Fudzilla over K10:
quote:
The real numbers are in Q4
Our well informed sources have confirmed that so far AMD didn't instruct its sales channels on how and when they are going to start selling Barcelonas.
As we are entering the last days of July our sources implicates that it is getting really unlikely that AMD will have any serious quantities of K10 CPUs in August. AMD has already officially said that this is the current plan.
We already told you that Barcelona comes in serious quantities in Q4, but AMD has to launch whatever it has in Q3 as this is what it promised investors back in December.
Even if they manage to get many Barcelona and Agena FX and Agena CPUs out, the biggest problem of them all will be to get significant sales and even in a best case scenario we would expect that K10 will be less than 10 to 15 percent of AMDs total processor shipments in Q4 with healthier numbers in 2008.
quote:
Promising scores but...
We learned that three to four percent of K10 Barcelona processors can actually hit speeds higher than 2.5 or 2.6 GHz. This is finally good news as AMD doesn’t need much more of these super fast CPUs to beat Intel in the high end. The trouble is that the majority of CPUs are doomed to life under 2.5 GHz.
These chips are reported to hit between 2.8 and some of them even 3.0 GHz which is exactly how much AMD needs to compete with Penryn and the current Intel 65 nm Quad cores.
The big question remains how are you going to utilize the performance of all four cores, as in most of the applications you won't see any benefit by going from dual core to quad core.
In current games you don’t see the any advantages which signals that in games the K10 Phenom FX won't be faster than the upcoming and much cheaper dual core Kuma CPU, but it will be faster clock to clock than the Core 2 Duo or Athlon X2 or FX.
quote:
K10 at the corner of a wafer doomed to 2 GHz
The curse of wafer
Some K10, Barcelonas can go over 2.6 and all the way to 2.8 or even 3.0 GHz but most of them at least the ones that are going to come as corner chips on the wafer and not going to work at much higher speed than 2 GHz if not less.
This can explain why the first Barcelona we pictured back at Computex worked at 1.6 GHz only. The second trouble is that not all the chips from wafer works and we know that AMD is currently not happy with yields of the chips and will likely need another revision and it can pray that everything including clocks, leakage and speed is going to be fine. Otherwise it won't be pretty.
Most of the Barcelona or Opteron 2200 or at later date Phenom FX CPUs won't actually work at 2.5 or more, they will rather be clocked to modest speed of only 2.0 GHz or lower.
2 GHz is far from being enough to dethrone Core 2 Duo who even at 3.0 GHz sells for around $270 and as we said numerous times Quad core won't bring much versus Dual core. In case it does Intel already has two iterations of 3 GHz or close to 3 GHz (2.93) so the pressure is certainly on DAAMIT.
Dus goede en slechte berichten. Het is ook zo dat AMD een erg grote stap heeft gemaakt qua ontwerp. De K10 is de meest geavanceerde cpu op dit moment. En de enige Native Quad core. Ze hebben ook elke core afzonderlijk nog aangepast. En dat kost ze nu de kop omdat ze te vernieuwend zijn. De Yields vallen tegen clockspeeds zijn daardoor slecht. En het product is niet zo goed als het had kunnen zijn. Volgens mij heeft intel daarom eerst gekozen om een snelle dual core te maken en daar er gewoon twee van aan elkaar te plakken. Resultaat snelle quad core. Met goede Yields.
Hopelijk gaat AMD dit straks ook doen... Want dat bespaart ontwikkeling kosten. Volgens geruchten van een tijd terug zouden ze twee Shanghai's aan elkaar gaan plakken. Om een Octa Core cpu te krijgen. Ik ben benieuwd.
Dus als het een beetje mee zit hebben we volgende maand benshmarks

Zou wel tijd worden.....