I'm almost certain that the Titanic2 will outperform anything on this planet in FPU performance.
But as long as Intel doesn't improve the ALU performance and bus design, it'll continue to produce lackluster Integer and memory performance, forcing Intel to further increase their already ridicules L2/L3 cache size. (Which in turn, produces a huge core with lower yields, shooting the Titanic's price into the "I sold my lungs for a dual machine" range.)
Intel is currently losing money on the Itanium, and lots of it and using the Xeon's huge profit margin to sustain it.
Never the less, the Itanium, though being a expensive-lack-luster performance from day one, did do one thing: It killed off most of the competition making room for the Xeon to become their main server CPU design.
The biggest Irony is that the Itanium is improving, It's actually a solid option for FPU intensive applications; However, now that Intel has competition in the enterprise market, I wonder how much longer they'll continue to finance this black-hole when they can no longer artificially sustain such huge profit margins on the Xeon.
Gilboa
On Mon, 2005-08-08 at 22:18 +0300, Shlomi Fish wrote:
On Sunday 07 August 2005 16:17, you wrote: > * The Itanium (1.4Ghz, Medison core?) has lousy Integer performance and > memory performance. Don't touch it. (Or you'll burn... literally...) > Are you talking about the original Itanium or about Itanium 2? Itanium 1 was very crappy: http://www.mail-archive.com/linux-il%40cs.huji.ac.il/msg20048.html However, Itanium 2 is rumoured to be very nice. Here is some lines from a conversation in OFTC's #offtopic channel a while ago: <<<<<<<<<<<<< <velco> rindolf: virtually everything beated the crap out of itanium, itanium2 is different though. <rindolf> _deep: so while Itanium can be considered cutting edge, it's still quite crappy. <rindolf> velco: is it better? I did not hear about it much. <rindolf> Yet. <rindolf> velco: it is possible that they simply used brain-dead algorithm in the Itanium 1 design, and managed to refactor them for Itanium2. <velco> rindolf: prolly only new alpahas (ev7?) can compete with it. dunno for power4. all the rest bite the dust. <ijuz> IA64 has some great features, x86-64 is in a lot of regards cheap shitt <rindolf> velco: are you talking at chips with the same clock-rate? <velco> nope. <rindolf> velco: I mean an alpha with 600 MHz and Itanium 2 with 600 MHz? <velco> well, dunno for sure, but I think curent alphas ans itenium2 are clocked virtually at the same rate. >>>>>>>>>>>>> (I'm "rindolf"). Regards, Shlomi Fish --------------------------------------------------------------------- Shlomi Fish [EMAIL PROTECTED] Homepage: http://www.shlomifish.org/ Tcl is LISP on drugs. Using strings instead of S-expressions for closures is Evil with one of those gigantic E's you can find at the beginning of paragraphs. ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
