What's coming up in hardware?

My P4 boxes are getting long in the tooth, so I'm lurking again and starting
to think of new builds. Maybe this winter.
The new Penryn class of processors look promising, once the price gets down
to the normal range, and I'm sure there will be a lot of new mainboard specs
and peripherals options once they are mainstreamed. Is AMD coming out with a
competing product?
So what else should I be keeping my eye on and researching for the coming
months? What do you know?

TVeblen wrote:
My P4 boxes are getting long in the tooth, so I'm lurking again and starting
to think of new builds. Maybe this winter.
The new Penryn class of processors look promising, once the price gets down
to the normal range, and I'm sure there will be a lot of new mainboard specs
and peripherals options once they are mainstreamed. Is AMD coming out with a
competing product?
So what else should I be keeping my eye on and researching for the coming
months? What do you know?



The next thing for Intel, is Nehalem. Nehalem is a socket change
for Intel.

This article clarifies the difference between Nehalem for
servers and desktops. The desktops get a "smaller" socket,
and no integrated memory controller. Desktop Socket H (LGA 715)
versus server Socket B (LGA 1336). The word "rumor" appears
in the title (so better, more recent articles may exist elsewhere).
So this could amount to an FSB change for the Intel desktop.

http://www.xbitlabs.com/news/cpu/dis...815122559.html

Another version of the article, here. I don't see how there
is room for an integrated memory controller, with only 715
contacts. Many of those are used for power and ground for Vcore.

http://www.dailytech.com/Gearing+Up+...rticle9823.htm

I think the safest bet is, the socket will change, and what exactly
it is, won't matter until it is released :-)

I think there might be a later AMD roadmap around than this one.
This one is from December 2007. AMD is working on 45nm technology,
but one article I read several months ago, mentioned their 45nm
approach is not a complete redesign, and more of just a shrink.
That should help their profitability with what they can sell,
by making the silicon die smaller. But in terms of overclocking,
unless a miracle were to occur, I wouldn't expect the top
clock to change so radically, as to overtake Intel. It may
also help, when using companies like Chartered, to aid in
volume production.

http://my.ocworkbench.com/bbs/attach...1&d=1197568327

Depending on what you are doing with your computer, I think the
existing, mature technology is pretty attractive. Cranking the
FSB doesn't do anything, unless the thing behind the FSB is
starving.

AMD cannot be as forthcoming, about their future plans. On the
one hand, they can display the developments that are imminent,
and likely to succeed (that is called "execution" and is important
to stock valuation). But for AMD to claw their way into real
competition, takes a tech transition, and I'm sure there are
some engineers losing sleep over that as we speak. If they cannot
rely on process technology to make them competitive, then it'll
have to be by architecture alone.

Just a guess,
Paul

"Paul" wrote in message ...
TVeblen wrote:
My P4 boxes are getting long in the tooth, so I'm lurking again and
starting to think of new builds. Maybe this winter.
The new Penryn class of processors look promising, once the price gets
down to the normal range, and I'm sure there will be a lot of new
mainboard specs and peripherals options once they are mainstreamed. Is
AMD coming out with a competing product?
So what else should I be keeping my eye on and researching for the coming
months? What do you know?

The next thing for Intel, is Nehalem. Nehalem is a socket change
for Intel.

This article clarifies the difference between Nehalem for
servers and desktops. The desktops get a "smaller" socket,
and no integrated memory controller. Desktop Socket H (LGA 715)
versus server Socket B (LGA 1336). The word "rumor" appears
in the title (so better, more recent articles may exist elsewhere).
So this could amount to an FSB change for the Intel desktop.

http://www.xbitlabs.com/news/cpu/dis...815122559.html

Another version of the article, here. I don't see how there
is room for an integrated memory controller, with only 715
contacts. Many of those are used for power and ground for Vcore.

http://www.dailytech.com/Gearing+Up+...rticle9823.htm

I think the safest bet is, the socket will change, and what exactly
it is, won't matter until it is released :-)

I think there might be a later AMD roadmap around than this one.
This one is from December 2007. AMD is working on 45nm technology,
but one article I read several months ago, mentioned their 45nm
approach is not a complete redesign, and more of just a shrink.
That should help their profitability with what they can sell,
by making the silicon die smaller. But in terms of overclocking,
unless a miracle were to occur, I wouldn't expect the top
clock to change so radically, as to overtake Intel. It may
also help, when using companies like Chartered, to aid in
volume production.

http://my.ocworkbench.com/bbs/attach...1&d=1197568327

Depending on what you are doing with your computer, I think the
existing, mature technology is pretty attractive. Cranking the
FSB doesn't do anything, unless the thing behind the FSB is
starving.

