1080 SLI vs Titan X Pascal

[Rizen]

Serious question: what "Fast twitch online games" don't run at max FPS with a decent single GPU?

Games like Overwatch, CS:GO, Starcraft, League, DotA, etc all can be maxed out easily with a mid-range GPU, let alone a Titan X. SLI is not needed.

 

[Trandana]

Serious question: what "Fast twitch online games" don't run at max FPS with a decent single GPU?

Games like Overwatch, CS:GO, Starcraft, League, DotA, etc all can be maxed out easily with a mid-range GPU, let alone a Titan X. SLI is not needed.

True, but not BF1, Titanfall 2, Crysis 3 (if you are into the multiplayer) or possibly the next COD. Also it depends on your definition of maxed, because dual 1080's can play more games in 5k 60hz than a Titan XP can.

In other news, the 1080ti specs have been released: GTX 1080 release date, price, benchmarks, and GTX 1080 TI speculation - everything we know
Looks like a 980ti vs Titan X battle again

 

[thesmokingman]

Those games suck ass and are not out aside from C3. Quick change the subject...

 

[Trandana]

Those games suck ass and are not out aside from C3. Quick change the subject...

(I agree about COD) but BF1 sucks ass? lol okay man whatever you say. It will receive top reviews when released I can guarantee that, so I don't think you understand
what makes a good video game.

 

[Rizen]

BF1 from the reports I have heard runs extremely well, similar to BF4, which I can run at over 100fps consistently @ 2560.

Competitive gamers don't play at 4K/5K resolutions so my assumption - perhaps wrong - was that you were playing at 2560 or under. Yes, if you want to play those games at 4K/5K resolution you might need more power, but I still think you're going to see a better experience overall with a single TXP than a pair of 1080s.

 

[thesmokingman]

(I agree about COD) but BF1 sucks ass? lol okay man whatever you say. It will receive top reviews when released I can guarantee that, so I don't think you understand
what makes a good video game.

So you got no meaningful reply and resort to ad hominem?

 

[Rizen]

Speculating performance on unreleased games is pointless at best.

 

[Trandana]

So you got no meaningful reply and resort to ad hominem?

BF4 sold extremely well, despite a rough release. Titanfall 1 sold very well, and got excellent reviews. I don't like COD, but it still sells like mad.
The BF1 beta reception was great, so what would possibly make you think that these big budget games will not do well?

BF1 from the reports I have heard runs extremely well, similar to BF4, which I can run at over 100fps consistently @ 2560.

Competitive gamers don't play at 4K/5K resolutions so my assumption - perhaps wrong - was that you were playing at 2560 or under. Yes, if you want to play those games at 4K/5K resolution you might need more power, but I still think you're going to see a better experience overall with a single TXP than a pair of 1080s.

"I still think you're going to see a better experience overall with a single TXP than a pair of 1080s"

Not at 5k like I mentioned. It depends on your preferences, but I personally haven't played in 1440p since 2013.
My point was that sli has less input lag REGARDLESS of the resolution you are running at, thus the benefit for pro gamers.
Whether you are playing at 1080p or 4k you still have less latency than a single card.
This argument keeps going in circles, with the same facts being brought up


"Competitive gamers don't play at 4K/5K resolutions"

Not 5k yet, but 4k yes many do. High refresh monitors will not make someone a better player, nor give someone a discernible competitive advantage.
There are only 3 monitors with less input lag than the ASUS mg24uq IPS 4k monitor. They are 2012/2013 monitors with only 1080p and 60hz and they are only 1 ms faster. They also do not support freesync like the mg24uq does.
Input lag and internet connection are far more important than high fps in terms of competition. As for enjoyment of the game, well that's subjective.

*Unless you play using an oldschool CRT monitor, which has lower input lag than any monitor

 

[Trandana]

Speculating performance on unreleased games is pointless at best.

They may be unreleased, but Titanfall 2 and BF1 have both been played and well tested. Few changes will be made before release. The beta
is for network testing and bug fixing, they are in the last phase of their product's development.

