Blace

seyumi
 
Quite simply, single core/thread performance. Almost every single PC game in existence relies heavily on 1 core while all the other 3~7 cores get a pathetic 15% usage or whatever. Even games that flaunt saying "Our game will use as many cores/threads you can throw at it," it still primarily just uses 1 core (Arma 3 for example). Direct X 12 is supposed to 'help' remedy this by "spreading out the load" on all the cores more evenly which is something we've been wanting several years ago ever since quad cores were introduced. I unfortunately doubt this will happen for a long time and would assume the game would have to be built from the ground up on DX12 and not simply just ports. Of course, we have to assume game publishers actually optimize PC games which we all know will never happen.
 
If you look at CPU gaming benchmarks, the $350 4970k almost always outbeats the $1000 5960x in practically 90%+ of all gaming benchmarks due to the higher core clock. You can realistically get a 4970k to about 4.9Ghz w/o much issue. The 5960x you're looking at around 4.4~4.5Ghz good overclock. This gap will become even larger once Intel release Skylake supposedly coming out in a few months. The i7 6700k will replace the 4970k with an estimated 5%~15% performance improvement. This means a 6700k at lets say 4.8ghz will be roughly the same as a 4970k at 5.04ghz~5.52Ghz. Potentially anywhere between 600mhz~1.1Ghz more than a 5960x which is actually quite a bit. For some reason, the mainstream platform always gets new technology anywhere between 6~24 months before the enthusiast "X" platform does. The X99 platform will probably just receive a +100mhz bump on the 5960x calling it the 5970x and will be stuck with this for another whole year.
 
Many older games such as flight simulators and MMO's are about 95% always 1 core games. I doubt we'll see any MMO's built from Direct X 12 from the ground up anytime soon. Even the PC Master Race game Star Citizen doesn't use cores that well. Of course it's still in pre-pre Alpha so not holding it to high standards just yet. Another example, Skyrim only uses 2 cores. For every +100mhz in CPU speed was like another 5FPS or something like that I forgot the exact #'s. For me personally, if I got rid of my 3rd Titan X and changed to a Skylake platform, there would be half a dozen games that will decrease in performance wanting that 3rd card for 4K resolution but at the same time I'll also have half a dozen games that will increase in performance that need more than my 4.4Ghz on 1 core. So it's really a wash but I end up saving thousands of dollars.
 
FYI: my source is I have a little USB monitor where I can see real-time usage on all my CPU cores and GPUs (% used, temps, current clock, ect.). I've tested dozens of games from 2013+ and it's all pretty much the same story. 1 core takes the most load and the rest are not really being used that much. I bought my system when there was no such thing as a Titan X but only 980's. You pretty much needed 4x 980's to get decent FPS on a 4K monitor. 2x Titan X's are roughly 3.5 980's in performance so more or less 4x 980 performance. If I bought my system when the Titan X's were already out, I probably would have went with a Z97 platform instead. Just my personal preference of course. I plan on waiting out Skylake to see what happens. There are games that do indeed benefit from a 3rd & 4th Titan X on 4K.

 

subn3t
seyumi's detailed explanation covers pretty much all the important points.
 
Just my $0.02 on the matter: From the research that I've done, I think 4.9GHz on a 4790K is closer to an "epic overclock" than "can hit it without much issue". 4.7 is a more realistic OC, and even then you can't always get that just because of the silicon lottery. Tip of the hat to anyone who went with a 5960X, but it's a very expensive processor and from a bang-for-the-buck perspective, most people go for the 5820K as it's the exact same as a 5930K for all intents and purposes until we start talking quad-SLI. The actual comparison you should be making is something like 4970K@4.7 vs 5820K@4.5, especially since they're similarly priced.
 
For pure gaming, the 4790K is a better processor, and a Z97 setup will cost less than X99 (though the difference is relatively small in the context of Titan X systems). However, it really is so close as to be "a wash" as seyumi said. However, when applications CAN utilize the extra cores, the 5820K comes out on top by a much more significant margin. 5820K > 4970K in BF4, for example. AnandTech has a great article that covers this from August 29, 2014. Sorry, my account is too new, and it won't let me make links.
 
Personally, I went with X99 just because 4790K felt like a sidegrade from my 2600K@4.2. 6-cores felt like a more worthy upgrade for my SLI Titan X's, and I do just enough non-gaming stuff that utilizes the extra cores for it to make sense. Also, all the reviews for the Rampage V Extreme were so glowing that I really wanted to get one!