AMD cannot be as forthcoming, about their future plans. On the
one hand, they can display the developments that are imminent,
and likely to succeed (that is called "execution" and is important
to stock valuation). But for AMD to claw their way into real
competition, takes a tech transition, and I'm sure there are
some engineers losing sleep over that as we speak. If they cannot
rely on process technology to make them competitive, then it'll
have to be by architecture alone.

Just a guess,
Paul

Thanks for that info Paul. Very interesting.
As I understood the article, the additional pins on the processor allow it
to communicated with the northbridge. With the memory controller on the chip
there is no need for that communication, so the pins can be reduced from 775
to 715.
Mmm, and they will be able to have more PCI-E slots. What do we do with
these? Are they making sound cards, NICs, or modems PCI-E? Curious.
I am a big fan of mature technology and OS's. Never liked the idea of being
the first dough-boy over the hill. Maybe with the release of the LGA715
processors and boards the 45 nm 775's will drop down into the bargain bin?

In message "TVeblen"
wrote:

Mmm, and they will be able to have more PCI-E slots. What do we do with
these? Are they making sound cards, NICs, or modems PCI-E? Curious.

Most modern motherboards implement more of the onboard components on the
PCI-E rather then on the PCI bus, my guess is that the additional PCI-E
lanes are to facilitate furthering this, in combination with allowing as
much bandwidth to additional PCI-E slots as possible.

There aren't many PCI-E cards out there beyond video cards, although a
few do exist.

"DevilsPGD" wrote in message
...
In message "TVeblen"
wrote:

Mmm, and they will be able to have more PCI-E slots. What do we do with
these? Are they making sound cards, NICs, or modems PCI-E? Curious.

Most modern motherboards implement more of the onboard components on the
PCI-E rather then on the PCI bus, my guess is that the additional PCI-E
lanes are to facilitate furthering this, in combination with allowing as
much bandwidth to additional PCI-E slots as possible.

There aren't many PCI-E cards out there beyond video cards, although a
few do exist.

Yes, but it sounds a bit like cart before the horse. I remember when we were
transitioning from ISA to PCI slots, it was done incrementally, but there
were always PCI cards available when those mainboards came out.
Question: are PCI-E slots backwards compatible? Could a PCI card be made to
work in one?
Do you think the overall transition to USB enabled devices and/or SATA
connections is pushing out the use of daughter cards overall? Sort of like
the push to kill off the floppy drive?

In message "TVeblen"
wrote:

Yes, but it sounds a bit like cart before the horse. I remember when we were
transitioning from ISA to PCI slots, it was done incrementally, but there
were always PCI cards available when those mainboards came out.

True -- However, with nearly everything being onboard these days, there
is less of a need.

Onboard sound used to be a joke, today it's all but the best option for
reasonable 7.1 analog or digital output for SOHO (or anything other then
high end professional needs)

Oh, and there are a few PCI-E sound cards out there now;

| http://www.memoryexpress.com/Product...)/Default.aspx

NICs are the same, onboard gig-E is actually performing well, faster
then PCI cards are even capable of doing, with 2-4 separate ethernet
adapters onboard on most modern hobbyist boards. There are some nice
PCI-E and PCI-X cards out there too.

There are already SATA II and RAID controllers using PCI-E, which makes
sense since they can use the bandwidth.

You can also find serial, parallel, USB, and ExpressCard adapter cards
that all support PCI-E;

http://www.memoryexpress.com/Product...chPhrase=pci-e

Modems are about the only card I know of anyone owning that can't be
found onboard or in PCI-E these days.

Question: are PCI-E slots backwards compatible? Could a PCI card be made to
work in one?

Not directly, although like with anything else, there are chips out
there designed to ease the transition. Some of the lower bandwidth
PCI-E cards (serial/parellel) were designed for PCI, and use a PCI-E
converter, similar to how nearly all (if not all) the IDE/ATA drives
today are really SATA drives under the hood, but with a IDE-SATA chip.

(Oh, and no affiliation with MemoryExpress, they're just a local shop
that I buy from regularly, and I know their stock reasonably well)

"TVeblen" wrote:

My P4 boxes are getting long in the tooth, so I'm lurking again
and starting to think of new builds. Maybe this winter.
The new Penryn class of processors look promising, once the price
gets down to the normal range, and I'm sure there will be a lot of
new mainboard specs and peripherals options once they are
mainstreamed. Is AMD coming out with a competing product?
So what else should I be keeping my eye on and researching for the
coming months?

After recently upgrading from single to dual core, I'm sure you'll
be impressed with any multiple core CPU. Apparently each additional
core is like 100% more gigaflops and that's what matters if there
are no bottlenecks. For upgrading CPU power, this looks like a great
era.



