That was a nice read Scoob, you make some awesome points as always. You might be right about AMD moving away from the "Bang for Buck" pricing strategy that they have been using until now.
Hi,
Thanks
It might make the market a little more interesting over the coming months too. Consider that traditionally - and I'll use 6000 vs 500 series as the prime example - we'd expect AMD's high-end single GPU part to trade blows with NV near high-end, so 6970 vs. 570 for example. This is in the "normal" situation of good HD (1920x1080) gaming many PC gamers kit supports.
Next we'd usually see AMD's high-end single GPU nipping at the heels of NV high-end single GPU card. Never quite as quick but a fine offering and often at a price point that makes it hard to resist. It appears to me that AMD's approach up to now has made it cheaper for them to produce their high-end offering, so they can price
very aggressively as needed.
If with their new generation of GPU's the production costs are UP per unit, AMD may not have quite the same degree of room for positioning their products at the "right" price-point based on performance offered. We of course need to wait to see what levels of performance, and of course price, the more-popular mid-range cards offer. Previous gens of AMD GPU's scaled exceptionally well, giving them very strong mid-range offereing for the masses - possibly due to the "more, but less complex" nature of AMD "cores" vs. NV's. NV did well at this with the 500 series in my mind, but had been a little behind previously. Typically a mid-range AMD card offered more for your money that a similar NV card - but that's just my view. However, the late-arriving 460 did redress the balance and begin to hint at the true potential of Fermi as an all-rounder (Price, performance, power, heat) that the 500 series later proved.
When it comes to DUAL GPU's on a single card however AMD have been showing everyone how it's done. The older generation "x2" cards were epic, if sometimes a little tricky to get working perfectly. It's here NV
really need to catch up as, to be quite frank, the 590 was a bit embarassing and I so wanted to like that card. I know later revisions (which came quite quickly if I recall) basically sorted things (like with the 295) and better drivers smoothed things out further, still, it wasn't such a great release in my mind. Considering the 500 series was a great step forward over the 400, the 590 wasn't quite what I'd hoped for. It's a shame it took ASUS with their MARS II to get it right...but considering the price they sorta lost that particular fight. But hey, MARS is a whole different marketing exercise really so we'll call that a special case
Looking back, my old GTX 275 was a great card, one-step behind the top single-GPU offering from NV, the 285, yet very powerful, overclocked well (for that generation) and is still going strong in my media box. The 460 (a late-comer to the 400 lineup) wasn't actually much faster in DX9 and 10 titles and DX11 didn't have much of a draw right away as nothing really used it's features. The 470 (a logical upgrade path and available long before the 460 of course) performed OK, but not
that much better and was expensive, hot and noisy, being as it was a slightly older design than the 460. The 480 was at the top of the performance tree but even hotter and noisier, so I avoided that one. The 570 nailed it for me upgrade-path wise. It gave me ~double performance, DX11, was cooler and quieter and used less power than my 275. Additionally the 570 had great overclocking headroom.
The reason I mention the above is because I want to see such a jump going to the 700 series. For me (if I were to upgrade) a 770 would likely be the level I'd consider - albeit as a pair for SLI. How I see things is a GPU upgrade usually occurs ever two generations or series releases. I.e. 275 to 570 gave me double the performance and some nice new features while being cooler and quieter and less power-hungry. So, those on say a 470/480 now would ideally like to see the 770/780 giving them double what they have now - something I thing 28nm and some architecture tweaks may well bring. We shall see.
I'd like to add that the 480 is a great GPU, my friend runs a pair under water and they are epic running his 30" screen at 2560x1600. I've said for a while that it's almost like these cards were made for water, they really are good. However, a little while back, I did have experience using an air cooled (stock) 480 for a while. It really is hot and so very noisy on that stock cooler - though I know the aftermarket offering improve this massively. If I'd opted for a 480, as could so easily have happened had the 500 series not materialised or arrived late, I think I'd be looking eagerly at Keplar right now and, if NV bugger it up,
knowing that AMD have a very strong offering availble NOW.
If Keplar turns out to be slightly less than we'd hope, AMD will rightfully take the lead with their excellent offering - caveat being how the rest of the line up...erm...lines up in the price/performance stakes of course
Worst-case I'd like NV offering to be "good enough" to keep AMD honest, the market busy and the bargains coming! I'll admit, I'd be very pleased if NV pull an "8000 series" coup with Keplar as that'd certainly shake the market up and give us even more options. Imagine, if you will, that NV's 780 is 30% faster than the 7970 on average. AMD would (traditionally speaking) then price more aggressively still. 7970 for £330? Damn, that's a lot of GPU for the money even if the 780 performs better but is say £450! The BIG question is of course, can AMD still do this with their new much more complex design? I'm seeing 7970's ranging from £480 to £515 currently online, which is really rather expensive. Isn't this higher than the 580 was actually available for at launch? I'm sure around £450 was normal with some brands being near £400 - I did look at the time before eventually deciding on my first 570 which was just £250 (+£10 P&P) but offered 85%+ the performance...
Interesting times in GPU land for sure
Scoob.
P.S. My cave evidently has internet so I guess I'll stick around to see how this one pans out