GTX 1070 or RX 480 dilema

at 1080p the rx 480 is just fine use the differense from the 1070 sell your cpu and buy new.
sorry for grammar
with 1 shot 2 birds:thumbup:

 

ok forget it enjoy your new hardware:thumbup:

 

I run a 2500K@4.2GHz with a 980 Ti (about the same perf as 1070.) The GPU is the bottleneck in most games, not the CPU. GPU reaches 99% at max boost clock while the CPU cores are chugging along at 40-60%.

The only case where the CPU is the bottleneck is when I drop the resolution to 720p or lower the graphics setting to medium or low. At 1080p, it's uncommon to hit max CPU. And at 1440p, it's very rare for the CPU to be the bottleneck.

Bottom line: The 1070 will bottleneck your CPU if you OC to 4.2GHz, not the other way around. There are exceptions to this, but this is what will happen most of the time.

 

That would suggest Kaby Lake is no faster than Sandy Bridge in gaming, if the Sandy Bridge CPU is only at 40-60 percent load max.

 

http://www.anandtech.com/bench/product/1749?vs=1731
average 85% faster
You were saying????? The 1070 utterly destroys the 480.

And higher resolution alleviates a CPU bottle neck by inducing a GPU bottle neck.

 

Price is never an indicator of speed.
Just look at CPUs, SSDs, cars, basically anything... You typically pay more for less performance increase as you climb up the tiers.

I too came from a 7950 -> 1070. The drivers are better, because they suit what I need them to do (custom resolutions, the games I play, the programs I use, recording).

Everyone's usage is different, someone mentioned their 2500K not bottlenecking the card. My friend with the same setup @ 4.4GHz is heavily bottlenecked in BF4/BF1/MEA, etc. I get roughly 10-20% more FPS (and less FPS dips) even with the same GPU and clocks. It all depends on the game.

 

@thatguy91 @RealNC
There's more than that to it to begin with.
What's "gaming" by the way ? It's just more or less heavy software, right ? So depending on the software, number of cores, architecture, instructions set, frequency, ram speed or latency (etc.) might or might not be a bottleneck. "40-60" percent load means nothing if one core is completely hammered, or if the DRAM bus is saturated, or if the cache lines are full, or simply if the software is unable to use the ressources at hand efficiently, and so on.

My 2700K @ 5GHz is definitely a limiting factor but depending on the software I'm using it either will or won't have an impact on usage.
In most current games that are CPU dependent, if I'm aiming for a stable 60fps, I'll be more than fine. If I'm aiming for 120 or 144FPS (and if the GPU isn't a limiting factor on this specific game), I'll most likely run into issues.
BF1 comes to mind. A 5GHz 7700K would be able to push 135FPS average with my GPU while my 5GHz 2700K will limit me to about 110FPS @ 1080p max settings (a 5820K or even a 6950X would get even more frames with lower clocks on this specific game).
With a 60-75Hz display -> no issue, and we can even crank up the eye candy like downsampling. With a 120-165Hz(or more) display, we're forced to use a frame limiter or have G-sync/FreeSync compatible hardware to compensate or it'll be pure hell to play.

PC "gaming" term can't be used as is when talking about hardware. It means nothing in a world where playing League of Legends, The Witcher 3, Forza Horizon 3 or Ghost Recon Wildlands are equally considered as "gaming", when the hardware requirements or performances from a software standpoint can't even be compared.

Sorry for that little novel, I was bored I guess :nerd:

 

Online games are kind of "special". They do use the CPU more. I assume due to the synchronization work with all other players? Or high-tickrate physics?

Anyway, I'm not aware of many single-player games that do that.

I was playing RotTR just now, and Witcher 3 yesterday. The GPU is constantly maxed at 99% when it doesn't hit the frame cap (for G-Sync.) And that's been my average experience with most games ever since I got that GPU.

I have to think hard to remember any games where the GPU wasn't maxed and instead the CPU was maxing out. Hm, Fallout 4. And it wasn't exactly the CPU maxing out, but memory was the bottleneck (neither CPU nor GPU were maxed. but the game couldn't reach its own 60FPS cap.) And that's about it I can only recall that one game where the GPU wasn't the bottleneck. The other such game is one I only heard about, never played it: Watchdogs 2. Apparently that's also heavy on CPU.

If you play lots of online games, then yeah, it might be different. I only play one online game (CS:GO) so I don't have much experience to share there. But for single-player games? A 1070 is going to be the bottleneck for sure. At *worst*, the 1070 will be an even match for an 3570@4.2GHz.

 

I was, of course, talking about a "supposedly 100%" GPU load case. If you ever tested or upgraded to a much faster platform while keeping your still powerful GPU, you may have noticed that while your minimum framerates are obviously improved, your averages and max are as well, but your GPU load is still 100%. And your older CPU wasn't running at 100% anyway so why would there be any bottleneck then ? That's what I was talking about earlier. It's really not as simple as to look at the load graphs.

I know about a few solo games having such a behavior in a very marked way. Generally speaking, all frostbite 2+ games are concerned, but also GTA V, Crysis 3, Deus Ex : MD (even though it's not as obvious as other games because of how GPU heavy it is) and Watch Dogs 2 indeed. I'm pretty sure there are plenty but I'm just a man and can't play everything
RotTR is definitely not one of them, or at least not at a meaningful scale.

Btw, since you're playing CS:GO, some competitive players playing at 300+ FPS could relate that while even a stock C2D E6600 from 2006 isn't anywhere near maxed by the game, it's a major bottleneck to achieve high framerates with a GPU more than capable to push these stable 300+FPS, yet GPU load reads 100% on both platforms..

 

All I was pointing out is out of the two, the GTX 1070 is the obvious choice for being faster, not only because it is, but for the fact you'd hope it would be considering the price difference. It will be interesting to see if AMD will fill that gap in the market, considering there is likely a huge gap in performance and price between the RX 580 and Vega.

 

In France right now the GTX1070 costs two times the price of the RX480 8GB. Now what does AMD has to offer against the 1070 atm? The aging FuryX? We need more competition. I'm very curious to see Vega as well.

 

Gonna try that. Thx for tips.

 

You talk as if this is something unique or unusual. Its the nature of GPU performance tiers/value. The higher up you go in performance, diminishing returns kick in. What argument do you have for 1080ti buyers vs 1070? Or FuryX vs 290 a couple years ago? Its the way its always been with graphics cards, even more so with CPUs. Why would anyone buy a ryzen 1800x vs a 1700, etc, etc?

 

Well after playing Mass Effect Andromeda for few days on my new Phoenix 1070 i noticed my system used almost 8gb of onboard RAM when gaming, and had minor freezes when turning around in game. Finnaly i decided to get another 8Gb of RAM and i can say it's a jolly experiece so far. The game runs smooth on high settings vith Vsync and Triple buffering on. The freezes are gone, and it's a brand new experience. Considering my quoted dilema, at the end i got the best deal :nerd:

 

Congratulations! For enjoyment, it's well worth the premium. Have fun!

 

I moved from a r290 to Evga 1070 gtx, totally happy with upgrade. I think 480 is good but not going to overpower 1070. So far great performance and EVGA keeps bios updated.

 

While I do like my 480 a 2 year old 980ti can still kick its butt!

I would love a 1070/80 though $$$

 

GTX 1070, much much better choice

 

I'd recommend a EVGA 1070 SC at least over the AMD graphics cards.