1440p and vram

[bystander36]

A 3.3GHz base / 3.7GHz turbo 4C/8T Haswell is a bottleneck at 1440P and 2GB vram isn't?

It all comes down to with what you want out of your system.

As RaulF has eluded to, if you want 8x MSAA, 4x SSAA, 200% resolution/downsampling and all the settings turned up extremely high, then 2Gb of Vram can be an issue.

If you are like Communism, and you like 80+ FPS, or at least never have dips below 60, and you use modest AA settings, which helps get you those FPS, then any stock clocked CPU is going to be an issue.

 

[Robster]

Quote: Intel Xeon E3-1230v3

You'll be CPU bottlenecked before anything matters.

Get a 4670k 4.2ghz+ or better before even thinking about getting more than 1x 770 GTX.

Everything else discussed is 100% pointless in the OP's situation.

Do you mean it would be pointless for me to get another GTX 770 because of my CPU? If I understood your opinion correctly, could you elaborate why?

 

[Carfax83]

Do you mean it would be pointless for me to get another GTX 770 because of my CPU? If I understood your opinion correctly, could you elaborate why?

SLI requires greater CPU usage, so your stock clocked processor would likely bottleneck two GTX 770s. Although it would depend on the type of game you're playing and your settings..

Although in truth, it's not so much that the CPU itself is too slow. It's mostly because of the overhead associated with DirectX, and the fact that most DirectX games don't use multithreaded renderers. Even if they do, they are still limited to two or three threads assuming they used manual threading, like Battlefield 4 for instance.. DX11 multithreading is better than manual threading, but only NVidia supports it.

DirectX 12 is supposed to fix all of the overhead and parallelization problems associated with the Direct3D API..

Anyway, can't you just overclock your CPU?

*Edit* I guess those Xeons don't have an unlocked multiplier.

 

[OCGuy]

How about i use what i want! :\

What about COD Ghost? Any comment on that one?

 

[Robster]

SLI requires greater CPU usage, so your stock clocked processor would likely bottleneck two GTX 770s. Although it would depend on the type of game you're playing and your settings..

Although in truth, it's not so much that the CPU itself is too slow. It's mostly because of the overhead associated with DirectX, and the fact that most DirectX games don't use multithreaded renderers. Even if they do, they are still limited to two or three threads assuming they used manual threading, like Battlefield 4 for instance.. DX11 multithreading is better than manual threading, but only NVidia supports it.

DirectX 12 is supposed to fix all of the overhead and parallelization problems associated with the Direct3D API..

Anyway, can't you just overclock your CPU?

Do you mean it would bottleneck in terms of minimum/average FPS? If so, what amount of difference would there be? Could someone provide empirical benchmark links for reference? I've looked for SLI tests with 4770k stock vs OC results to no avail.

Gaming would occur in 1440p with whichever optimal max setting. Some examples are BF4, Far Cry 3, Max Payne 3, Crysis series, Metro series and maybe AC4.

 

[Carfax83]

Do you mean it would bottleneck in terms of minimum/average FPS?

Yes..

If so, what amount of difference would there be? Could someone provide empirical benchmark links for reference? I've looked for SLI tests with 4700k stock vs OC results to no avail.

Depends. Does max turbo affect all cores on Haswell or is it just two cores? I know that's how Sandy Bridge works. Max turbo clock speed for your CPU is 3.7ghz, which should be enough for most games to easily maintain playable frame rates. But if you want to get the most out of an high end SLI setup, an overclocked CPU is practically a necessity.

The faster the GPUs, the more important the CPU becomes to maintain high frame rates..

Gaming would occur in 1440p with whichever optimal max setting. Some examples are BF4, Far Cry 3, Arma 3, Max Payne 3, Crysis series, Metro series and maybe AC4.

Out of everything you listed, only BF4 and Crysis 3 have multithreaded renderers. The rest of them use only a single thread for rendering, which makes them more CPU bound and heavily reliant on high frequency, high IPC cores for maximum performance..

The more GPU bound the settings though (and the game itself), the less important the CPU becomes for performance..

 

[Robster]

I still don't understand why it would be pointless to buy another 770, unless the difference from a higher clocked CPU is vast.

