My M18x turned into a zombie...so now I'm building that monster desktop I always talked about.

Stable but learning that LLC on 'Extreme' can have a discrepancy in measured voltage on the board and that displayed by .1 V. Back to screwing with it again, thought I was done with this!

[Official] Asus Rampage IV Black Edition Owners Club! - Page 811

 

Well another update, sorry to drag on a dead thread, but I have more good news to report. I managed to dial in 4.6Ghz at 1.408V stable and about a 400 point Physics increase in 3DMark 11, good temps.

Oh and for those who play Planetside 2, OH MY GOD, this game runs about 300-400% smoother now, I am not exaggerating, flying an ESF is no longer a slide show, everything is so smooth it is surreal. I can even turn shadow quality up to Ultra and VSync on and it doesn't make any change to the frames at all. Game still says I am CPU bound lol.

I am looking back at my previous benches to get a feel for what kind of improvement I have seen all said and told and it is remarkable to say the least. Comparing my scores back to what my M18x R2 was putting out with multipliers that didn't require 50+ Flex VID (45,45,44,43) and its about a 40% improvement in both 3DMark 11 and Firestrike Physics scores with Vantage showing a 50% CPU score improvement:

M18X R2 3DMark 11 @ 4.45GHz (45,45,44,43): 10k Physics: NVIDIA GeForce GTX 680M video card benchmark result - Intel Core i7-3920XM Processor Extreme Edition,Alienware M18xR2

RIVBE+4930 3DMark 11 @ 4.6 Ghz with 1.408V stable: 14.5k Physics: NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

M18X R2 Firestrike @ 4.45Ghz: 11.3k Physics: NVIDIA GeForce GTX 680M video card benchmark result - Intel Core i7-3920XM Processor Extreme Edition,Alienware M18xR2

RIVBE+4930 Firestrike @ 4.6GHz w/ 1.408V stable: 16.9k Physics: NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

M18X R2 Vantage @ 4.45Ghz: 29.5k CPU: NVIDIA GeForce GTX 680M video card benchmark result - Intel Core i7-3920XM Processor Extreme Edition,Alienware M18xR2

RIVBE+4930 Vantage @ 4.6Ghz: 46.5k: NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

Oh and the RIVBE+4930 temps for all of the benches above were usually around 60 C, for both the GPU and CPU's.

So to conclude this thread, I have to say that i am extremely satisfied with my decision to migrate to desktop for only a little more than the cost of upgrading my mobile GPU's. Can I set up desktop in a cafe? No. But if you can afford to do so, I highly suggest enjoying the best of both worlds. Aside from the obvious improvement in synthetic bench performance I can say that the desktop boots quicker, stays cooler (no game or normal program gets the temps over 50 C, not kidding, GPU doesn't even see 60 C, and I can't even hear it with my headphones on) and with a single GPU a few games run a lot better. And this thing was very fun to put together, and I learned a whole lot in the process, both about the hardware and Windows. Not just installing Windows (twice) but getting it just where I like it with a custom theme and everything else that I am sure you all do yourselves. And the motherboard's features are not superfluous as I thought they would be, I have used quite a few of them, just today I went out and picked up a digital multi-meter to measure the voltage my CPU is seeing on the motherboard versus what sensors, Windows and software is reporting as depending on the motherboard and CPU, very high levels of LLC can create a discrepancy between observed and actual voltage:

Load-Line Calibration: why overclockers should care - CPUs, Motherboards, and Memory - Linus Tech Tips

And I was happy to discover that not only was 'Extreme' LLC not pushing my VCORE into extremely unsafe levels but that I also had a little more headroom to try for 4.6GHz (well not really since I am over 1.4V, however so slightly). And the entire process was fun. If you guys think the M18xR2 is fun, wait until you build your first desktop, it is a blast.

[Official] Asus Rampage IV Black Edition Owners Club! - Page 812

Sitting here about an hour or so ago trying Planetside 2 for the first time on the new system, with no cacophonous din from the fans, looking at the temps on OSD, with my CPU sitting at 45C and my GPU at 55 C in a full 48 on 48 fight, I was smiling ear to ear knowing that I had made the right decision. I felt compelled to come back here and conclude my experience with this. Yeah sure, you will say that I should have waited for Haswell E and Maxwell, but there will always be something better right around the corner, that is the sad nature of Moore's Law (which CPU seem to be mysteriously ignoring, having yet to breach the 5.0GHz barrier for the past three years now). Three things happened in a row tonight that really lifted my mood enough to come and relay my experience as being very positive overall, I got through 35 minutes of Prime95 Torture at 4.6GHz with 1.408V with max temps between 75-80C, this 100Mhz improvement correlated to nearly half a thousand improvement in 3Dmark 11 and a 500 improvement in Firestrike, and I finally fixed my USB Creative X-Fi 5.1 sound-card which I have been trying to get working right for the past three days, I had mistakenly downloaded the wrong driver update and I was associating the error "unable to locate device on your system" with something else and must have uninstalled, clean booted, cleaned the registry in safe-mode etc. 15-20 times trying to get it to work.

Well I suppose I will go see what kind of improvement Vantage yields, if I break 47k CPU I will come back and update.