 

[Rizen]

My point was that sli has less input lag REGARDLESS of the resolution you are running at, thus the benefit for pro gamers.
Whether you are playing at 1080p or 4k you still have less latency than a single card.
This argument keeps going in circles, with the same facts being brought up

Well to be fair, that "fact" you listed is generally false. Frame variance and frame times are almost always worse on SLI setups even at the same or better frame rates. You usually need a lot more performance out of the SLI setup to equal the more powerful single card's smoothness.

Input lag is primarily a function of your monitor and the technology you are using (V-Sync, FastSync, GSync, etc), SLI has no bearing on input lag (I'm being pedantic here but deal with it).

 

[Trandana]

Well to be fair, that "fact" you listed is generally false. Frame variance and frame times are almost always worse on SLI setups even at the same or better frame rates. You usually need a lot more performance out of the SLI setup to equal the more powerful single card's smoothness.

Input lag is primarily a function of your monitor and the technology you are using (V-Sync, FastSync, GSync, etc), SLI has no bearing on input lag (I'm being pedantic here but deal with it).

Did you not read any of the graphs I posted on the earlier pages? Please go back to review them, and you will see that latency is in fact lower when using SLI.
Those FCAT measurements display the amount of milliseconds it takes to render the frame and
send it to the screen, and on average it is at least 5-7 ms faster than single gpus.

"Input lag is primarily a function of your monitor"
Not only your monitor, there are many factors involved, including GPU, CPU, memory, clock speeds, motherboard, monitor processing speed, internet latency, even the
buld quality and capacitors on your circuit boards all play a factor. Your mouse and keyboard too
are taken into account by pro gamers. harcore twitch gamers still use wired mice and keyboards rather than wireless due to the lower (albeit very little difference) latency.
On wall street, finance companies and day traders rely on computers with exceptionally fast memory speeds, because those minuscule fractions of a second can
mean the difference between winning and losing a lot of money. The low latency is what successful stock brokers rely upon, sometimes with thousands of stock micro transactions
in a single day.

When you buy your ram, you divide the timing by the mhz frequency. A CAS timing of 15 / 2400 mhz
Gives a 0.00625 second response time, while a timing of 14/3200mhz provides latency of 0.004375
Thus the 14/3200mhz ram is faster.

 

[thesmokingman]

On wall street, finance companies and day traders rely on computers with exception memory speeds, because those minuscule fractions of a second can
mean the difference between winning and losing a lot of money. The low latency is what successful stock brokers rely upon, sometimes with thousands of stock micro transactions
in a single day.

Seriously, what the hell does that have to do with SLI? How are brokers using SLI???

 

[Rizen]

Did you not read any of the graphs I posted on the earlier pages? Please go back to review them, and you will see that latency is in fact lower when using SLI.
Even when using two cards against a single card of the same performance, the SLI setup gets lower latency.Those FCAT measurements display the amount of milliseconds it takes to render the frame and
send it to the screen, and on average it is at least 5-7 ms faster than single gpus.

Okay so here's the problem. What I said was "frame variance and frame times are almost always worse on SLI setups even at the same or better frame rates". You posted benchmarks from two games - BF Hardline and GTA V. I would hardly call that a representative sample, but fair enough, let's go with that.

Both of these games are using D3D11, which has explicit SLI support as part of the API. And in the case of the Frostbite engine used in Hardline, has significant attention spent to making mGPU support work well. In D3D12, there is no explicit SLI support, developers have to program for it themselves.

If you buy two 1080s over a Titan XP, you are rolling the dice that 1) developers bother to implement SLI support in their games, and 2) that SLI support is implemented well.

I've owned many SLI and CFX setups. This is nothing new to me and I'm not just spouting off BS - I used 4850 CFX, 5870 CFX, 580 SLI, 680 SLI, 290X CFX etc. I switched to a single Titan XP because I was tired of getting half-assed SLI / CFX support, or having to wait months for it to be fixed/implemented properly in games that did offer it.

So sure, if literally every game you play is an AAA FPS game from a major publisher who always implements SLI - then yes, in your specific usage, 1080 SLI is probably a better solution. I don't think that's true for most people, and it wasn't true in my experience, which is why I recommended a single TXP over 1080 SLI.

 

[Trandana]

Seriously, what the hell does that have to do with SLI? How are brokers using SLI???

