Colossus, Cray and Blue Gene: The History of Supercomputers

PC Plus ^ | June 19, 2009 | Staff

[texas booster]

I was able to use an i7 (2.66 gHz) for a month, and I ended up putting on 1 GPU and 7 console versions.

That baby could crunch some numbers! I think that I was averaging 7,000 ppd just from that one system.

I need to start another round of upgrades for my systems. With no i7 price drop coming this summer I might as well get started, with free upgrades to Windows 7.

[texas booster]

I just bought my daughter a Macbook. Haven't delved into that world yet.

Generally it seems that Macs are well supported in the F@H world.

[texas booster]

Thank you!

[Oatka]

And I think there was a similar article that said something like the old 286-12 computers had more power than the computers on the Apollo missions.
Oh hell, your digital watch has more processing power than those old Apollo IBM computers.

I didn't realize they were so low-powered - or that the digital watches had that much ooomph. Know of any equivalents or what system they used?

[HangThemHigh]

Action report:
The ASUS card doesn’t support my monitor at 1920x1080. Also, you apparently need a 2nd monitor or dummy plug attached to an unused display adapter to use it for folding. So the ASUS card is out unless I can locate an updated driver. I tried the ATI drivers a while ago, and I don’t recall the problem, but they didn’t work well. You mention the ATI 185 driver, do you have an ASUS card?

I installed the GPU client on the onboard video alongside the SMP client. It looks like the GPU is running about 200 ppd based on 1% finished.

I was wrong about choosing the number of cores running for SMP, the command line parameter I entered that I thought was doing this was being ignored and all 4 cores run. I changed the CPU usage to 75%, but it still runs 4 cores, but not as hard.

I am going to try installing the SMP as a service when my current WU finishes. I don’t want to lose a days work. The GPU is installed as a service, but I wanted to finsh this note before rebooting and testing it.

[texas booster]

First, I was wrong. Too many systems that I mess with. The i7 system had an ATI 4850 from ASUS (IIRC) which crunched splendidly. The 185 driver is for an nVidia card that I also use.

The need for a dummy plus is a problem, but being in the business I have a garage full of ... last year's equipment. I never even paid it any attention.

The low ppd on the ASUS is probably due to not having a CPU core that will keep it fed. Being a motherboard chip, it may not hit the higher numbers of a dedicated GPU card, but it should be better than that.

I like your idea to reinstall as a service and keep running SMP. If you are going to get a better GPU card it will cost $100+, for a decent folding card.

[HangThemHigh]

I rebooted now and GPU folding started and shut back down in about 2 min. I tried to restart the service manually, which started the client, but no folding??? I am having to run it manually.

As far as keeping the GPU fed, the processor is running about 26% without SMP running. It look like 1 core would be about right or run SMP at 75%. I can’t tell if it is running better this way, from past experience, I think not.

I have a couple of old monitors in the basement, I just can’t bring myself to cut the plug off while they still work.

I eventually want to get a new video card, but I probably need to get some better cooling first. When I put that ASUS card in, the fan started making a lot of noise trying to keep up. This PC is for my home theater, so noise is a bother.

Thanks for the help.

[SunkenCiv]

:’) 8” floppy?

[texas booster]

For the aggravation levels verses the points gained I agree that either GPU isn't worth it.

SMP running as fast as you want will crank out lots of points, and we know that it works now as a service.

Erase the GPU, check in services to ensure that it is disabled, and glom all of those SMP points and productions.

We will tell over the next couple of days whether we have increased your contributions. Certainly will be higher if power issues compound over a hot summer.

Just set it and forget it!

[texas booster]

The ones with the write protect notch in the front?

Yep. Plus somewhere in the Smithsonian Texas Wing, a couple of 220 volt 8” drives. Another bright idea to buy those and replace the transformer with a 110 unit, and grab some fast storage.

[SunkenCiv]

One must admit though, floppy drives were sooooo much faster and convenient than cassette. :’)

[Robert A Cook PE]

And I think there was a similar article that said something like the old 286-12 computers had more power than the computers on the Apollo missions.
Oh hell, your digital watch has more processing power than those old Apollo IBM computers.

I didn't realize they were so low-powered - or that the digital watches had that much ooomph. Know of any equivalents or what system they used?

Look at the Apollo capsule interiors at the Smithsonian, Cape Canaveral, or Huntsville AL: That “computer” dead center of the three astronauts - in a capsule where 1/10 of an oz was a highly prized weight gain! - there is a four function calculator. 16 buttons. Multiple display “folded” tube lights for each number, in a box about 10x10x10. THAT was the “computer” sent into space by the best of what we had then.

[texas booster]

I had a friend at TI that actually used paper tape on a 99/4 home computer. An assembly compiler IIRC.

Why? Probably to prove that he was the geekiest of us all.

[SunkenCiv]

I remember a system that was touted by Nibble (an Apple II programming mag published by Mike Harvey; there’s a nice website with archival material now). It consisted of a scanner and some software, and the programs (BASIC, binary, whatever) were printed out in strips. The strip reader device would scan in the strip and you’d have a flawless program in memory without typing (or typos). Also, the user could output programs as strips, to a printer.

[Oatka]

Look at the Apollo capsule interiors at the Smithsonian, Cape Canaveral, or Huntsville AL: That “computer” dead center of the three astronauts - in a capsule where 1/10 of an oz was a highly prized weight gain! - there is a four function calculator. 16 buttons. Multiple display “folded” tube lights for each number, in a box about 10x10x10. THAT was the “computer” sent into space by the best of what we had then.

OK, thanks. I'm a little rusty here. Was the flight controlled from earth or did the astronauts "fly" this thing. It seems to me there was some pressure to have the crew pilot the capsule, unlike the Russians who basically shot up a boiler with a guy in it and did all the controlling from earth. Guess I gotta do some research as I thought they had an early computer on board.

[HangThemHigh]

Hope you are right about SMP as service, GPU is running 150 ppd (all by itself) and won’t run as service. A single folder would be better. Your post is pretty much where I’m at, which is to say, back where I started.

Keep on keepin’ on.
Thanks again.

[Robert A Cook PE]

The gross calc were done on earth, with the actual firing - the button pushing - beiong done locally. (For example, the retro rockets were fired by the astronauts returning to earth from earth orbit, the lunar orbit firing was done behind the moon so it could not be controlled from earth.)

A little bit dramatic, but there were a few astronaut controlled firings: look at the Apollo 13 hand controled burn to orient it back towards earth. There, they aimed by controlling the attitude of the capsule by aiming the window of the LEM at earth. The timing of the burn (how long the rockets were fired) was predicted on earth (fire for 33 seconds, for example, at 1432 hours).

[ShadowAce]

[martin_fierro]

And so we've seem supercomputers come and go

Proofreaders, too.

[texas booster]

Good catch!

I read the article a couple of times before deciding to post it.

Blame it on bad eyes or over-reliance on technology.