Sign in with
Sign up | Sign in
Your question
Solved

gtx 980 SLI bottleneck?

Tags:
  • Gtx
  • Bottleneck
  • SLI
Last response: in Graphics & Displays
Share
September 19, 2014 11:50:39 AM

I have a 3570k and I'm wondering if this would bottleneck 980 in SLI

More about : gtx 980 sli bottleneck

September 19, 2014 11:58:36 AM

Yes it definitely would if you are doing 2+ way SLI. You really need a top of the line i7 to get max performance with such a setup.
m
0
l
September 19, 2014 12:14:08 PM

That kinda sucks didn't think a 3570k would bottleneck ;/
m
0
l
Related resources
September 19, 2014 12:39:08 PM

It might be a little bit, but probably not as much as tambeshakunt infers. You might get 5% more performance with the top end processor, but I doubt you'd notice the difference.
m
0
l
September 19, 2014 12:46:47 PM

With a decent overclock on that 3570k you should be just fine.
m
0
l
September 19, 2014 12:49:53 PM

great as i don't plan to change my CPU till skylake

what overclock would you recommend?
m
0
l
September 19, 2014 1:01:36 PM

The higher the better. I'm not sure where you'll start seeing diminishing returns because I haven't played with gtx980s. I know that with my 780s and my 4770k I saw improvements up to 4.2ghz in gaming. What resolution are you gaming at?
m
0
l
September 19, 2014 1:08:33 PM

1440p at the moment. my 3570k is clocked@ 4.5.

But i plan on getting a 1080p 144hz G sync monitor when one comes out. which is why i want two 980s so i can take advantage of the high refresh rate almost all the time.
m
0
l
September 19, 2014 1:10:17 PM

wip99gt said:
The higher the better. I'm not sure where you'll start seeing diminishing returns because I haven't played with gtx980s. I know that with my 780s and my 4770k I saw improvements up to 4.2ghz in gaming. What resolution are you gaming at?


Also to my knowledge (i could be wrong here) but my 3570k@4.2 is pretty much the same as your cpu @stock (for gaming)

have you seen any bottlenecks with your current SLI setup?
m
0
l
September 19, 2014 1:22:42 PM

My graphics cards are my bottleneck. I do run triple 1440p monitors though and my 4770k is at 4.4ghz. Sli GTX980s will be overkill for a 1080p monitor at any refresh rate. If I was in your shoes I'd wait on the new monitor and see what comes out in the 4k resolution in the next while. Your getting a serious amount of gpu power and should get the proper screen to push those things. You could try tweaking your monitor right now. I was able to get my Asus ones up to 90hz.
m
0
l
September 19, 2014 1:25:00 PM

would you say a single 980 is enough for 1080p 144hz? i am getting the 980 "free"
m
0
l
September 19, 2014 1:31:45 PM

Probably. Those things look like they overclock incredibly well too. Try just the one first. It's easy to add another later on.
m
0
l
September 19, 2014 1:33:04 PM

Cool
m
0
l
September 19, 2014 1:37:36 PM

wtf? there is no known bottlenecks with any sandy bridge or higher i5. not sure who would think so or what benchmark someone is looking at.
m
0
l
September 19, 2014 1:54:13 PM

nikoli707 said:
wtf? there is no known bottlenecks with any sandy bridge or higher i5. not sure who would think so or what benchmark someone is looking at.


I thought the same before i posted here lol
m
0
l

Best solution

September 19, 2014 2:34:52 PM

nikoli707 said:
wtf? there is no known bottlenecks with any sandy bridge or higher i5. not sure who would think so or what benchmark someone is looking at.


yeah, there are. If you run an i5 with a pair of any of the higher end (GTX680 or higher) GPUs, it will not quite run as fast as a higher end processor. The difference isn't big, but it is there.

