GTX 1080 - CPU requirments?

[Recon-UK]

I mean... I... I just dont know how to respond to that... my mind is blown.

Having enough FPS for a game can still have a bottleneck. It has nothing to do with elitism and everything to do with getting the most out of what you buy... removing a glass ceiling if you will.

It's a kick in the pants(wallet), to me, to buy something and not use it to its full potential regardless if it's 'good enough'. This is why I buy modern intel cpus, to get the most out of my system/gpu.

Don't think you understand, it's massively game dependent also these consoles are a huge part of why we are seeing more CPU bottlenecks because GPU's now are too far ahead of these underpowered calculators we call XBOX ONE and PS4.

No other console gen has had a console not beat top tier PC hardware except this gen of consoles.

 

[EarthDog]

I get its game dependent.. I said it.

 

[Artas1984]

According to digital-foundry gaming benchmarks, an overclocked I5 2500K to 4.6 GHz is about as fast as a stock I5 6500. It's hard to imagine that your I5 3570K OCED to 4.8 GHz would be a bottleneck...

That also means that I5 4670K should also bottleneck, since it's performance is identical to I5 3570K:

http://ocaholic.ch/modules/smartsection/item.php?itemid=1158

 

[NdMk2o1o]

Hey guys, I've made a decision based on answers here and on two other forums. It's the i7, next weekend I'll plug her up.

I can't vouch for this, but one guy who seems to "know his stuff," said that with the advent of Pascal (1070 1080 and up) i5 just can't handle pascal. He has seen some videos and tests and I believe him. Another senior member backed this up. BTW they both said it will get worse with time. i7 XXXXk is the standard now and i5 just doesn't do the the Pascal 1070 and higher justice. That's according to them. I didn't know this, and some of you may dis-agree. Fair enough.
I think they are right though simply from what I have observed, even with my juicy OC.

No one, anywhere has recommended sticking with the i5. So many people say go for the i7, a few say try it out. No one says don't bother.

So thermal paste out at the weekend.

I want to thank everyone who posted here. All the posts have been great and useful. So thank you all.

Almost forgot:


Amazingly it never throttles. I use H100i water twin fan cooler, and offset voltage, just a tiny above stock, and lower than stock on pll. A few years ago I got it running at over 5GHz. That freaked me out so I didn't go any further, but I will now. (No crashes after Prime 95. Intel Burn test. Realbench and more. Oh 3DMark 7 too. hours at over 5GHz max usage, not one crash, never any throttling.) It was years ago so memory is fuzzy but it was over 5GHz, possibly 5.1GHz. I'll understand if anyone finds that hard to believe. I do too. But done it, seen it. Going to frame it in solid gold, display on the wall over my monitor. I will never sell this chip. I love that silicon gold.

Once again, a big thank you for all the great input. Cheers.

I'm not saying stick with one or the other, however unless it's a CPU intensive game as some have been mentioned before (ARK/ARMA etc) then an extra 4 thread won't gain you anything as they have the same IPC so saying an i5 **** can't feed pascal where as an i7 can in the same scenario is quite simply not true at all. That said my personal opinion would be to go with the i7 as it won't cost you any FPS, will gain you some in CPU intensive games that can take advantage of 4+ threads and will also perform better in general in other tasks compared to the i5 where more threads can be used.

 

[Vayra86]

Hey, Thanks EarthDog! Yes, I am fully aware about the physics test. I know those benches are not really highly relevant to everyday gaming but I included that to illustrate the point.

Your last sentence "it depends on the title." is what interests me. I have been under the impression that games still don't effectively utilize more than 4 cores. But that may be changing already?

BSOD, as I said, clearly. I always erase dumps. I could probably find the error code, but I have no intention of getting back into that. Please understand, I carefully check and search all big errors or BSOD,
but when I have a clear answer. (not always possible for me) I don't revisit it, just depressing. Luckily for me this was clear. Of course you don't need to believe that.

If you run a 3570k @ 4.8 right now I can safely say its best of worlds compared to the i7 @ 4.5 for pure gaming

This will change dramatically as soon as you also use applications alongside your game, be it FRAPS or any recording software, but also stuff like Skype, a browser. Steam Overlay. At that point the i7's HT will be kicking anytime you need it, and that will probably diminish the increased single core performance you get from the i5.

All things considered, with i5 running game only = best perf
i7 running game only = questionable if you have noticeable perf difference
i7 running game + apps in background = almost guaranteed performance stability, while the i5 may very occasionally show dips or even stutter.

The only reason to really figure it out, is by gaming in your typical 'use case'

Whatever proc feels best to you, is the proc you want to keep.