If youre on the fence about putting together a solid PC and have ~$2k to spare, I say jump! Feel free to use my build as some kind of rough template.

"I'll be back...."

 

Go go push the GPU harder. We want to see 19k GPU score.

 

PlanetSide 2 runs a lot smoother for you? Huh, that's weird. It didn't seem that way when I talked to you in-game right after you built your rig and there hasn't been a patch since then.

Game has been running like absolute garbage for me since the Implants update, FPS slashed in half across the board and CPU-bound all over the place. I drop from 60 down to 20-30 just from getting shot at or when I kill someone and the notification pops up. Needless to say, it's not very playable in this state and no amount of overclocking or underclocking or changing graphics settings seems to do anything on any of my systems.

Guess it's time for me to take another couple months off from the game until performance gets fixed again. I swear, no game I've ever played has ever had performance issues as off-putting as this one, and it's not because my hardware is inadequate (it's barely getting used...) or the game is ridiculously graphics-intensive or something (uh, the graphics suck...).

 

Zombie..... as in a server?

 

Server based platform but hardly a server......

 

So.....what are you doing with the M18x?

 

UPDATE!!

Unbeknownst to me, having only 2x8GB sticks of memory was what was resulting in my lower than usual Physics scores. If you remember me commenting on them, I said that they should be around 15k not 14k, well I added 2x8GB more Corsair Vengeance Pro at 2133MHz XMP and PICKED UP OVER 1K PHYSICS in 3DMark 11.

Quad Channel (15,297 Physics):

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

Dual Channel (14,235 Physics):

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

Oh and I also got 'Offset Voltage' working correctly, the issue simply happened to be the minimum processor state in the Windows Power Settings for both Balanced and High Performance; Balanced needs to be set to Min Proc State: 0 and High Performance from 100% to 10% (so that EIST works).

I also figured out how to create shortcuts for both Balanced and High Performance power plans on the desktop, assign hotkeys to them (Ctrl+Alt+1 for High Performance etc.) and then hide the icons. Very slick.

And I also managed to dial back the voltage 4.6GHz requires from 4.445 (as measured off the board with a digital multi-meter) to 4.04V by increasing CPU PLL from 1.8 to 1.875V. and it is Prime95 stable.

Now I just need to figure out how to "upgrade" from Windows 7 Home Premium to Professional so that I can use more than 16GB of memory, seeing as I want to allocate the excess 20GB or so to RAM Disk. I just now learned that the limit for Premium is 16GB, pretty lame if you ask me, I've already given Micro$oft plenty of money having one separately purchased instance of Windows on the three computers that I own and I don't feel like contributing any more money to Bill Gates' Eugenics program. So if anyone knows of a sure-fire, virus-free, less than legal way of upgrading to Professional or Ultimate simply to use all 32GB of memory please feel free to PM me.

I'm not pushing this thing over default voltage. The default EVGA SC vbios is actually very robust with default core of 1006 MHz and boost of 1150 Mhz. This is good enough for most games, and then if not I just go the +120/+450 as that is mostly stable with default voltage. I want this GPU to last so I'm not upping the voltage.

I really don't think PS2 is optimized for SLI, I am not getting any of the stuttering I used to get in first-person while driving a Lightning or Magrider or piloting a Scythe as I was with 680M SLI overclocked to the hilt. Even then GPU utilization was only around 50%. All of that is gone. It is truly a night and day difference with this single 780Ti. I don't believe it is the additional VRAM either as I never saw more than 75% of memory utilization with 680M. And I don't think it is the CPU, unless all 6 cores are truly being utilized, although I could be mistaken. The performance improvement, which is profound (60FPS everywhere except 48+ fights at Tech Plants, which then gets my frames down to ~45) is either attributed to switching to single GPU or the extra CPU power or a combination of both. But I am not exaggerating, it is about 4-5 times smoother.

As in although my 680M's are still strong on default clocks, they refuse to take an overclock of +250/+500.

I might sell it, it is still in good condition and the GPU's are still strong. Looking to get $2k for it (considering I paid $4300 for it new and taking into account the potential cost of replacing both GPU's with 780M SLI, CPU is still solid). Specs in signature.

 

Scratch what I said earlier. After this week's patch, performance is right up where it used to be again.

I'll never understand the monkeys that are coding this game.

(Of course it's not optimized for SLI or fast GPU's in general because it's so CPU-bound. The game is just not optimized, period).

 

After adding the two additional 2x8G sticks of Corsair Vengeance Pro 2133MHz I decided to allocate about half of that to RAM Disk with a program that came bundled with the mobo, ROG RAMDisk. I don't know how but its not just the load times that are a few seconds quicker (very noticeable in cumulative, especially when dealing with terminals, changing character class, redeploying etc, it all adds up) but I am hardly ever seeing less than 60 FPS everywhere now, it is incredibly smooth.

Nice thing about this program is that it loads whatever you have allocated to RAM Disk as Windows is loading so you don't have to go and put it on RAM Disk every time. And it isn't an additional 2-3 minutes, its maybe 20-30 extra seconds to Windows boot time.