--
Currently filtering out most Google Groups posts/branches.

"DevilsPGD" wrote in message
...
In message "TVeblen"
wrote:

Yes, but it sounds a bit like cart before the horse. I remember when we
were
transitioning from ISA to PCI slots, it was done incrementally, but there
were always PCI cards available when those mainboards came out.

True -- However, with nearly everything being onboard these days, there
is less of a need.

Onboard sound used to be a joke, today it's all but the best option for
reasonable 7.1 analog or digital output for SOHO (or anything other then
high end professional needs)

Oh, and there are a few PCI-E sound cards out there now;

|
http://www.memoryexpress.com/Product...)/Default.aspx

NICs are the same, onboard gig-E is actually performing well, faster
then PCI cards are even capable of doing, with 2-4 separate ethernet
adapters onboard on most modern hobbyist boards. There are some nice
PCI-E and PCI-X cards out there too.

There are already SATA II and RAID controllers using PCI-E, which makes
sense since they can use the bandwidth.

You can also find serial, parallel, USB, and ExpressCard adapter cards
that all support PCI-E;

http://www.memoryexpress.com/Product...chPhrase=pci-e

Modems are about the only card I know of anyone owning that can't be
found onboard or in PCI-E these days.

Question: are PCI-E slots backwards compatible? Could a PCI card be made
to
work in one?

Not directly, although like with anything else, there are chips out
there designed to ease the transition. Some of the lower bandwidth
PCI-E cards (serial/parellel) were designed for PCI, and use a PCI-E
converter, similar to how nearly all (if not all) the IDE/ATA drives
today are really SATA drives under the hood, but with a IDE-SATA chip.

Thanks for the good intel. It sounds like the transition is well underway,
and since these new mainboard designs are not due to be announced until this
fall I'm sure there will be even more stuff ready to go by then. So I'm
going to treat the PCI slot like an ISA while I contemplate my specs. Might
be good to have one or two for "legacy" devices (I've got a bag of PCI
NICs).

Paul wrote:
TVeblen wrote:
My P4 boxes are getting long in the tooth, so I'm lurking again and
starting to think of new builds. Maybe this winter.
The new Penryn class of processors look promising, once the price gets
down to the normal range, and I'm sure there will be a lot of new
mainboard specs and peripherals options once they are mainstreamed. Is
AMD coming out with a competing product?
So what else should I be keeping my eye on and researching for the
coming months? What do you know?


The next thing for Intel, is Nehalem. Nehalem is a socket change
for Intel.

This article clarifies the difference between Nehalem for
servers and desktops. The desktops get a "smaller" socket,
and no integrated memory controller. Desktop Socket H (LGA 715)
versus server Socket B (LGA 1336). The word "rumor" appears
in the title (so better, more recent articles may exist elsewhere).
So this could amount to an FSB change for the Intel desktop.

http://www.xbitlabs.com/news/cpu/dis...815122559.html

Another version of the article, here. I don't see how there
is room for an integrated memory controller, with only 715
contacts. Many of those are used for power and ground for Vcore.

http://www.dailytech.com/Gearing+Up+...rticle9823.htm

I think the safest bet is, the socket will change, and what exactly
it is, won't matter until it is released :-)

I think there might be a later AMD roadmap around than this one.
This one is from December 2007. AMD is working on 45nm technology,
but one article I read several months ago, mentioned their 45nm
approach is not a complete redesign, and more of just a shrink.
That should help their profitability with what they can sell,
by making the silicon die smaller. But in terms of overclocking,
unless a miracle were to occur, I wouldn't expect the top
clock to change so radically, as to overtake Intel. It may
also help, when using companies like Chartered, to aid in
volume production.

http://my.ocworkbench.com/bbs/attach...1&d=1197568327


Depending on what you are doing with your computer, I think the
existing, mature technology is pretty attractive. Cranking the
FSB doesn't do anything, unless the thing behind the FSB is
starving.

AMD cannot be as forthcoming, about their future plans. On the
one hand, they can display the developments that are imminent,
and likely to succeed (that is called "execution" and is important
to stock valuation). But for AMD to claw their way into real
competition, takes a tech transition, and I'm sure there are
some engineers losing sleep over that as we speak. If they cannot
rely on process technology to make them competitive, then it'll
have to be by architecture alone.

Just a guess,
Paul

Some real products are shown here, for Nehalem in desktop and
server. I'd say the rumor articles weren't quite correct, but there
aren't enough details in this article to be sure. It is possible,
that other enthusiast sites, who sent reporters to Computex, will have
more details. Like the actual pin count on the desktop socket.

http://www.theinquirer.net/gb/inquir...008-great-wall

Paul