Core i5-4570 vs FX-8350 for next-gen games (consoles)

rubidium · Sep 8, 2013

A lot of people and reviewers says that FX-8350 is predicted to beat Intel core i5 in next-gen games due to 8 cores as in PS4 with even clock boost, will this difference be significant with my CPU(listed downwards)? I thought most important thing in gaming was GPU, and the MOST IMPORTANT function of CPU is NOT to bottleneck the GPU and limit its potential, so why will it differ when it comes to number of cores or even clock speed if it wont bottleneck the GPU? Thanks in advance.

Here is the rest of my build if you are interested in knowing it:
CPU: Intel core i5-4570
GPU: EVGA GTX 760 SC /w ACX cooler 2GB
Motherboard: ASUS Z87-k(doesnt support SLI, was sad to know this)
PSU: SeaSonic M12II Bronze 620W 80 PLUS
Case: NZXT Lexa S
RAM: 8GB(2x4GB) G.Skill Ripjaws-X DDR3 1600 CL9 1.5v
Cooling system: i dont have any at the moment actually as my budget ran out before buying it, and i am not thinking of overclocking(as it may be obvious from mobo and CPU), and the case had some pretty good supplied fans

BSim500 · Sep 8, 2013

Both Intel & AMD's are likely to perform better than now if new future games thread better, but no game is going to magically scale 100% perfectly going from 4 cores to 8 (just as the difference between i3's & i5's at same clock is usually about 5-40% - not 100%). If you run most games with resource monitor CPU graph showing load per core in the background, you'll see even on an i5 although the game uses 4 cores, often those 4 cores are still hovering as low as 30-50% - simply because there's nothing left for them to do per second. 8 threaded games doesn't mean quad cores will run slower if they can still do far more each second due to much higher IPC - for exactly the same reason i3's can do just as much as the "hex core" FX 6300 per second:-

FX-8350 8C = 86.7
i5-3470 4C = 85.8
i3-3220 2C = 85.6
FX-6300 6C = 83.0
http://www.techspot.com/review/586-amd-fx-8350-fx-6300/page6.html

What's more at least 1-2 cores of the new consoles are said to be used for "background tasks" (ie, downloading). The new Jaguar chips on consoles are also only running at 1.5-1.75GHz vs 3.5-4.5GHz of modern i5's, and game engines will typically be written for that level and ported across to PC. It would cost far too much to write one engine for consoles and a different one for PC's for the same game. And yes, above a certain baseline GPU's make far more difference than CPU's.

The green & purple lines are rarely achievable on any platform (outside of easily paralleled stuff like video encoding), etc:-

rubidium · Sep 10, 2013

First of all,thanks and i understood everything from the first time i read it, but the graph... i stared at it for 5 mins xD to know whats its even about but i have a question, is this speed up a percentage or a ratio i mean 2x or 2% and thanks. (and who the hell has 65536 cores, this makes me feel ill

)

palladin9479 · Sep 10, 2013

It's about software engineering. Basically there is a finite amount of code that can be made to run in parallel within a single task. Adding more cores doesn't mean anything if there isn't something for those cores to do.

That being said, there are hundreds of background tasks happening on your computer, everything from virus scanning to firewalls to your web browser / email client. Very few people strip down their computer during actual operation, and that's the environment most benchmarking takes place in. Those two cores allocated for "background tasks" will be doing much more then downloading, their running the OS, handling memory management / resource allocation and anything else the developers ask them to do. The other six cores are dedicated to whatever current software is being run, typically a "game". Also consoles are bare metal environments and software engineers can program directly for the hardware, this gives them an insane performance advantage, typically in the 80~150% range, over modern PC's. Modern PC's have many layers of abstraction between the hardware and the software, the game is written for a generic DX API which abstracts the code to the WDDM driver layer which then translates the code into whatever language that specific piece of hardware use's. All that creates quite a bit of inefficiency but allows the software to run on lots of different hardware configurations.

