Super computer

39 public posts in this discussion.

Post 1

:shock:

Fujitsu VPP300 Vector Supercomputer
13 vector processors @ 143 MHz, 16 wide pipe for a total of 2.2 GFlops per processor. Total peak performance is 28.6 GFlops
Full crossbar interconnect @ 570 MB/s per link, 5 micro-second latency
14 GB of memory 5 x 2 GB + 8 x 512 MB, 512 way interleaving
4 Cabinets + 1 power supply
Limited internal disks
No software or Operating System
Documentation

[Image: http://ebay2.ipixmedia.com/abc/M28/_EBAY_b1c91f6a3a70ba7a49aeb73caf3489f0/i-1.JPG]

http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&item=3405888342

Someone buy for me :D

Post 2

Got some folding to do eh?
:lol:

Post 3

Must

Run Folding and seti@ home on super computer :shock:

Post 4

fuck carting that around to a LAN :P

Post 5

Are vector processors good or something? 143Mh X 13? Aint THAT much...

Post 6

Quote:
2.2 GFlops per processor.

looks like it. Of course you'd need an OS for it (and no, sorry, Windows only runs on the x86 architecture - which I don't think that is)

Post 7

This lot would be better for folding :D
[Image: http://members.optushome.com.au/generalmartok/shuttles.jpg]

Post 8

lol wow thats alot of shuttle ccases
there cool little thing i recon
p-air has a couple that he uses for servers and they are really nice and light and can have the same shit as u would in a midi tower
but nonly 1 driver and hdd :P

Post 9

I'm getting one soon for lanning, gonna have my 100GB WD drive in there, getting a 48x Liteon and an LS240 disk drive.

If your worried about only having one HDD and one cdrom, you just get a burner and a high capacity HDD.

Post 10

Quote from ruin:
Are vector processors good or something? 143Mh X 13? Aint THAT much...

It depends on what you're trying to do.

A lot of mathematical - ie computable - problems can be expressed as a matrix, or as vectors. Vectors and matrices are closely related mathematical objects - you can deal with them differently, but they are deeply linked at a level that lowly first-year-uni-mathematicians like me never learnt.

Some problems, expressed as a vector or a matrix, are very quick and easy to solve, because of the mathematical properties of matrices and vectors. Other problems are easier to solve in the standard algebraic form. Other kinds of problems are well suited to be expressed as partial differential equations ... and you get the idea. Different horses, different courses.

Except that there are very large areas of the physical world which just sing and dance mathematically when expressed as matrices or vectors. But, alas, your average processor is more or less serial in nature. If you look at a nice clean assembler code like MIPS, you can see that most instructions act on one, two or three pieces of information.

This makes working on matrices very slow, and vectors slow to compute as well. It turns out that this is on the order of O(n^3) complexity, that is, the computing time rises cubically with the size of the computation. The upshot is that big physical experiments run painfully slowly on standard ISAs like the x86 or MIPS, because they more or less run through things element by element, more than once.

Enter vector processors. These little beauties can do all the neat things that a serial processor can do, except that they do them to matrices and vectors. So you use them to add, subtract, multiply, divide, remainder and logically compare vectors and matricies. They do it at an effective O(n) speed, which gives them a linear, rather than cubic increase in delay as the size of the problem increases.

To compare:

Let's have four vectors, of 1 element, 2 elements, 3 elements and 4 elements.

The vector processor takes a linear time-hit for each increase in vector size. Assuming that each increase takes one processor cycle, we can see that the first vector will take 1 cycle, then 2, then 3 and then 4. No problems,

The serial processor's load, however, rises cubically, becoming: 1, 8, 27, 64 ...

You can see why, in the large areas where matricies and vectors are expressive, a vector processor would be attractive. Note that very large chunks of physics are either vector-like or matrix-like in nature, all the way from Newtownian kinematics down to Quantum Chromodynamics.

Traditionally supercomputers were pure vector machines: the Crays were for as long as Seymour Cray was alive. Fast forward to the creation of Beowulf Linux, and massive clusters of cheap serial systems became the rage, mainly because they could make quite dent in the problem for a lot less than the prevailing vector systems.

However vector systems have made a very strong comeback in the Earth Simulator, a supercomputing project in Japan. This massively parallel vector system is approximately five times faster than anything else in the world, mainly because it moved away from the American trend of massive clusters towards a pure vector model.

I note also that the PowerPC chips gained on-die vector coprocessing with the G4, at least from Motorola ("AltiVec" or "VMX"). Now that IBM has created its own implementation of AltiVec, Apple will probably move away from embedded-market-focus Motorola to high-end-breeding IBM for chip supplies. Also, in cluster super-computing, for centres that cannot afford to buy POWER4-based systems, the PowerPCs may form a very attractive middle-ground of affordability, size, heat dissipation and performance on serial and vector tasks.

Anyway, that's the story about vector processing.

Post 11

Ahh, cool. Very interesting.

Post 12

if u could get and os on it would it be useful for gaming?

Post 13

It would probably use Solaris or some other specialist OS, most likely not designed for running games, although it might be good for hosting a whole bunch of servers

Post 14

ROFL....someone bid " for it. ill take it for $2 :D

Post 15

Well its currently at $3100 AUD with the reserve not met, and I'd say for something like that there is quite alot more bidding to go before the reserve is met :)

It would probably be cheaper, easier and quieter to set up a really big cluster of P3 tualatins and cheaper still to set up a cluster of Athlons (but maybe not quieter :))

Post 16

Yeah but noone would do that because P3s are crap

Post 17

Quote from stealthninja:
Yeah but noone would do that because P3s are crap
Give me one good reason why their crap, and next time think with your mouth closed

Post 18

Goddamn the Tualatin P3s rock, but then, thats OT.