Thus I'd appreciate if any empirical results could be shown.

 

[Carfax83]

I still don't understand why it would be pointless to buy another 770, unless the difference vast.

Thus I'd appreciate if any empirical results could be shown.

That's complicated, because it really depends on the game and your settings.

For example, I did some tests once myself with BF4 at 1080p ultra settings as a comparison to AMD's Mantle test. My CPU is typically overclocked to 4.5ghz, with stock clock at 3.2ghz and max turbo at 3.8ghz; but only on two cores. Anyway, I benchmarked the most CPU heavy area of the SP campaign (the part when they're on the sinking ship) with my CPU at 4.5ghz and at stock clocks.

With the overclock, I was about 10 FPS higher on average than I was with stock clock, and I never dipped below 100 FPS, whereas with stock clock, I dipped below a 100 a few times.

I was surprised because I thought there would be more of a difference. But when I thought about it, BF4 does use multithreaded rendering so frequency would be less important than actual number of cores.

But if I had tested a game like Arma 3 which uses a single thread only for rendering, the difference would likely have been much more than 10 FPS as core frequency would be much more important.

Anyway, this is a forum for PC hardware enthusiasts, and knowing enthusiasts, they typically want to tweak their rigs to the max to get the most performance out of them..

So when you ask us for advice, you're probably going to be told that yeah, your CPU is limiting your performance. But does that mean that you're going to have crappy frame rates due to your CPU?

No it doesn't.. Your frame rates should still be very playable. It just won't be as high and steady as if your CPU was overclocked..

 

[24601]

http://translate.googleusercontent....3.html&usg=ALkJrhiBV31F2Ga3maFFpj3h011fm8qyzg

It's always deceptive, as 100% of the increase of Average FPS is from decreasing the maximum frame times in that test.

Increase of Average FPS by 15% (what the tests show) is a decrease of maximum frame times of at minimum 30% since that's the only place where the test is deriving the increase of Average FPS from.

And that's from tests with 780 Ti and 290x.

770 SLI is far more CPU bottlenecked than those two cards.

 

[Robster]

That's complicated, because it really depends on the game and your settings.

For example, I did some tests once myself with BF4 at 1080p ultra settings as a comparison to AMD's Mantle test. My CPU is typically overclocked to 4.5ghz, with stock clock at 3.2ghz and max turbo at 3.8ghz; but only on two cores. Anyway, I benchmarked the most CPU heavy area of the SP campaign (the part when they're on the sinking ship) with my CPU at 4.5ghz and at stock clocks.

With the overclock, I was about 10 FPS higher on average than I was with stock clock, and I never dipped below 100 FPS, whereas with stock clock, I dipped below a 100 a few times.

I was surprised because I thought there would be more of a difference. But when I thought about it, BF4 does use multithreaded rendering so frequency would be less important than actual number of cores.

But if I had tested a game like Arma 3 which uses a single thread only for rendering, the difference would likely have been much more than 10 FPS as core frequency would be much more important.

Anyway, this is a forum for PC hardware enthusiasts, and knowing enthusiasts, they typically want to tweak their rigs to the max to get the most performance out of them..

So when you ask us for advice, you're probably going to be told that yeah, your CPU is limiting your performance. But does that mean that you're going to have crappy frame rates due to your CPU?

No it doesn't.. Your frame rates should still be very playable. It just won't be as high and steady as if your CPU was overclocked..

I'm aware of this forum as I'm soon hitting my 10-year mark Ironic as it sounds I've never used SLI or cared to, even though I'm an enthusiast myself. And taking this inexperience to notice, hearing Communism's words as adding another 770 would be pointless threw me off. But I think I understand now what he was getting at that getting the most out of SLI with a locked Xeon would not happen. But would still only benefit and would not be pointless.

I stated earlier that I'd sit on it unless I catch a good deal. Maxwell is not far ahead.

 

[24601]

I'm aware of this forum as I'm soon hitting my 10-year mark Ironic as it sounds I've never used SLI or cared to, even though I'm an enthusiast myself. And taking this inexperience to notice, hearing Communism's words as adding another 770 would be pointless threw me off. But I think I understand now what he was getting at that getting the most out of SLI with a locked Xeon would not happen. But would still only benefit and would not be pointless.