Share
September 19, 2014 2:54:52 PM

dgingeri said:
nikoli707 said:
wtf? there is no known bottlenecks with any sandy bridge or higher i5. not sure who would think so or what benchmark someone is looking at.


yeah, there are. If you run an i5 with a pair of any of the higher end (GTX680 or higher) GPUs, it will not quite run as fast as a higher end processor. The difference isn't big, but it is there.



so what type of scaling would you expect on average with my 3570k@4.5 with a pair of 980s
m
0
l
September 19, 2014 3:03:27 PM

show me a game that is known to be highly threaded that can give you a quantifiable boost in fps when going with a faster processor. not horribly coded watchdogs with ridiculous driver overhead. i can definitely see small but barely noticable difference if your not running max settings at a low resolution. but in a high end game with extreme gpu demand, there just isn't enough calls to the cpu for any amount of work. ivy 4c4t@4.5ghz compared to what?.... a 5960k@5.0ghz.... maybe 4% performance loss at best in the worst case 2 way sli/cfx 980/290x setup. some games/benchmarks actually run slightly slower on x99 than they do on z87/z97, though again this is a very small hit... something like 1%.
m
0
l
September 19, 2014 3:10:00 PM

zqa20 said:
dgingeri said:
nikoli707 said:
wtf? there is no known bottlenecks with any sandy bridge or higher i5. not sure who would think so or what benchmark someone is looking at.


yeah, there are. If you run an i5 with a pair of any of the higher end (GTX680 or higher) GPUs, it will not quite run as fast as a higher end processor. The difference isn't big, but it is there.



so what type of scaling would you expect on average with my 3570k@4.5 with a pair of 980s


It's really not going to be enough to worry about. Any cost of buying the next processor up would just be too much for the minor increase, but if you really want, you'll probably get between 5 and 7 percent higher speeds out of a 3770k, depending on the game. MMOs, like World of Warcraft and Star Trek Online see more than that because they're more CPU bound anyway.

I know for certain, from recent experience, that World of Warcraft sees better than 10% improvement between HT off and on with a 4790k and dual GTX 680s. (My own system, recent experimentation.) Star Trek online saw slightly less, but the difference was noticeable. Starcraft II saw slightly better than 10%. Diablo III saw no difference at all. I don't have any recent FPS games to test, though. My most recent first person shooter is Doom 3, and I haven't even bothered to install it since my last system rebuild. I'm just not into those.

It's not a review about these directly, but it does compare the new devil's canyon chips on a system with SLi, which should see about the same difference between them: http://www.anandtech.com/show/8227/devils-canyon-review... There's a slight difference, and the higher end chips do see a little boost, but the difference really isn't much.
m
0
l
September 19, 2014 3:12:42 PM

nikoli707 said:
show me a game that is known to be highly threaded that can give you a quantifiable boost in fps when going with a faster processor. not horribly coded watchdogs with ridiculous driver overhead. i can definitely see small but barely noticable difference if your not running max settings at a low resolution. but in a high end game with extreme gpu demand, there just isn't enough calls to the cpu for any amount of work. ivy 4c4t@4.5ghz compared to what?.... a 5960k@5.0ghz.... maybe 4% performance loss at best in the worst case 2 way sli/cfx 980/290x setup. some games/benchmarks actually run slightly slower on x99 than they do on z87/z97, though again this is a very small hit... something like 1%.


http://www.anandtech.com/show/8227/devils-canyon-review...

It's small, but it is there.
m
0
l
September 19, 2014 3:24:58 PM

dgingeri said:
nikoli707 said:
show me a game that is known to be highly threaded that can give you a quantifiable boost in fps when going with a faster processor. not horribly coded watchdogs with ridiculous driver overhead. i can definitely see small but barely noticable difference if your not running max settings at a low resolution. but in a high end game with extreme gpu demand, there just isn't enough calls to the cpu for any amount of work. ivy 4c4t@4.5ghz compared to what?.... a 5960k@5.0ghz.... maybe 4% performance loss at best in the worst case 2 way sli/cfx 980/290x setup. some games/benchmarks actually run slightly slower on x99 than they do on z87/z97, though again this is a very small hit... something like 1%.


http://www.anandtech.com/show/8227/devils-canyon-review...