 

[slozomby]

No other console gen has had a console not beat top tier PC hardware except this gen of consoles.

i have a hard time with this statement. the GeForce 3 ti500 was as fast as the xbox when they were both released within a month of each other. having a hard time getting texture fill rates for consoles older than that.

the big difference in image quality/fps was for a long time consoles were stuck at 640x480 ( ntsc spec) where pc's were mostly running at 1024x768. now that there is consistency in resolution between monitors and TV's a direct comparison is showing the deficiencies.

 

[Recon-UK]

i have a hard time with this statement. the GeForce 3 ti500 was as fast as the xbox when they were both released within a month of each other. having a hard time getting texture fill rates for consoles older than that.

the big difference in image quality/fps was for a long time consoles were stuck at 640x480 ( ntsc spec) where pc's were mostly running at 1024x768. now that there is consistency in resolution between monitors and TV's a direct comparison is showing the deficiencies.

Oh you say it matched it a month after? we don't even get that this gen mate the console hardware is far below what was even launched in 2011.

XBOX GPU was a hybrid and had much more going on.

https://en.wikipedia.org/wiki/Xbox_technical_specifications


It had what GeForce 3 had just not in a raw grunt kind of way, it had wider paths so to speak.

http://www.overclockers.com/forums/showthread.php/30105-GeForce-3-vs-XBOX-NV2A/page2

OC-Master will tell you.

The OG XBOX ran some games at 720P and 1080i native, no upscaling.

http://www.gamespot.com/articles/inside-the-xbox-gpu/1100-2764159/

"In fact, the NV2A is a faster GeForce3 with one major addition to the hardware: a second vertex shader pipeline. Since the chip design was recently finalized for production, Nvidia could confirm that it will run at the faster speed of 250MHz. This puts the pixel fill rate at 1,000 megapixels and the texture fill rate at 2,000 megatexels--much lower than the Microsoft's original official spec of over 4,000 megapixels and 4,000 megatexels, but fill rate isn't the most important spec. Fill rate is not a determining factor because, as a console, the Xbox is naturally destined to run at the low resolution of consumer TVs. Also, most of the Xbox's special capabilities, such as vertex shaders or full-scene antialiasing, occur at different points in the pipeline and are not at all dependent on fill rate. More important is the NV2A's bandwidth to the 64MB of shared memory. The memory in the console is 200MHz double-data rate (DDR) SDRAM, which provides a maximum of 6.4GB per second of throughput to both the CPU and graphics chip. The custom CPU uses up to 1GB per second of this shared bandwidth. This leaves less headroom than the GeForce3 has with its 7.36GB per second bandwidth, but again television's lower display resolution means this should be plenty. The shared memory design also avoids the bottlenecks PC developers see when sending a lot of data from a PC's main memory over the AGP bus to the graphics card."

"Even more interesting than the speed boost is the second vertex pipeline. This part of the chip allows game developers to create amazing real-time lighting and animation effects, and it is highly programmable, meaning that more and more effects will become available over time. While such vertex effects are similar to techniques used in prerendered computer animation, it's likely that it will take developers some time to fully take advantage of these features. The Xbox hardware has to stay constant for several years, so it's definitely a good thing to have extra power to grow into. One reason we won't see Nvidia's PC products match this capability in the near term is simply because DirectX 8 can't address the additional vertex pipeline. But Microsoft has modified the Xbox's version of DirectX to give developers access to the hardware's full capabilities."


A worthy read.

https://forum.beyond3d.com/threads/in-what-major-ways-is-the-nv2a-different-than-the-geforce-4.2118/




Now consoles hold things back.

 

[slozomby]

The OG XBOX ran some games at 720P and 1080i native, no upscaling.

so would the geforce 3

Oh you say it matched it a month after? we don't even get that this gen mate the console hardware is far below what was even launched in 2011.

its not like the previous gen (xbox360/ps3) could keep up with a pc either.
the main popularity of consoles is they've always been significantly less expensive than a gaming pc. but that difference is rapidly shrinking

 

[Recon-UK]

so would the geforce 3


its not like the previous gen (xbox360/ps3) could keep up with a pc either.

That was a far longer generation of consoles and a big reason why consoles now are so crap performance wise.
Fail argument, if you actually saw what was around at 360 launch nothing kept up for a while.

We were sporting Fermi GeForce GTX 580's when Crysis 2 hit... it's beyond insane how much faster Fermi is to the PS3 / 360.