I stated earlier that I'd sit on it unless I catch a good deal. Maxwell is not far ahead.

Depends on if you are some kind of 5 dimensional creature that perceives fluidity with average FPS instead of maximum frame time like mere 4 dimensional humans do.

One would need to perceive 2 dimensional time for average FPS to be the determinator of fluidity instead of maximum frame time.

 

[Carfax83]

I'm aware of this forum as I'm soon hitting my 10-year mark Ironic as it sounds I've never used SLI or cared to, even though I'm an enthusiast myself. And taking this inexperience to notice, hearing Communism's words as adding another 770 would be pointless threw me off. But I think I understand now what he was getting at that getting the most out of SLI with a locked Xeon would not happen. But would still only benefit and would not be pointless.

Yeah, Communism's last link of multiplayer BF4 shows how important the CPU is for SLI. When I did my test, it was during the SP campaign (as I never play multiplayer); albeit the most CPU limited area. It was still worth a good 10 FPS though, and I had no substantial frame rate dips.

If I had done a similar test on a fully loaded 64 player server, the difference would have been much greater though to be sure..

I stated earlier that I'd sit on it unless I catch a good deal. Maxwell is not far ahead.

I wouldn't jump on Maxwell until it's on 20nm, but that's just me.

 

[Robster]

I wouldn't jump on Maxwell until it's on 20nm, but that's just me.

That's the one I've been thinking of. Late 2014/early 2015 was it? It was mere speculation though what I read.

 

[Enigmoid]

You have to remember that using a 60 Hz monitor as long as the minimums are over 60 there are no visual improvements. Getting 100+ fps only matters for 120 Hz.

 

[bystander36]

You have to remember that using a 60 Hz monitor as long as the minimums are over 60 there are no visual improvements. Getting 100+ fps only matters for 120 Hz.

There is a visual difference still, if v-sync is off. There is a bit of a latency difference too. Without V-sync, faster FPS allows for part of the image to update during a refresh. It causes tearing, but it also updates that part of the screen a bit earlier.

 

[Robster]

You have to remember that using a 60 Hz monitor as long as the minimums are over 60 there are no visual improvements. Getting 100+ fps only matters for 120 Hz.

Am using a QNIX 2710 which is overclockeable without frame skipping. Highly recommend. :thumbsup:

 

[BrightCandle]

Am using a QNIX 2710 which is overclockeable without frame skipping. Highly recommend. :thumbsup:

http://www.tftcentral.co.uk/reviews/qnix_qx2710.htm

Tftcentral did a review and determined that overclocking skipped frames and it never did better than 60hz. How did you determine there were no skipped frames?

We tested these over both DL-DVI and DisplayPort. Although the image was displayed, frames were unfortunately dropped in all cases, meaning this was all rather pointless. Some artefacts were also visible in some situations. It didn't seem to allow any useful overclocking beyond 60Hz sadly. Maybe results will vary with other models in the range of single input variants as some people have reported. The specified ability to overclock up to 120Hz is somewhat misleading for this particular model though.

 

[Face2Face]

http://www.tftcentral.co.uk/reviews/qnix_qx2710.htm

Tftcentral did a review and determined that overclocking skipped frames and it never did better than 60hz. How did you determine there were no skipped frames?

I used this site http://www.testufo.com/#test=framerates

Then took a picture with my DSLR camera at a very low shutter speed. After doing so you will be able to determine if you are experiencing frame skipping or not. The model they tested has a full array of inputs, which means it has an scaler. These models do not perform as well in overclocking.

 

[Robster]

http://www.tftcentral.co.uk/reviews/qnix_qx2710.htm

Tftcentral did a review and determined that overclocking skipped frames and it never did better than 60hz. How did you determine there were no skipped frames?

Exactly as F2F said. The model that they tested was a multi input variant, not the single DVI-D which does not have a scaler and does not skip frames when overclocked. I'd recommend you a review by NCX of the correct model: http://wecravegamestoo.com/forums/m...eview-2560x1440-matte-overclock-able-pls.html .

edit
NVM he already explained the low shutter method

 

[BrightCandle]

We need to careful then about just saying the QNIX 2710 because there are clearly multiple versions of this monitor out there and you need to be very explicit that the version that has been tested and is overclockable is the one without a scalar and only supports single inputs. This is essential otherwise people are going to get the wrong one and end up with a really poor experience.

