AMD VS INTEL IN COST PER FRAME RATE (LOTS OF DATA)

[i_build_nanosuits]

@i_build_nanosuits linked the 4160, not the 4130. The clock increase makes quite a difference in minimum frames with the i3 chips. You posted two videos about the wrong chip.

i judged her sources so sketchy and at the same time so unrepresentative of real world performance (passmark..come on!...check passmark single-threaded CPU results @Nena360) i didn't even bothered formulating a response...at this point it's no longer even worth it but thanks for pointing this out.

What this witcher videos show is that witcher 3 is not CPU intensive at all in this tutorial section AND they are in a GPU limited scenario.

[Prysin]

AMD started becoming a serious competitor for Intel with K7, they had an absolute golden age with K8 where they beat Intel into the dust in every conceivable way, they were very competitive with K10, and Kaveri is still very competitive at the low end.

 

This then, is the key: Always name your microarchitectures in such a way that those names start with K, and for crying out loud, avoid earth moving equipment.

and if you plan to make a product that is aimed at future performance (ref. their GPUs w/dx12).... then at least keep up while doing so. Dont lag behind

[SuperCactus]

Great list, but I immediately noticed that the 4790 beat the 4790k... which means the 4790k wasn't overclocked at all... 

 

The "k" versions need to be overclocked to at least a reasonable degree to show value. For example, 4790k@4.7 turbo or such.

[Edgar R. Zakarian]

860K = FX-4300

FX-4300 > Pentium

FX-6300 = i3

FX-9590 < i5

Also, about this topic...it's just plain wrong on so many levels...you don't buy into a platform for gaming thinking FPS per dollar and forget about everything that must surround it (ex. motherboards/overclocking ready hardware/aftermarket CPU cooling etc. as mentionned already, cost of the CPU alone won't mean much quite often) and more often than not when it comes to gaming the CPU will have the most impact on SMOOTNESS OF ANIMATION and STUTTER FREE animation and not necesseraly on FPS...average FPS on top of that...those only highlight a very small fraction of what's going on...what about minimum FPS? average frametime variance? 99th percentile latency etc.

I was just looking at this.

 

I can see that the FX6300 DOES dip down at some points. But I'm really not fully convinced that it's worth double the price yet haha, sorry :lol:

The cooler isn't for overclocking as mentioned, it's just because the stock amd cooler sucks, and it's really loud.

[Guest trucker2]

The archetecture does not limit OC on steamroller.It is the material or manufacturing process the had to use to fit 2.41 billion transisters on a chip. Forget the name of it.It was in one of the reviews. We have never seen a pure steamroller, just a neutered Kaveri.

Software lags hardware by a decade or so. That was true when we had the first Pentium and were still waiting on a 32 bit OS and worried about legacy programs. Also PC gaming is still newly resuscitated, Was only a few years ago I got laughed at the local GameStop when I asked about PC games. Also game development takes 4 years. @ years ago some game developers where polled about the I5 or the 8350 for futureproofing at Eurogamer. I take what software people say seriously.

   On fps per dollar or bang for buck many times local sales trump allot. See a hot deal on AMD or intel JUMP on it. I would not touch a two core cpu with a ten foot pole, with or without hyperthreading, 4 cores minimum IMHO. As game developers get better at writing games for jaguar you will see the 8 core CPU's destroy the 4 core ones. No longer will games be written for VLIW and ported to x86 as an afterthought

   Frankly just buy what you want and can afford. When I first played Crysis 3 on my barefoot apu I was like "wow.....it handles Crysis 3!" Later I was disappointed cause Crysis 3 won't let me shoot the annoying guy helping me. I like to shoot everybody:).   

[Edgar R. Zakarian]

860K

http://www.cpubenchmark.net/compare.php?cmp[]=2362&cmp[]=1814

FX-6300

http://www.cpubenchmark.net/compare.php?cmp[]=1781&cmp[]=2317

FX 6300 doesn't seem to bad here?

[Edgar R. Zakarian]

Ehm sorry but i dont know what you are trying to say here?

its still a matter of silicon luck.

 

 

yes. Silicon luck is one thing. But the architecture. To achieve that increased IPC, AMD had to sacrifice something, and that was overclocking headroom. The steamroller architecture simply do not OC that well and simply do not get GOOD RETURNS from OCing...

 

There was tests with old Piledriver FM2+ APUs vs NEW FM2+ APUs

 

 

Guys, I won't be overclocking, AT ALL, so this is pointless

[Prysin]

Guys, I won't be overclocking, AT ALL, so this is pointless

who are you?

