Received: by maskin (mbox dlanor)
 (with Cubic Circle's cucipop (v1.31 1998/05/13) Tue Apr  6 16:28:01 1999)
X-From_: vxp@imaginet.fr Tue Apr  6 11:56:35 1999
Received: (from uucp@localhost)
	by maskin.ettnet.se (8.9.1a/8.8.8) id LAA22756
	for <dlanor@oden.se>; Tue, 6 Apr 1999 11:56:34 +0200 (MET DST)
Received: from UNKNOWN(194.51.83.171), claiming to be "relay-smtp.imaginet.net"
 via SMTP by maskin, id smtpdAAAa005ZX; Tue Apr  6 11:56:26 1999
Received: from imaginet.fr (sophocle.imaginet.fr [195.68.0.10])
	by relay-smtp.imaginet.net (8.8.8/8.8.8) with ESMTP id LAA08248;
	Tue, 6 Apr 1999 11:49:20 +0200 (MET DST)
Received: from vxp.imaginet.fr (cyber27.creil.imaginet.fr [195.68.8.27])
	by imaginet.fr (8.8.8/8.8.8) with SMTP id LAA20680;
	Tue, 6 Apr 1999 11:45:43 +0200 (MET DST)
Message-ID: <001b01be8013$12e3b740$1b0844c3@vxp.imaginet.fr>
From: "centek" <vxp@imaginet.fr>
To: <Undisclosed.Recipients@imaginet.fr>
Subject: WEB Site update about the PHENIX 060
Date: Tue, 6 Apr 1999 12:00:39 +0200
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 4.72.2106.4
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.2106.4

PHENIX 060 - Version R1

We enjoy to announce you an updating of our site, particularly important
in the domain of the PHENIX, our computer.

Here are the last changes :

1 - the 040 version  of the PHENIX is abandoned !
The PHOENIX will be therefore a computer basicly provided with the
mythical 68060 at 80MHz, giving 120 MIPS
( 040 provides 32 MIPS at 40 MHz!).
Don't be affraid, prices of this 060 have quite changed and we are
therefore in measure                         to propose a 3 or 4 times
more powerful machine for only 2000 FF more in the full configuration.
This tariff could decrease according to providing of 060 because this
tariff takes account of the worst offer of our suppliers.

2 - LINUX 060 already exists!
Whereas with one or two 040, LINUX would have been just usable, the 060
offers the necessary strength for a comfortable utilization...
Let's not speak of two 060 (240 MIPS).

3 - the DSP56301 works at 100 MHz and its SRAM descends to 12ns. A light
increase of cost for 20 MIPS more(by DSP!).
The addition of a daughter card fitted  with 3 DSP at 100 MHz is more
that ever of actuality with recent and new objectives of some of our
customers and partners.
This daughter card will be connected in the PCI slot #3 but it will be
possible to add an identical second of it in the PCI  slot #2 (the #1 is
occupied by the graphic card).
Total, 7 DSP at 100 MIPS each (700 MIPS) and 3 DSP connectors offering
each 4 high bi-directional debit serial channels (either 12)  of
33Mbits/s for all numeric application ways...

4 - The utilization of SDRAM instead of EDO is essentially reported to
an ulterior version of the PHENIX 060 for cost reasons and time of
development. The SDRAM supports are 4 times cheaper than the EDO
supports (so 60 FFS more) and besides  it's necessary to harmonize SDRAM
signals in 3.3 V and signals of the 060 bus in 5V with  very speed
buffers (2ns) and costly (60 FF of supplement).
Total, the added 120 FF on the motherboard compensate the gap of tariff
between a 32 Mo SIMM  and a 32 Mo SDRAM.
For those that know the 060, it is true that he is compatible with
signals in 3.3V, but our PCI bridge is not: what obliges us to choose 5
V for all components of the 060 local bus.

Until 40 MHz as regards to performances, the SDRAM doesn't offer ANY
advantage on the EDO 60ns: accesses are made on all sides in 1 cycle.
Between 40 and 80 MHz, the SDRAM makes difference because she always
permits BURST accesses in 1 cycle whereas the EDO obliges accesses in 2
cycles (that why the SDRAM imposed itself on computers where processsors
reach 66, 80 MHz, and more...

On the PHENIX 060, the only way to take profit from the SDRAM is to pass
the 060 in ' FULL SPEED' mode on its bus.
It means that instead of accessing the bus to an half frequency of its
execution units (in our case 40 and 80 MHz), the 060 would access the
bus to the same frequency than its units, so 80 MHz.
The gain is immediate: with EDO 60 nses, we can access at 80 MHz in
BURST in 2 cycles, whereas the SDRAM offers 1 cycle.
It means that the 060 bandwith in RAM is 2 times more important !
Unfortunately, our PCI bridge can't be  matched with the 060 beyond 45
MHz (verified timing for CENTEK by manufactory), what obliges us 060
accesses at half speed (HALF-SPEED mode) in bridge (toward the PCI bus).
We have a solution to this problem, but it will be for a little bit
later, even though the challenge to wire a 060 in full speed on the
SDRAM is exciting enough because that has never been achieved!
Imagine that the 060 in FULL-SPEED mode at 80 MHz is able" to swallow "
320 Mo/s with BURST of 1 cycle 80 MHz!
On Phenix R1, we should be sufficient with BURST 1 cycle 40 MHz, so 160
Mo/s... It is already excellent.


The PHENIX should be produced from JUNE 99.


For all question, we recommand you to visit our site before addressing
you to our services, because we are already overflowed by numerous
questions...

Thank you for your attention.

Centek
Tour Blanche de Castille
4, avenue Saint Exupery
F-60180 Nogent-sur-Oise
France
Tel: +(33)-344-746-330
Fax: +(33)-344-746-340
http://www.centek.fr
vxp@imaginet.fr






.
