Computers!!!!
- Thread starter shadowdragon
- Start date
No, until the upgrade, they used 1975 era zilog z80's... now I understand they are using 486 class chips...
The average astronaut's PDA has as much processing power as the entire flight dynamics computer network pre-upgrade, when counting flops. And far more ram.
Ruggedization is only part. Out of date tech is another.
One soldier I was talking to was complaining about the slowness of the PII based field laptops.
No, looking at US milspec (which is typically quite paranoid) isn't going to give you reasonable numbers. But looking at high end workstations, they still are about 4-8 cubic meters, counting operator space. High capacity data strage still eats up volume.
And I doubt that the guys at GDW were not up on computer tech at the time. (It became evident that they fell behind later.)
The average astronaut's PDA has as much processing power as the entire flight dynamics computer network pre-upgrade, when counting flops. And far more ram.
Ruggedization is only part. Out of date tech is another.
One soldier I was talking to was complaining about the slowness of the PII based field laptops.
No, looking at US milspec (which is typically quite paranoid) isn't going to give you reasonable numbers. But looking at high end workstations, they still are about 4-8 cubic meters, counting operator space. High capacity data strage still eats up volume.
And I doubt that the guys at GDW were not up on computer tech at the time. (It became evident that they fell behind later.)
The space shuttles main computer system was not Zilog Z80s, but a customized version of the IBM System/360 mainframe: link http://history.nasa.gov/sts25th/pages/computer.html
Someone may have been thinking of the GRiD computer carried in the early flights. These were actually Intel 8086 machines, but were not used for the main system. They were used for monitoring the experiments, and that kind of necessary, but not critical functionality.
And, yes, all those systems were hardened and "flybridized" (NASA's term). Think about it: a standard CDRW, for example, will not work in orbit. It depends upon gravity to hold the CD in place while in use.
Another impact does come from using semi-conductor memory as opposed to magnetic core, which military computers of the 70's used. If power goes off, semi-conductor memory loses it all, but not mag core. The battery technology of today makes up for that issue now.
The only things I think are broken about the computer rules are (1) storage is way too small as compared to cpu (LBB2 and HG rules); (2) the idea that a discrete transistor or vaccum tube computer would be used in a starship; (3) the fact that TL does not give you anything but bigger computers, which is a general fault with Traveller anyway.
Just IMO
Someone may have been thinking of the GRiD computer carried in the early flights. These were actually Intel 8086 machines, but were not used for the main system. They were used for monitoring the experiments, and that kind of necessary, but not critical functionality.
And, yes, all those systems were hardened and "flybridized" (NASA's term). Think about it: a standard CDRW, for example, will not work in orbit. It depends upon gravity to hold the CD in place while in use.
Another impact does come from using semi-conductor memory as opposed to magnetic core, which military computers of the 70's used. If power goes off, semi-conductor memory loses it all, but not mag core. The battery technology of today makes up for that issue now.
The only things I think are broken about the computer rules are (1) storage is way too small as compared to cpu (LBB2 and HG rules); (2) the idea that a discrete transistor or vaccum tube computer would be used in a starship; (3) the fact that TL does not give you anything but bigger computers, which is a general fault with Traveller anyway.
Just IMO
Don't sweat the big stuff
There's really no point in worrying about computer technology when ships can travel faster than light without seriously fscking up causality. If you've studied general relativity, the monumental size of this handwave makes just about any other seem a mere twitch of the pinky by comparison.
--Devin
(Please don't break causality. I need it to do stuff.)
There's really no point in worrying about computer technology when ships can travel faster than light without seriously fscking up causality. If you've studied general relativity, the monumental size of this handwave makes just about any other seem a mere twitch of the pinky by comparison.
--Devin
(Please don't break causality. I need it to do stuff.)
I submit that the complexities of the tasks and uses that computers are used for in ships do not scale with their size.
I do not believe the complexity of whatever physics are involved to calculate a ship jump change in any dramatic way with the size of the ship. The math used to navigate the Voyager Probe is identical to that of the Space shuttle. The Space shuttle might have a few more thrusters to play with.
Specifically, I don't think it takes "1MB of memory per dTon of ship displacement" to calculate a jump, or whatever measure you want.
I do not believe that the computer size is in any dramatic way tied to the displacement of the ship. For example, I'm confident that the there are more computers on a modern naval destroyer than on a Super Tanker, despite the fact the latter vastly out displaces the former.
