I think most people on this forum can attest to the fact that I recommended GTX460 and specifically for its strong overclocking headroom. I was fully aware that GTX460's power consumption skyrocketed but for the price it still offered great value I felt.
That's great RS. I'm not sure what it has to do with what we're discussing, but I'm sure you meant well.
Now if you look at the power consumption difference between GTX470/480 and 5850/5870, it becomes obvious the gap was gargantuan! How much faster was a stock GTX470/480 over 5850/5870 on average? 10-15%?
It's around the difference between an overclocked 7970 and an overclocked 680, less or more whatever you want to say is fine. If we look at modern reviews with more DX11 samples within those samples we see the 480 is 36% faster at stock including Crysis 2 where AMD disabled tess. When AMD wasn't disabling tess via the drivers it was over 100% faster (BM:AC). You won't find that performance advantage with the 7970 even with it's similar increase in power. So we go back to the route of this exchange. Why was that so bad, yet this is clearly worse?
http://forums.anandtech.com/showthread.php?t=2224543
What about if you overclocked 5850? It was still well under 480's power consumption with performance at least as fast, but cost hundreds of dollars less. Sure, there were cases where 480 owned 5870 but there are cases where 7970 GE owns 680 by 15-20% as well. On average though 7970 GE is faster for not much more power consumption while 480 was faster and uses 90-100W more, not to mention GTX480 cost more $ too and 7970 GE isn't exactly $130 more expensive.
Wut? The 5870 isn't even on par with the 470 in modern games, let alone trying to add in overclocking which GF100 thumped 5xxx series in without question. Your definition of "owned" is grossly overstated, owned is 113% faster in BM:AC, owned is having to disable tess via drivers, owned is 36% faster overall, 7970 does none of that.
And here is the difference between HD7970 Vapor-X vs. GTX680 at 1137mhz
There you go again, showing an older card vs a new one. I guess when you talked about fab maturity, and silicon improvement helping the 7970 you forgot that isn't a one way street. Perhaps the tunnel vision is to blame, perhaps not. Their 7970 was reviewed Thu, Oct 18, 2012. When was their "TOP" card reviewed? Do they even have a TOP card? They show no listing for a NVIDIA 680 TOP in their reviews.
Funny how that works isn't it?
27W. You just said 7970 uses 100W more than a GTX680. :hmm:
I didn't cherry pick, I went with the two most recent video cards reviewed on [H], the Matrix and the Gigabyte SOC. I didn't want to skew my perspective into being more favorable to my point of view.
Performance with Cats 12.11s speaks for itself. You can pick up a 1.05-1.1ghz HD7970 GE on Newegg for $450 with 3 free games. Asus GTX680 TOP costs more than $500 on Newegg. Other 680s cost less but they are still slower, come with no games, have worse overclocking headroom. Heck, skip the GE altogether and get one of these for
$387. If 1.05ghz 7970 GE is faster than 680, then a 1Ghz 7970 should be at least as fast and still cost way less than a 680. You can twist and turn, GTX680 is overpriced now and so is the rest of NV's line. Expect NV to respond with its own driver or game bundles because we know NV hates dropping prices above all.
Your misconception of how fast AMD's cards really are is quite humor to read.
Even without an overclock, 680 is worse value - costs more, performs worse. Same for HD7950 vs. GTX660Ti and 7970 1Ghz vs. GTX670. In the case of 7950 vs. 660Ti, it's not even a contest. With overclocking and 7950 will find itself going against 670 OC, not 660Ti OC.
Yet people keep buying the Nvidia cards, despite your opinion. And I call it an opinion because most people don't play every game benched, they're looking at the titles they play and even if you were to go by overall the performance difference between each is laughable.
There is no large power consumption difference. It's 20-30W. An 1150mhz HD7970 still uses less power than a GTX580. ()
There you go again comparing aftermarket cards to release day 680s, must be nice with those blinders on. Probably helps when you use last gen full node larger cards as your point of reference.
Did you also not see
my post where I said you don't need to run the GPU at 1.25V to reach HD7970 1.05ghz clocks? There are so many people online getting 1150-1200mhz overlcocks at well under 1.25V that AMD specifies.
Your idea of "stable" isn't anything close to AMD's, nor can AMD possibly run through every card and stability test each to their QA levels like indivdual users. Neither can Nvidia for that matter, this whole concept you've created is a moot point.
What about this guy who got
1230mhz at 1.174V on his 7970? You are saying all of us have golden samples? HD7970 doesn't need 1.25V to run 1.05ghz.
What does that even matter? Can all of you do it? What kind of statement/question is that? My 470 does 918 core on air with 2100 memory in benchmarks who cares? It's not important this isn't a benchmark competition.
The performance difference isn't minute. It's 10-15% faster at 1600P which is about as much as GTX580 had over HD6970 (if that). 580 cost $499 and 6970 was $369. HD7970 GE costs very close to the cheapest 680s but is actually faster and you get full voltage control + free games you can use or sell. See the difference?
It's not fast enough for 1600p and AMD is a bad choice for MGPU so where does that leave us? Does that make it's 10-15% performance advantage moot? I think it does, considering it's more like 10%, and 2% at 1080. 580 wasn't that fast at 1600p, I'd say they were pretty close, I'd also say like then 1600p doesn't matter it was a dual card resolution and still is and AMD is still the overwhelming wrong choice in that situation as it was back then. You get what you pay for.
GTX470 ~ GTX560Ti. On
this graph, that's far away from HD7970 GE / GTX680 cards. I'd say it's not bad for "mid-range" HD7970 GE/680s to be almost 2x faster than a stock GTX470. You can't have it both ways. First you say power consumption is a big deal, then you overclock your 470s to 900mhz in SLI.
I never said it was a big deal, you did. You said it was a big deal with 4xx vs 5xxx but now somehow it isn't, even though I've showed 4xx did far more with it's power than 7xxx does and that was my point. Do you understand now?
My on Air 470 does 900+, my SLI 470s did 970, my Tri 470s do 950. What exactly is your point? I did not have a problem with power when the performance backed it up, my point is 7970 does less with it's extra power vs the 680 than the 470 does against the 5870. I don't even care much about it, only asked why you said what you said. Which you never answered in this huge TL;DR post that virtually said nothing new and just repeated everything you've been saying.
Didn't you just say power consumption was a big deal for you? How does that work when a stock 470 uses more power than an HD7970? You realize 7970 GE is nearly 2x faster than a stock 470 for maybe 10-15W more power? :hmm:
No I didn't, you did. There you go again, true to form.
Keep giving half the story, avoiding the truth and manipulating the data to fit your position instead of changing your views with the information. :thumbsup:
Facebook
Twitter