Originally Posted by wedge
The tests were very poorly run. Or perhaps it was done on purpose to make the 4890 look better.
Just look at what they did with the clock speeds. They overclocked both gpu's to 900MHz to keep it even. But they also decreased the memory speed of the 6790 to 3900MHz just to match the stock memory speed of the 4890.
With an overclocked gpu and underclocked memory, they created an artificial bottleneck in the 6790. Memory bandwidth is actually more important on that card since it's a "dual core" while the 4890 is a "single core". The poor little 6790 gpu was being data starved.
At stock gpu speeds, the 6790 would have a 10MHz disadvantage. But I bet that card could win at that speed, if it had full use of its memory capacity. In most tests it was only 1 or 2 fps behind, so it would at least be a tie.
That would show clear progress in power efficiency, price/performance, and image quality over the 2 year difference between those cards.
Correct, but HD4890 is a tiny bit more powerful than HD6790 even when both @ Stock.
If you look HERE
, you'll see that HD5830 > HD6790, and look HERE
to see that HD4890 > HD5830. Driver updates and games used might play a small role here but it should be HD4890 > HD6790.
Originally Posted by pursuinginsanity
But it's not a good comparison. The 4890 isn't even on the market right now, and it's common sense that new midrange cards will be able to keep up with 2 year old flagships. This has always been the case. This comparison isn't telling us anything we didn't already know, and certainly not anything valuable.
Correct but its not always about the performance or which card to buy, no one will suggest anybody to buy a HD4890 over HD6790, its about how much improvements.