[Question] CPU or GPU intensive?

Cheers.. Just have a quick question. What is this game most reliant on? CPU or GPU?
I am currently getting 30 FPS at lowest GFX settings at 1980*1200. And was wodering if i should upgrade my CPU or my GFX card.. my CPU is i7 920 and GFX card is ATI HD 4890. So what to do?

CPU but not by much. Your 4890 is your current bottleneck for low fps.

wait for 2.0 before updating anything.

Oh okay.. BTW forgot to mention the 6GB of ram. But i was thinking about changing the GFX card anyway..

I'll have to disagree with Zien. I'm having a significantly slower CPU than you and I'm getting a better framerate with most settings at high, 4x antialiasing and 1920x1080 on my GTX460.

Your GPU is your bottleneck and your CPU is probably far from being one.

Dual core i5 2.3Mhz and a GT330M
usually @18..25FPS on below standard settings
GPU permanently at 100%
CPU usually 50/60%

Not the best system I know, but for me GPU is the bottleneck by a long way for me.

That is the most hard core 2.3MHz CPU I have ever seen!

Anyway, yeah. GT330M is not exactly a high end card.

i7 at 2.7 GHz is probably not too far from your 2.5 i5 in per-core performance, and the i7 920 has 4 cores. I have a Q9450@3.2GHz myself (probably weaker than a 920@2.7), and I know that if I lower the graphics settings so that my GPU doesn't limit me anymore, I can easily do 60fps almost everywhere in the game.

GPU...never goes below 99% lol...well I have the AO turned on with my 470.

right now definitely more intense on GPU, its only singlethreaded game so you could have best CPU in the world and it still wont run 100%. My advice wait until 2.0 before upgraded as this game is poorly optimized to run on anything well. In 2.0 you should be able to play the game fine. As of right now the way its set up this game never utilizes more than 8 GB's of RAM, and I have yet to test out how much GPU power it uses but it never uses more than 10% of my GPU's power cause of a huge bottleneck.

Quote:

---

Originally Posted by Mirage

That is the most hard core 2.3MHz CPU I have ever seen!

---

Sadly I'm actually old enough to remember 1 MHz CPU's. But I take your point!

Quote:

---

Originally Posted by Norack

right now definitely more intense on GPU, its only singlethreaded game so you could have best CPU in the world and it still wont run 100%. My advice wait until 2.0 before upgraded as this game is poorly optimized to run on anything well. In 2.0 you should be able to play the game fine. As of right now the way its set up this game never utilizes more than 8 GB's of RAM, and I have yet to test out how much GPU power it uses but it never uses more than 10% of my GPU's power cause of a huge bottleneck.

---

I'm pretty sure FF14 can make use of several cores. I have 4 cores myself, which means 100% utilization of a single core would show up at 25% total CPU usage in task manager. I have however seen FF14 use around 50%, which would mean it makes use of either all of two cores, or one full core and half of two other cores. Or any other combination that adds up to 50% total.

I also think you'll find that the vast majority of current PC games do not use anywhere near 8 GB RAM.

-edit-
1920x1080 4xaa medium-high graphics settings, 30fps near uldah's aetheryte with a ton of characters on screen. blur because i am a paranoid mother fucker

FFXIV runs on X64 by utilizing WoWx64 because of its single thread usage. That has nothing to do with the amount of cores it can use.

Quote:

---

Originally Posted by Norack

FFXIV runs on X64 by utilizing WoWx64 because of its single thread usage. That has nothing to do with the amount of cores it can use.

---

Process Explorer tells me that it's created 48 separate threads on my system.

Quote:

---

Originally Posted by Norack

FFXIV runs on X64 by utilizing WoWx64 because of its single thread usage. That has nothing to do with the amount of cores it can use.

---

Wow64 lets 32-bit programs run natively in 64-bit CPU modes. This doesn't really impact the number of threads the game can use, or have any significant impact on how much of the CPU it can utilize.

I'm not exactly sure what you're trying to say. Could you elaborate?

the game can create as many threads as it wants, 32 bit has only 1 core thread to go through the operating system, 64 bit OS's have 2 threads to go through. Other people see it though as the handler just cutting the pointers down by half so whichever way you wanna look at it. Either 2 core OS threads or 1 core OS thread cut in half from the 32 bit version.

So it behaves exactly like every other 32bit game running on windows x64 then. If it doesn't affect the amount of usable cores, as you're saying, why even bring this up?

Because a 32 bit OS only has 1 thread through the North bridge of the MB. A 64 bit OS has 2 threads through the North bride of the MB. 2 > 1, that being the case that will allow more datum transfer, theoretically loosening up the CPU bottleneck letting the game run 2X faster depending upon GPU architecture of the new engine.

I'm not sure what Norack is saying. Instead I looked up 64 bit on Wikipedia.

Quote:

---

Originally Posted by Norack

Because a 32 bit OS only has 1 thread through the North bridge of the MB. A 64 bit OS has 2 threads through the North bride of the MB. 2 > 1, that being the case that will allow more datum transfer, theoretically loosening up the CPU bottleneck letting the game run 2X faster depending upon GPU architecture of the new engine.

---

So basically you're saying a native 64bit application performs 2x faster than a 32bit app running on an x64 CPU.

Well, that's not exactly what real life examples show. 64bit photoshop does not apply my effect filters twice as fast as it does in 32bit mode, for example.

"Two things are infinite: the universe and human stupidity; and I'm not sure about the universe." Einstein

That's all I have to say.