 

[slozomby]

That was a far longer generation of consoles and a big reason why consoles now are so crap performance wise.
Fail argument, if you actually saw what was around at 360 launch nothing kept up for a while.

We were sporting Fermi GeForce GTX 580's when Crysis 2 hit... it's beyond insane how much faster Fermi is to the PS3 / 360.

see the oblivion video.

 

[Recon-UK]

Yeah i see cut backs in graphics fidelity, as i stated the consoles got too long a life cycle in that generation, and now that PS4 / ONE is here we have had the same crap happen because those consoles are so weak.
It's why i mentioned a reason for bottlenecking of the 1080 happening.
Games just are not advanced or technically demanding enough to feed a PASCAL GPU like the 1080 to full whack constantly.
We need another generation entirely right now but that won't happen.

The lowest denominator will always hold us back unless it is a fully fledged PC port because consoles are the number 1 money maker for companies.

 

[Kanan]

Consoles always were better in 1 point - price to performance. That's it. They were never any better in graphics fidelity, raw power. But the whole point of consoles are games and not how good the graphics are. And actually they are releasing consoles now at a faster pace, the new PS4/Xbone for VR with increased/partly doubled specs are released soon. No wonder they are releasing new consoles now faster, the PS3 and Xbone were much more expensive and simply better at their time compared to Xbone/PS4 at release. PS3 sported a rather expensive IBM CPU (called "Cell") with 8 cores, 2 of which were deactivated for better yield - that was many years ago. Also a bluray drive in a time where it was something new and special. Xbox sported a similar CPU (actually they got the same design that was co-produced by Sony and IBM for the PS3 and changed it a bit - 2 less cores (1 deactivated for yield) but SMT instead) with a better GPU that had first time ever unified shaders, something not even the PC had back then. The GPU of the Xbox 360 could do tesselation in DX9 (special Xbox 360 version of it) too. Compared to the Xbone or PS4 it's not comparable, with their cheap CPUs and pretty normal GPU's that were also for sale for PCs at the same time (HD 7870/7850 - Xbone: HD 7770/7790). They wanted cheaper consoles, because they made no money with them - Sony actually lost money on every console sold. Also they had problems with heat - the PS4/Xbone's consumption is so low that it couldn't happen again - or not so often at least, it's a rare thing now, burning consoles. The money makers were games and additional hardware like controllers, cameras, etc. Now they make some cash with the consoles too.

 

[AndyGawg]

Right. Tested and compared both CPUs 1. i5 3570 @ 4.8GHz. 2. i7 3770k @ 4.5GHz. Everything else is the same. (16GB DDR3 at 1,840. GTX 1080 ftu overclocked.) Fallout 4: I have heavily modified the ini files. uGrids 9, Shadows and shadow maps doubled. LOD, grass, fade distance, objects in game, decals and more all +50% to 75%. This requires 4.2GB of VRAM - tried and tested that twice. System memory is at 7GB, so taking away 2GB (it's less, no back grd apps) It uses 5GB DDR3. Over 4GB vram. Over 5GB DDR3) Note this game commits a devs sin - it links the physics to the refresh rate which is rare, and very bad practice. (Maybe Beth had some special reason?) As a result for smooth crash free proper physics, I've locked it at 60fps. (The max it is programmed for. Can go higher, but things start moving very quickly as the physics go wonky.

1. Bench marks - Firestrike, Firestrike extreme, Time spy, Heaven. As expected a large jump in total score, especially with firestrike. Everyone expected this as firestrike includes a large chunk of the total score as the cpu speed. Not real world, boring, but good for stability testing.

2. Fallout 4 Nothing amazing. But, i5, occasional drops below 60fps. HW monitor showed that the CPU all cores (CPU total) had hit 100% - When the frame rate dropped it was probably because the engine was requesting more than 100% of the i5 total processing power)
But, with the i7, the game looked the same and ran the same (to be expected) but it never reduced fps below 60fps. This was a five hour test. HW monitor backed this up perfectly as the i7 had never had all 4+4 at 100%. In other words, it had extra to spare, but the i5 didn't even though i5 is 300MHz faster. To be honest I was somewhat surprised. So, until I totally upgrade my platform I'll stick with the i7.

All the above is tested fact. I am not offering any opinions here. Anyhow, you guys are smart enough to make of it what you will.

It was interesting, took quite a bit of time, fun too, but results were nothing that hasn't already be stated. I don't know yet about other games. Only FO4, and benchmarks tested. (FO4 One, 5 hour non stop session, and One, 2 hour session. 7 hours, never ever below 60fps with i7. With the i5 7 hours would have yielded multiple drops, but not game breaking or a problem really, the fps drops with the i5 were usually 5 - 20 fps, and short duration (look down and back to 60fps for instance) so the i5 can keep it over 30, but not consistently 60fps. The i7 can keep a solid 60fps even with the hugh tweaking, and as some have stated, maybe poor coding in that game? But I'm not sure about that. That's all folks.