I am not swayed by the "space ship computer are super duper uber hardened" argument. If you want to apply that to YTU, that's fine, but be consistent and effectively eliminate all other electronics on your ships, weapons (sorry, that gauss rifle doesn't work in space -- the firing computer crashed, nor does that holo sight), combat armor, vehicles, vacc suits etc. as they are NOT hardened. Or if they are, then they manage to do it in small spaces.
There's a nice "adventure" moment. "Looks like some remote solar flare fried out the air raft again..."
Computers scale to the tasks they perform. If the task is roughly the same, the size will be roughly the same. For example, do you think the Engine Control Unit in your car is vastly different in size or capacity than the one in, say, a motorcycle? or a large Kenworth tractor? All of these will fit in your hand. If the Kenworths is larger, it's because it can be vs say the one on the motorcycle. Perhaps the Kenworth one is cheaper than the motorcycle one because the motorcycle one must be smaller as space is at a premium vs a Kenworth. But the problem they are all solving is effectively identical. Mapping fuel injector solenoids to an air sensor, RPM, etc. I'd argue that the ECU for that 14 cylinder, 100K HP diesel container ship motor (1.5M in^3 displacement) is not dramatically larger than the ECU for a diesel truck. More wire, perhaps some more sensors, more displays, but the computer driving it all is 1 to 2 square inches big with a couple RAM chips. Amazing.
What does scale with displacement? Well, wires, cables, conduit, etc. for one. These certainly consume space. I have conduit in my house, but it's contained within the walls, so conduit costs ME (effectively) zero space. Ships might have solid bulkheads, thus requiring such cables to consume extra space. Or, perhaps they can be routed along with the assortment of air ducts and other life support infrastructure.
What else might scale? Well, obviously the demands of a single turret may well be different than 100, or 1000 turrets (though I'd argue not by much). I'd also argue that the control functions would be locally driven by the mount (fire control computer tell turret computer (the one the size of a stick of gum) "point turret to 14 degrees", turret replies "turning", "done", or "stuck"). Mind, number of turrets are only indirectly related to displacement. While a scout ship can't mount 1000 turrets, and large ship could only mount a single turret.
Lets take as an example a ship that operates as a tug. Alone, the ship isn't very big. Some oversized thrusters and maybe extra fuel. But as soon as it hooks up to a large payload (a large, dumb payload mind), all of a sudden "the ship" suddenly grows to 10 times it's normal displacement. Should it suddenly need a larger computer? Has it's problem domain really dramatically changed? That block it's latched on to is dead weight. Maybe, like a tow truck, someone will drag some tail lights to the other end of the payload. But did this ship become ten times more complicated? Hardly. Merely fatter and heavier.
A harbor tug does not adopt the complexity of an aircraft carrier simply because it's pushing it out or towing it in. Harbor tugs are bone stupid. Aircraft carriers are not.
Finally: http://www.sun.com/products/sunmd/s20/index.jsp
A CRAP load of MIPs and storage in 2dTons. Today. Right now. Off the shelf.
Everything in the ship is a computer: the gun mounts, the weapons, the sensors, the doors, the toilets, the refridgerators, the power plants, door bells and landing lights. The space consumed by "the computer" is interconnecting them all, and workstations/displays to manage it, making the ship a single combined whole. I could run the back office of a 200 person company for 5 years off of an iPhone (not just the CPU, RAM and storage as well). All of the space will be taken up by the printers and terminals. The CPU is invisible.
Controls are all that matter. Those are the things that need to be counted.
I do not believe the complexity of whatever physics are involved to calculate a ship jump change in any dramatic way with the size of the ship. The math used to navigate the Voyager Probe is identical to that of the Space shuttle. The Space shuttle might have a few more thrusters to play with.
Specifically, I don't think it takes "1MB of memory per dTon of ship displacement" to calculate a jump, or whatever measure you want.
I do not believe that the computer size is in any dramatic way tied to the displacement of the ship. For example, I'm confident that the there are more computers on a modern naval destroyer than on a Super Tanker, despite the fact the latter vastly out displaces the former.
I am not swayed by the "space ship computer are super duper uber hardened" argument. If you want to apply that to YTU, that's fine, but be consistent and effectively eliminate all other electronics on your ships, weapons (sorry, that gauss rifle doesn't work in space -- the firing computer crashed, nor does that holo sight), combat armor, vehicles, vacc suits etc. as they are NOT hardened. Or if they are, then they manage to do it in small spaces.
