401 said:
Question:

How come in the GPUz charts accompanying the OC results, the GPU's only doing ~63% load?

What did you use for stability testing? Was vsync on?


As we have for years now we use extended periods of gameplay to determine gaming stability.
 
If anyone buys this video card, would you mind reading the BIOS using GPU-z and submitting it to techpowerup?

I'm interested in taking a look at it.
 
Bought this card based on the review.
I simply wanted a good factory overclocked card that would be stable and still put out a little extra.
I went from two R9 290 cards that I had a lot of problems with. Never any problems in games but there has not been a stable desktop experience with those cards since the 14.4 driver.
With this card all my troubles disappeared and the performance is comparable from a single card.
I did a little overclocking and though my ambient is about 22-23 C it does run up to 57 C under load which is ok. I have the radiator entirely outside the case too, so no "case-warm" air going through.

My overclocking does not come close to those in the article though.

I got it working close, but to get it stable for a long time I had to have voltage at +87. Power limit at 107%, Core at +120 and Memory at +365. I could probably get a little more out of memory by fine tuning it in time, but +385 was too much, and core at +125 craps out after about 10 minutes of Fallout 4.
Edit: I might add that I am a reluctant overclocker normally, so I don't know all the tricks in the book.
Also, long time since I had a Nvidia card. The driver keeps renumbering my screens, one day it is numbered 3,2,1 the next 1,3,2 I have even had 2,1,3 Since I use mixed outputs, DP,HDMI, and DVI in order, I would think this would be a no-brainer to keep track on. Last two days it has been on 3,2,1 though, so I hope it has determined that is the way it's gonna be. Is this a normal Nvidia behaviour? I remember in the early Eyefinity days this was a problem too, but the next driver solved that and it never was a problem again. But for Nvidia showing this still is a bit strange.
 
Last edited:
Me too got one. i always wanted a AIO cooled GPU after I put one on my CPU. Very nice card, I using only for folding/distributed computing. What is really impressive are the low temps and fan. 60C with 33%. Never got that with my 780 or 970 on air.
 
I'm running mine at [email protected]...

Core Voltage: +15mV
Power Limit: 107%
Core Clock: +82
Memory:+460
Fan: 33%

Temps: idle @ 23-25c and load @ 50-55c using a low rpm push/pull setup.

All 20 of my case fans are running @ the lowest rpm possible. I originally struggled with the loudness of an Asus STRIX 980ti, when those fans on the card ramped up to 50% it was by far the loudest thing in my case. I returned it to Amazon and purchased this MSI Seahawk and WOW, night and day difference, the small fan on the card stays silent and never passes 33% and the 2x120s on the radiator running at 500 rpm keeps this card incredibly cool and very silent!

I'm running at a solid 120fps using a custom sweetFX config and Battlefornt looks incredible.
https://youtu.be/FC2CW4RPALI?t=37s

NBybbRb.jpg
 
Last edited:
I'm playing with it now, it runs really well, my temps are higher though. It idles at 27, under load 38, overclocked i've only tested a few times so far. This is all at 33% fans.
 
Back
Top