Since you are having trouble counting
The 480 scales better than older cards and well... is using 6 months newer drivers than anything you've managed to find as well.
Crimson 16.6.2 - Radeon
GeForce Game Ready 368.39 WHQL - GeForce
Considering 47.59% gamers are still on dual-cores according to Steam Survey, which includes Core i3 and slower Pentium/Celeron CPUs, it's a shame more reviewers didn't test this new mainstream card with <$200 processors.
Considering the criticism I endured, when defending my current CPU in my gaming rig from the QCP, saying that "dual-core gamers should be burned", and "real gamers have quad-cores" - is it any wonder with that kind of sentiment in the forums, that a website wouldn't even bother to test with dual-cores? And yes, I am in the target market for an RX480.
You're reading the situation wrong.
The reason why (Since Sandy Bridge came out) you should have never gotten anything less than a 4 threaded Intel Sandybridge @ stock 2500k level of single threaded performance is what these tests blatantly show.
The marketers on the forums that try to say otherwise are simply lying to people for various reasons and agendas.
Yup, that's why this thread is a joke. The OP is trying to prove that AMD cards are CPU bottlenecked with garbage i3, FX and 3200mhz i5s, while blatantly ignoring that those same CPUs still bottleneck the 970. The argument is flawed since it positions the argument that one shouldn't buy an R9 390/390X/480 with garbage CPUs but none of those CPUs even allow anything faster than a 960 to be fully utilized....
The real takeaway from this article isn't that AMD cards are more CPU bottlenecked, but that for RX 480/390/970, you want nothing slower than a 2500K OC and for 1070/980Ti/Fury X, nothing slower than a 2600K OC.
That's why to me any 2016 gaming PC without at least an i5 6600K is garbage. I see people with i5 2400 or i7 980X looking to get a 1070/1080 and I just have to laugh.
Another reason why the argument against RX 480 is flawed is that there isn't a single NV card that has 4GB of VRAM and good DX12 performance at $199. Chances are someone who has an FX6300/i3 can't afford a $250-275 GPU.
If I were building a new PC in 2016, and I couldn't afford at least a 6600K, I'd either go into the used market for 3770-4790K or just save up.
If someone cannot afford a $240 6600K to keep for 5 years, they are likely pirating PC games. The TCO even if one sells the CPU for $90 in 5 years is only $150/5 = $30 a year! Most people in Brazil, Russia, Asia who buy the garbage APUs, i3, FX, etc. straight up pirate PC games because it's not logical that someone can afford $30-60 USD PC games but they can only afford a $50-100 CPU over 5 years? Please.
What we should be doing is guiding PC gamers to spend more upfront to build a good foundation for a PC (CPU / good PSU, good monitor). The videocard can always be upgraded 2-3X over 5-6 years.
The argument in favour of 970 is completely illogical because if someone could only afford an i3 6100 instead of an i5 6400, then how the hell is this gamer going to have the extra $ for a 970 over the 480 4GB? Secondly, this thread straight up ignores the awful DX12 performance of the 970.
It sounds like a pre-launch 1060 3GB NV PR to get people to skip the 480 in favour of a VRAM gimped NV card. IF NV releases a $199 1060 6GB, then we can talk. Until then, a $199 480 4GB is the best mainstream GPU.
It's funny that you don't even (appear to) realize who I'm talking about.
All AMD/ATI GPUs going back to DX9 at minimum have massive driver overhead compared to Nvidia cards.
This thread is pointless because if people say they don't know that by now, they either have a vested interest in now saying that they don't know that, or they are dense enough to not know something for over 10 years, and anything said about it now isn't going to get through to them. Essentially vested interest and/or Dunning-Kruger effect.
And you have a short memory, apparently. Nobody is disputing Core i5 and above can handle all these cards just fine, you don't need to post the same benchmarks three times in a row.
It's funny that you don't even (appear to) realize who I'm talking about.
All AMD/ATI GPUs going back to DX9 at minimum have massive driver overhead compared to Nvidia cards.
