NCAR Supercomputer Center and bison

[ke0me]

We took our oldest grandson on a camping trip as part of his high school graduation present.

He is totally into computer programming so I was stuck for a while on where to take him.

By accident, I found out that NCAR (National Center for Atmospheric Research) has a super computer center in Cheyenne, Wyoming.

They offered a tour into the super computer area on Fridays.

We arrived and the Assistant Operations Manager took us on a tour. We were the only people on the tour. Our grandson is SCARY smart, and I assure you he was asking questions about computers where I couldn't even understand the question!

The tour guide (Jim) was super and answered all his questions. He did a super job explaining the more technical aspects so us lesser mortals could understand.

This is a huge facility, even larger than some of the biggest telecom switches I used to work on.

To get an idea of what they do, imagine dividing the surface of the earth into small cubes from sea level to say 50,000 feet. Then load each cube with a large amount of data, pressure, temperature, wind, moisture, etc.

Now take the equations that some really smart folks have developed, and crunch the numbers to predict and analyze the weather for the entire earth.


Just a few facts my brain retained:

faster computer equals more heat. The computer cabinets are both LIQUID and AIR cooled. The pipes in the lower level that carry the cooling fluid are 2 feet in diameter.

They are talking about operation speeds in Petaflops/second! I would have to ask Carl how big a Peta is. I only go up to "Tera".


The "tape library" that stores their archive data has around 10,000 tapes. Each tape holds 4 Terabytes of data, and is retrieved using a robotic system that selects the tape from storage and mounts it in a drive bay. In 20 seconds.

The facility is very energy efficient , recycling a lot of the heat from cooling the supercomputers for other uses.

OH, and the bison.

We stayed at Terry Bison Ranch and RV Park, south of Cheyenne. They run a bison herd of about 4000 head. They have a funky "train" that takes you on tour where you can feed the bison. Great fun for the kids.

The campground is full hookup, but average in other respects. No pool, no cable.

So we were able to experience the old west and the latest technology in one trip.

www.terrybisonranch.com

http://www.nwsc.ucar.edu/

[NN5I]

Peta is 1000 tera.

I've been responsible for some pretty large mainframe systems, in the teraflops range, but none in the petaflops range. "Flops", by the way, is "floating point operations per second", an obsolescent measure of CPU speed. Petaflops would be faster than anything there was, I think, in 2001 when I retired -- even at NSA.

Quite a trick, by the way, dividing a surface into cubes of any size.

I'm a little incredulous of those two-foot-diameter cooling-water tubes, too. Two inches, maybe; or perhaps those could have been cooling-air conduits. Of course my experience is 15 years old, which in the supercomputer world as in the PC world is a long time. But super-high speed is achieved mostly by making the internal parts (transistors and such) smaller, so I can't imagine that there'd be any place to use the amount of liquid coolant carried by a two-foot pipe. I'd guess that the two-foot tubes, if any, would carry chilled air to heat exchangers inside the cabinet, which in turn would cool a liquid (probably distilled water) that then cooled the circuitry.

Distilled water, which has a very high volume specific heat, is a better coolant than any other liquid -- about twice as good as, for example, liquid mercury. If you could build car engines with stainless-steel blocks. and make efficient stainless-steel radiators, pumps etc., distilled water would be a better coolant than any water-and-glycol mixture. Water is amazing stuff.

[N3LYT]

Large radio/TV transmitting tubes are water cooled (heat exchanger). Yes water by it's self cools much better than any mix. It has taken the RR industry a long time to deal with coolant as they cool the engines with just water and they just let the engines run in freezing weather. It slowly dawned on them that running a 3,000HP engine was a little costly so they now mostly use a small diesel gen set and use electric heaters to keep the big one warm and the batteries charged.

[wa8yxm]

Another interesting Visit place for Radio types is the Voice of America station outside of Dayton (Visit before/after Hamvention) This has been turned into a museum.. They too had no boiler/furnace. they used transmitter heat to keep warm.. had to install a heating system when they decomissioned... Much of the original installation is still there.. Antennas are gone however (how nice it woudl be to hook our transmitters into their arrays).

Ham Nation or Amateurlogic.TV or both have done tours of this facility with cameras rolling.

[NN5I]

Large radio/TV transmitting tubes are water cooled (heat exchanger). Yes water by it's self cools much better than any mix. It has taken the RR industry a long time to deal with coolant as they cool the engines with just water and they just let the engines run in freezing weather. It slowly dawned on them that running a 3,000HP engine was a little costly so they now mostly use a small diesel gen set and use electric heaters to keep the big one warm and the batteries charged.

Interesting. I had read about just letting them idle so they wouldn't freeze -- but I didn't know they had wised up and found better ways to keep them warm. Glad to hear it.