He said: "Input lag is primarily a function of your monitor and the technology you are using (V-Sync, FastSync, GSync, etc), SLI has no bearing on input lag (I'm being pedantic here but deal with it)."
I was showing him why he is incorrect about input lag...

 

[Rizen]

The graphs you showed were measuring frame time and not input latency. So again, we don't need to be discussing input latency here. Especially since you're talking about playing on 4K/5K monitors, of which there aren't even any 120hz/144hz consumer models available.

 

[Trandana]

Okay so here's the problem. What I said was "frame variance and frame times are almost always worse on SLI setups even at the same or better frame rates". You posted benchmarks from two games - BF Hardline and GTA V. I would hardly call that a representative sample, but fair enough, let's go with that.

Both of these games are using D3D11, which has explicit SLI support as part of the API. And in the case of the Frostbite engine used in Hardline, has significant attention spent to making mGPU support work well. In D3D12, there is no explicit SLI support, developers have to program for it themselves.

If you buy two 1080s over a Titan XP, you are rolling the dice that 1) developers bother to implement SLI support in their games, and 2) that SLI support is implemented well.

I've owned many SLI and CFX setups. This is nothing new to me and I'm not just spouting off BS - I used 4850 CFX, 5870 CFX, 580 SLI, 680 SLI, 290X CFX etc. I switched to a single Titan XP because I was tired of getting half-assed SLI / CFX support, or having to wait months for it to be fixed/implemented properly in games that did offer it.

So sure, if literally every game you play is an AAA FPS game from a major publisher who always implements SLI - then yes, in your specific usage, 1080 SLI is probably a better solution. I don't think that's true for most people, and it wasn't true in my experience, which is why I recommended a single TXP over 1080 SLI.

"Okay so here's the problem. What I said was "frame variance and frame times are almost always worse on SLI setups even at the same or better frame rates". You posted benchmarks from two games - BF Hardline and GTA V. "

Here are more sli FCAT measurements with different cards, you will see that sli has less input lag in almost every single one. The SLI
setups are beating the single card setups.

"Both of these games are using D3D11, which has explicit SLI support as part of the API. And in the case of the Frostbite engine used in Hardline, has significant attention spent to making mGPU support work well. In D3D12, there is no explicit SLI support, developers have to program for it themselves."

Yes, there is explicit dx12 support for multi gpu's. You have it backwards. in dx12 there is explicit support for mGpu's whereas dx11 the hardware wasn't. DX12 support for sli will provide MORE effective scaling, not worse.
This is taken directly from Microsoft:

Recap: What is Explicit MultiGPU and how does it make things better?
Explicit mGPU refers to the explicit control over graphics hardware (only possible in DX12) by the game itself as opposed to an implicit (DX11) implementation which by nature is mostly invisible to the app.

Explicit mGPU comes in two flavors: homogeneous, and heterogeneous.

Homogeneous mGPU refers to a hardware configuration in which you have multiple GPUs that are identical (and linked). Currently, this is what most people think of when ‘MultiGPU’ is mentioned. Right now, this is effectively direct DX12 control over Crossfire/SLI systems. This type of mGPU is also the main focus of this post.

Heterogeneous mGPU differs in that the GPUs in the system are different in some way; whether it be vendor, capabilities, etc. This is a more novel but exciting concept that game developers are still learning about. This opens up doors to many more opportunities to using all of the silicon in your system. For more information on heterogenous mGPU, you can read our blog posts here and here.

In both cases, MultiGPU in DX12 exposes the ability for a game developer to use 100% of the GPU silicon in the system as opposed to a more closed-box and bug prone implicit implementation.

Explicit control over work submission, memory management, and synchronization gives game developers the power to provide you the fastest, highest quality gaming experience possible; something only achievable with a well thought out game implementation, an efficient API, and a large amount of GPU power.

With Rise of the Tomb Raider, Nixxes leads the way on showing how to effectively transform CPU usage savings into maximum GPU performance using DX12 on mGPU systems.

 

[Trandana]

The graphs you showed were measuring frame time and not input latency. So again, we don't need to be discussing input latency here. Especially since you're talking about playing on 4K/5K monitors, of which there aren't even any 120hz/144hz consumer models available.