Anyhow, it's always been about cost vs performance, what performance do you need and how much are you willing to spend.

BSim500 · Sep 11, 2013

rubidium :

The graph basically demonstrates there's a limited amount of code that can be threaded on any system. Horizontal = number of processors. Vertical = x speeds (eg, 2.00 = 2x speed up or 100% boost, etc). Each line shows the differing relationship between piling on more cores, and seeing the return from that - given different types of software that can be threaded at different efficiencies.

The green line (95% threaded) is virtually impossible to achieve outside of things like video editing (which has the luxury of being fed a continuously predictable "offline" data structure and can "break it up" easily (eg, dividing a video file into 4 equal lengths and encoding each chunk on each core)). But stuff like games real-time AI engines, collision detection, etc, is much harder. Today, even the best "multi-core optimized" game is basically 50-60% faster on an i5 vs an i3 (between the blue and red line). The 65k processors simply shows that unless software is very heavily threaded, each additional thread added won't run faster with extra cores because it can easily run on an existing core which is only at quarter-to-half load with no slowdown (precisely because threading is only 50% efficient).

Here's an example of threadable code:-
1. A = 5+2
2. B = 7x2

One core will run A and calculate the answer A=7 at the same time as another core can run B and calculate the answer B=14.

Here's an example of non-threadable code:-
1. A = B+2
2. B = 7x2

This can't be easily threaded - because A needs B to finish before it can start otherwise it won't know what the value of "B" is for its own calculation. One core will run B and calculate the answer B=14, whilst the other core has to wait for B to finish before it can calculate A=16 (14+2). Computers do millions of these per second, some calculations are like the first - easily threaded - and others like the latter - and this is why it's impossible for many games / general apps to scale anywhere near 100%, and why an FX-8350 isn't going to magically jump up 100% in framerates over an i5.

Like Palladin said - some background tasks on your CPU will use more. And consoles can theoretically be coded "direct to metal", but the issue there is the increasing cost of cross-platform "low level coding" development & development project time-limitations. Often they'll just code to the lowest common denominator purely for cost purposes (as we've all seen with some rather bad PC ports over the years...)

rubidium · Sep 11, 2013

Ok guys, i'm really grateful for the VERY useful information you provided me with(that example of threadable and non was really simple and illustrating, as well as the OS and RAM management that makes some cores busy). And about my budget, the thing is that my rig is brand new(just bought it from 3 weeks), i know its not a high-end but we have horrible prices over here and i spent A LOT on this compared with builds that common people have, but thanks anyway for trying to help with upgrade ^^
Is there any "voting up system" or something over here, i'm new to toms hardware and i wanted to vote for you both for that info i rarely get on any other forum

MEC-777 · Sep 11, 2013

^You can click the "select best answer" button at the bottom of the post you found most useful. but you can only do that once I believe.

Great info in this thread. Thanks for taking the time to answer rubidium's question in such detail. I've seen similar questions pop up a lot and many are claiming the 8-cores will become more relevant in gaming in the near future.

One way to sum it up is that 8 less-powerful cores will only benefit if the application coding is specifically written to be optimized for that piece of hardware, and even still, not all tasks can be "split up" or divided among the cores. Where as 4 more-powerful cores can still benefit even if the software is written for more cores because it can rip through the threads faster in series instead of parallel. (speaking in terms of gaming, specifically).

From what I gathered, it's safe to say the Intel i5's will continue to be the most capable gaming CPU's on the market with the exception of some applications being more optimized for the FX-8350's. We've seen some of that happening now, but I wouldn't worry about the i5's going "obsolete". Not for a VERY long time.

rubidium, your system is a very nice build. You'll be gaming on that for a long time yet before needing an upgrade. Don't worry. Enjoy it.

Bill Reinhardt · Sep 11, 2013

BSim500 is my new favorite person on this site.

rubidium · Sep 13, 2013

MEC-777 :

^You can click the "select best answer" button at the bottom of the post you found most useful. but you can only do that once I believe.

