I would like some input.
does anyone think this system will be able to complete a WU?
after 36 hours its about 3% done with an 6.03 enhanced WU!

Specs:
75mhz Pentium
48MB ram
1.6GB HD
Win 98 SE

its kinda for laughs but i'm hoping it will pull through!

also anyone else with some SUPER vintage hardware?

I would like some input.
does anyone think this system will be able to complete a WU?
after 36 hours its about 3% done with an 6.03 enhanced WU!

Specs:
75mhz Pentium
48MB ram
1.6GB HD
Win 98 SE

its kinda for laughs but i'm hoping it will pull through!

also anyone else with some SUPER vintage hardware?

If you run 24/7, and do NOT do AstroPulse, then you should be able to complete tasks on time (barely). My old P90 was doing fine until it died. The warning is that AP tasks have much tighter deadlines, however, you can turn off AP tasks on the web under your account SETI@Home settings.

---

BOINC WIKI

<snip>

also anyone else with some SUPER vintage hardware?

Astro made a grins and giggles run with his P90 a while back. I've been digging through the old post catacombs for the thread but haven't gotten to it yet.

IIRC, it did make it back in time to get credit, but I'm not sure if that was within the deadline or if he managed to beat the third wingman back.

Don't know if you could bank on that now, it was a lot safer bet back in those days. ;-)

Alinator

I would like some input.
does anyone think this system will be able to complete a WU?
after 36 hours its about 3% done with an 6.03 enhanced WU!

Specs:
75mhz Pentium
48MB ram
1.6GB HD
Win 98 SE

its kinda for laughs but i'm hoping it will pull through!

also anyone else with some SUPER vintage hardware?

A few seconds of head-math comes up with a crunch time of about 45-50 days. Better have a looong deadline!

---

Pure mathematics is, in its way, the poetry of logical ideas.

Albert Einstein

LOL...

Found it!!

And apologies to Astro (if he still even lurks here anymore)....

That was the 'Powerful P60'! :-D

The whole epic saga can be found here.

Alinator

About 8 year ago before BOINC some Dutch Team was entering kit for the Most Useless SETI Crunch, my contribution was a 486SX 25MHz with 8MB Ram.
It sat whirring and paging and doing software FFTs for 28days flat without managing a checkpoint before I switched it off.

About 8 year ago before BOINC some Dutch Team was entering kit for the Most Useless SETI Crunch, my contribution was a 486SX 25MHz with 8MB Ram.
It sat whirring and paging and doing software FFTs for 28days flat without managing a checkpoint before I switched it off.

Got you beat. I ran Seti Classic on a 386 SX-2 16mhz with 5mb RAM. Estimated to take 3 years per wu. The SX series did not have a math-co processor, and naturally at the time they were expensive. When I was toying with Seti Classic, however, you could get the 80387 co-processor on Ebay for 10 bucks including shipping. Once I plugged it in, the crunch time dropped from 3 years to only about 11 months. :-P

---

You will be assimilated...bunghole!

Don't confuse the SX series of the 386 with the SX series of the 486.

The 386 chip was originally designed to be a 32bit chip from the ground up, but when a manufacturing bug was encountered in the 32bit function of some of the first 386 chips, they were rebadged as "SX" series and marked "16bit software only". Later, Intel simply released the SX series as a lower cost part. The DX series then indicated a fully functioning 32bit processor.

It was the 486 series that had an integrated math coprocessor. Then later, Intel released a 486 without the math coprocessor as a lower cost part, and dubbed it the 486SX. Many 486SX chips simply had the math coprocessor disabled internally, and the 487 was simply a 486DX chip that disabled computation on the main 486SX chip entirely!

386SX = 16bit only, math copro optional
386DX = 16/32bit, math copro optional
486SX = 16/32bit, no math copro
486DX = 16/32bit, math copro integrated

---

The oldest machine that I ever ran S@H on was a 386-20 with a 387 co-processor. It actually completed work successfully (because of the math co-processor mostly).