[Radio]

In the winters in WWII tank crews started the engines at intervals all night so they would not only not freeze, but start quickly if needed.

I have worked in much, much smaller facilities that had only cooling, no heat, because of the equipment inside. And we ran the cooling all winter! Never occurred to anyone to pump that heat through the building to save a buck or two. Of course, it was only an Atlanta winter...

[N3LYT]

The Yankee building on Mt Washington has no heat the transmitters keep the building warm but I guess that depends on the occupant. I have been in the building in sub-freezing weather and it was warm with a light jacket on but never in the dead of winter all though there was living space in there years ago. I guess I should encourage the club members to use our UHF repeater up there more often in the winter!

[N3LYT]

Interesting. I had read about just letting them idle so they wouldn't freeze -- but I didn't know they had wised up and found better ways to keep them warm. Glad to hear it.
A good deal of that came from residents near the rail yards. We had a rail head with in 100' of our buildings and once they left 5 engines running for some thing close to 48 hours and we were pulling diesel smoke into our fresh air intakes I found the number in Boston that seem to get results a couple of years before and gave it to the Sheriff (his office was in the building) they were moved about 3 hours latter. Since then I was told that the systems have become automated because old habits die hard and the engineers would not turn them off so they did that them self's after a certain idle time.

[NN5I]

In the winters in WWII tank crews started the engines at intervals all night so they would not only not freeze, but start quickly if needed....

Many WW2 tanks had air-cooled radial engines nearly identical to aircraft engines. No freezing worries, but you'd sure like to be able to start them quickly. This was true of some Sherman tanks (Wright R-675), but other Shermans had Ford V8s or any of a considerable variety of other engines, some (such as the Chrysler multibank) really bizarre.

Shermans weren't the only tanks with radial engines, either.

[ke0me]

Additional info on NCAR supercomputer facility

The first post was so long I didn't want to go into additional detail, but since you'all asked, here goes:

The 2 foot pipes do carry chilled water solution for cooling the computer cabinets.
In the lower floor beneath the cabinets, those large 2 foot pipes have valves where a 2" (appx) dia hose goes through the floor to the liquid cooling system in each cabinet.
An interesting thing was that the big pipes had only 45 degree bends, too much turbulence or restriction with a 90 degree bend.

Wish I had some pictures to attach, but alas, we were too intent on listening to the guide.

Anyway, it was a great tour and was free.

[N3LYT]

Many WW2 tanks had air-cooled radial engines nearly identical to aircraft engines. No freezing worries, but you'd sure like to be able to start them quickly. This was true of some Sherman tanks (Wright R-675), but other Shermans had Ford V8s or any of a considerable variety of other engines, some (such as the Chrysler multibank) really bizarre.

Shermans weren't the only tanks with radial engines, either.

I have a friend with a Canadian tank it has a 9 cylinder radial air craft engine and it's not some thing you want to start right up. The drill was to hand crank it over 50 times after sitting over night! (crank in took kit) because all the oil would leak into the bottom cylinder!

[NN5I]

I have a friend with a Canadian tank. It has a 9 cylinder radial aircraft engine and it's not something you want to start right up. The drill was to hand crank it over 50 times after sitting overnight! (crank in tool kit) because all the oil would leak into the bottom cylinder!

Yes, that's the drill with radial engines generally. On airplanes it's done by pulling the prop through by hand -- which on something as big as a B-17 would require from two to eight guys pulling the prop blades through three or more revolutions. Then you can begin the somewhat involved process of starting the engines.

[electricflyer]

I remember waiting to get on a DC-6 at the Dayton OH airport and looking out the window there was a Connie that had just started the engines and they were idling, then one of the bottom cylinders would load up with fuel and there would be a big explosion and a big cloud of smoke and the prop would make a couple fast turns and then back to idle. Same thing happened with a Stearman at a flying breakfast I went to years ago. Sad thing about that was that as he was getting to pull on the taxiway from the grass one of the tires dropped in a hole at the edge of the hard surface and it was enough to let the prop hit the concrete and it bent the tip of one of the blades and there was someone there from the FAA and they wouldn't let him leave until the prop was repaired.

[NN5I]

Same thing happened with a Stearman at a flying breakfast I went to years ago. Sad thing about that was that as he was getting to pull on the taxiway from the grass one of the tires dropped in a hole at the edge of the hard surface and it was enough to let the prop hit the concrete and it bent the tip of one of the blades and there was someone there from the FAA and they wouldn't let him leave until the prop was repaired.

They were right. If it didn't lose a blade in flight, which would probably kill him, the vibration might shake the whole engine loose, which would also kill him. What a bargain -- killed twice for the price of one.

And there went a bunch of bucks for that prop, too.