This thread is pointless because if people say they don't know that by now, they either have a vested interest in now saying that they don't know that, or they are dense enough to not know something for over 10 years, and anything said about it now isn't going to get through to them. Essentially vested interest and/or Dunning-Kruger effect.
EDIT: Why do I bother. 95% of the people posting on the crucial subforums are marketers now (as opposed to probably 75% from just a couple of years ago). You win, I need to stop wasting my time here.
Yet you keep posting i5 results while saying that they don't matter...
Did you not notice that the i5 is ALSO a bottleneck when not overclocked?
Are you seriously trying to tell me that isn't within margin of error
Generally it isn't. But as you can see in the Fallout 4 benchmark you just quoted, the Radeon suffers more when stepping down from Core i5 to Core i3:
'Geforce GTX 970 goes from 43.4/39 -> 40.8/33 with Core i3 - 6.3%/18.1% faster'
'Radeon RX 480 with Core i5 goes from 38/34 -> 34.2/27 with Core i3 - 11.1%/25.9% faster'
No, it's a well known fact.
Generally it isn't. But as you can see in the Fallout 4 benchmark you just quoted, the Radeon suffers more when stepping down from Core i5 to Core i3
Tts obvious you didn't post this thread to have a meaningful discussion when you posted 4 results, I posted 4 and you called them all invalid even though computerbase does the best testing around and stress tests their cards not just 5 minute testing periods and done.
Not going to both posting here anymore.
You might wanna check his posts again, I counted 4 games just like the OP provided.You posted Radeon RX 480 results from two games, one of which can yield very different results depending on the scene tested, as already explained.
You might wanna check his posts again, I counted 4 games just like the OP provided.
Yup, that's why this thread is a joke. The OP is trying to prove that AMD cards are CPU bottlenecked with garbage i3, FX and 3200mhz i5s, while blatantly ignoring that those same CPUs still bottleneck the 970. The argument is flawed since it positions the argument that one shouldn't buy an R9 390/390X/480 with garbage CPUs but none of those CPUs even allow anything faster than a 960 to be fully utilized....
Snip
^ Nailed it in one. I don't want an i7 or +130w AMD in my HTPC if the only reason it's 'needed' is to brute force through a driver bottleneck that occurs predominantly on only one brand of card. Of course that extra cost will have to be factored in. Aside from HTPC use it'll be used predominantly for older DX11 gaming. (Like you I'm not interested in DX12 AAA stuff or VR either). An i3 does bottleneck the 970 on average fps but nowhere near the same extent as same i3 bottlenecks even much lower end cards like the 7790 range on min fps. I simply don't need to believe what I've already measured in person (in addition to 4 separate tech sites commenting on it) and will continue to take my observations over others "how dare you even raise the question!" strong emotions.The question of CPU overhead is still relevant, because whether it meets your standards or not, a lot of people are still gaming on the "garbage" cpus that you mention above. They could well be looking to upgrade their gpu only, in which case cpu overhead is a factor to consider. It is also irrelevant if an nVidia card is bottle-necked, because an AMD card with more cpu overhead would be bottle-necked at a lower level.
Now I am talking more in general terms, not so much 970 vs 480 specifically. Between those two, if they are the same price, I think it comes down to whether you are more interested in current and recent fairly demanding DX11 games that tend to favor nVidia or in anticipated DX12 games that should favor the 480.
Edit: Personally, and I know I am in the minority, but there are currently zero DX 12 games that I am anticipating being interested in. I mainly intend to play my GOG and steam backlogged games, and maybe purchase some 2015 era games like GTAV when they get cheap on a steam sale. I am content to stick with my HD7770 for now until the new cards sort themselves out and the prices stabilize. If it comes closer to the promised efficiency than the somewhat disappointing (power wise) 480, I might consider a 470.
Radeon RX 480 would only do worse with 2C/2T CPUs.