---

BOINC WIKI

That's pretty old alright, There was also the 80186 and the 80286, I wonder if anyone had tried to run S@H on those cpus or not?

I did the 386 just for giggles, but I remember the big bottleneck was the RAM. 5mb was barely enough to run Windows 95, which was the minimum for the Windows version of Seti Classic. As a 286 or 8086 / 8088 would not even have 5mb, they would be unable to do it.

On the other hand, I never tried any of the Linux clients back then...so it's possible a Linux client could work in a sub-386 box.

---

You will be assimilated...bunghole!

That's pretty old alright, There was also the 80186 and the 80286, I wonder if anyone had tried to run S@H on those cpus or not?

I did the 386 just for giggles, but I remember the big bottleneck was the RAM. 5mb was barely enough to run Windows 95, which was the minimum for the Windows version of Seti Classic. As a 286 or 8086 / 8088 would not even have 5mb, they would be unable to do it.

Then you would have had to have the 386DX chip and not the 386SX chip. As I mentioned earlier, the 386SX was 16bit only, and Windows 95 was a mix 16/32bit OS requiring a 32bit processor, which is why the minimum requirement for Windows 95 was a 386DX CPU (no speed grade was specified).

Incidentally, my 286 has 4MB of RAM, but it lacks the 32bit capabilities to run any SETI client ever released.

On the other hand, I never tried any of the Linux clients back then...so it's possible a Linux client could work in a sub-386 box.

Any sub-386 CPU would be 16bit only, and therefore would still be incapable of running SETI unless someone coded a 16bit app specifically for whatever OS.

If I'm not mistaken, Linux was designed to run on 32bit chips only, like Windows 95, and thus also required at least a 386DX chip or newer to run.

---

Then you would have had to have the 386DX chip and not the 386SX chip. As I mentioned earlier, the 386SX was 16bit only, and Windows 95 was a mix 16/32bit OS requiring a 32bit processor, which is why the minimum requirement for Windows 95 was a 386DX CPU (no speed grade was specified).

Nope, definitely 386SX. 386SX-2 16mhz with optional co-processor (which I added at a later date).

---

You will be assimilated...bunghole!

Don't confuse the SX series of the 386 with the SX series of the 486.

The 386 chip was originally designed to be a 32bit chip from the ground up, but when a manufacturing bug was encountered in the 32bit function of some of the first 386 chips, they were rebadged as "SX" series and marked "16bit software only". Later, Intel simply released the SX series as a lower cost part. The DX series then indicated a fully functioning 32bit processor.

386SX = 16bit only, math copro optional

Some errors to correct here.

1. The 32-bit multiply problem of the 80386 was a design problem, not a manufacturing bug. It was a marginality, not a hard logic error, so a portion (about half) of the parameter distribution actually worked. Without an adequate test, parts unable to to handle 32-bit multiply properly in all cases were inadvertently shipped until a few months into 1987. Material shipped up to that date was recalled and retested, with the failures eventually offered for 16-bit software only use to selected applications. This was not, however, in any sense an 80386SX. I was involved: in fact I was the guy who found the first test which was used to start screening material within three days of my return to Intel in 1987, and was also the guy who designed the circuit amendment, carried two of three existing first mask reticles in my carry-on luggage to a ion mill in Beverly Massachusetts, and oversaw their modifications. We blind converted 100% of the fab work in process at Livermore to my modified reticles sight unseen. I did not sleep especially well for the week it took to move the first lot to wafer test.

2. Just as the 8088 was not software limited compared to the 8086, the 80386SX was not, in fact, in any sense a 16-bit software chip. It did have a narrower external data bus, which could lower performance. While the original 80386 (later called DX) had 32 address pins, no one expected one would ever actually 4 Gigabytes of RAM, and with the 80386SX the decision was taken to bring out only 24 address pins, limiting one to a maximum of 16 Megabytes of RAM (this was perceived as abundant at the time, not as an intentional product differentiation restriction). I argued for two or four more pins, and got Pat Gelsinger's support, but to no avail.