The Akasa low-profile memory heat-spreaders arrived a few hours ago and I have to say that they are very good quality, the difference in weight between them and the decorative aluminum heat-spreaders that are on the Vengeance Pro is very noticeable. Should keep the temps a lot lower.

Back to Planetside 2, it sucks because now that I am having no issues with performance I feel that the game just isn't as fun as it was say a year ago. Part of it is the new Pay to Win Implant system and the other part, at least from my personal perspective, is that the VS feel underpowered and the population imbalance, especially the invasion of nothing but TR from Briggs, is a reflection of this fact. I don't think "no bullet drop" beyond 100 M is a trait that compensates for on average lower DPS, ROF, and damage-per-magazine especially now that the TR received a pretty significant arsenal-wise buff in the form of reduced horizontal recoil. Where is the commensurate buff to the VS? The TR's T9 Carv has been described by man who play in both factions as a "100 round magazine Orion". If you go on down the list its hard to make the argument that the VS are balanced against the other factions:

MBT, aside from strafing to successfully dodge incoming cannon fire from beyond say 200 meters, the Magrider feels like it is a distinct disadvantage under 100 meters, I find that I lose nearly all 1 vs. 1 engagements with the other MBT under 100 meters. The fact that the Saron is continually nerfed just adds to this, you can't even aim up to defend yourself against enemy ESF anymore, and Infantry? Forget it! So now everyone is using the PPA which makes them defensible against infantry but if an MBT shows up youre dead!

Empire Specific Heavy Assault Weapon:

It goes like this: MCG = Jackhammer > Lasher. Nuff said. Lasher sucks. The COF nerf it received makes it so you can't lay down suppressive fire beyond say 20 meters. It needs either the COF nerf reverted or a ROF and DPS buff or all three. I am not kidding. And what did SOE do last "balance patch"? They made the MCG absurdly OP! As though that was an issue!

Empire Specific Missile Launcher:

Here is what everyone will say if you say the Lancer feels anemic. "You have to use it in a squad". Yeah? What about the Phoenix or the Striker? They don't have to be used in squads? It is nearly impossible to assemble a Lancer squad even when running in a public platoon, they only happen spontaneously and then its only 2-3 heavies with it.

Empire Specific Sniper Rifle:

Phaseshift. Nuff said. "Don't go AFK or I will shoot you with my Phaseshift!"

VS Small Arms:

A few decent weapons, Pulsar Carbine, Polaris, Sirius (not my personal favorite but many seem to like it) but typically speaking, the NC and TR variants hit harder, have a lower TTK and larger magazines. Again, supposedly "no bullet drop" a benefit completely absent from weapons that would actually benefit from it (Sniper Rifles, Magrider main cannon etc.), and that does not seem to factor into engagements under 100 meters (90% of them) is a compensation for lower DPS. I say BS.

I think the developers are aware of the weapon imbalance but the fact that the player-base is thinning out and that possibly more higher BR players are among the VS, their refusal to address this issue could be justified from a purely financial perspective. I think they want veteran VS players, most of whom are no longer spending real money buying weapons etc, to become so frustrated they leave the faction for the NC or TR, knowing that they will no longer be at a distinct disadvantage. Once in the new faction and faced with the Sisyphean cert-grind, they will pay for premium subscriptions and purchase new weapons to get that whole process going along quicker.

If you look at the populations dynamics over the past 6-12 months, this actually seems to be the trend. The VS are losing players, even now, yet the NC and TR are picking up players. How many of these new players are completely new to the game and how many are disenchanted VS players tired of losing to overwhelming population and weapon imbalance?

If you understand SOE, you know that most of the decisions they make that seem to make no sense are carried out because of financial imperative. I am talking about releasing new weapons that are slightly OP so that they are attractive, players buy them in droves, then a few months later they announce that they are nerfing the weapons, a few weeks go by, then they are releasing a NEW weapon that is slightly OP, same players buy in droves, a few months go by and they announce that the weapon is OP and that they have to nerf, a few weeks go by and SOE releases new weapons that are slightly OP.....

It should be fairly simple to recognize this pattern and extrapolate it on to the factions themselves. Make one faction slightly underpowered, players leave for stronger faction, spend real money on premium memberships and weapons etc, a few quarters go by, SOE then buffs another faction etc. My question is, when will the VS get their turn at being OP? Because prior to PU01 or "Operation: Make NC OP" the TR were ridiculously OP (remember pre-nerf Vulcan and Striker?) and the NC, admittedly, were slightly under-powered with VS somewhere in the middle with pre-nerf Saron, Orion, SVA-88 and ZOE (not to mention that the PPA used to do damage against everything, including armor). Now? Same thing, TR is OP, NC about tied with it depending on the situation and VS need 65% of the population to win any fight. 50/50 population? VS will lose 100% of the time irrespective of the terrain of location (where that "no bullet drop" trait might come into play). Go ahead, next time you play take a mental note of the population, you'll see that the VS lose 100% of the time with equal population. If that isn't indicative of gross underlying weapon imbalance and problems I don't know what is.

So the point is, I am not interested in giving SOE more money or doing the cert grid again or joining an already overpopulated faction and the game is just not very fun with a distinct weapon and population disadvantage. So what do I do? I just stop playing the game as much. I used to play 4-6 hours a day, now I will get on and some days log-off within half-an-hour in disgust, others I might play for two hours, some days I don't play at all.

