rec.autos.simulators

by **John Metco** » Thu, 30 May 2002 10:52:58

also

"........The first spin of DDR technology doubled the
performance
of standard synchronous DRAM by pumping data bits on
both
the rising and falling edge of the clock cycle. Macri
said the
revised version doubles the total bandwidth again by
increasing
the data fetch from 2 to 4 bits. As a result, the same
100-MHz
SDRAM core that became a 200-Mbit/s-per-pin DDR device
will
jump to 400 Mbits/s per pin with DDR-II. "

by **someon** » Fri, 31 May 2002 19:49:45

umm isnt 4 bits per clock cycle called QDR? just like the QDR FSB that the
P4 uses.

> also

> "........The first spin of DDR technology doubled the
> performance
> of standard synchronous DRAM by pumping data bits on
> both
> the rising and falling edge of the clock cycle. Macri
> said the
> revised version doubles the total bandwidth again by
> increasing
> the data fetch from 2 to 4 bits. As a result, the same
> 100-MHz
> SDRAM core that became a 200-Mbit/s-per-pin DDR device
> will
> jump to 400 Mbits/s per pin with DDR-II. "

> > this IS VERY INTERESTING :=

> > ".......KOREAN DRAM GIANT Samsung will announce today
> > it has created a 512Mbit DDR-II memory chip,
> conforming
> > with the JEDEC DDR-II standard set in March.
> > And IBM has developed a DDR-II memory interface chip,
> > meaning the high speed platform is almost ready to
> > roll.

> > The memory chip can transfer data at 533Mbps, claimed
> > Samsung, but that can be extended to 666Mbps for
> > special environments and for networks, the company
> > said.

> > The chip uses a 60 ball grid array package, and
> > includes off chip driver calibration, on die
> > termination and posted CAS for better bus efficiency.

> > The firm will say that products will go into volume
> in
> > the third quarter of 2003*.........."

by **Chuck C** » Fri, 31 May 2002 20:18:56

>>The memory chip can transfer data at 533Mbps, claimed
>>Samsung, but that can be extended to 666Mbps for
>>special environments and for networks, the company
>>said.

> I wonder what they mean by "Special" environments?
> Hmm...

> The night was black was no use holding back
> Cos I just had to see was someone watching me?
> In the mist dark figures move and twist
> was all this for real or some kind of hell?
> 666 megahertz RAM of The Beast
> Hell and fire was spawned to be released!

> Torches blazed and sacred chants were praised
> as they start to cry hands held to the sky
> In the night the fires burning bright
> the ritual has begun Satan's work is done
> 666 megahertz RAM of The Beast
> Sacrifice is going on tonight!

> This can't go on I must inform the law
> Can this still be real or some crazy dream?
> but I feel drawn towards the evil chanting hordes
> they seem to mesmerize me...can't avoid their eyes!
> 666 megahertz RAM of The Beast
> 666 the one for you and me!

but will it make you a Powerslave or will you be caught Somewhere in
Time?

Chuck

--
Beer is proof that God loves us and wants us to be happy.
Benjamin Franklin

by me » Fri, 31 May 2002 20:38:58

Internal clock multiplier, no magic involved.

by me » Fri, 31 May 2002 20:43:18

The issue is cost, not speed. There is memory that runs in the multi ghz
range with true random access instead of
syncronous memory access which is what everyone is using in the PC world.
The PC world says the memory runs at 400Mhz
and 533Mhz, that means it runs at a clocked out speed of that fast for ONLY
successive memory accesses which are a single
data step higher or lower. It doesn't mean that it works at true random
access speeds that fast. IF you want to take random
access speeds into consideration, todays memory when using completely random
addresses runs in the order of roughly 20-50Mhz
speed only if you were to apply a frequency counter to the access lines of
today's memory.Not fast at all, but extremely fast
compared to the very slow memory in the 70's and 80's. There is faster
though, much fast but it costs out the ass and has low
density or high density but extremely tempermental heating issues.

> > DDR2 is called QDR memory. It stands for Quad Data Rate.

> oh, aha! ok. I looked a bit at Samsung's websites and got the
> impression they weren't the same.

> > There is many
> > memory technologies coming out and you don't EVER have to fear that
> > there won't be memory fast enough as there is memory that can clock
> > out at the rate of gigabits per second when referring to bitwide
> > devices which are what memory cells are composed of.

> bah. nothing can ever be fast enough. :)

> thanks for clearing it up, "me", whoever you are. :)

> --
> Gunnar
> #31 SUCKS#015 Tupperware MC#002 DoD#0x1B DoDRT#003 DoD:CT#4,8 Kibo: 2
> to err is human -- to forgive is bovine.

by me » Fri, 31 May 2002 20:45:04

What's going to hold a light packet?
It's not light that's the issue, it's holding the light in a light form
because when you convert it to another
format, then it gets reduced back down to near the speed of light instead of
the speed of light. In reality it
isn't even near the speed of light because of capacitance. It's reduced more
into the THZ and GHZ ranges.
Paralell processing is the trick. Doesn't matter how fast you go in
sequence, it's how wide you can get. :)

> Wait until we go to *photonics* using photons(light) for carrying signals
> instead of *electronics* where we use electrons. Warp 10 years ahead
please!

