Originally posted by MacCoaster
I never said that it was depenedent on MHz. I simply stated 2000 sounds puny to 4000... FROM a marketing standpoint.
Agreed, 2000 vs 4000 sounds bad from a marketing performance viewpoint. But my issue was why are we getting in to comparing the numbers? This way of comparing manages to exclude anything about hardware build quality or software quality. I think the Mac community should be trying some agenda setting of its own regarding quality, rather than having to get defensive about numbers. All these number comparisons avoid many significant aspects and I was reacting to seeing another sub-thread starting on just a numbers comparison. For example, I've spent a while reading Cocoa documentation (not quite enough free time to really get down to interesting coding unfortunately though). So, for basically no effort, every text widget can benefit from text-to-speech and textual summarise capabilities - does MS offer this? These have no direct relationship to the performance numbers, but they strike me as pretty significant (at least noteworthy) capabilities. Likewise, when I'm typing, I get a voice telling me that Mozilla needs my attention and the focus remains precisely where it is until I decide to go and see what was upsetting Mozilla. Heck, I've tried that last item on a range of "generic" users and all have them have responded that they'd like to have that feature available. but nowhere in their reaction was there anything close to 2000 vs 4000 comparisons.
So my argument is that, by only talking about how the numbers are not what they seem, Mac users are ignoring the thing that really makes a difference, namely quality.
See post directly above for AMD Hammer. As for Intel's Itanium, I'm not sure. Let me check. I'll post the number in an edit.
Okay. Intel's Itanium has 6.4GB/sec memory bandwidth as stated here: http://www.intel.com/products/serve...um2/index.htm?iid=ipp_srvr_proc+itanium2srvr& so Apple and IBM are behind the pack already.
The memory bandwidth issue is going to be an interesting debate for a good few months to come. Elsewhere, we had a figure of about 19GB/S for Hammer, but spread across three buses, which makes for only fractionally higher peak bandwidth that either PPC970 or Itanium are talking about. It is
very believable that the issue of how fast the CPU is is going to be largely irrelevant because future memory systems are going to struggle to keep up.