 

Update:

Curious to see whether or not the 14GB or so I have allocated to RAM Disk (out of 32GB) has an impact on memory performance I ran 3DMark 11 and it posted a higher score than with no RAM Disk! Talk about having your cake and eating it too!

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

Can I put this one up on the Best Benchmarks page?

 

Dang I'll never talk bad about RAM Disks again! Also my friend has that same board with SLI 780s and absolutely loves it

 

You will find many Benchmarks actually score higher with effectively less RAM XTU is a perfect example of this

 

Being able to have a big surplus of RAM for tinkering around with caching is the only reason I have 32GB of RAM. Having 16GB was more than adequate for ordinary gaming and benching. There are several benchmarks that get higher scores using cached memory, but as Brother scracy points out, it depends on what you're doing.

 

Update:

I've added that second 780 Ti and have made a few other changes. I'm seeing 22.5k GPU in Firestrike on default clocks and 24.5K GPU @ +100 core/+300 memory with default voltage:

https://www.youtube.com/watch?v=XGtlksj-Fdc

Default clocks: NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

+100 core/+300 memory: http://www.3dmark.com/fs/2319124

 

Nice! Good job.

It really sucks that Fire Strike gives so little weight to CPU performance. The more I run that benchmark the less I like it. The new Sky Diver benchmark seems just as demanding and does not de-emphasize the importance of CPU performance to the extent Fire Strike does. The reason I think Fire Strike sucks is because the difference in our physics scores alone (4.5K) even if the graphics scores were the same should put a huge distance between our results, but the overall score is barely 2K higher. I guess maybe it's a "self esteem friendly" benchmark for people with crappy CPUs to not to feel so defeated, LOL. 3DMark11 and Vantage are much more meaningful benchmarks, IMHO. If you have mediocre hardware you don't walk away from 3DMark11 or Vantage with an impressive benchmark result.

 

Yes I have to say that I absolutely love this board, I've used many of the features (secondary BIOS, DMM contacts on the front of the board, MemOk!, DirectBIOS) and the programs are actually really awesome (AI Suite 3, Fan Expert 2, ROG RAMDisk 2.1, ROG GameFirst 2 etc.) surprisingly the one thing I haven't used at all is the cherry on top, the 'OC Panel', it does look nice in front of the case though.

Could you elaborate?

Could you point me to a tutorial on how to do this?

You know what, youre absolutely right, Firestrike sucks. You can have a solid GPU array and a mediocre CPU and still attain decent scores.

I think 3DMark 11 is probably the most comprehensive in terms of reflecting actual system-wide performance potential, it is the only test that also really showed a significant difference between dual and quad channel memory, Firestrike didn't even register it when I upgraded to quad-channel; it resulted in a nearly identical CPU score while 3DMark 11 showed a 1k CPU improvement.

Here are my recent 3DMark 11 and Vantage runs with the GPU's on stock clocks (1006 core/1150 boost/1750 memory).

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

I noticed poor GPU utilization during Calico, certain scenes were showing 80-90 FPS yet GPU utilization was only 30% each! CPU was only 18% What gives?

It also doesn't seem to scale 780 Ti well, both '11 and Firestrike show roughly 90% improvement with 2x SLI (i.e. from 12k GPU single 780 Ti to 22.5k 780 Ti SLI) yet GPU in Vantage went from 47k to 71k? Huh?

And 3x SLI doesn't seem to scale well either, I think Vantage might be getting a little long in the tooth, it is also a total chore to have to sit through as it feels considerably longer than '11 and especially Firestrike. I think I will do a few more Firestrike Extreme runs when I get my 2560x1440 monitor (hopefully next month, fingers crossed ROG Swift's TN is as good as they say it is!) but will probably only rely on '11 from here on.

Oh and Unigine Valley and Heaven or enjoyable alternatives, but unfortunately they don't stress the CPU. It has been a few years now, Futuremark will probably be releasing a new benchmark, hopefully this time around it combines the impressive GPU test of Firestrike with a more demanding CPU test.

 

Benchmarks such as XTU,HWBOT Prime for some reason will give you a better result with around 8GB of RAM and 4GB of RAM with HWBOT both those benchmarks benefit greatly with tighter RAM timings,XTU particularly benefits from tighter secondary timings

 

I just realized the video I posted was set to private, well its public now, this is my PC in its finished state if anyone is interested:

https://www.youtube.com/watch?v=XGtlksj-Fdc

 

Hey everyone, I wanted to update this thread, I did manage to get in on the first wave of ROG Swift that was just released a few days ago (which you know, unless youre living in a cave) I will be updating my post-build video on youtube shortly, here's a few posts summarizing some of my initial impressions:

[Official] ASUS ROG Swift PG278Q Owner's club - Page 13

Newegg rocketed this thing to me in one day with 3 day shipping from Southern California to Northern Nevada by ground, extremely impressive; I ordered it yesterday morning and had it around noon today.