There's a nice "adventure" moment. "Looks like some remote solar flare fried out the air raft again..."
Computers scale to the tasks they perform. If the task is roughly the same, the size will be roughly the same. For example, do you think the Engine Control Unit in your car is vastly different in size or capacity than the one in, say, a motorcycle? or a large Kenworth tractor? All of these will fit in your hand. If the Kenworths is larger, it's because it can be vs say the one on the motorcycle. Perhaps the Kenworth one is cheaper than the motorcycle one because the motorcycle one must be smaller as space is at a premium vs a Kenworth. But the problem they are all solving is effectively identical. Mapping fuel injector solenoids to an air sensor, RPM, etc. I'd argue that the ECU for that 14 cylinder, 100K HP diesel container ship motor (1.5M in^3 displacement) is not dramatically larger than the ECU for a diesel truck. More wire, perhaps some more sensors, more displays, but the computer driving it all is 1 to 2 square inches big with a couple RAM chips. Amazing.
What does scale with displacement? Well, wires, cables, conduit, etc. for one. These certainly consume space. I have conduit in my house, but it's contained within the walls, so conduit costs ME (effectively) zero space. Ships might have solid bulkheads, thus requiring such cables to consume extra space. Or, perhaps they can be routed along with the assortment of air ducts and other life support infrastructure.
What else might scale? Well, obviously the demands of a single turret may well be different than 100, or 1000 turrets (though I'd argue not by much). I'd also argue that the control functions would be locally driven by the mount (fire control computer tell turret computer (the one the size of a stick of gum) "point turret to 14 degrees", turret replies "turning", "done", or "stuck"). Mind, number of turrets are only indirectly related to displacement. While a scout ship can't mount 1000 turrets, and large ship could only mount a single turret.
Lets take as an example a ship that operates as a tug. Alone, the ship isn't very big. Some oversized thrusters and maybe extra fuel. But as soon as it hooks up to a large payload (a large, dumb payload mind), all of a sudden "the ship" suddenly grows to 10 times it's normal displacement. Should it suddenly need a larger computer? Has it's problem domain really dramatically changed? That block it's latched on to is dead weight. Maybe, like a tow truck, someone will drag some tail lights to the other end of the payload. But did this ship become ten times more complicated? Hardly. Merely fatter and heavier.
A harbor tug does not adopt the complexity of an aircraft carrier simply because it's pushing it out or towing it in. Harbor tugs are bone stupid. Aircraft carriers are not.
Finally: http://www.sun.com/products/sunmd/s20/index.jsp
A CRAP load of MIPs and storage in 2dTons. Today. Right now. Off the shelf.
Everything in the ship is a computer: the gun mounts, the weapons, the sensors, the doors, the toilets, the refridgerators, the power plants, door bells and landing lights. The space consumed by "the computer" is interconnecting them all, and workstations/displays to manage it, making the ship a single combined whole. I could run the back office of a 200 person company for 5 years off of an iPhone (not just the CPU, RAM and storage as well). All of the space will be taken up by the printers and terminals. The CPU is invisible.
Controls are all that matter. Those are the things that need to be counted.
Interesting you should use the SUN box as an example. I've used them before as exemplars of high end computing.
The SUN Modular Datacentre is closer to 4 dTon (roughly being 3x3x6 metres - especially if you start counting venting space for the air conditioning) and doesn't actually operate that well in the middle of a hot desert or in any polar regions, but is isn't a bad start for a facility on a space vessel. It can be equipped with 8 racks, and has 280 RU available for IT equipment.
An RU for anyone who doesn't know is about an inch and a half. So a 1 RU component is approximately the size of a large pizza box. Modern datacentre computers get that small, especially the PC equivalents though 2 RU is still more slightly more common in my experience. There may be another 2 to 4 RU interface platter (normally a basic keyboard/screen/mouse interface that folds into a rack), though the interface is likely to be shared by many components.
As a real world example I run a system in the real world. It wouldn't leave much space spare in that datacentre. What is it I hear you cry? A nuclear reactor control station? Major power infrastructure? Analysis for a synchrotron? Control for a Hospital?
No. A web server. One that (probably) gets a few more hits then this site, but all in all it is still a freaking web server.