[i_build_nanosuits]

I can see that the FX6300 DOES dip down at some points. But I'm really not fully convinced that it's worth double the price yet haha, sorry :lol:

you kidding me right? the i5 is consistently outputing better framerates and this is with a GTX770...that goes to show that the FX-6300 is already a bottleneck to even a GTX770...what do you think would happen if you where to run a stronger graphics card? what about future GPU upgrade for this rig? nada.

[PCPartPickerPro9171787]

 

@PCPartPickerPro9171787

 

generally great post. BUT

 

you do not always use the same GPU... meaning the cost of the system is then imbalanced... Sure not all CPUs can reach a certain FPS with that specific GPU. But in those cases, they should be left out, due to not making it..

 

It simply makes no sense, like one of them when you added a 980 to a 860k.... WHY? that GPU cost like nearly twice that of the other GPUs used in the test.. Of course the price to performance would be worse.... Either use the same GPU, everywhere. or use 2-3 GPUs..

 

High 980

mid 280x

low 750Ti

 

 

cut out the crap with some 260Xs, 280s, some 770s, some 290Xs.... it just causes imbalanced charts that are not representative of actual cost to performance.

 

I do not care if intel or AMD or Nvidia gives you the best result. but these results arent even GOOD. they are all over the place and not representative, nor comparable.

 

the point of a test is consistency, consistency, consistency and the ability to replicate the results by another party....

 

this test as a whole is bullshit. it has too many variables, from games to GPU brand to CPU brand and how you mix those three.... it proves or disproves nothing. It is but a mere pointer...

 

I don't think you understand what is going on with the charts, or the methods I used to come to my conclusions. No offence

[Sintezza]

@i_build_nanosuits linked the 4160, not the 4130. The clock increase makes quite a difference in minimum frames with the i3 chips. You posted two videos about the wrong chip.

 

lol that 200mhz diffrence on just 2 cores isnt going to make any significant diffrence.

with a midrange gpu like a 960.

[Sintezza]

They perform the same:

http://cpu.userbenchmark.com/Compare/AMD-FX-4300-vs-AMD-Athlon-II-X4-860K/2879vs3265

both are rated as being ''Very Poor''

end of discussion.

 

 

Steamroller definitely is an improvement over vishera architectual based.

but in the real world its not realy something to get excited about.

Would have been nice if AMD made a 8 core chip based on steamroller.

But still they fall behind intels per core performance obviously.

So they dont botter.

But ZEN might be promissing.

 

 

 

 

[Prysin]

I don't think you understand what is going on with the charts, or the methods I used to come to my conclusions. No offence

Listen.

The whole deal with a benchmark, of any sort, is for others to have a chance to replicate the results, easily.

I know that you did your best to AVOID a GPU bottleneck where you simply do not want it.. you are looking to have 100% CPU bottleneck all the time.

problem with that aproach is that you also make the whole test hard or impossible for others without some 10 different GPUs lying around, to replicate. Your results is exceedingly hard to reproduce.

If we cannot easily reproduce, we cannot verify.

There is also the cost of your setups. They vary too much to be realistic.

If you want to verify in cost vs performance. Then you should either select a certin GPU for a certain FPS range, then just black out CPUs that cannot do it. Instead, you add CPU and GPu combos that is way too expensive, in several cases you should know when to stop. Back to my former example of a 860k + 980 combo... the 860k cannot feed that GPU data at all, it will bottleneck.

Some of the i3 combos also just cost way too much to make sense. At some point, one must just look at ones tests and just see that this wont work...

You do not go buy a ridiculously expensive GPU, and a 70 dollar CPU... you always look at some balance between the two... This is common knowledge.

Unfortunatly, your results are too broken up, fragmented. They do not represent actual builds or potential builds, and several important GPU combinations were left out.

Why no GTX 970? it is a hugely popular and widely used GPU.

Why no HD 7970 GHz? Sure its basically a 280X, but then again, drivers DO make a difference.

One the CPU side.

Why no FX 6300? If anything, it is one of those FXs that actually make sense to buy.

Why no Phenom IIs? The GTX 770 was in sale during the time you could still buy Phenom IIs. A lot of people still rock their X4 or X6 unlocked Phenoms, which are decent CPUs, even today.

Then there is the wildcard of OC.

Some things simply make a lot more sense if you OC it, but OC costs more due to cooling. How do you plan do account for that?

I do not dispute your results as is.

I dispute the test and how you laid it out in the first place. It is simply too fragmented, and DO lack quite a bit of logic behind it. In some cases, it is borderline moronically done in terms of combinations selected.