Great info in this thread. Thanks for taking the time to answer rubidium's question in such detail. I've seen similar questions pop up a lot and many are claiming the 8-cores will become more relevant in gaming in the near future.

One way to sum it up is that 8 less-powerful cores will only benefit if the application coding is specifically written to be optimized for that piece of hardware, and even still, not all tasks can be "split up" or divided among the cores. Where as 4 more-powerful cores can still benefit even if the software is written for more cores because it can rip through the threads faster in series instead of parallel. (speaking in terms of gaming, specifically).

From what I gathered, it's safe to say the Intel i5's will continue to be the most capable gaming CPU's on the market with the exception of some applications being more optimized for the FX-8350's. We've seen some of that happening now, but I wouldn't worry about the i5's going "obsolete". Not for a VERY long time.

rubidium, your system is a very nice build. You'll be gaming on that for a long time yet before needing an upgrade. Don't worry. Enjoy it.

Thanks Mec for the complement, and sorry for Bsim, i wished i could choose your answer to be the solution but i couldnt

and palladin as well and thanks guys for the help.

8350rocks · Sep 13, 2013

BSim500 :

Both Intel & AMD's are likely to perform better than now if new future games thread better, but no game is going to magically scale 100% perfectly going from 4 cores to 8 (just as the difference between i3's & i5's at same clock is usually about 5-40% - not 100%). If you run most games with resource monitor CPU graph showing load per core in the background, you'll see even on an i5 although the game uses 4 cores, often those 4 cores are still hovering as low as 30-50% - simply because there's nothing left for them to do per second. 8 threaded games doesn't mean quad cores will run slower if they can still do far more each second due to much higher IPC - for exactly the same reason i3's can do just as much as the "hex core" FX 6300 per second:-

FX-8350 8C = 86.7
i5-3470 4C = 85.8
i3-3220 2C = 85.6
FX-6300 6C = 83.0
http://www.techspot.com/review/586-amd-fx-8350-fx-6300/page6.html

What's more at least 1-2 cores of the new consoles are said to be used for "background tasks" (ie, downloading). The new Jaguar chips on consoles are also only running at 1.5-1.75GHz vs 3.5-4.5GHz of modern i5's, and game engines will typically be written for that level and ported across to PC. It would cost far too much to write one engine for consoles and a different one for PC's for the same game. And yes, above a certain baseline GPU's make far more difference than CPU's.

The green & purple lines are rarely achievable on any platform (outside of easily paralleled stuff like video encoding), etc:-

Wanted to correct a couple of things...

The consoles are running @ ~2 GHz for XBone and PS4 may be clocked slightly higher.

In single threaded applications the i3 is comparable to the FX 63XX series...however, in heavily threaded applications and games...(see Crysis 3 for a good example) the i3 is no match in any way, shape or form.

palladin9479 · Sep 14, 2013

rubidium :

MEC-777 :

Thanks Mec for the complement, and sorry for Bsim, i wished i could choose your answer to be the solution but i couldnt

and palladin as well and thanks guys for the help.

I changed the post type so you should be able to chose Best Answer or Solution. If you have any issues PM with your choice and I'll manually select it. We try to encourage BA picks as it enhances the quality of the forum.

rubidium · Sep 14, 2013

palladin9479 :

rubidium :

I changed the post type so you should be able to chose Best Answer or Solution. If you have any issues PM with your choice and I'll manually select it. We try to encourage BA picks as it enhances the quality of the forum.

Thanks a lot for the help

Search

Core i5-4570 vs FX-8350 for next-gen games (consoles)

rubidium

Honorable

BSim500

BSim500

Honorable

rubidium

Honorable

palladin9479

Splendid

BSim500

Honorable

rubidium

Honorable

MEC-777

Honorable

Bill Reinhardt

Distinguished

rubidium

Honorable

8350rocks

Distinguished

palladin9479

Splendid

rubidium

Honorable

TRENDING THREADS

Latest posts

Moderators online

Share this page