^ Nailed it in one. I don't want an i7 or +130w AMD in my HTPC if the only reason it's 'needed' is to brute force through a driver bottleneck that occurs predominantly on only one brand of card. Of course that extra cost will have to be factored in. Aside from HTPC use it'll be used predominantly for older DX11 gaming. (Like you I'm not interested in DX12 AAA stuff or VR either). An i3 does bottleneck the 970 on average fps but nowhere near the same extent as same i3 bottlenecks even much lower end cards like the 7790 range on min fps. I simply don't need to believe what I've already measured in person (in addition to 4 separate tech sites commenting on it) and will continue to take my observations over others "how dare you even raise the question!" strong emotions.
Like you, I'm in the target market for a 480 / 1060 (or more likely a 470 / 1050) that'll be used predominantly for older DX11 games. What I want to see is more real-world testing in general, including RX 460/470/480 and GTX 1050/1060 cards on lower end CPU's. Not a particularly outrageous concept to want to see if the rate of drop off is the same. I'm astounded though, how over-emotional and hyper-reactionary some get though when people "dare" to question AMD scaling issues on non top end CPU's (and 2 of my past 3 cards have been AMD, along with a roughly 50/50 ATI/nVidia split and 45/45/10 Intel/AMD/Other CPU split going back much further to ye olde 286's, so I'm hardly some "newb" or brand-addict).
As usual, this stuff flies straight over some people's heads who seem to persistently struggle to understand why people don't match up $350 CPU's with sub $200 GPU's in secondary light gaming rigs...
It's really hilarious - are you seriously suggesting to consider such a GPU for dualcore CPUs?
Like you, I'm in the target market for a 480 / 1060 (or more likely a 470 / 1050) that'll be used predominantly for older DX11 games. What I want to see is more real-world testing in general, including RX 460/470/480 and GTX 1050/1060 cards on lower end CPU's. Not a particularly outrageous concept to want to see if the rate of drop off is the same. I'm astounded though, how over-emotional and hyper-reactionary some get though when people "dare" to question AMD scaling issues on non top end CPU's (and 2 of my past 3 cards have been AMD, along with a roughly 50/50 ATI/nVidia split and 45/45/10 Intel/AMD/Other CPU split going back much further to ye olde 286's, so I'm hardly some "newb" or brand-addict).
As usual, this stuff flies straight over some people's heads who seem to persistently struggle to understand why people don't match up $350 CPU's with sub $200 GPU's in secondary light gaming rigs...
No, you are making this up. This thread highlights the importance of strong CPU performance. The problem is, not everyone will pair Core i5+ processors with Radeon RX 480/470. As you can clearly see on Steam Survey, a huge amount of people are still using dual-cores, and Polaris 10 comes out as a cheap upgrade to them.
Ideally we would have both. That's what I (and many others) are simply calling for - more in depth tech reviews rather than just the same 20-25 tech sites doing copy-paste stuff. Yet when some sites do that, there's this "clan" that pops up and endlessly attacks them for portraying the "wrong image", even childish calls for "permanent bans" for merely asking a question (see page 1)... Of course, we see the same review disparities in general in the PC industry, eg, "950 PRO is 3x faster than 850 EVO" = 14.7s vs 14.8s game load times, but for some reason anything that involves a certain brand of CPU / GPU and it "touches a raw nerve" for some, and out come the forum pitchforks...We would need a review testing those older DX11 games, not new games then to make your point valid.
The biggest takeaway from this thread is everyone seems to want to talk about everything else except the main topic of this thread (how much is the driver overhead, has it changed / improved, what steps are AMD taking to minimize it, etc)... As usual it's ended up "If x wants to talk about something, then attack their need to do it" which isn't answering the question, it's running away from it. :sneaky: Regardless of what I (or others) own or intend to buy, the discussion / question of how much is the NV vs AMD disparity of performance scaling of i7 -> i5 -> i3 is still an entirely valid metric to measure, even if some people really do not want it discussed...That's the biggest takeaway from this thread: