Interestingly, Foster, the Pentium 4-based revision to the Xeon server processor family, will be launched on the same day. Foster was to have been launched today, but Intel delayed it to the end of the month because of a problem with the chip's in-system packaging.
We wonder if Itanium was to be have been launched today too. That might explain IBM UK's announcement last week that it's going to launch Foster and Itanium-based systems, the M- and Z-Pro Intellistation machines, on 10 May. That would time IBM's event just after Intel's main Foster launch.
IBM will apparently roll-out a dual-processor M-Pro workstation, based on two 1.7GHz Xeons - Intel is dropping the 'P' word - and the Z-Pro, an Itanium-based box powered by two 800MHz chips. We hear there's also a uni-processor Xeon M-Pro box based on a 1.4GHz part. The Xeon machines use Intel 860 mobos. The Itanium system will, we reckon, be based on an Intel 460GX motherboard with 4MB of L3 cache.
IBM has already been demo'ing the boxes, running 64-bit Windows XP on the Itanium machine. The OS was described as "quite stable" by our correspondent.
Sources who've seen IBM's boxes tell us the company is claiming it will ship them on 29 May, at an Intel Itanium 'coming out' (in the tradition sense...) party, by the sounds of it.
Itanium has, of course, been doing the rounds for a while now, under Intel's IA-64 pilot programme. We imagine the 29 May event will mark its official entry into the product line. ®
>what applications can take advantage of 64bit processing?
Scientific & engineering stuff can take advantage of it because they can now efficiently use long double variable types. This will reduce error growth due to machine precision issues during large calculations.
In theory, there is no difference between theory and practice.
In practice, there is.
Obviously there are huge advantages for people using large amounts of RAM or high precision mathematics, however, for SOHO servers, the advantages will be next to nothing. On everyday tasks such as web serving (with 512MB of RAM or so), the first dual Itanium systems have been show to perform worse than a pair of 866MHz PIII Xeon (1MB).
I'm not sure about the pricing, but I guess if you want 50 processors in one system, it might be an option. I think server farms of smaller systems are becoming more popular, however, or high bandwidth cluster networks based on beowulf principles.
Maybe Merced will be faster. In the mean time, I guess the best choice for (most) people will be the Foster Xeon, or Northwood Xeon. That is if you aren't trying to calculate pi to 4 billion decimal places
errr....as I understand, long doubles are floating point...which already hit 80 bits with the 486, and hit 128 bits with SSE/SSE2. Is there some internal architectural difference that 64-bit provides for floating-point?
I suppose it would be nice to be able to toss around integers larger than 2.1E9 though.
64-bit would be interesting for fixed-point, except that fixed-point went out of style after the Pentium's debut. :wink:
bash-2.04$ kill -9 1
init: Just what do you think you're doing, Dave?
In C/C++ a float is 32-bits. A double is 64-bits. However, all floating point arithmetic on x86 processors are done in 80-bit. The results get truncated to fit the size of the required type.
A 64-bit CPU would provide for 64-bit integer registers and 64-bit instructions for adding, multiplying, etc. It would also supply 64-bit addresses, in addition to other features.
= The views stated herein are my personal views, and not necessarily the views of my employer. =
Hp has been making Workstations and Servers with 64-bit RISC processors for years, yet nobody ever mentions them when praising the Itanium. In my opinion, I think our outlook as consumers should be " Where've you been all this time Intel? " instead of " Wow I can't wait to get an Itanium and run a 64-bit version of XP ".