 

[RealNeil]

As expected, the i7 brings a little more grunt to the party. And as you state, nothing really wrong with the i5 part.
Thanks for sharing your results with us.

Neil

 

[EarthDog]

You tested what was already known. Glad you confirmed it!

 

[Kanan]

It was interesting, took quite a bit of time, fun too, but results were nothing that hasn't already be stated. I don't know yet about other games. Only FO4, and benchmarks tested. (FO4 One, 5 hour non stop session, and One, 2 hour session. 7 hours, never ever below 60fps with i7. With the i5 7 hours would have yielded multiple drops, but not game breaking or a problem really, the fps drops with the i5 were usually 5 - 20 fps, and short duration (look down and back to 60fps for instance) so the i5 can keep it over 30, but not consistently 60fps. The i7 can keep a solid 60fps even with the hugh tweaking, and as some have stated, maybe poor coding in that game? But I'm not sure about that. That's all folks.

It's a good example for a game that needs more than what a i5 can offer. Very CPU heavy. If you want more heavy games for testing go for City Skylines or Battlefield (4 or 1). 300 MHz on the other hand is nothing - HTT can increase performance up to 20% (realistically, in 3DMark it's up to 50%), but 300 MHz on top of 4500 isn't even 10%.

 

[AndyGawg]

Thanks for the replies everyone. I don't really feel like swapping out the CPUs again. Costing me a fortune in thermal grease. Ha Ha. So I will stay with the i7 3770 k until system upgrade.

The limited, but thourgh testing I did goes to prove beyond doubt that in those scenarios tested, what the majority of people have said here is correct.

It's good to confirm this stuff for ones self even if the results are similar to what was forecast. Seems to me that members here know their stuff well as these tests bore out what had been said/predicted.

---------------------------------------------------
Anyway, I probably will upgrade with the next gen of CPUs after skylake. It's partly to do with all the extra chipset goodies, DDR4 etc, and of course I'll expect a performance boost overall.

Haven't really looked into it in detail, so my next question to you fine people is. When should I upgrade whole system cpu chipset etc. Next gen CPU (Kaby ?) or longer. This year is basically impossible for me for other reasons.
Cheers everyone.

EDIT; Just thought, that CPU question maybe should be in another thread? Not sure, I am a newb!!

Earthdog, just move it if must be moved. Sorry for the trivia.

 

[AndyGawg]

It's a good example for a game that needs more than what a i5 can offer. Very CPU heavy. If you want more heavy games for testing go for City Skylines or Battlefield (4 or 1). 300 MHz on the other hand is nothing - HTT can increase performance up to 20% (realistically, in 3DMark it's up to 50%), but 300 MHz on top of 4500 isn't even 10%.

Agreed!

 

[Kanan]

Thanks for the replies everyone. I don't really feel like swapping out the CPUs again. Costing me a fortune in thermal grease. Ha Ha. So I will stay with the i7 3770 k until system upgrade.

The limited, but thourgh testing I did goes to prove beyond doubt that in those scenarios tested, what the majority of people have said here is correct.

It's good to confirm this stuff for ones self even if the results are similar to what was forecast. Seems to me that members here know their stuff well as these tests bore out what had been said/predicted.

---------------------------------------------------
Anyway, I probably will upgrade with the next gen of CPUs after skylake. It's partly to do with all the extra chipset goodies, DDR4 etc, and of course I'll expect a performance boost overall.

Haven't really looked into it in detail, so my next question to you fine people is. When should I upgrade whole system cpu chipset etc. Next gen CPU (Kaby ?) or longer. This year is basically impossible for me for other reasons.
Cheers everyone.

EDIT; Just thought, that CPU question maybe should be in another thread? Not sure, I am a newb!!

Earthdog, just move it if must be moved. Sorry for the trivia.

Just when you need more performance or more features like you said on the motherboard like m2 slot usb3.1 etc. Sata express too. First reason of course performance. The other things are trivial atm.

And because that's highly dependable I can't tell you now which cpu that is supposed to be. It depends on time and which cpus are best then (maybe zen?).

 

[AndyGawg]

True, Kanan. I am certainly in no rush, it won't be skylake, something after that. Zen could well be it, but, as you said can't decide now for something that will happen probably 1st quarter, maybe later than that, next year.