If you haven't worked in high end computing infrastructure this might surprise you. I don't know if I can explain it (it might breach some terms of employment) but high levels of redundancy, monitoring and storage really start adding up in terms of space.
The thing is you can get somewhat arcane in designing these things.
You have 280 RU. 248 if you have an interface in each rack. 208 if you have duplicated datacentres of half the size. A rack disappears for a in datacentre power system (a UPS or similar) and another for a tape/data storage facility. A SAN along with associated control and logic systems take up another 16 RU or more, again duplicated in both locations. Load balancing high intelligence routers take up another 20 RU.
This leaves you with 36 RU at each datacentre remaining, and we haven't even started looking at the monitoring gear. You can probably cram that into as little as 4 RU at each location. Database control takes a couple of computers, and each of these needs a test and qa environment (as does nearly everything else) for 8 computers across both locations. You then add secondary systems - a search engine, a mail server, data server. Again each take up another 8 computers (2x2 live systems, 1x2 qa, 1x2 test).
This leaves about 12 RU at each site. Enough for 12 computers to do web serving.
The infrastructure around them would (in todays terms) be able to route a small city, and have more storage space then would conceivably be used by most non data intensive endeavors. But it is damn near bullet proof. The loss of any single component can be safely ignored (though you would be informed) with no loss of service. "A datacentre" is a single component. Loss of power for longer then X (where X is the time taken to drain the UPS) would be a problem - as would shared poor environmental conditions.
This is from an environment where person time massively outcosts hardware and space, and money is very available if it increases security or reliability.
There are different drivers on board a space craft.
far-trader
SOC-14 10K
RU is shorthand for "Rack Unit" a standard modular interface space in large computer mountings (the racks). Each "Rack" is about 2 feet square and 5 feet high iirc. Give or take, not counting space above and below for connecting and ventilation. Each "Rack Unit" is about 1.5 inches high.
An RU is like a big empty slot that can accommodate a large variety of different modules to allow quick and easy mainframe customization to varied tasks.
The interface RU modules are especially neat now that lcd screens have replaced crts. The keyboard module is typically 1RU and simply slides out (installed at the appropriate height, usually for standing) to be used. And the lcd monitor is in the next slot above, slides out and flips up to be viewed (temporarily blocking access to the RUs directly above but that's no biggie)
It's how I've long imagined Traveller computers being set up. And I've often pictured Traveller's "Programs" and "Slots" as units in RUs. For example that Jump-2 "program" that requires 2 slots in the computer is a 2RU cartridge that takes up 2RU of space in the computer mount when engaged. Or can be pulled to make room for some other "Program" module.
Last edited:
pendragonman
Absent Friends Margrave
Excellent idea. One I should have been using for a while, but you know onld age is such a problem 
Icosahedron
SOC-14 1K
Yeah, look at it like the slots on the front of your PC tower where you fit CD drives, etc.
The racks in question are generally 19'' wide (rail to rail) and can be anywhere from about 3 to 6 feet tall depending on where you're putting them. They take up about 60cm by 60 cm of floor space. Think of the components in a rack as bolt in shelves with a common mounting arrangement. Each of these has a height measured in Rack Units or RUs. If you have a count of RU for equipment plus other bits and know how many racks (of a given capacity of RU) you can put in the room, it's easy to see how enjoyable your job will be. 3 x 48 RU racks and 180 RU of equipment means you will be hating life. Same racks and 120 RU required means you're OK until you need to add more stuff.
Let's see if I can give another sizing example.
Well, it was the Christmas 2005 ish and these guys outsourced me so what the hell. I was working for a mid-sized outfit (about 80 people) and we moved, thus I had to fit out a new 'computer room.' I'll leave out my rant about architects and managers.
We ran three servers ~ 5 RU on their side, so there is a third of a rack. Network switches and hardware took another half a rack. We had a couple of extra boxes and were deploying a VOIP phone system, taking a second rack. The patch panels for UTP network cabling (about 200-300 ports) took most of a rack.
Once I convinced the powers that be that two former coffee prep areas of 2x2 meters each weren't going to do the job, I was able to get an area of 4x5 meters for the 'data center.' One wall was taken by the two AC units. [1] Another wall had the door and a third was on the support beam we used to hold the UPS. [2] Once the racks were deployed in the center with walking around space in front and back, we had pretty much filled the room. This works out to about 8 or 9 DT to support 80 people not doing a whole lot. A company doing real work would need a lot more space.