It measures the input latency in terms of the gpu! That graph shows how big of a variation there is between frames yes, but ALSO shows how LONG it took to render that particular frame. You can see the millisecond scale on the left side of the graph.

"Especially since you're talking about playing on 4K/5K monitors, of which there aren't even any 120hz/144hz consumer models available."
So?? 1) you can overclock 4k monitors, 2) 120/144hz 4k are right around the corner, in fact i beleive one was presented at CES this year. and you said competitive gamers don't play at 4k.....but they do.
Did you forget what you just wrote? That is why I wrote that the best 4k monitors on the market have the same input latency IF NOT BETTER than any 1440p monitor and the majority of 1080p monitors, including those with high refresh rates. Look at this website for input measurements for monitors. Best TVs and Monitors for Gaming: Input Lag Database


Think about it like this: when you switch from single channel ram to dual or quad channel, you are effectively reducing memory latency due to interleaving across mutiple channels at once, rather than over a single module.

 

[Rizen]

The graphs you posted were comparing the same GPUs in single GPU vs dual GPU configurations. Yes, obviously you are going to get lower frametimes there. That's not what we are discussing here. A Titan XP is much faster than a 1080. It's not a simple doubling of performance. I looked for TXP vs 1080 SLI frametimes but couldn't really find anything.

Regarding SLI in DirectX11 vs DirectX12, perhaps I should have used different wording. In DirectX 11, SLI support was implicit, and required less work for the developers because it was controlled by the API and the vendor drivers. In DirectX 12, multiple GPU support is explicit, which means game developers have to program multiple GPU support into the engine themselves. This does give the potential for better performance under DirectX12, but it shifts the burden to the developers, not NVIDIA/AMD, which means that it remains to be seen how this plays out in reality because many developers may just say "Fuck it", as multiple GPU customers are a tiny fraction of gamers, and the cost of developing good mGPU support may not be worth it. More info on this here: Rise of the Tomb Raider, Explicit DirectX 12 MultiGPU, and a peek into the future

Explicit mGPU refers to the explicit control over graphics hardware (only possible in DX12) by the game itself as opposed to an implicit (DX11) implementation which by nature is mostly invisible to the app.

Explicit mGPU comes in two flavors: homogeneous, and heterogeneous.

Homogeneous mGPU refers to a hardware configuration in which you have multiple GPUs that are identical (and linked). Currently, this is what most people think of when ‘MultiGPU’ is mentioned. Right now, this is effectively direct DX12 control over Crossfire/SLI systems. This type of mGPU is also the main focus of this post.

Heterogeneous mGPU differs in that the GPUs in the system are different in some way; whether it be vendor, capabilities, etc. This is a more novel but exciting concept that game developers are still learning about. This opens up doors to many more opportunities to using all of the silicon in your system. For more information on heterogenous mGPU, you can read our blog posts here and here.

In both cases, MultiGPU in DX12 exposes the ability for a game developer to use 100% of the GPU silicon in the system as opposed to a more closed-box and bug prone implicit implementation.

Explicit control over work submission, memory management, and synchronization gives game developers the power to provide you the fastest, highest quality gaming experience possible; something only achievable with a well thought out game implementation, an efficient API, and a large amount of GPU power.

 

[Araxie]

a big LOL, this guy Trandana is really confusing Frame Times with input lag?.. damn... that's bad and sad at the same time. man, the lower frametime the higher framerate that's why the VR test on [H] OCP are tested via frametime, their target is 11.1ms which is 90FPS, the same as 6.66ms is for example 144hz. you can be at a flat line 6.66ms and still suffer from horrible input lag.. lot of people confuse this with frametimes and FPS due the game responsiveness, the higher FPS the more responsive a game is up to certain points. however the input lag is an entire different thing and for sure as a past SLI user SLI add some input lag which for some guys can be terrible and for other simply acceptable as much as the FPS is high enough. But damn, never make the mistake to confuse frame times with input lag again..

 

[Rizen]

It measures the input latency in terms of the gpu! That graph shows how big of a variation there is between frames yes, but ALSO shows how LONG it took to render that particular frame. You can see the millisecond scale on the left side of the graph.