 

[Robster]

We need to careful then about just saying the QNIX 2710 because there are clearly multiple versions of this monitor out there and you need to be very explicit that the version that has been tested and is overclockable is the one without a scalar and only supports single inputs. This is essential otherwise people are going to get the wrong one and end up with a really poor experience.

Good point.

As this monitor isn't very known to the normal consumer, most people who are interested find these monitors through various forums/threads dedicated to just korean panels and these threads usually provide all the info they need for selecting the correct monitor. A well known thread is http://www.overclock.net/t/1384767/official-the-qnix-x-star-1440p-monitor-club-read-1st-post . I recommend reading this for anyone who is interested in purchasing a very affordable korean PLS panel.

The correct model name of the QNIX is as stated in my signature (QNIX QX2710 Evolution II), therefor I didn't bother typing it. The multi-input ones are easy to spot as they include Multi, True 10 or DPmulti in the title. These are to be avoided if DVI-D suffices and one is interested in overclocking.

There have been various reports regarding pwm dimming on this model. Some reporting it uses it and others which don't.

All the info one needs on these monitors can be found in that thread. :thumbsup:

 

[Lepton87]

Yes..

Depends. Does max turbo affect all cores on Haswell or is it just two cores? I know that's how Sandy Bridge works. Max turbo clock speed for your CPU is 3.7ghz, which should be enough for most games to easily maintain playable frame rates. But if you want to get the most out of an high end SLI setup, an overclocked CPU is practically a necessity.

The turbo works exactly the same on Haswell as on SandyBridge. Max Turbo is typically a single core turbo, however a lot of motherboards have MCE (multi core enhancement) what it does is it gives you max turbo frequency even when all cores are fully loaded.

 

[blastingcap]

http://www.eurogamer.net/articles/digitalfoundry-2014-titanfall-tech-interview

Baker cites compute shader technology here, reeling off a list of examples where tasks traditionally run on CPU are instead run on the graphics hardware via DirectCompute. Bearing in mind the lack of DX11-only engines there were before next-gen console arrived, the transition to the newer API is welcome, and even older DX11-capable graphics cards should benefit. But Titanfall also sees one area where perhaps the vast unified memory of the consoles could have implications for the PC experience. At its very highest settings, the Titanfall beta's "insane" texture setting crippled performance on any GPU with less than 3GB of RAM. So what was the score there? Did "insane" mean something really bonkers, like running with completely uncompressed assets?

"No, we have a lot of high-res textures. We talked a lot about having some kind of mega-texture support or streaming but it's a multiplayer game, you drop in anywhere, you move around quickly," Baker says. "If we added streaming, we could use a lot less RAM for the same visual quality - if you were sitting still and looking in the same direction. But given that you're spinning around, wall-running and jumping through windows all over the place all the time, we didn't have a whole lot of confidence in texture-streaming without getting textures delayed a lot."
In the beta, ramping up texture quality to the insane level on a 2GB card could result in enormous "swapping" problems, with reduced performance as texture data switched between system DDR3 and GPU RAM.
"For the shipping game we've changed it quite a bit - what textures are various resolutions," explains Baker. "In the beta, if you switched it down one level below insane, all the textures were a quarter of the res."

So in beta they let the horses run wild, reining them in for the shipping game in a nod to gamers with 2GB cards. But will it always be that way going forward? One can imagine future games with Insane modes uncapped, and those with 2GB can simply turn down their graphics a notch down.

 

[KingFatty]

I would like to see a side-by-side example/comparison between using the super-huge insane textures vs. the 1/4-size textures one level below.

I guess you'd not really notice except when you approach near to objects and see the texture up close?

 

[blastingcap]

I would like to see a side-by-side example/comparison between using the super-huge insane textures vs. the 1/4-size textures one level below.

I guess you'd not really notice except when you approach near to objects and see the texture up close?

Even if some people don't want the option for super high rez textures, many people do want it. Just look at people loving the high-rez texture packs and mods on Skyrim, Crysis, and numerous other games.