Initially, coming from an Eizo Foris 23" IPS I was non-plussed with the viewing angles, and what another owner describes as crystallization that is pretty noticeable on a white background, and now I am realizing that I have somewhat moderate light-bleed in the lower right-hand corner. After calibrating it to death (Brightness: 30, Contrast: 40, "User defined Color Mode" with R,G,B at their default setting) I am somewhat near the image quality/color fidelity of the Eizo Foris but until there is a DIY way for removing the matte coating and fixing the backlight bleed am not exactly 100% thrilled with this monitor.

I was playing Metro Last Light at 120 FPS and the smoothness is amazing, but with everything graphically maxed and 2x MSAA I am starting to see 90%+ core utilization (memory, excepting Titanfall, is usually around 75% surprisingly) in this and many other games with 780 Ti SC w/ACX SLI (factory clocks).

Titanfall was hands-down the most night and day difference, with V-Sync off and presumably 120 FPS (FRAPS and MSI Afterburner not showing FPS for me) it is so much more smooth and responsive.

Assassins Creed Black Flag, I can't seem to squeeze more than 60 FPS out of it, even setting the framerate to 120 FPS in the .ini and turning off V-Sync doesn't let me get beyond 60 FPS. Anyone know a fix for this?

Overall I give it a 7 out of 10. If it had less backlight bleed and no matte coating associated crystalization I would say 9 out of 10. I can live with the viewing angles, but the aforementioned two issues kinda dampen the ROG Swift experience.

...

I'm warming up to this monitor, playing Metro Last Light all settings maxed (2x MSAA) and then Crysis 3 all settings maxed (FXAA) at ~120 and ~90 FPS respectively is a whole new experience. I have to say that that is the biggest improvement, how smooth games are with G-Sync; the step up in resolution is nice but it feels nearly identical to 1920x1080 on a 23" monitor with a viewing distance of ~2 ft.

Oh and the improvement in Titanfall's performance is absolutely amazing, before with V-Sync enabled there was noticeable input lag and stutter, with it off tearing, now it feels like a buttery 120 FPS on all maps except Demeter (VRAM limited, funny how both Metro LL and Crysis 3 completely maxed out only use 1.5GB of VRAM at 2560x1440 yet Titanfall with its extremely low res textures requires double that, this and Watch Dogs are two examples of horrendous PC porting and should NOT be interpreted as the level of VRAM games will be demanding in the mid-term). What's amazing is that it is maintaining ~120 FPS at 2560x1440 on mostly the primary GPU as there is still no proper SLI optimization (secondary usually sees 15% or so utilization, primary around 95%)

I've also made a few more changes that have improved the image quality somewhat, based on the provided settings of another owner here I tried brightness at 25 and contrast at 50, "User Mode" color temp with R,G and B at their default settings: I like a slightly warmer tone and it seems everyone is simply turning down R.

I think I'm gonna stick with it, even with the moderate light-bleed emanating from the right-middle of the screen. I tried applying some pressure in this area and on the adjacent bezel area but it didn't alleviate the issue.

Yes I have to say, this monitor has mostly lived up to the hype.

Oh and I managed a Firestrike Extreme run, I could have done it a long time ago but I wanted to wait to do it on the ROG Swift as it is basically Firestrike at 2560x1440:

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

That's on default clocks.

Although my case has exceptional airflow, I have come to the realization that I may need to play with it completely unbuttoned as the temps can get up there now. I just ran around in the beginning of Crysis 3 for a while, about up until you pick up your bow and the utilization didn't drop below 80% on both cards, it stayed at 98% for most of the time at around 90 FPS avg.

Max temps were 71 and 64 C with a pretty cool ambient, feels like 60 F and I did see 76 C primary in Tomb Raider fully maxed earlier today with an avg FPS of 120. I think I am going to need to figure out a way to limit the FPS with some of these games as although 120 is better than 90 FPS the additional work-load is really pumping up my temps. Any suggestions? Funny, I've never actually had to limit my FPS, this is a first!

I am somewhat wishing I had a pair of Hydro-Copper Titans right about now, but I do like having both of my kidneys. I hate to say it, having gone the air-cooled route, but I believe liquid cooling may be a pre-requisite with this monitor. Here is my rig if anyone is interested, I will probably be adding another update tomorrow, I'm thinking of heading the subtitle "A Monitor Fit for the PC Master Race"

https://www.youtube.com/watch?v=XGtlksj-Fdc


Hope everyone is doing well, its late in the morning where I am, I will be back on here tomorrow to catch up.

 

Happy your enjoying yourself Brother Vulcan. I was reading through your post wondering if you'd privilige us with a benchmark.. and you did I'm curious, what does your machine pull on a 3d Mark 11 run?

 

TBONE! I miss you dude, here are my '11 benches with and without default GPU clocks:

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

+100 core:

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

What's new friend?

 

Just found this one, slightly better at +100 core:

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

 

Lol at 780ti sli running a 1280x720 benchmark.

 

What's 1280x720?

Well here's Firestrike Extreme (2560x1440) to make it official:

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

BTW, Metro LL at 120 FPS is absolutely eye melting, you guys gotta try this one day.

 

3dmark 11.

Do an official fire strike extreme run and vantage extreme (this is a great stability test).

 

3DMark11 Extreme (X) is 1080p and Performance (P) is 720p.