Quote:

---

Originally Posted by Mirage

64bit photoshop does not apply my effect filters twice as fast as it does in 32bit mode, for example.

---

32 bit Photoshop doesn't have a CPU block on it like video games do. The simplest way I think I could explain it is like this. on X86 Photoshop is already running at 100%, so if u take Photoshop and bring it over to X64 it is still going to run at 100%. If you play a game on X86 and it runs at about 20% speed because of the necking it receives, that means that its max on a X86 is 20%, but it can still be improved. If you take that game over to a X64 it will be improved and run at 21% efficiency or higher.

Cool. Do you have any benchmarks showing this huge difference in game performance between 32bit and 64bit games?

I mean, if it's as gigantic as you make it sound like, surely, there must be some gaming enthusiasts out there making use of it. No one in their right mind would say no to a potential 2x performance boost.

I mean, maybe you are right, technically. I don't know enough about processor microarchitectures to say you're dead wrong, but I can't understand how this wouldn't have been on every hardcore PC gamer's lips if they could indeed squeeze so much extra performance out of their games in 64bit mode. Of course, assuming that they're actually CPU limited, which is extremely unlikely in this day and age, unless graphics settings are turned down to a minimum.

So if you really think that upping a game to 64bit mode is going to increase actual performance and not just theoretical performance under certain circumstances, I would say some unbiased benchmarks are needed.

I'm also really thinking that there is something wrong with your computer's setup though if FF14 is never using more than 20% of your GPU's power. I'm pretty sure you could put on a lot more antialiasing, ambient occlusion, and various other effects if that was the case. My 460 (but the same is the case for people I know with 560s) is most of the time running at 95% load or more, even with my old, inferior Q9450.

Nope, im sure some ppl have benchmarked it though. I'm only talking out of MB to CPU architecture theory.

did a quick like 1 second google search and found this outta millions of results
http://www.iinuu.eu/en/it-guru/windo...ance-benchmark

Remember its theorized and the results do show significant improvement. Quite frankly im surprised the 64 bit improved as much as it did because the hardware they used if you check it out at the beginning of the test they used all hardware that a X86 processor can take advantage of, and when they switched over and tested it on X64 they didnt change the hardware at all. I don't really understand the point of benchmarking something on X86 versus X64 when you use the same hardware. For example, why not upgrade your RAM? I can understand why there computer only used 4 Gigs of RAM when they where running X86 cause thats all it can utilize. But X64 can utilize 192 Gigs of Ram, depending upon which version you have. So why not upgrade the RAM? You know what I mean, its like "hey on the X86 where using 100% of our RAM so lets switch over to X64 and use the exact same RAM and see if we can use more than 4 GB's of RAM when we only have 4 GB's of RAM in there..............by the way of magick?" XD well thats a little over board but hopefully you understand what I mean.

EDIT:
------------------------------------------------------------------------------------------------------------------------
Have everything on FFXIV maxed out and only 20% which is why im assuming its necked out. So hopefully the 2.0 new architectural changes it to where it can utilize more of my GPU. I think really the biggest change between X86 and X64 in this game would just be the RAM usage as of right now. Id love to see how much funner this game can be from running using only 4 Gigs of RAM like I am now to all of a sudden running it using 36 Gigs of RAM. Lol I know thats never gonna happen and would never need that much but............. son of a just thinking about it XD.

That benchmark is showing negative numbers in several of the tests, and doesn't even mention gaming performance.

The few tests that involved GPU usage showed a very small decrease in performance for some tests, no change in some other tests, and a slight positive increase in some of the tests. As you can see, the increase in performance for GPGPU tests were all less than 5%. That's not very impressive, really.

A few specific tests that heavily utilized a small number of the new 64bit registers on 64bit CPUs saw a huge increase. I don't see any tests there that replicate the type of work load a game would commonly have.

I know like I said I don't really see the point in benchmarking that stuff when you keep the same hardware. Its pointless but apparently the person that did that article wanted to see something. The hardware they where using a X86 can use like I don't know lets just say 98% of all its resources. I don't understand the point of upgrading to hopefully get the extra 2% for your ROI. I don't know the actual numbers for that one but when it comes to computers its a good rule of thumb that you will never get more than 70% of what your hardware is capable of doing. Least that's what all my colleagues and professors always go by.

Look no further than here, Crysis 64-bit vs Crysis 32-bit: http://blog.tune-up.com/windows-insi...e-performance/

Have to scroll down a bit

Thank you.

to OP. new GPU will give the best "bang for the buck"

Quote:

---

Originally Posted by Ayla

Look no further than here, Crysis 64-bit vs Crysis 32-bit: http://blog.tune-up.com/windows-insi...e-performance/

---

You really can't rely on benchmarks like this to tell you anything. For example, we know nothing of the host system. How well the software has been tuned for use of the additional registers X64 provides. And so forth. A benchmark like this may look very different in a year running different software. Another comparison perhaps would be Quicktime 7 and Quicktime 10 for Mac OS X. Quicktime X is *much* faster than Quicktime 7. How much of the speed increase is do to the fine-tuning for 64bit architecture is unknown.

Ideally, one should run 32bit in a 32bit OS. This will always yield better performance than running 32bit software within the confines of a 64bit kernel or 64bit software on top of 32bit kernel. It may not be as easily distinguishable as one would expect but there can be a considerable performance decrease (5~10% perhaps? I kind of pulled that figure out of thin-air). Of course this is often impractical for today's desktops and laptops. So this should not be a considered a recommendation.