Those other cards have two GPUs in case you didn't notice. SLI/CF-fps != Single-GPU fps.
In the case of GTX780 SLI vs. Titan, there is no compromise. Titan OC can not beat GTX780 OC by more than 5% since the Titan OC runs into a TDP bottleneck. Therefore, you get 95% of the performance of the Titan when SLI is broken and 180-190% of the Titan when it works. All that for only $300 more. Not only that but after market 780s have superior coolers than the reference Titan cooler.
47/35=135%
It IS half a generational jump, there is no denying it.
Yes, but only 12 fps more in a game. You pay $1000 for a GPU that outperforms a $400 GPU by 12 fps.
The 780 is a way better deal, that much is clear. Still your attempt to badmouth the performance itself is disgustingly biased. Percentages are valid - nobody forces you to play at (for your taste!) unplayable fps or settings. They can be adjusted...
You keep focusing on the %. Let's say 1 GPU performs at 20 fps and another at 27 fps. That's still a 35% increase which you would call half a generational. The difference is 35 and 47 fps are still slow. 12 fps more over a $400 GPU is laughable when it costs $600 to get that unless you make $1000 a day, or are a gaming addict.
You forget important information tidbits (like that Titan is already 4 months old)
The Titan being 4 months old is worse for its price, not better. You seem to be defending the Titan because it is 4 months old? I do not get your point. Now that the 780 came out, the Titan's price needs an adjustment. You seem to ignore this.
compare SGPU with AFR without a second thought
I compared the Titan to both GTX690 and HD7990. You can look at the graphs and those dual GPUs outperform it by 29-30%. In the case of the 690, it's micro-stutter is excellent.
, generalize about compute
Like? What I've said about compute is that GCN architecture is superior in compute than Kepler. This has been proven in many titles now. If we look at 561mm2 die size of GTX780/Titan and compare it to 365mm2 Tahiti in titles that use Compute shaders or compute for SSAA, global illumination, etc. it can be clearly seen that GCN is superior to Kepler. You have denied it for the last 18 months.
As far as generalizing compute for Titan, it's an advantage for the card. If someone is using professional programs that take advantage of many CUDA cores, the Titan is justified. Otherwise, it's a waste. For multi-monitors, the Titan can't take advantage of 6GB of VRAM since it runs out of GPU power.
and ignore important facts like specs
Huh? When did I ignore specs? Over the last 4-5 GPU generations between NV and AMD, I have proven with facts that you cannot compare GPU specs on paper between AMD and NV without looking at real world performance. Case and point pixel / ROP performance of HD7970 vs. 6970, which proves that you cannot even accurately compare AMD to AMD cards on paper. Therefore, I do not even understand your point on specs.
compare OC vs non-OC cards etc.pp.
Since when? For as long as I remember, I linked GTX460 OC vs. 5850 OC, GTX680 OC vs. 7970 OC, and I even go out of my way to show people benches for them like 760 OC vs. 770 OC vs. 780 OC. Additionally, I have linked comparisons of 7950 OC vs. 760 OC and 7950 OC vs. 7970 OC, as well as compared GTX670 OC vs. 7970 GE stock to show that 670's power consumption and performance matched a stock 7970GE once the 670 is overclocked.
There is absolutely nothing wrong with comparing OC vs. OC and OC vs. stock depending on the context of what's being discussed. If we are talking about how much power increases on a stock card to reach a performance of the next tier card and how much overclock is required, then it 100% makes sense to look at slower card A overclocked vs. faster card B stock. If you are talking about max performance, then we look at Card A overclocked vs. Card B overclocked. I think you have not been able to pay attention to my posts carefully.
Even recently, I linked reviews of all overclocked cards
here and
here.
This has become a habit of yours, and a bad one at that in my opinion. You should work on that.
Not it has not. What has become a habit is NV owners defending price/performance of their overpriced purchases. Every time I use real world performance and mathematics to show NV cards in negative light, I get accused of "bad" habits, or working for AMD.
And BTW, even Computerbase has already proved that the Titan is irrelevant for games since the cooler is worse than on after-market 780s and after-market 780s are faster than the reference Titan.
http://www.computerbase.de/artikel/...it-geforce-gtx-780-super-jetstream-im-test/2/
Anyone who is spending $1,000 on the Titan can easily afford $1,300 after-market 780s. For this reason, the Titan as a gaming card is a complete and total waste of $ unless someone is buying 3-4 of them because they are loaded and price to them is a non-factor. Despite all this, the Titan's performance in Crysis 3 is very disappointing, same for other games like COH2, GRID2, etc.
Facebook
Twitter