 

Mr. Fox I love your new avatar I see what you guys are talking about now, I will do an X run, for now here is a Firestrike Extreme at +100 core/+300 memory with default voltage/vbios:

NVIDIA GeForce GTX 780 Ti video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION

I was playing Assassins Creed: Black Flag tonight on the new monitor and was consistently seeing core utilization above 75% so I upped the clocks by this amount; this game is absolutely gorgeous on this monitor, I was in awe the entire 4 hours, but it's been like that with every game I have returned to, I am about to go through the DLC for Bioschock: Infinite at 120Hz So I was curious to see what kind of Firestrike Extreme it would pull with this OC on the GPU's, getting close to 10k!


Update:

Unlike Firestrike, I only have the freeware version of '11. It is the same preset I used for the benches in my sig though with the M18x R2.

 

Is the bench above official?

 

Signature updated

 

Thank you. I never realized how hard it was to find a really cool looking fox image for an avatar until I started looking for one. Most were cuddly and sweet-looking or just silly cartoons, but I really do like this one.

 

LOL! This is the best so far...

 

I've made a few substantial changes to my desktop, basically I'm running a hybrid AIO cooling solution etc. if anyone is interested:

https://www.youtube.com/watch?v=zSE8SYdFXwY

This was one of the limitations I was frustrated with with the M18x R2 and I was really hoping Asetek would have released their liquid cooling solution.

 

Yeah, gonna have to disagree with the end of that video. lol

 

I'd try using that rear fan as an intake instead of exhaust, especially since your top fans are intake, so basically your rear fan is sucking fresh air straight out the back lol. Another benefit is that it'll bring some fresh air over the motherboard VRMs, which can get a bit toasty if you're running some very computation heavy stuff. (I do some molecular modelling on the side so it really gives those VRMs a good workout)

 

Should know also with Intel Smart cache that minimum framerates are higher with more cache. That's why I went with the 3930K 2 years ago. It's not really shown in most reviews.

Anyway I miss the hell out of the M18XR2. I've been lurking on the M18 page a lot. Can't wait to see the 980M overclocked benchmarks.

 

Seems to me that at the point where you're running 3 AIO's on your system, it'd probably be better to go down the custom loop route...you could fit a rad on every intake/exhaust, probably giving significantly better temperatures. You'd be able to push that CPU quite a bit higher as well.

As others suggested, you'll probably want to flip that rear exhaust into an intake. You want it to be working with your rad push-pulls, but as an exhaust it's working against them by pulling air away.

 

I suspect you would have better temps if your fans in the front were intakes. A little warm air flow across the gpus and motherboard would be fine. More so than very little or no air flow. That side windows fan isn't going to do much because you air flow is not proper. Gotta keep those vrms on the MB cool man... I think you could have easily did a custom loop for the cost of all the aios and windows fan...

Nice system brother.

 

What part? The H60 in push-pull handling a 4930 or my recommendation to skip the Kraken G10 altogether in favor of a semi custom loop?

Thanks for watching it all the way through.

I did experiment with the rear fan as intake, as I also see that the top rear 140mm fan's airflow may be going right out the back with it as exhaust BUT that heat-bridge gets really hot with 180W going into the CPU, the airflow is doing some work running over the top of it and out the back. I may again experiment with it as intake, with both configurations there is very little to go by in terms of CPU and GPU temps as to what is better, but the bridge there to the left of the memory is noticeably cooler with it as exhaust and I believe that it is evacuating heat accumulated by the copper heat-sinks surrounding the AIO bracket on the primary GPU.

Same, I am also excited to see 980M SLI up and running in the R2.

Ok you guys have me convinced, I will return rear fan to intake again. As far as pushing the CPU any further, I've overclocked the snot out of it, of done everything under the sun to squeeze more out of it, including experimentation with loosening the timings and upping the CPU PLL voltage to its limit AND I even had DRAM voltage at 1.675V where I did have 4.6GHz stable for a time at 1.4V as measured on the board with a multi-meter BUT after upgrading to Nvidia 3D Vision that stability was lost as I believe the new consistently higher GPU loads were really seeking out any weakness with the CPU OC and after trying 1.425V, a big no no as far as the voltage limits of 4930 is concerned, and not being able to reign 4.6GHz back in I felt compelled to dial back the GHz to 4.5, and in the process was able to reduce VCORE from 1.425V down to 1.37V and DRAM from 1.675 to 1.55V, as win if you ask me.

TLDR: The issue isn't the inability of the small 120mm H60 AIO to handle heat generated, the issue is that I am at the voltage limit.

Thanks for the feedback but I would have to humbly disagree in this instance, the weakness of the Kraken G10 cooling solution is that the VRM's are not benefiting from the liquid cooling with many reviews showing an actual increase in VRM temperatures on reference Kepler cards with no VRM cooling mid-plate (unlike the EVGA SC ACX, Asus DCU2 and MSI Lightning 780 Ti which have a VRM cooling mid-plate to dissipate heat in conjunction with the G10 92mm fan and show a decrease in VRM temps).

Having the radiators as intake, directly in front of the GPU's, and blowing that HOT and I mean HOT exhaust right into the VRM's would literally be a death wish.