 
Wow so many posts already XD I backtracked your responses and just wanted to thank you both for your detailed input. I agree that if you OC both then the Z97 CPU's will win on single core performance any day, but I guess personally that difference is negligible, as long as I get the 5930k to the 4.5Ghz area. I'm going to go for thatone because I'm planning to fully utilize the PCI-e lanes with a couple of additions like a PCI-e SSD and perhaps an M.2 version in the future (not possible with the 5820k in some configurations).

Outside of gaming the video/photo rendering/editing speed with a 6 or 8core just wins it for me. I might upgrade it to an extreme edition once broadwell-e shows up, to make the switch to an 8core. If these AIO coolers prove their worth (I'm sure they will), I might also go for a triple Titan X setup in the future.

More on topic: thanks for all these temp posts guys! Truly awesome results. I guess some of the difference people are mentioning are just due to hot air being 'stuck' between the 2nd card and the PSU, or the 2nd card and the 1st card. Also, some folks have a wider cone of air blowing onto the 2nd card from one of the radiators (when they are intake ofcourse). As long as it is a 5 degrees difference on average I don't see it as a problem.

I really do hope that if they release a Titan X shroud, that it looks as badass as the current one on the Titan X. I would pay a little extra for it to be metal like the original one.

Edit: also, seyumi, could you tell me at what fps numbers you are getting microstutter with fc3 and bioshock infinite? and with what monitor? I'm kinda curious about 4k, but I'll stick with my 144Hz Gsync panel for now, and in some cases Gsync is supposed to eliminate microstutter.

Nitemare3219
 
You guys with these small cramped cases are making me cringe! My 750D has tons of room and I love it!
 
Anyway, are any of you replacing the pre-applied thermal paste with anything else? I grabbed some IC Diamond because it has always treated me well and was planning on using it, but a lot of you are getting great temps as is! My only concern with pre-applied is the possibility to air bubbles getting trapped during the mounting process.
 
I'm surprised more people aren't OC'ing harder with custom BIOS! Just back yours up and you should be good to go... I'm aiming for 1500 MHz core and 8000 Mhz memory!
 

seyumi

subn3t
Which brings me to my question - I think I'm going to get Corsair SP120s and 140s to replace the stock stuff that came with my case. Should I get the quiet edition or the performance edition? I think I would really like a quiet PC, but if it's going to make a big diff in temperatures, I'll go with performance. Any thoughts?

There's no reason why you would need anything more than the quiet editions. Your temps are already really good it'll probably go down even more on the quiet edition fans. You probably know that those fans are more or less the best radiator fans in existence and beats all the competition by a mile on sound/air push ratio's. The high performance editions might drop your temps a few C at the cost of probably double to triple the noise. Not really worth it considering these GPU's don't throttle unless they hit 80C or 90C or something. I got borked airflow on mine and 1 of my GPU's gets up to 69C. Doesn't bother me one bit.

Also, I see a lot of you recommending 2x SP120 fans. That is fine and all, but Linus actually got his best results with SP120 in push, and AF120 in pull - just food for though, and it makes sense! This is what I'll be doing when I install tonight or tomorrow, along with testing intake and exhaust at the bottom of my 750D.
 

subn3t
** also, I have no idea how you guys are getting 1490 out of your GPUs. [H]ard|OCP reviews said they got +220 core out of theirs with the fans running at 100%. I ran my OC benchmark at 220, but 250 caused me to crash.

 
Have you increased your voltage? Also, keep in mind, for example, a +200 MHz increase does not include the stock boost clocks. +200 MHz typically translates to ~1400 MHz actual, bearing stock boost based on temps/power target.

seyumi

Blace
Edit: also, seyumi, could you tell me at what fps numbers you are getting microstutter with fc3 and bioshock infinite? and with what monitor? I'm kinda curious about 4k, but I'll stick with my 144Hz Gsync panel for now, and in some cases Gsync is supposed to eliminate microstutter.

I run Bioshock Infinite at vsync 60FPS on my 4K monitor. There were a few scenes where my 2 Titan X's would go full 99% load and the FPS would dip below 60FPS so then I put a 3rd Titan X in. I immediately noticed a horrible difference. I was now getting a 100% consistent 60FPS but the game felt horrible like I was getting 30FPS or something at ALL times even though I was technically at 60FPS. After some research it looks like it's some form of SLI latency and people recommended using a frame limiter. This actually worked for me. I had to use a 3rd party program and limited the frames to 58FPS (needs to be below the monitor hz to counter vsync lag). The game was now smooth again just like when I had 2 cards. There were times where that 30FPS feeling came in but it was only for a few seconds. Not as consistent as just using 2 cards but was at least bearable. Farcry 3 was similar but my FPS was just all over the place when I went from 2 cards to 3 cards. The frame limiter helped but it was still more consistent on 2 cards. I'm sure this is just a problem with these games and their SLI profile to be honest. I've played about half a dozen other games and don't experience this (FFXIV, Tomb Raider, RE6, Sniper Ghost Warrior 2, Saints Row IV). I hope I don't run into anymore of these issues on my dozens of other games I haven't played yet.

