SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?


To: Paul Engel who wrote (72960)2/5/1999 2:15:00 AM
From: Tenchusatsu  Read Replies (1) | Respond to of 186894
 
<IA64 architecture: A chip set under development for Netfinity servers, based around the Merced 8-way symmetric multiprocessor, that contains copper wiring for significantly faster processing.>

Minor correction: Merced, like Xeon, is 4-way symmetric, not 8-way. The 8-way server will have to resort to at least two separate processor buses.

Anyway, this is good news for Merced.

Tenchusatsu



To: Paul Engel who wrote (72960)2/5/1999 12:02:00 PM
From: Tony Viola  Read Replies (1) | Respond to of 186894
 
Paul, Re: "IBM will use Copper Metallization in a Merced Chip Set !"

Great article. Some comments:

But it also mentions a new chip set that IBM is developing internally for their
upcoming 8-way SMP Merced servers ! And it will use their copper metallization
process !


Sounds like IBM wants to be out there real early with Merced servers. They must be getting early Merceds to hook up with this copper connected chipset (this fall-winter?).

Trying to gain an edge in the voraciously competitive PC-based server market,
IBM on Thursday will launch
its Netfinity 99.9 Percent Availability Program that guarantees corporate users
99.9 percent uptime on
selected servers.


This is breakthrough for NT servers. 99.9% availability (the A in RAS) is right at what IBM rates their S390 servers. 24X7X52 = 8,736 hours in a year. 0.1% of that is 8.7 hours a year of unavailability. That usually even includes a certain amount of upgrades, which can be done concurrently, over an internet link nowadays (no downtime). Or, hardware upgrades are hot-pluggable in many cases, as stated. They mention PCI, most likely would also include power supplies and hard drives. Also included in the unavailability, of course, would be downtime due to failures. 8.7 hours unavailability for all reasons is extremely low in a year. Bottom line is that this is as agressive as it gets right now, right up there with the flagship of IBM, S390. All this and on a Windows operating system no less. Ov Vay! This product will bear very close watching.

More, OnForever: A feature which is intended to extend the high availability capabilities
of Hot PlugPCI to
processors and memory offering online real time diagnostics


What is offered in S390, e.g., is that as many CPU's as can be populated on the system board are put there by Manufacturing at IBM, regardless of how many customers may order. When the customer has the system, and wants to add CPUs over and above what he initially paid for, IBM configures them dynamically, over the link, and the customer pays for them at that point. How would you like this strategy to be used for Intel based servers, i.e., four or eight Merceds bought and paid for by IBM for each and every NT server motherboard they build? I don't know if they'll incorporate this into NT servers, but when they started talking hot plug CPUs for S390, this is what it turned into.

All good stuff for Intel here.

Tony