As I note below the video, if this were a waterblock set-up, intaking the radiators would be somewhat feasible.

I don't know if youve ever felt the exhaust from a 300-400W GPU but I can tell you that it is much, MUCH warmer than ambient. Think hair dryer. You would NOT want that flowing over your GPU's.

As far as airflow is concerned, I neglected to show in the video, but I am still running another 140mm fan in the PCIE bracket area which is definitely helping because under load the aiflow coming out of it is noticeably warm.




And as I note in the video, I completely agree, I could and should have done a semi custom loop. The allure of the G10's is their low price to performance ratio, but what I've spent here, including replacing the CPU air cooler with the Corsair H60, 3x SP120 LED fans, and all of the copper heat-sinks put me only $100 under a pair of EK water blocks, a Coolermaster Glacer 240L and another 360mm radiator. I didn't know this starting out, had I known I wouldn't be saving a whole lot I would have gone the semi custom loop route.

Thanks for all the feed-back everyone! I got more here than on OC.net lol.

 

The part where you say,
"The performance concerning your vram temps."
Here are some of mine using 780 ti's and 980's in various configurations.






I did add a fan on top of the cards to see if anything changed but nothing changed. In the 4 card setup the msi & gigabyte cards are venting the heat straight up. (Or inside of a case. If I had one.) While the 780 ti's are venting heat outside the case. (Again, If i had one) This is getting far hotter than your setup and I do not have anything extra on the vram chips. The only thing making them run worse is over clocking them too high and you are not raising vram voltage. Which is completely separate from core voltage. If you were running kingpin classified 780 ti's and evbot, then you would have control. I would be a bit more concerned about vregs than the vram

 

Yeah I've felt the hot air, ram my three kingpin cards on air for a few weeks to test before building my system/loop.

I was talking about airflow over the mothetboard mosfets etc.. And yeah some warm air over the gpu and mothetboard is still beneficial and more so than no air flow even if under water..

 

I apologize for the confusion, I actually am saying the issue with the Kraken G10 solution is the VRM temps, not the VRAM as the following Puget Systems article demonstrates by adding the G10 to a reference Titan:

NZXT Kraken G10 Review - Puget Custom Computers

Further, in the comment section, you will find those who undertook the review commenting that they saw an 8-9 C reduction in VRM temps using a 200mm side panel fan, similarly to what I have done (italics mine):

"We actually tested a couple of different configurations before we decided to use this fan setup. Double check the "Test Setup" section, I think you might have missed the little bit of text where we said that we have both a top fan (which vents the heat) as well as a side fan blowing directly onto the card. This setup ended up raising the GPU core temperature a couple of degrees, but we found that it also great reduced the VRM temperatures. Removing the side fan (which makes the overall intake/exhaust levels more equal) did result in lower core temperatures, but the VRM was about 8-9 C hotter. We didn't try with the front radiator fan as an exhaust, but we've done plenty of testing with closed loop CPU coolers, and having them set as an intake drops the temperature of whatever they are cooling by a good 5-7 C in most cases.

As for the passive MOSFET/VRM heatsinks, there are two reasons we didn't use them. First, NZXT doesn't supply them with the cooler, and since we are testing the cooler as they supply it, we didn't want to start adding additional parts to fix the problems. Second, those heatsinks are almost always going to be a permanent addition which means you likely cannot switch to a different cooler (or even back to the stock cooler) in the future if you want to."

In my opinion, the only way to safely use the G10's, is if you already have a non-reference card with a VRM cooling mid-plate to work in conjunction with the G10's 92mm fan and the addition of copper heat-sinks in the VRM area, otherwise this is really not a recommended cooling solution, especially if you intend to overclock.

If anyone is interested in the G10's for their desktop cards, I highly recommend the following min-guide by "Faceman" here, where he also points to the testing done confirming a 20C reduction in VRM temps using a non-reference 780 Ti with a VRM cooling mid-plate. If youre using reference Kepler, you can probably expect an increase in temps as show in the Puget Systems article above. From what I gather, sustained VRM temps above 85C will diminish the life of the card.

http://linustechtips.com/main/topic/232654-kraken-g10-help/

 

And how is this working out for you?
http://www.3dmark.com/compare/fs/3118847/fs/2733742
Looks like some efficiency is being lost somewhere.
NVM, I see why. You ran windows 7 instead of windows 8.

 

Btw I echo EvilCorsair's sentiment that when you have 3 AIOs in your unit, perhaps it really is time to think about getting a custom loop going.

You don't need that much radiator if you're not aiming for a 10C water delta. It's quite amazing what a single 280mm rad is capable of. Yes that's THREE 780s in tow with a 3930K @ 4GHz, all on ONE 280mm radiator. Granted it is a 60mm thick rad and one of the best you can buy, but I think the example is illustrative.

 

Exactly. AIO's are great if you're looking for easy CPU cooling without messing around with a custom loop, but having three of them is going to be very inefficient compared to a custom loop.

If you swapped those AIO's for a custom loop with a good 360mm radiator, you'd have awesome temps across your entire system. Plus, if I remember correctly, waterblocks for the GPUs will also provide much better cooling to the VRM and VRAM. And if you wanted it to be whisper quiet (and had the money to burn), you could swap all of those fans for Noctua's. You'd have a system running very cool even under 100% load while likely being inaudible without you pushing your ear against the side of the case.