It's small, but it is there.


it definitely is. but also like you said, definitely not worth uprooting an already powerful lga1155 rig to go to lga1150 for a mere 2 or 5fps, especially if your at or over 60fps@60hz. if your on a 120/144hz refresh, 105fps or 110fps will never be noticeable with the actual in game per second fps jumping from around, a gsync monitor would be much more worth it at this point.
m
0
l
September 19, 2014 3:35:48 PM

Well, with a pair of GTX 980s, it would be a bit more pronounced. I still wouldn't think it would be worth upgrading the CPU, though.
m
0
l
September 19, 2014 3:50:48 PM

sigh, how much of a performance scaling loss do you think i'd see then with a 3570k@4.5?

I still don't fully understand how my 3570k could bottleneck that bad. From my understanding haswel is almost the same as ivy in terms of gaming. A 3570k@4.2 is just as fast as a 4770k so surely @4.5 I'm going to get the same experience as a haswel refresh right? I might be wrong but that's what i thought.
m
0
l
September 19, 2014 3:57:51 PM

dgingeri said:
Well, with a pair of GTX 980s, it would be a bit more pronounced. I still wouldn't think it would be worth upgrading the CPU, though.


zqa20 said:
sigh, how much of a performance scaling loss do you think i'd see then with a 3570k@4.5?

I still don't fully understand how my 3570k could bottleneck that bad. From my understanding haswel is almost the same as ivy in terms of gaming. A 3570k@4.2 is just as fast as a 4770k so surely @4.5 I'm going to get the same experience as a haswel refresh right? I might be wrong but that's what i thought.


i wouldn't worry about it, your 3570k@4.5 is very very fast by any standard.
m
0
l
September 19, 2014 4:02:38 PM

(double post)
m
0
l
September 19, 2014 4:02:38 PM

nikoli707 said:
dgingeri said:
Well, with a pair of GTX 980s, it would be a bit more pronounced. I still wouldn't think it would be worth upgrading the CPU, though.


zqa20 said:
sigh, how much of a performance scaling loss do you think i'd see then with a 3570k@4.5?

I still don't fully understand how my 3570k could bottleneck that bad. From my understanding haswel is almost the same as ivy in terms of gaming. A 3570k@4.2 is just as fast as a 4770k so surely @4.5 I'm going to get the same experience as a haswel refresh right? I might be wrong but that's what i thought.


i wouldn't worry about it, your 3570k@4.5 is very very fast by any standard.


Thanks, do you think i'd need two 980s to take advantage a 144hz 1440p g sync monitor? or should i just go for a single 980 on a 1080p 144hz monitor?
m
0
l
September 19, 2014 4:08:16 PM

depends on the game. you dont have to hit or sustain 144fps to enjoy the benefits of a 144hz refresh. if its battlefield 4 in a 64 man server going hard, you would need two if you want to keep your minimum fps above 90-100fps@1440p, probably about 110-120fps@1080p. averages will be much higher. im on a 60hz 1440p monitor and im spoiled by ips. i cannot go back to a tn based panel. if your serious about getting all this gaming horsepower, i would look into when the strobbed backlight ips high refresh 1440p gysnc panels will be coming out.
m
0
l
September 19, 2014 6:21:56 PM

If your planning on gaming on a 1440p 120hz+ monitor then yeah 2 gtx980 would be best. 1 will do plenty good on it's own but if you have the budget and really want to then get the 2.
m
0
l
October 17, 2014 8:54:14 AM

Now that I've switched from the I5 3570k to the i7 4790k, (Gtx 980 stock reference) I'm getting a good amount more in fps in some games IE Watchdogs. The 3570k does bottleneck it a bit, but I had the cpu at stock clocks. Increasing the clock might help.
m
0
l
!