MrImSoGood

bcavnaugh

Nitemare3219
 
You guys with these small cramped cases are making me cringe! My 750D has tons of room and I love it!
 
Anyway, are any of you replacing the pre-applied thermal paste with anything else? I grabbed some IC Diamond because it has always treated me well and was planning on using it, but a lot of you are getting great temps as is! My only concern with pre-applied is the possibility to air bubbles getting trapped during the mounting process.
 
I'm surprised more people aren't OC'ing harder with custom BIOS! Just back yours up and you should be good to go... I'm aiming for 1500 MHz core and 8000 Mhz memory!
 

seyumi

subn3t
Which brings me to my question - I think I'm going to get Corsair SP120s and 140s to replace the stock stuff that came with my case. Should I get the quiet edition or the performance edition? I think I would really like a quiet PC, but if it's going to make a big diff in temperatures, I'll go with performance. Any thoughts?

There's no reason why you would need anything more than the quiet editions. Your temps are already really good it'll probably go down even more on the quiet edition fans. You probably know that those fans are more or less the best radiator fans in existence and beats all the competition by a mile on sound/air push ratio's. The high performance editions might drop your temps a few C at the cost of probably double to triple the noise. Not really worth it considering these GPU's don't throttle unless they hit 80C or 90C or something. I got borked airflow on mine and 1 of my GPU's gets up to 69C. Doesn't bother me one bit.

Also, I see a lot of you recommending 2x SP120 fans. That is fine and all, but Linus actually got his best results with SP120 in push, and AF120 in pull - just food for though, and it makes sense! This is what I'll be doing when I install tonight or tomorrow, along with testing intake and exhaust at the bottom of my 750D.
 

subn3t
** also, I have no idea how you guys are getting 1490 out of your GPUs. [H]ard|OCP reviews said they got +220 core out of theirs with the fans running at 100%. I ran my OC benchmark at 220, but 250 caused me to crash.

 
Have you increased your voltage? Also, keep in mind, for example, a +200 MHz increase does not include the stock boost clocks. +200 MHz typically translates to ~1400 MHz actual, bearing stock boost based on temps/power target.

IC Diamond may cut into the GPU, I use MX-4.
On this same subject is their a cover on the preinstalled TIM that needs to be removed be you install it?
I wonder because of the high temps that astrallite is reporting "Idle about 60C, load 91C"

Cover? I'm interested in what you mean by that

bcavnaugh

subn3t
Wow, I can't even keep up with this thread anymore. As for OC w/ AIO, I have been doing a modest overclock. I'm waiting on 4 fans to arrive from Amazon on Thursday, and also my EVGA backplates. Once I get those installed, it'll be time for the real overclocking.
 
As for big case vs. little case, I actually really like having everything in a tight, compact case! Makes the cable management a challenge, that's for sure!

Yes, maybe the Overclocking should go under a different Thread.
A Title might be " Titan X AIO HYBRID Overclocking"

Avenger411
Hey
 
Nope, no k-boost, no voltage settings have been changed. I get 1380mhz  (fire strike : 17000) core stable with 47c load with the AiO 980 cooler :) ultra silent too! Have to say i do not have the humming people seem to talk about. Could it be because my rad is placed horizontaly and NOT verticaly ? this could be a fix :) I recall my original h100i fans used to rattle while placed horizontaly and would be silent verticaly. I ended up changing them for SP120's white led which do not rattle while being in a horizontal position. So bottom line, for those of you that have a humming rad, is your rad vertical or horizontal ?
 
Max

Avenger411
Hey
 
i think you were referring to this
 
 
 
Indeed, food for thoughts, now i have to change my fans on my order ;D
 
Max

bcavnaugh
Nice Video but I do not think the extra Cost for the Fans are worth the 2 or 3 Degrees.
The one item he did not talk about was Fan Noise with each of fan types.
But he did talk about Case Air Flow Fans and Pressure Fans for use on Radiators

tackle70
So.... any idea how long a wait we have before getting a version of this cooler with a Titan X branded shroud? We talking weeks or months?
 
I'm tempted to just buy these but can't tell if I will become OCD about having two cards with Titan X backplates but GTX 980 cooler shrouds