Whats the fastest and most powerful Computer (or processor) on Earth ?

---

> Whats the fastest and most powerful Computer (or processor) on Earth ?
>

Your brain :o)

---

> > Whats the fastest and most powerful Computer (or processor) on Earth ?
> >
>
> Your brain :o)
>
>

Nope. MY brain. :-P

---

You will be assimilated...bunghole!

> > Whats the fastest and most powerful Computer (or processor) on Earth ?
> >
>
> Your brain :o)
>
>

OK, seriously now :o)

I think the Nec SX6 is one of the fastest CPUs

Regards Hans

---

> OK, seriously now :o)
>
> I think the Nec SX6 is one of the fastest CPUs
>

Sorry, Hans - the NEC is third. We talked about that a while back in this thread:

C

---

Join Team MacNN

> Sorry, Hans - the NEC is third. We talked about that a while back in this <a> href="http://setiweb.ssl.berkeley.edu/forum_thread.php?id=7869#63088">thread:
> [/url]
>
> C
>

[Nitpick Mode]
Yep, but the SX6 has the most Flops per CPU :o)
[/Nitpick Mode]

Regards Hans

---

>
> [Nitpick Mode]
> Yep, but the SX6 has the most Flops per CPU :o)
> [/Nitpick Mode]
>
> Regards Hans
>

That's why I added that column tot he other data...thought it looked interesting. Too bad that NEC just didn't think "bigger" - only another 5100 or so processors and they could have been #1!!


C

---

Join Team MacNN

See here for the top500 list. BlueGene is king of the pack... for now at least!

---

Neil Munday



http://boinc.mundayweb.com

The great thing about IBM's baby is that it's scalable. This image says it all.

I'm not sure why it's running sub-GHz, but I know that the PowerPC 440s could go way higher on the clock (I'd say that 2.1GHz isn't too far-fetched).

---

> The great thing about IBM's baby is that it's scalable. This <a> href="http://www.llnl.gov/asci/platforms/bluegenel/images/bgl_slide2.gif">image[/url]
> says it all.
>
> I'm not sure why it's running sub-GHz, but I know that the PowerPC 440s could
> go way higher on the clock (I'd say that 2.1GHz isn't too far-fetched).
>

I guess they matched the CPU frequency with the available memory and communication bandwith.

By doing so you get the most MIPS/Watt.
[edit] A tad below 10W/CPU :o) [/edit]

Regards Hans

---

Still, it could have been clock-doubled to 1.4GHz... but you're probably right - it makes sense.

---

The SETI@Home team should make a fund raiser to buy one these awfully fast computers... sounds crazy, but it might work :)

---

> The SETI@Home team should make a fund raiser to buy one these awfully fast
> computers... sounds crazy, but it might work :)
>
No, they might decide they don't need us! LOL

---

<img src='http://www.boincsynergy.com/images/stats/comb-912.jpg'>

> Still, it could have been clock-doubled to 1.4GHz... but you're probably right
> - it makes sense.

One of the reasons that IBM systems have fewer problems than many others is that they have always been conservative on design parameters. By running below what could be achieved, they guarantee better system reliability.

Another factor is that it is easier to cool slower processors. Cray research, old time builder of the fastest verctor machines had more patents in cooling technology than anyting else.

One of their later designs used a blood substitute (I forget the name) as a coolent and the boards were fully imersed in the fluid as cold plate cooling was not getting them there ...

---

Makes you wonder whatever happened to liquid nitrogen and superconductivity, doesn't it?

Looking back at the entire PowerPC series, I've never felt that they were "conservative on design parameters". But after re-reading TOP500's BlueGene page, the word "beta" showed up in its description.

I guess I owe you a bagel.

---

> One of their later designs used a blood substitute (I forget the name) as a
> coolent and the boards were fully imersed in the fluid as cold plate cooling
> was not getting them there ...
>

Not sure about that. I read they were using something called Flourinert.

BTW They hit a brick wall when they had to run the cooling so vigorously,
that the fluid started erasing traces on their PCBs :o).

Regards Hans

---

> One of their later designs used a blood substitute (I forget the name) as a
> coolent and the boards were fully imersed in the fluid as cold plate > cooling was not getting them there ...
>
>


fluorinert & the waterfall!

Florinert, that was the name ...

If you look at almost all of the IBM hardware you find that the machines were very definate candidates for overclocking. I forget the exact numbers, but the IBM pcs could be easily overclocked and were done so often that they had to start to void warrenties if the clock was changed, there we even a ROM patch later on I think to detect clock changes ...

One of the field service stories was that if you ordered a faster printer from IBM the tech would come out, spend all day, seemingly changing tons of stuff, but in reality all he did was to change a belt from one sized pully to another ...

The part that I find interesting with the current generation of super-computers is they are all using the "slice it up" and run on lots of small nodes. The difficulty there is that there are some problems that are not easily solved that way. Though the mathemeticians are getting clever about it ...

---

> Florinert, that was the name ...
>

Paul, you were right about fluorinert being used as blood substitue.
I would never have guessed that one can survive getting a liter of fluorocarbon per IV :o).

Fluorinert can carry a lot of oxygen, you can even breathe that stuff!

Regards Hans

---

Hans,

> Paul, you were right about fluorinert being used as blood substitue.
> I would never have guessed that one can survive getting a liter of
> fluorocarbon per IV :o).
>
> Fluorinert can carry a lot of oxygen, you can even breathe that stuff!

I just was always fascinated by the supercomputers. I got a tour of the Supercomputer center at UCSD where they had a Cray. They used an IBM Mainframe just to schedule and hold work for the Cray. There were also several specialized machines that executed the Cray's instruction set that were used to test the programs before they would be run on the Cray to avoid wasting time on bad programs.

---