[Dabombinable]

Listen.

The whole deal with a benchmark, of any sort, is for others to have a chance to replicate the results, easily.

I know that you did your best to AVOID a GPU bottleneck where you simply do not want it.. you are looking to have 100% CPU bottleneck all the time.

problem with that aproach is that you also make the whole test hard or impossible for others without some 10 different GPUs lying around, to replicate. Your results is exceedingly hard to reproduce.

If we cannot easily reproduce, we cannot verify.

There is also the cost of your setups. They vary too much to be realistic.

If you want to verify in cost vs performance. Then you should either select a certin GPU for a certain FPS range, then just black out CPUs that cannot do it. Instead, you add CPU and GPu combos that is way too expensive, in several cases you should know when to stop. Back to my former example of a 860k + 980 combo... the 860k cannot feed that GPU data at all, it will bottleneck.

Some of the i3 combos also just cost way too much to make sense. At some point, one must just look at ones tests and just see that this wont work...

You do not go buy a ridiculously expensive GPU, and a 70 dollar CPU... you always look at some balance between the two... This is common knowledge.

Unfortunatly, your results are too broken up, fragmented. They do not represent actual builds or potential builds, and several important GPU combinations were left out.

Why no GTX 970? it is a hugely popular and widely used GPU.

Why no HD 7970 GHz? Sure its basically a 280X, but then again, drivers DO make a difference.

One the CPU side.

Why no FX 6300? If anything, it is one of those FXs that actually make sense to buy.

Why no Phenom IIs? The GTX 770 was in sale during the time you could still buy Phenom IIs. A lot of people still rock their X4 or X6 unlocked Phenoms, which are decent CPUs, even today.

Then there is the wildcard of OC.

Some things simply make a lot more sense if you OC it, but OC costs more due to cooling. How do you plan do account for that?

I do not dispute your results as is.

I dispute the test and how you laid it out in the first place. It is simply too fragmented, and DO lack quite a bit of logic behind it. In some cases, it is borderline moronically done in terms of combinations selected.

So done as if he were fanboying? Because you just described everything that they do.

[Nena Trinity]

@i_build_nanosuits linked the 4160, not the 4130. The clock increase makes quite a difference in minimum frames with the i3 chips. You posted two videos about the wrong chip.

The highest i3 I find on Youtube is i3-4150...

Still the FX wins at stock speed so what is your excuse this time?

[i_build_nanosuits]

The highest i3 I find on Youtube is i3-4150...

Still the FX wins at stock speed so what is your excuse this time?

how many of these videos of all those other games on youtube you think i can link in which the little i3 flatout outperform the FX-6300 or FX-8350?!

You cherry pick the few games that run at acceptable level on AMD ancient CPU's...come on!

[Nena Trinity]

how many of these videos of other games on youtube you think i can link in which the little i3 flatout outperform the FX-6300 or FX-8350?!

You cherry pick the few games that run at acceptable level on AMD ancient CPU's...come on!

I posted many show me some if you find... 

[PlayStation 2]

Exactly! Steamroller = Piledriver II or Bulldozer III!

No...

honestly, the 860K is a much better all-rounder CPU than the G3258.

[Nena Trinity]

No...

honestly, the 860K is a much better all-rounder CPU than the G3258.

Yup 860K humiliate the Pentium!

[PlayStation 2]

Yup 860K humiliate the Pentium!

The 860K humiliates it in newer games and in streaming... and in any multithreaded workloads

The G3258 humiliates the 860K in older games, emulation, anything that needs that single core performance.

[Nena Trinity]

The 860K humiliates it in newer games and in streaming... and in any multithreaded workloads

The G3258 humiliates the 860K in older games, emulation, anything that needs that single core performance.

But anything can play old games...

[PlayStation 2]

But anything can play old games...

I kinda meant more of the "shitty console port" games rather than "old" old games.

[Nena Trinity]

I kinda meant more of the "shitty console port" games rather than "old" old games.

Erm wha? 860K should be able to play that!

[79wjd]

But anything can play old games...

MMOs/MMORPGs is where the G3258 really makes sense (as a long term cpu). Older games and indie games -- games which make up most of the games that use <4 threads -- are usually fairly easy to run anyway. 

[Nena Trinity]

MMOs/MMORPGs is where the G3258 really makes sense (as a long term cpu). Older games and indie games -- games which make up most of the games that use <4 threads -- are usually fairly easy to run anyway. 

War Thunder, World of Warcraft and other MMOs I have tested all ran above 60 FPS on a FX chip and these games should run better in "theory" on a 860K...