I think there must be some rationale for AMD's numbering scheme on their RADEON HD series.
the card is typically numbered with 4 digits, here represented by 4 characters: SXYY
S- Series. Every new series follows on the next. The most recently anounced is the 7000 series with only the 7970 being available. Prior to this there was the 6000 series cards. Prior to that the 5000 series.. you get the picture.
X - ????. Architecture?
YY - Model. Typically Radeon has the 90, 70, and 50 models within each grouping. They will have a 6890 and the 6790 within the 6000 series. This is where the X above becomes relevant, but I don't know in which way. In the 6000 series, they also have the 6970, 6870, 6770, 6670, and 6570.
I suspect that the number YY here may also perhaps be indicative of Gaming, enthusiast and value, for instance, but I'd like to know.
When I look at benchmarks, it is very confusing, since often a previous series card is ranked very close to a current series card.
S - Series, simplest thing. But you can't compare different series like how you try to do .
X - Pretty much architecture, like Cayman is 69xx, Barts is 68xx, Juniper is 67xx.
However there are some exception, like 6790, which is actually barts, or turks which is both 66xx and 65xx. I know it's pretty confusing.
But we don't to be confused with this, so what we do is basically - the more X gets, the better the card is (again, in the same series).
YY - X indicator - or 30, 50, 70, 90, Indicates what is faster from the same X's, let's say 67xx series, the 6790 and 6770. Because of YY we know that 6790 would be better than 6770. But again, you can't compare the YY without the same X. Let's say 6950 and 6790, you can't compare and 6950 is clearly faster.
Hope this helps and clear out your confusion about AMD cards . Notice also that the variables depend on the variable before it.
I wouldnt bother looking at the numbering scheme or trying to decipher it. Yes it is good to know which is supposedly low, med, or high cards - but benchmarks and graphs are better at telling them apart (as well as of course price, all together).
The numbering/naming schemes are flawed for both companies (amd and nvidia). The 6770 is pretty much identical to the 5770, for example. What they do sometimes is take the last generation medium GPU and make it either a low end "current" gen gpu, or rebadge it as a "medium" GPU, even though technically, they are practically the same.
I was wondering if the YY-designation denoted strength in particular aspects of the card's performance, more than others. Maybe shaders, ROPS, whatever. I'm not 100% up to speed on all terminology used.
If one is looking for price/performance, it seems to me that one may find a older generation card of a higher rating and come out ahead.
But if power consumption/heat is a factor, then I guess the most recent architecture is most likely to run cool and cheap?
Well, you probably can't go with the last cards... Because most of the time they are out of stock at newegg (like the 5xxx series). Even though it will be a good deal, But it's not always cheaper than the newer generation card which has better architecture and usually better dirextX and openGL support. Sometimes old card are not worth it. Also the efficiency and temperature should be lower on older graphics.
Also YY doesn't look out for specialized strength, usually they just have higher clocks or more shaders, VRAM, etc.