If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart. <P ID="edit"><FONT SIZE=-1><EM>Edited by darko21 on 08/17/04 09:35 AM.</EM></FONT></P>
oh great quad core cpus, so now what? no advancement in frequency and speed but just more cores. I dunno , but are we gonna be stuck at 3ghz for the rest of our lives? Using multi-cores is a last resort to gain performance. CPU companies are hitting a wall in frequency. This will slow down CPU improvement and development.
All is not well in AMDville though. The dual core chips do lack one thing, a second memory bus, <Snip> The two chips will share a single 128-bit memory channel. If you look at the PR ratings when going from a socket 754 to a socket 939 Athlon64, you will see that values this at about 1-200 PR points.
But we all know that the <i>actual</i> gains a single-core K8 get from dual channel are very slight, because it simply doesn't take advantage of the available bandwidth. but a <i>dual-core</i> K8 would surely (for well enough coded apps) be able to take advantage of this, so would dual channel actually make a comparatively large difference for certain tasks on the AMD systems?
I could be wrong, but the thought popped into my head while reading that.
I'd expect bus contention on dual-core chips to occur anyway, be it in the form of dual K8's single memory controller or crudely halving available FSB resources per core with dual-core P4s.
I wonder if the special "arbiter chip" that they developed for Montecito - whose design has been complete for over two months now - can be used with other cores (not IA64 ones)?
Also, a good approach in solving the bus contention problem would be using (duh) dual core dothans instead of P4s. The 1066Mhz FSB that will be available will be twice as fast as the current dothan bus - it wouldn't slow the cores down as compared to current designs. Of course, it seems increasingly unlikely that Intel would choose dothan for dual-core desktop chips, given the added amount of information that points towards netburst...
<i>Edit: Oh, and BTW, the usefulness of multicore is currently greatly exaggerated. It will only make any difference for the average Joe in at least two years or more. Quad-core is hype only, right now. I mean, I'd love to have a quad-core 2Ghz AMD64 system, but it will probably take at the very least until 2007 for multicore to really show any potential.
And add to that that a quad-core A64 in 2007 isn't so surprising; remember, Intel's quad-core Xeon/Whitefield is made of completely new cores, not netburst ones, and is due in 2006.
Let's wait for multicore to make any real and widespread difference before getting excited about it. </i>My thoughts on the subject...<P ID="edit"><FONT SIZE=-1><EM>Edited by Mephistopheles on 08/17/04 04:34 PM.</EM></FONT></P>
I don't think it will be a major bottleneck. Remember, there are 2 way opteron boards where memory slots are only connected to one socket (cpu). This means a 2 way opteron on this board would see only half the bandwith of a "normal" opteron board where each cpu gets its own DIMM slots. The funny thing is, that when benchmarked, the difference is barely measureable. Therefore, I think a dual core opteron with just one memory controller will do just fine..
= The views stated herein are my personal views, and not necessarily the views of my wife. =