"Especially since you're talking about playing on 4K/5K monitors, of which there aren't even any 120hz/144hz consumer models available."
So?? 1) you can overclock 4k monitors, 2) 120/144hz 4k are right around the corner, in fact i beleive one was presented at CES this year. and you said competitive gamers don't play at 4k.....but they do.
Did you forget what you just wrote? That is why I wrote that the best 4k monitors on the market have the same input latency IF NOT BETTER than any 1440p monitor and the majority of 1080p monitors, including those with high refresh rates. Look at this website for input measurements for monitors. Best TVs and Monitors for Gaming: Input Lag Database


Think about it like this: when you switch from single channel ram to dual or quad channel, you are effectively reducing memory latency due to interleaving across mutiple channels at once, rather than over a single module.

I'm really getting tired of arguing this with you, lol. The 4K 144Hz monitor shown at CES will retail for $4999. Go for it, if you've got that type of money to spend on a single display. As far as this conversation is concerned that's pretty much irrelevant.

You're mixing up frame times and input lag here quite a bit and I'm getting tired of repeating myself so I'm gonna bow out of this one.

 

[thesmokingman]

You get all kinds...

 

[Trandana]

a big LOL, this guy Trandana is really confusing Frame Times with input lag?.. damn... that's bad and sad at the same time. man, the lower frametime the higher framerate that's why the VR test on [H] OCP are tested via frametime, their target is 11.1ms which is 90FPS, the same as 6.66ms is for example 144hz. you can be at a flat line 6.66ms and still suffer from horrible input lag.. lot of people confuse this with frametimes and FPS due the game responsiveness, the higher FPS the more responsive a game is up to certain points. however the input lag is an entire different thing and for sure as a past SLI user SLI add some input lag which for some guys can be terrible and for other simply acceptable as much as the FPS is high enough. But damn, never make the mistake to confuse frame times with input lag again..

Wrong.
Single GPU's have a best time frame latency of 1 ms, and a worst frame time latency of 2 ms.
MultiGPU's have best time frame latency of 1ms, and worst time frame time latency of 1.5 ms.

When the frame times are better in an SLI setup the SLI latency will be the same or lower, than the single cards latency. And in the case of 1080sli vs titan xp they are.

In the case of comparing say two 980 or 970's to a titan xp you would be absolutely right, there would be more latency but in this case you are incorrect.

 

[Trandana]

Lmao you posted the same source as me? Why?

You proved yourself wrong. Do you see your circular arguement?
When I prove you wrong in one thing you change the subject.

We all see the pros and cons of SLI vs single GPU, but you continue to go back and refute or debate topics we already discussed

The SLI frame times are in fact lower than the Titan XP, look at guru3D's fcat analysis

Even the 1070 sli achieved lower frame times, not by much granted, but its something.

 

[Araxie]

Lol are you going to keep trolling?.. you are just embarrassing yourself with those ignorant statements, the lower frame times are due higher frame rates you moron.. stop the trolling and get out of [H]ere, the higher framerate the lower frametime, and is not being related to a panel/monitor/tv input lag, you are just trolling yourself as you have no freaking idea of what are you talking about, please stop with that...

 

[thesmokingman]

Lol are you going to keep trolling?.. you are just embarrassing yourself with those ignorant statements, the lower frame times are due higher frame rates you moron.. stop the trolling and get out of [H]ere, the higher framerate the lower frametime, and is not being related to a panel/monitor/tv input lag, you are just trolling yourself as you have no freaking idea of what are you talking about, please stop with that...

Look at his join date/post history... smh.

 

[Trandana]

Lol are you going to keep trolling?.. you are just embarrassing yourself with those ignorant statements, the lower frame times are due higher frame rates you moron.. stop the trolling and get out of [H]ere, the higher framerate the lower frametime, and is not being related to a panel/monitor/tv input lag, you are just trolling yourself as you have no freaking idea of what are you talking about, please stop with that...

"and is not being related to a panel/monitor/tv input lag"
I NEVER associated the input lag of a monitor to a gpu.

Let me try to explain it to you simpler:

The GPU takes time to render each frame. Depending on many factors, such as the gpu's clock speed
the textures being rendered, the number of cores utilized etc it will take longer or shorter.
The time it takes to render that frame and send it to the screen can be seen as the GPU's processing time,
or in other words the LATENCY. I'm NOT saying the monitor's input lag, I am saying the processing time
to render the frame by the GPU(s). I am referring to the time data takes to send from the CPU/memory to the GPU(s) and finish the render.

