Let me start off by first saying (To avoid confusion) that the below picture is not of the future or any upcoming PPC Adapter card. It is in actual fact a Slot 1 PPGA Socket CPU adapter for PC's that require one. So why put this here? This adapter offers a graphical interpretation of what to expect for the BoXer in months to come. I'm not one to theorize on things or speculate from what we know, which is nothing of the PPC forth BoXer but one thing is certain - This can't be far of. So far we know that the BoXers PPC Socket is the 3rd Dimm or 1st Simm socket on the motherboard. Depending on which way you want to look at it.
Typically a Sdramm/Dimm Memory socket is 168pins wide.
This equates to 13.3cm in length x 2.9cm height.
The Slot one Card below (from my PC) is 12.6cm in length x 7.2cm in
height.
So with a little extrapolation we can basically assume the Slot 1 card
below would make an ideal pictorial example of what to expect - complete
with CPU processor socket. Simply tack on a G4 and CPU fan or heatsink.
While little is known about the future of PPC for the BoXer we can all agree it is of great importance. We do know that the current bus speed on the BoXer prototype of the PPC socket is 66mhz and may eventually be 100mhz on the final release board. Im personally hoping so as why have a fast CPU being bottlenecked by a slow bus. I'll be adding more on Bus Speeds and why speed is important at a later date.
Allot of people argue that a PPC Socket (Ziff Socket as its called)
should be directly available on the board. Well time permitting, space
permitting and even more so development time permitting this could be possible.
But look at it this way, why not have a common (100mhz Bus) CPU Socket.
This would then allow cards of varying features. By maintaining this standard
we could in the future see Multiple G3 or even Single and Multiple G4 cards.
Yes the current Amiga OS should support this but by opting for this method
the BoXer will be capable of featuring other possible CPU cards featuring
Mips, Strongarm, Intel and a host of others. Software will not be an issue
in the future as the new TAO/Amiga OS will be able to directly support
not only a variety of CPU's but also a combination of multiples.
So how realist is this card or its concepts?
Personally I think it would be safe to assume this is a very realistic
assumption. Of course I could expect see to see things like Jumpers on
the board to accommodate various Bus settings ranging from 66-100 and even
higher. This would theoretically allow for overclocking but I'm sure there
would be more to it. Im afraid we will have to wait on this for more information
from AntiGravity.
The only other possibility for variance is to cut down on time and
costs is produce a card without a Ziff Socket and have the CPU directly
mounted on the card. Thus lowering development time (if any), overall costing
and of more importance - weight. But at the cost of flexibility. For myself
id like the option to be able to use a Socketable card that would allow
me to say for example upgrade my existing G3 300 to a 400 or higher, if
I had access to one that is. Just set a few jumpers and you would have
a CPU card with extreme flexibility.
Costing and when?
When is always the biggest question isn't it. I guess we still have
to wait for the BoXer itself but I know (basing on the emailes I've received
over the years) that many like myself would not purchase a BoXer board
or system unless there was a PPC accelerator available.
After when, how much would be the next biggest question.
When considering costs you have to factor in development time and component
costs and a reasonable profit margins for those involved. Both for the
designer and for the reseller. Take into account the CPU's are available
from Manufacturers at varying amounts and then in some cases can only purchased
in certain quantities. The larger the batch ordered - the lower the price.
being that the BoXer will not sell in massive quantities initially it would
be unrealistic for the designers of the PPC accellerator card to purchase
large quantities of processors.
But if we look at the complexity (or the lack of) this will be the most
uncomplex (is there such a word) card that ever existed for the Amiga.
it would simply be a bread board with a CPU either socket or surface mounted
directly. Perhaps a jumper for power, a few for clock settings and some
capacitors and resistors. In fact the card should be simple in itself.
Hence the lower costs and overall development time. In fact the only
component cost would really be the CPU itself.
So I expect over all costs would depend upon the CPU used. Add a few
dollars for extra components and time and labour and profit margins?
I will be adding a New Poll to the Polls Page shortly to guage how much would you, the future boxer owner be willing to pay for a PPC card.
One last thing I'd like to mention is that I am concerned about the
support. if a card of this type is created how stable will it be. Typically
the memory cards used for these sockets are of only a few centimeters in
height. A Slot 1 card as shown above uses pegs or side legs with caps to
hold it in place. This is due to the height and physical weight of the
card with the CPU and heatsink and fan. The cards become very lop sided.
Hence the stronger mounting features.
Question is how will the PPC card adapter cope with the limiting support
features of a standard Sdramm/Dimm socket. There are only 2 small locking
clips on the socket.
Last Updated: 26th April 2000