Computers - again :)

[teinedraig]

Hi,

I know this has been discussed a lot in this forum but as I am trying to get a unified design system up and running I thought I would post this to check my logic.

Basically, if you look at the Max PP of a ship computer you can find the model in the computer gen rules. If you assume that each ship computer is actually 4 linked (or one big one) you end up with :


</font><blockquote>code:</font><hr /><pre style="font-size:x-small; font-family: monospace;">
Ship Min Max Max PP
Model TL Jump PP Skill Model CPU TL x4 Model
1 5 1 28 11 m1 1000 5 4000 m2
2 7 2 35 12 m2 3000 5 12000 m4
3 9 3 42 12 m4 10000 5 40000 m8
4 10 4 49 13 m4 10000 5 40000 m8
5 11 5 49 13 m4 10000 5 40000 m8
6 12 6 57 13 m5 15000 5 60000 e0
7 13 6 57 13 m5 15000 5 60000 e0
8 14 6 57 13 m5 15000 5 60000 e0
9 15 6 65 14 m6 21000 5 84000 e2
</pre>[/QUOTE]Have I got this right. If so you can then design any computer you like to meet the requirements of the model and away you go.

It gets more complex if you decide that the ship PP (col4) is what is left after you have added in the Command and Logic PP demand.

For example : Low AI, Full Verbal = 30PP so a model 5 ship computer would now need 79 PP making it 4xM8 or a E6 and model 9 sc needs 95 PP making it a 4xM9 or a E8 or E9

Am I on track here?? Does this make sense??

 

[Psion]

May not be too pertinent too you, but I assume that much of a computer's weight and mass on a ship are consumed with starship specific sensory and control gear, which is fairly realistic. A far future computer of that size in the 57th century, however, is a bit less plausible.

 

[teinedraig]

Maybe it answers the "what is the 20tons of bridge for" question if you down size computers

.

 

[RainOfSteel]

65 EP? Hmmm, what was that conversion? 250 Megawatts/EP? 65 x 250 = 16.25 Gigawatts.

Well, www.top500.org lists Earth Simulator as the current world's fastest.

It has 640 Nodes (8 CPUs a piece). Rates at 35.86 Teraflops. Consumes 20 kVA/node. So, that would be 12800 kVA (guessing on that one)? How does that compare to the 16.25 Gigawatts? Power-wise, not weight/size wise?

EDIT------------------
Oh! I just saw that it has 2GB of RAM per CPU, or 10TB of RAM total . . .

 

[teinedraig]

Where is 65EP from

If that is derived from the computer design sequence then it uses the vehicle power EP which I factor 1/50 of a ship EP

Given that 65 EPv = 1.3 EP = 325

 

[RainOfSteel]

Oops! I interpreted PP as Power Points/Energy Points.



I've gamed in so many systems sometimes the acronyms bite me.

 

[veltyen]

For synchronous phase power VA is watts. For DC power they are the same thing. Phase lag changes how many watts are available. The equation is VAcos(phaseangle) = watts if that helps.

So 12800kVA would be at maximum 12.8 MegaWatt, but probably less then that unless you are using a perfect power supply.

Isn't the EP conversion 10 to 1 vehicle to ship?

Getting back to the original discussion, there is a comment about starship computers being ruggedised (taking 10 times the space and power) and there being 4 nodes, I had always assumed that 3 of the nodes where dedicated to sensors, communications and avionics, and the last one was where all the free computing (and jump calculation) was done.

 

[RainOfSteel]

veltyen,

In T20 design sequences, the mass/volume and power requirements for sensors, communications, and avionics are accounted for separately.

 

[veltyen]

G'day RainOfSteel,

I am aware that the accessory components are accounted for seperately, there is a difference between the accesory gear and the computer analysing these imputs.

The sensors (for example) are made up of a set of information collectors (telescopes, neutrino detectors, grav distortion detectors and so on) The computer system can support the sensor gear up to a rating dependent on the level of the core computer (in this case the 4 sub-computers making up the core system). The sensor gear by itself is almost completely useless without enough computing grunt behind it to analyse these inputs and create a meaningful data set for the sensor operator to look over. Ditto for the avionics and communications gear. Jumping by comparisson appears a lot simpler, because you can use a handcomp to calculate a jump without the main computer being online at all, something that just isn't possible with the sensor, communication and avionics gear.

As for using 4 identical computers for this purpose I see no problem with this as it is how I would design the systems. The spare/Master would be the system used to debug and double check the outcomes from the other systems, and it allows the opurtunity in a catastrophic event to push the resources onto another platform as a type of live DRP. It would allow you to rotate a 'dodgy' system to the least nessersary position on ship until it could be replaced. Adding FIB allows yet another full system for disaster recovery, dedicated and at an alternative location (as alternative as possible given both are on the same ship) this allows rapid transition and recovery given a catastrophic event.

Anyway, that is just my take on the situation.

For humor value the default "catastrophic event" used arround here is a direct meteorite strike on my companies main computing facility. Which is far less humorous on an active starship where such an event is more likely.

I guess I just outed myself as a computing geek here though. Oh well.