Why yes, I am a Systems Admin. Glass half full / half empty? No, I want to know what idiot under-specified the backup capacity.
[1] Wall mount split systems. Mounted high, but you don't want to put anything below a unit that can leak water.
[2] Second floor, ~800 kg of UPS w/ batteries. The electrical engineer had an excessive safety margin. I think we could run the whole room for about five hours after failure.
Let's see if I can give another sizing example.
Well, it was the Christmas 2005 ish and these guys outsourced me so what the hell. I was working for a mid-sized outfit (about 80 people) and we moved, thus I had to fit out a new 'computer room.' I'll leave out my rant about architects and managers.
We ran three servers ~ 5 RU on their side, so there is a third of a rack. Network switches and hardware took another half a rack. We had a couple of extra boxes and were deploying a VOIP phone system, taking a second rack. The patch panels for UTP network cabling (about 200-300 ports) took most of a rack.
Once I convinced the powers that be that two former coffee prep areas of 2x2 meters each weren't going to do the job, I was able to get an area of 4x5 meters for the 'data center.' One wall was taken by the two AC units. [1] Another wall had the door and a third was on the support beam we used to hold the UPS. [2] Once the racks were deployed in the center with walking around space in front and back, we had pretty much filled the room. This works out to about 8 or 9 DT to support 80 people not doing a whole lot. A company doing real work would need a lot more space.
Why yes, I am a Systems Admin. Glass half full / half empty? No, I want to know what idiot under-specified the backup capacity.
[1] Wall mount split systems. Mounted high, but you don't want to put anything below a unit that can leak water.
[2] Second floor, ~800 kg of UPS w/ batteries. The electrical engineer had an excessive safety margin. I think we could run the whole room for about five hours after failure.
Thanks for all of the information.
The basic concept of triple redundancy on spacecraft computers plus the need for possible access from all sides creates a very different concept of what a starship computer room looks like.
Looking at a 2 dTon computer room (adequate for a CT model 2 computer on a Far Trader). The room is roughly a 3 meter cube, which seems huge when compared to the PC sitting on your desk. If we imagine the Model 2 Computer to be a 1 m cube, then that seems more reasonable for such a minimal starship system. If we assume that the ship's computer is really three computers, then we can stack them into a 1m x 1m floor area unit that is 3 meters tall. It would then make sense that a model 2 computer should easily fit in 0.5 dTon space (a 1.5 x 1.5 x 3m square on the declplan).
This is where I would normally be tempted to fill the other 1.5 dTons with desks and chairs and reference books. Thanks to your Rack Image of starship computers, my 1mx1mx3m tower might actually require a 1 meter access/walking space around it. Suddenly I need a 3m x 3m x 3m (2 dTon) space for my triple redundancy model 2 computer rack.
If each of the redundant computers required its own 1x1x3 meter rack, then the room (including access) might need to be 3m wide x 3 meters tall x 7 meters long or about 4.5 dTons (perhaps a model 5 computer).
The space is also to stop failure jumping (electrical arcing and fires) and for ventilation - not just access. You also need space to wire everything together, and this could include cooling pipes and anti-fire dumping systems. Traditionally this was a snake pit (a false floor over cables) lately I've seen more overhead solutions, but either way you need some space for that.
Some real world images for perspective.
Raytheons F16 computer upgrade with a mean time between failure of 8 weeks - or in other words every 4 jumps everyone dies. This is more appropriate for a smallcraft computer system obviously.
Wikipedia : data center Note the pull out interface in the second image down that we've been talking about.
MTBF is simply hours of operation between detected significant errors.
MTBF on most drives is only a few hundred hours... but 99.9% of the time, a failure is not even significant enough to be brought to user attention, and can be worked around by the drive and/or the OS.
MTBF on most drives is only a few hundred hours... but 99.9% of the time, a failure is not even significant enough to be brought to user attention, and can be worked around by the drive and/or the OS.
Gallowglacht
SOC-12
How much room do computers take up in real life on the likes of an Aegis cruiser?
Hard to tell. I was looking for that information for the examples above - most of the material I could find would suggest multiple small cabinets distributed over the entirety of the ship rather then a single facility.