---

Don't confuse the SX series of the 386 with the SX series of the 486.

The 386 chip was originally designed to be a 32bit chip from the ground up, but when a manufacturing bug was encountered in the 32bit function of some of the first 386 chips, they were rebadged as "SX" series and marked "16bit software only". Later, Intel simply released the SX series as a lower cost part. The DX series then indicated a fully functioning 32bit processor.

386SX = 16bit only, math copro optional

Some errors to correct here.

You're right, and I knew that. For some reason, the bit of information that sticks in my mind is an old article I read quite some time ago that is erroneous.

---

Then you would have had to have the 386DX chip and not the 386SX chip. As I mentioned earlier, the 386SX was 16bit only, and Windows 95 was a mix 16/32bit OS requiring a 32bit processor, which is why the minimum requirement for Windows 95 was a 386DX CPU (no speed grade was specified).

Nope, definitely 386SX. 386SX-2 16mhz with optional co-processor (which I added at a later date).

...and there was no software detection routine that prevented you from installing it on an SX processor?

---

Then you would have had to have the 386DX chip and not the 386SX chip. As I mentioned earlier, the 386SX was 16bit only, and Windows 95 was a mix 16/32bit OS requiring a 32bit processor, which is why the minimum requirement for Windows 95 was a 386DX CPU (no speed grade was specified).

Nope, definitely 386SX. 386SX-2 16mhz with optional co-processor (which I added at a later date).

...and there was no software detection routine that prevented you from installing it on an SX processor?

Never had any issues with that...just installed Windows 95 and loaded the Seti CLI client. A few months later I popped in the 80387 co-processor and went from there.

---

You will be assimilated...bunghole!

There was also the 80186 and the 80286, I wonder if anyone had tried to run S@H on those cpus or not?

IF someone did get one crunching, it is probably still on it's first WU . . . LOL

I have moved my account from an 800MHz P3, to a 1.2Ghz Athelon CPU and have gone from about 1 WU a day and a typical thruput score of 22, to about 1 WU every 2 weeks and a typical thruput score of 8.X

Obviously, the work units have grown in size and I'm not running AP work. I was happy with the smaller WUs because if I had a bad one, the machine didn't crunch for a month for nothing.

What I was really hoping, was that if I doubled my CPU speed and all others being equal, I'd do two small WUs per day.

Now, I'm trying to figure out what I did wrong.....

Any takers on offering advice??

Bob
------------

---

So Yer the Guy Byte Magazine mentioned back then, Wow, Cool.

So how much more ram would the chip have handled with the 2 to 4 more pins?

Was that a rhetorical question? ;-)

26 bits would have been 64 Meg, 30 a Gig.

Oh and BTW, it was possible to run 95 on an 386SX. It's just that the experience was worse than trying to run 3.0 on an 8088. :-D

Alinator

I have moved my account from an 800MHz P3, to a 1.2Ghz Athelon CPU and have gone from about 1 WU a day and a typical thruput score of 22, to about 1 WU every 2 weeks and a typical thruput score of 8.X

What I was really hoping, was that if I doubled my CPU speed and all others being equal, I'd do two small WUs per day.

Now, I'm trying to figure out what I did wrong.....

First, not all workunits are created equal. Different Angle Ranges take varing amounts of time to process, which can have a severe effect on one's RAC at such lower CPU speeds.

Second, I notice you went from using Windows 2000 Professional SP3 to Windows XP without a Service Pack. Numerous flaws were discovered in Windows XP pre-SP1 that allowed a hacker to compromise the system within minutes using a bot, turning the machine into a "zombie" (useful for a mass spam-email relay machine, consuming nearly all CPU cycles in the machine). You may want to consider upgrading to SP3 (which will require you to install SP1 or SP1a first) and then run an anti-malware and anti-virus scan to see if your system has been compromised.

Hope that helps.

---