I doubt it'd run Solaris (as only x86 and SPARC builds exist). Since vector processors can also do what serial processors can do it may run a variant of Unix or Linux, but its more likely that it uses its own specialised OS.

I could be wrong, but I doubt the OS for it would be something anyone here has heard of.

Post 19

Considering what it does it'd be more probably a unix rather than linux flavour. Hmmm let's see.

Geez guys don't guess. Google it!. :P

http://www.top500.org/ORSC/1996/node31.html

Operating system UXP/VPP (a V5.4 based variant of Unix)

Post 20

har har har i havent laughed so much in my life.... is it just me or does that say GigaFlops
DAMN thats alot of flops.

please explain to me what a flop is so i can stop laughing.

Post 21

floating-point operation

n : an arithmetic operation performed on floating-point numbers; "this computer can perform a million flops per second" [syn: flop]

Post 22

What el presidente forgot to mention is that SSE and 3DNow! are also matrix instruction extensions, only for your x86 architecture rather than PPC. This is why there was so much hype surrounding those two extensions in the context of 3D games and graphics.

You can express a 3D transformation and rotation as a 4x4 matrix.

Post 23

Quote from Aravi:
Goddamn the Tualatin P3s rock,
Damn straight, but then all these people just dismiss them because they aren't smart enough to realise there was a core change :)

Post 24

Quote from cthulhu:
What el presidente forgot to mention is that SSE and 3DNow! are also matrix instruction extensions, only for your x86 architecture rather than PPC. This is why there was so much hype surrounding those two extensions in the context of 3D games and graphics.

You can express a 3D transformation and rotation as a 4x4 matrix.

Thankyou for pointing that out. Intel sold it as "SIMD", or Single Instruction, Multiple Data. In true Intel style you need to switch context to use those instructions, and there's about a kerjillion of them, and they wound up needing to retrofit the whole thing anyway ... more than once. :)

At the risk of starting a pointless flame-war, I'd say that Motorola got it right the first time: they optimised the common case, gave the VPUs lots of big vector registers and the whole thing a nice wide data-path. This is why, inter alia, matrix-heavy Photoshop filters are still caning on 1Ghz G4s with crappy 133Mhz FSBs vs humungo 2Ghz P4s with the FSB running at 266Mhz+.

But it's as though many people actually use Macs, now is it? :)

Post 25

Quote from Bond007:
13 vector processors @ 143 MHz, 16 wide pipe for a total of 2.2 GFlops per processor. Total peak performance is 28.6 GFlops

Quote from Apple.com:
Dual PowerPC G4 processors - up to 1.42GHz - have 2MB of L3 cache for the ultimate in performance and productivity, delivering up to 21 gigaflops of power.

Post 26

Quote:
Measured Performances: The first VPP300 systems will be delivered in the first quarter of 1996 (first only with the 10 ns clock). Therefore, no performance figures are available yet.

That granted it will still do quite a few things heck of a lot faster than machines now.

Post 27

What about the machines sitting down the bottom of the Pentagon?

Post 28

Quote from Odyssey:
[quote:4d8a64d1d5="Bond007"]13 vector processors @ 143 MHz, 16 wide pipe for a total of 2.2 GFlops per processor. Total peak performance is 28.6 GFlops

Quote from Apple.com:
Dual PowerPC G4 processors - up to 1.42GHz - have 2MB of L3 cache for the ultimate in performance and productivity, delivering up to 21 gigaflops of power.[/quote:4d8a64d1d5]

2200/1430 = 1.53846 Kiloflops/clock
21000/14200 = 1.47887 Kiloflops/clock

And the Mac looses the clock for clock battle

Post 29

But Apples don't have fans on their HS :D

Post 30

Quote from Odyssey:
But Apples don't have fans on their HS :D

and even though it is an apple it is probably cheaper too ;)

plus.... it comes with Mac OS X :drool:

Post 31

yeah $8000 for a state of the art G4 that looks sexy on a table, as opposed to a few million for a big arse piece of crap that needs its own building to be stored in.

Post 32

Quote from El Presidente:

At the risk of starting a pointless flame-war.....

No chance mate.
Half of us dont even know what you just said.

Post 33

but apples look homosexual..

Think about it. If you went to lan with a mac everyone would think your gay :roll:

Post 34

Quote from Flammable:

Think about it. If you went to lan with a mac everyone would think your gay :roll:

No, they would think you had a lot more money then they do.

Right now my dream PowerMac fit-out hits "7k (top-spec all components, plus 2x 23" screens), and if you get silly and add the LAN-leeching-equipment-from-hell, it jumps to $40k (a 2.52 terabyte fibre-channel array of harddrives).

Me wantie.

Post 35

When Flam says “Macs look homosexual”, I’m assuming he is talking about the G3 iMac. They are for office, schools and general Internet browsing, not games. They have a 16mb Ati Rage 128 and its not upgradeable. A gaming mac would be a G4 PoweMac with its sexy tower case. All PCI slots are 66MHz 32bit, the case has one fan and its silent and it has carry handles, now that would be nice.
G3 iMac
[Image: http://a248.e.akamai.net/7/248/2041/316/store.apple.com/Catalog/Australia/Images/productshot_imac_snow.jpg]

G4 PowerMac
[Image: http://a248.e.akamai.net/7/248/2041/316/store.apple.com/Catalog/Australia/Images/productshotpowermac.jpg]

Post 36

and those 23" HD Cinema Displays are just amazing...

Post 37

errrrrrrrrr benchmark ;)

lol i want me a G3 server not sure what i would do with it :shock:

Post 38

Mac's bloddy rock.

Post 39

I found a place where you can upgrade G3 iMac processors to G4s 8)