I was able to find the rundown on the communications and sensor systems on a Canadian warship. Most seemed to be small dedicated lockers (similar to the Raytheon F16 computer) with one or two per system. Considering there were many systems listed and each seemed to have its own dedicated computer system it would add up to ... a lot.
After extensive time rummaging:
Halifax class frigate radios and systems was the link.
Peter Schutze
SOC-12
>most of the material I could find would suggest multiple small cabinets distributed
warships usually rely on 'distributed computing' and not only for redundancy. note this quote halfway down the webpage:
"HALIFAX Class frigate distributed processing architecture"
The cabinets you're referring to are 'the systems' not THE 'computer' and thus integral parts of that set of hardware.
Their writeup makes it sound like the "AN/UYK-507" is the actual equivalent of the traveller "computer". Certainly agrees with what my boss discribed on the Aussie navy destroyer he served on.
>and each seemed to have its own dedicated computer system it would add up to ... a lot
those should all be counted as part of the system's space and maybe the workstation space(s). All of the electronics in a standard traveller starship does add up to a lot of computing power .... if you include big chunks of the workstation and sensors etc etc volume
warships usually rely on 'distributed computing' and not only for redundancy. note this quote halfway down the webpage:
"HALIFAX Class frigate distributed processing architecture"
The cabinets you're referring to are 'the systems' not THE 'computer' and thus integral parts of that set of hardware.
Their writeup makes it sound like the "AN/UYK-507" is the actual equivalent of the traveller "computer". Certainly agrees with what my boss discribed on the Aussie navy destroyer he served on.
>and each seemed to have its own dedicated computer system it would add up to ... a lot
those should all be counted as part of the system's space and maybe the workstation space(s). All of the electronics in a standard traveller starship does add up to a lot of computing power .... if you include big chunks of the workstation and sensors etc etc volume
far-trader
SOC-14 10K
I'm more inclined to picture all that as part of the Bridge tonnage, or in certain places, like Engineering and Weapons, as part of the space of the specific installation.
The "Computer" is what ties it all together and makes it a single entity, rather than separate parts.
Aegis Mk-7
The answer to the size of the Aegis system - systems, really - is hard to find. The original core computing element was the AN/UYK-7 computer, a militarized version of the Univac 1108. I worked on these during my navy career.
The YUK (as it was know) is modular, available in up to 4 bays. Each is roughly 1 m high x 1 m length x 1/2 m wide. This gave an i/o processor, cpu, and 16 memory bays (16kb each). The memory was magnetic core, so in the event of power loss, the programming would not lost.
See this link: http://www.harpoonhq.com/waypoint/articles/Article_044.pdf
The full Aegis suite consisted of the AN/SPY-1, the Aegis Command System, and the Mk 86 Fire Control SYstem. The SPY-1 required 16 single-bay YUKs, and 11 AN/UYK-20 (16-bit versions of the UYK-7). The Command system was a 4-bay YUK. The Mk86 was a single bay YUK (IIRC - not sure on this one). Most of this equipment was concentrated in two areas - the radar control room, and the CIC.
Now all this info is outdated. The AN/UYK-43 began replacing the 7's sometime in the mid-80's. I understand they are all semi-conductor versions of the 7. I imagone those are being replaced now by something newer and faster, but I didn't look.
So, there you go.....
The answer to the size of the Aegis system - systems, really - is hard to find. The original core computing element was the AN/UYK-7 computer, a militarized version of the Univac 1108. I worked on these during my navy career.
The YUK (as it was know) is modular, available in up to 4 bays. Each is roughly 1 m high x 1 m length x 1/2 m wide. This gave an i/o processor, cpu, and 16 memory bays (16kb each). The memory was magnetic core, so in the event of power loss, the programming would not lost.
See this link: http://www.harpoonhq.com/waypoint/articles/Article_044.pdf
The full Aegis suite consisted of the AN/SPY-1, the Aegis Command System, and the Mk 86 Fire Control SYstem. The SPY-1 required 16 single-bay YUKs, and 11 AN/UYK-20 (16-bit versions of the UYK-7). The Command system was a 4-bay YUK. The Mk86 was a single bay YUK (IIRC - not sure on this one). Most of this equipment was concentrated in two areas - the radar control room, and the CIC.
Now all this info is outdated. The AN/UYK-43 began replacing the 7's sometime in the mid-80's. I understand they are all semi-conductor versions of the 7. I imagone those are being replaced now by something newer and faster, but I didn't look.
So, there you go.....