> > this IS VERY INTERESTING :=

> > ".......KOREAN DRAM GIANT Samsung will announce today
> > it has created a 512Mbit DDR-II memory chip, conforming
> > with the JEDEC DDR-II standard set in March.
> > And IBM has developed a DDR-II memory interface chip,
> > meaning the high speed platform is almost ready to
> > roll.

> > The memory chip can transfer data at 533Mbps, claimed
> > Samsung, but that can be extended to 666Mbps for
> > special environments and for networks, the company
> > said.

> > The chip uses a 60 ball grid array package, and
> > includes off chip driver calibration, on die
> > termination and posted CAS for better bus efficiency.

> > The firm will say that products will go into volume in
> > the third quarter of 2003*.........."

by me » Fri, 31 May 2002 20:45:45

It is. It is QDR technology also. It's a memory technology with a built in
clock multiplier. No magic here.

by me » Fri, 31 May 2002 20:49:35

Hehe, how creative. :)
Also, 533Mhz is to 400Mhz what 666Mhz is to 533Mhz.
In other words, it helps, but is not an earth shattering difference.
On the other hand, 666Mhz compared to 400Mhz is a good leap, just not
a leap from current 533Mhz. I'm running at 533Mhz. All it takes is a new
memory which has the exact same RDRAM rating except for one timing signal
which is 5ns tighter and it isn't even a memory cell speed issue, it's more
of a MUX timing difference that allows us to jump in speed. I consider this
not fair. If RDRAM cells could do the faster timing but it was restricted by
another timing, then I consider RDRAM as ripping us off as they should make
every cell accessable as fast as possible instead of selling us the same
thing
except with another timing issue just 5ns faster. That really sucks man
having
to change memory for such a small memory issue not related to memory cell
speed. :/

by **Tim** » Sat, 01 Jun 2002 08:44:09

>> This can't go on I must inform the law
>> Can this still be real or some crazy dream?
>> but I feel drawn towards the evil chanting hordes
>> they seem to mesmerize me...can't avoid their eyes!
>> 666 megahertz RAM of The Beast
>> 666 the one for you and me!

>but will it make you a Powerslave or will you be caught Somewhere in
>Time?

Chuck, I didn't think *anyone* was going to get that reference. :-)

Tim

by **Haqsa** » Sat, 01 Jun 2002 08:56:46

As a practical matter though, aren't most memory accesses done in large
chunks? Is the true random access speed really significant?

by **Lenn** » Sat, 01 Jun 2002 18:29:10

That's not certain at all.

Video card memory access is traditionally done in a rather predictable
manner. Read a certain (part of a) texture into the cache, spit out a bundle
of pixels, etc. However, pixel shader programs with dependable texture reads
etc can scatter memory accesses considerably. CPU memory access patterns
depend ENTIRELY on the program being executed. Some programs are cache
friendly and hardly touch memory at all, some jump all over the place.

Yes. :-) The faster you get to your data the faster you can begin to process
it.

by **First of On** » Sun, 02 Jun 2002 15:01:23

533 Mbps? As in 0.533 GB/s?

--
First of One, Primary Adjunctive Grid Alpha-01
Remove "NOT" from Email Address to Reply
Mech Arena: http://daxin.netfirms.com/
FormulaSAE Racing: http://fsae.utoronto.ca/

by **Chuck C** » Sun, 02 Jun 2002 21:20:50

Oh, there's still a few of us dinosaurs out here!!! I'm still pissed
the US didn't get a tour with "Brave New World" when Bruce came back!

Chuck
--
Beer is proof that God loves us and wants us to be happy.
Benjamin Franklin

by **Nick** » Tue, 04 Jun 2002 08:03:03

You don't mean that *gradually* PC's are entering the future of SIMD with
ultra fast buses and minimal caches? You know, like where AGP is a minimal
bus speed increase to a huge 64Mb (for example) chunk of video memory,
rather than using the same few display instructions (such as translate) on a
huge stream of similar data (such as vertices) along a bus - which seems to
me to be a massively better way of doing things from a theoretical point of
view. Of course synchronising a stream is a lot harder than 'brute-forcing'
by downloading all the image data to video ram and altering it there, but
why go for a neat, potentially-challenging-to-the-programmer solution, when
the brute force is available? It's the PC/Microsoft/Intel way! Why develop
this 533Mbps transfer potential when nothing is going to use it to its
maximum? Might as well just stick a few more huge caches around the place
and be done with it.

by **Alun** » Mon, 10 Jun 2002 10:22:58

>>> This can't go on I must inform the law
>>> Can this still be real or some crazy dream?
>>> but I feel drawn towards the evil chanting hordes
>>> they seem to mesmerize me...can't avoid their eyes!
>>> 666 megahertz RAM of The Beast
>>> 666 the one for you and me!

>>but will it make you a Powerslave or will you be caught Somewhere in
>>Time?

>Chuck, I didn't think *anyone* was going to get that reference. :-)

>Tim

Apart from me (sad old NWOBHM fan)...
--
Alun
I read a report that said the typical symptoms of stress were eating too much,
drinking too much, impulse buying, and driving too fast. Who are they kidding?
That's my idea of a perfect day.

rec.autos.simulators

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps

DDR-2 ON THE WAY => 533Mbps