 

Full cover blocks will if designed correctly be much better yes, of course to a lot of people a full on custom water loop is intimidating.

 

It would be intimidating to anyone doing it for the first time. Just as it was intimidating for me, and most likely many others, to open up a computer for the first time, or build one for the first time. Everyone starts somewhere, and if you follow proper guidelines and the advice of those with experience, you'll be fine.

 

I started out the same way, but then i made something like this after reading about what others were doing to control temps.
Two GTX295's on water and one on air. That's 6 gpus.


This was a long time ago though....

 

I'm looking to upgrade to 10 when it comes out, as much as I am sick of Micro$haft's shenanigans, if DX12 doesn't trickle down to 7 I don't very much have a choice with a hex-core CPU. Is 8.1 any better than 7 overall? I hear it's faster, but a lot of games seem to demonstrate poorer performance with it vs. 7.

That's fairly impressive for a single 280mm radiator, although that is still a lot of rad surface area (280x60mm).

Yes, I agree 100%, as I state in the video, had I known from the outset that the total cost of the finished hybrid cooling set-up would run me only slightly less than a custom loop I wouldn't have gone that route. I also wasn't aware that there were expandable AIO's such as the Swiftech H220 and Coolermaster Glacer 220L at the time of planning the system and I was actually intimidated by the complexity of a custom loop. I had already contacted EKWB and priced out a full custom loop but in the end there was some uncertainty as to where I would put the water-pump / reservoir combo, one option was in the back compartment of the case where all the PSU wires are jumbled. Ultimately the price, around $600 pushed me to the AIO route, as I was under the assumption I could accomplish comparable performance for half the price (~$100 per card, G10 bracket and AIO etc.) In the end, with the cost of the custom side panel, literally half a pound of copper heat-sinks, the 3x SP 120 LED fan's and the replacement of the Phanteks air-cooler with the Corsair H60 I am sitting right around $500:

2x NZXT Kraken G10 White: $60
2x Corsair H55: $100.
1x Corsair H60: $60.
4x Enzotech BMCR-1 copper heat-sinks: $68.
3x Enzotech BCC9 copper heat-sinks: $51.
3x Corsair SP120 LED Fans: $33.
Gelid GC Extreme thermal compound: $12.

Bitfenix Spectre Pro 230mm: $22.
Custom side-panel fabrication: $80.

Total: $488.

But I should point out that the cost of the last item, the custom side-panel fan, might have been added to the cost of the semi custom loop route, I really like it and it would help feed air to whatever radiator I had positioned in the front of the case.

Again, had I known from the outset I wouldn't be saving very much with all of the above I would have gone with water-blocks.

All of that said the performance of the G10 mated with an H55 is phenomenal. Peak temps in Heaven and Valley dropped from 90 to 55C at 1254 core and 2k memory with a Skyn3t vbios, a reduction of 35C. And although I haven't taken an IR thermometer to the VRM's, testing by cards with VRM cooling mid-plates ALONE, not with a plethora of copper heat-sinks, such as Asus DCU2 and MSI Lightning show a 15-20C reduction in VRM temps. I'm probably sitting 25C lower than with the default coolers, especially with the introduction of the side-panel fan.

So it may be fair to say that in the end I am seeing comparable performance to a loop with maybe 10-20C higher VRM temps (VRM temps with water blocks are exactly the same as the GPU core, typically peak load around 45-50C with a healthy OC) and the saving grace of this set up is that unlike water-blocks, this system, specifically the G10 brackets, are forward compatible with not only Nvidia GTX 970 and 980 but also AMD cards whereas with water blocks I would need to add the cost of new water-blocks whenever I do upgrade three to five years hence.

So in the long run the savings will begin to appear, if I do stay with this system.

The only real regret I have is that I could have more surface area had I known the two aforementioned expandable AIO's were even available when I inititally planning the system. I could have a 240mm rad in the ceiling along with a 360mm rad in the front of the case, nearly double the surface area.

Budget allowing, I may actually take the hit and replace the H60 and H55's but to be honest I am seeing stellar performance even with the limited rad surface area.

Peak avg. Prime95 temps under 75C (small FFT's) and typical peak load temps of 55C on the GPU's is absolutely amazing for 120mm AIO's. Seriously though, the H60 (with two fans in push-pull, essentially making it an H75) has completely shattered my expectations of AIO performance. I kind of new from the outset that I could expect performance 3-4C behind that of an H80i with an H75 and avoid the myriad reported issues of the Corsair Link software and the fact that both the H80i and H100i have BIOS in the pump itself but I am still pleasantly surprised that this cute little AIO is handling my 180W TDP beast of a CPU without issue.

6 GPU's, are those R295x2's?

 

As to windows 8 and 7. I tend to start with no tweaking and then only make one change at a time to see if things get better or worse. For games, I just let games decide. Then tinker from there if need be.

As to the cards.
EVGA GTX295 Red Edition GPU's

I was testing to see if it all 6 gpus could be used, but only 4 could be used in sli at any given time. I used the other one for physics.