Now, when you have a higher frame rate, this means it is taking LESS time to pump out a single frame, and the GPU(s)
can pump out more of them every second.

In the above examples, all those charts showing frame times, show the number of milliseconds it has taken for the GPU(s)
to render that specific frame. The charts are showing a 30 second period, over which the frame delivery times are analyzed.

In those charts, there are a few spikes which means it took slightly longer to render that specific frame or group of frames. This is known
as micro stutter, and it can be perceived as lag or skips if the lag is long enough.

As all these charts show, the time it is taking to deliver the frames, on AVERAGE, is lower than that of single card of lesser performance.
The key here is LESSER performance. Notice how the delivery time between the Titan XP and the 1070 sli is almost identical? That is because
a single titan xp is roughly equal to that of sli 1070s in terms of how fast it can display a frame to the screen.

Now, if we were talking about two sli cards that combined are slower or even equal to a single titan xp, then yes absolutely the Titan XP would have less input lag,
as SLI has a 1-1.5 frame latency, and it would deliver the frames faster.


HOWEVER, in setups where the dual cards are considerably faster than the single card, the dual cards will process the frames at a faster rate, and there
will be less processing time as compared to the single card.

Example; When you buy two 1080's INSTEAD of one 1080 you are not adding an extra frame or so of latency. The same is true when you buy dual 1080's vs Titan XP
but to a slightly lesser extent.

You seem to be assuming that a multi-GPU setup when compared to a single powerful are pushing the same frame rate, but they ARE NOT. The 1080's are faster.
The exception is if you play a game where a single card can match the FPS of the SLI cards, but in this case it can't.

 

[Rizen]

"and is not being related to a panel/monitor/tv input lag"
I NEVER associated the input lag of a monitor to a gpu.

Let me try to explain it to you simpler:

The GPU takes time to render each frame. Depending on many factors, such as the gpu's clock speed
the textures being rendered, the number of cores utilized etc it will take longer or shorter.
The time it takes to render that frame and send it to the screen can be seen as the GPU's processing time,
or in other words the LATENCY. I'm NOT saying the monitor's input lag, I am saying the processing time
to render the frame by the GPU(s). I am referring to the time from the CPU/memory to the GPU(s).

Now, when you have a higher frame rate, this means it is taking LESS time to pump out a single frame, and the GPU(s)
can pump out more of them every second.

In the above examples, all those charts showing frame times, show the number of milliseconds it has taken for the GPU(s)
to render that specific frame. The charts are showing a 30 second period, over which the frame delivery times are analyzed.

In those charts, there are a few spikes which means it took slightly longer to render that specific frame or group of frames. This is known
as micro stutter, and it can be perceived as lag or skips if the lag is long enough.

As all these charts show, the time it is taking to deliver the frames, on AVERAGE, is lower than that of single card of lesser performance.
The key here is LESSER performance. Notice how the delivery time between the Titan XP and the 1070 sli is almost identical? That is because
a single titan xp is roughly equal to that of sli 1070s in terms of how fast it can display a frame to the screen.

Now, if we were talking about two sli cards that combined are slower or even equal to a single titan xp, then yes absolutely the Titan XP would have less input lag,
as SLI has a 1-1.5 frame latency, and it would deliver the frames faster.


HOWEVER, in setups where the dual cards are considerably faster than the single card, the dual cards will process the frames at a faster rate, and there
will be less processing time as compared to the single card.

Example; When you buy two 1080's INSTEAD of one 1080 you are not adding an extra frame or so of latency. The same is true when you buy dual 1080's vs Titan XP
but to a slightly lesser extent.

You seem to be assuming that a multi-GPU setup when compared to a single powerful are pushing the same frame rate, but they ARE NOT. The 1080's are faster.
The exception is if you play a game where a single card can match the FPS of the SLI cards, but in this case it can't.

I agree with all of this, the issue I have with SLI and CrossFire is that the scaling does not always work out to be in the benefit of the two slower cards. So yes, if you get 80-90% scaling then it's great, if you're getting 50-60% scaling then it's not so great. NVIDIA has a better track record than AMD in this regard, but not perfect. So it really comes down to what your personal tolerance is for microstutter, fiddling with drivers and settings, and the games you actually play.

