i5-4690k vs Athlon II X4 651K question

[Diamond-HP]

I just started my new build (in my sig) which is an upgrade from an AMD Athlon II X4 651K set up I had up until yesterday.

Before I dismantled it I took some final benches to compare and I am using the same GPU.

So lets just say my i5 is currently on stock but at the current comparisons I actually respect my old Athlon, all I've ever heard from Intel fans is how superior Intel are to AMD and now I'm an owner of both so can make a trusted judgment.

I had the Athlon OC'ed to 3.4Ghz my i5 is on stock 3.5 Ghz - 3.9Turbo.

Results:

Athlon II X4 651K - 3.4Ghz

Heaven – Ultra 768p – 75.1fps

Rome II – Extreme 768p – 37.2fps

Valley – Ultra 768p – 54.9fps

Valley – Ultra 1080p – 39.4fps


i5-4690k – stock

Heaven – Ultra 768p – 78.2fps

Rome II – Extreme 768p – 45.2fps

Valley – Ultra 768p – 71.2fps

Valley – Ultra 1080p – 42.5fps

So apart from power consumtion and heat what is so special about the Intel architecture? I will say the heat is amazing with my i5, full Intel burner bench on very high and it didn't even break 55c however I now have a Phanteks so that may explain it?

Judging from my results is it mainly Ghz speed that affects performance gains because then it would make much more sense as there is only 1 Ghz difference or a 4Ghz at turbo.

How would my i5 fare against a AMD FX8350 Black Edition 8, it makes me wonder?

 

[Tanner Fredrickson]

The performance difference is negligible because you were never CPU bottlenecked to begin with. That's why the difference is larger at 768p - because the lower resolution removes some of the GPU bottleneck.

The FX-8350 would lose to the i5-4690k in every way, of course. Assuming you had a strong video card, which you don't.

You're not going to see much difference no matter which CPU you add, simply because your HD 7850 is going to hold them all back and make them perform almost the same. You could also stick in an i3-4130 and there'd barely be a difference from the FX-9590 with your GPU.

 

[Diamond-HP]

Yes i did notice my i5 had pushed my 7850 to the limit memory wise (mine only has 1GB VRAM) but Rome is supposed to be very CPU reliant.

but your probably right overall plus my i5 isn't OC'ed yet I'm looking to go to 4.4Ghz or 4.2Ghz but I haven't saved any settings yet in aid of this benchmark test.

........ anyway I'll be buying a GTX 970 as soon as I get my 1080p TV so no worries there.

I'm very behind on resolution atm, I use my bedroom TV as my monitor and I'm still on my 720p LED Samsung while some mofo's are moving on to 4k, bit embarrassing.

 

[Diamond-HP]

Just lastly my Asus z97 came with a switch on the Mobo Called GPU Turbo boost, I left it turned off because I prefer manual OC and because I'm guessing you can do the same auto OC from the Bios is that right?

I've never come across this switch before.

EDIT: Not to worry found this, glad I didn't switch it on.
http://www.tomshardware.co.uk/answers/id-2210129/gpu-boost-switch-asus-motherboards.html

 

[fudoka711]

The performance difference is negligible because you were never CPU bottlenecked to begin with. That's why the difference is larger at 768p - because the lower resolution removes some of the GPU bottleneck.

The FX-8350 would lose to the i5-4690k in every way, of course. Assuming you had a strong video card, which you don't.

You're not going to see much difference no matter which CPU you add, simply because your HD 7850 is going to hold them all back and make them perform almost the same. You could also stick in an i3-4130 and there'd barely be a difference from the FX-9590 with your GPU.

Exactly this.

Also, most games are gpu bound, which is why sites that review graphics cards usually put the best possible cpu in and just varry the gpu's.

And for your question about speed making a difference - yes, it does. However, what's more important is how much work each cpu core/processor does each cycle. Intel cpu's do much more work per cycle than their current AMD counterparts. Your i5 will beat an FX 8350 in most programs that uses 4 cores or less (which is almost every program we use today).

 

[Tanner Fredrickson]

The i5-4690k also beats the FX-8350 in most games that use 8 cores, not just 4 or less.

The architecture of the FX-8350 is set up in such a way that it only gets full returns up to 4 cores; after that each additional core adds latency and steals cache away from the first 4. That's the modular nature of the Piledriver architecture at work.

 

[slyu9213]

I own a 8350 and I know that it will be behind an i5 in pretty much every game except the ones that utilizes the 8-cores decently. So your 651K which should be based on Llano can't be even compared to an Haswell i5. My guess is you have these CPU paired with a mid to low end GPU where the CPU is not much of a bottleneck at all. Pair both CPUs with something like a R9 280 and above and differences should be enormous. Even if an FX 8350 was OCed to FX 9590 speeds and more the gains is not enough to make the power consumption of a 1.5V+ 4.8/5GHz+ FX 8-Core sound any better

 

[Diamond-HP]

Yeah scratch that I just ran Rome in Ultra (one step below Extreme) and it scored ......

Rome II – Ultra 768p – 81.1fps

That is bloody massive, my Athlon used to get 60 something fps but only on High or very high, this i5 test was a level or two up and beat it by a big margin.

I guess me turning it down one level alleviated the bottle neck somewhat on the GPU, that result is def improvement ...... and that was still on stock.

Very happy.

 

[Diamond-HP]

One final question am I best to have the CPU on auto or Core sync until I manually OC it?

What's the benefits of both?

These Asus Intel boards have so many options on them.

 

[fudoka711]

In my experience, its better to leave the cpu on auto until you can manually overclock it. I've usually run into problems using the auto-overclocking utilities found on various motherboards.