 

[Ninjawithagun]

Too much nuttiness in this thread and from ppl who have little real world experience using mgpu for years.

Not referring to me...I have over 10+ years of experience running SLI in multiple gaming systems. I replaced my two GTX980Ti cards with a single Titan XP and the experience has been phenomenal. I get about the same frame rates in games at WQHD and 4K UHD resolutions with a single Titan XP as I did with both my 'old' GTX980Ti cards in SLI, both at stock speeds and when overclocked. I was able to overlock my GTX980Ti cards to 1540Mhz/4100Mhz, so a very nice overclock for sure. My Titan XP overclocks to 2086Mhz/11000Mhz, so that is also very good. I am using a custom watercooling solution that remained the same for both setups. I was actually a bit surprised how well the Titan XP held up against the GTX980Ti SLI configuration, because the synthetic benchmarks would lead you to believe otherwise. However, in real world gaming, the margin is closed for the vast majority of games with a couple of exceptions. Overall, gameplay is smoother and watercooled temps are 10C lower. So, for me it was a win-win upgrade. I came into some money and so I'm now thinking about adding a second Titan XP, but the return of performance for the cost drops off a bit. Decisions, decisions ;-)

 

[dchobaby]

^ no coil whine for you? i have terrible on both of my cards

 

[thesmokingman]

^ no coil whine for you? i have terrible on both of my cards

Coil whine starts at the source in most cases. The odds of both cards whining on their own is improbable and may be indicative of their power source being the root.

 

[dchobaby]

Coil whine starts at the source in most cases. The odds of both cards whining on their own is improbable and may be indicative of their power source being the root.

i have an evga p2 1200 and also tried it on a corsair ax1500i. both whine. seems like there are a handful of owners on overclock.net with whining pascals as well.

 

[thesmokingman]

i have an evga p2 1200 and also tried it on a corsair ax1500i. both whine. seems like there are a handful of owners on overclock.net with whining pascals as well.

Ah that sucks. Have you tried testing with a single card? Have you gotten down and put your ear to the psu? No line conditioner or UPS in line I assume?

 

[ElBerryKM13]

i have an evga p2 1200 and also tried it on a corsair ax1500i. both whine. seems like there are a handful of owners on overclock.net with whining pascals as well.

Have two 980ti with a corsair ax1500i never hear any whine. Is this specific to the titan x?

 

[chenw]

Might even be down to specific cards themselves.

I have had to MSI Gaming X 970's before and neither whined unless I subjected them to unrealistic conditions (rendering at 300+fps for example), and there was quite a few complaints about 970's and 980's whining.

My MSI Gaming X 1080 whines, but it's quiet enough that I don't notice it half of the time.

 

[dchobaby]

Have two 980ti with a corsair ax1500i never hear any whine. Is this specific to the titan x?

not just the Titan, 1080 users are reporting it as well.

 

[dchobaby]

Might even be down to specific cards themselves.

I have had to MSI Gaming X 970's before and neither whined unless I subjected them to unrealistic conditions (rendering at 300+fps for example), and there was quite a few complaints about 970's and 980's whining.

My MSI Gaming X 1080 whines, but it's quiet enough that I don't notice it half of the time.

Yeah I have an open case (core p3) and now that they're pushing 100hz the whining is painfully obvious. sighs

 

[thesmokingman]

Yeah I have an open case (core p3) and now that they're pushing 100hz the whining is painfully obvious. sighs

I have one and run triple 144hz with no noise. This room is also a fresh drop, ie the wiring was new, new circuit (remodeled den). I also run with ups. Coil whine is always a mystery, some get it some do not. It doesn't follow an obvious order because if its a manufacturing tolerance issue, there should have been plenty of tolerance built in from the start to prevent issues like these. At least I would think that they had designed it so?

 

[Supercharged_Z06]

For me at least, moving to the single most powerful card available was pretty much driven by VR. Dual 1080's in SLI do squat all for VR gaming.

 

[dchobaby]

Troller, why are you still posting? And why did you send a pic of Chinese garbage?

took a while to get back to you but here they are: