slowest machine for DC projects

[Turbonium]

What's the slowest machine I should put to work and still expect to get WUs done "in time"? Cost isn't a problem here (so ignore cost effectiveness; power usage and system setup are non-issues).

Wikipedia states the following:

...the minimum system requirement for Folding@home is a Pentium 3 450 MHz CPU with Streaming SIMD Extensions (SSE). However, work units for high-performance clients have a much shorter deadline than those for the uniprocessor client, as a major part of the scientific benefit is dependent on rapidly completing simulations.

Is this really true? Is anything slower than a 450 MHz PIII worthless? Again, ignore cost of running the machine, and just consider the processing power (or lack thereof).

 

[GLeeM]

Is this really true? Is anything slower than a 450 MHz PIII worthless?

Yes it is true. There are only a few WUs that rig could finish in time ... like three months for 300 points ... what it that 3 points per day?

I wouldn't use anything older than Pentium 4 w/HT even if electricity were free.

Some else will have to help with GPU info.

Edit: I am still crunching with a GTX275, at least a little longer.

 

[Pokey]

Part of this issue is that Stanford is constantly pushing the envelope to do more complex projects and do them faster.

Some platforms have been left behind, like PS3s and older processors.

Personally it is not very satisfying folding on systems that meet "minimum requirements".

Also, while you may can do FAH on the P3 now; how long will it be before it is no longer supported?

Just my 2 cents.

 

[Icecold]

For me, folding doesn't seem worth it on anything slower than a Core2Quad. A lower end Core2Quad or Phenom II quad core from what I've seen only gets 3-4k PPD, while a ~$75 GTX 460 gets ~$15k PPD. Mid-range Core2duo's only seem to get 1-1.5k PPD. Take these numbers with a grain of salt though as I have been focusing more on BOINC projects lately with the exception of a couple machines.

BOINC projects are a lot more forgiving with slower processors. I still wouldn't be able to justify running anything slower than a Core2Duo, but an Athlon II quad core(which you could probably buy used for $50 or less), low end Core2Quad, etc. still turn in tasks reasonably fast depending on the project.

 

[VirtualLarry]

DC runs great on any Core2Duo on up, or Athlon II X2 on up, although I much prefer a quad- or hex-core CPU for decent DC performance.

I probably wouldn't run F@H on anything less than a quad, unless you were doing GPU WUs. (I had a quad-GPU machine with an AM2 dual-core CPU in it, at one point I think it was running F@H.)

TBH, If Stanford no longer supports the ATI X19xx series video cards for F@H, and they were orders of magnitude faster than any single-core CPU, then I'm surprised that they even still allow using a P3 or P4 CPU. I guess those slower WUs don't matter too much for their research, or something.

 

[sangyup81]

If you don't get unlimited electricity, I wouldn't run on anything less than a dual core CPU

 

[Turbonium]

DTBH, If Stanford no longer supports the ATI X19xx series video cards for F@H, and they were orders of magnitude faster than any single-core CPU, then I'm surprised that they even still allow using a P3 or P4 CPU. I guess those slower WUs don't matter too much for their research, or something.

I'm guessing it has to do with the sheer number of PIII and P4 CPUs out there that are contributing. For every single ATI card or whatever that's out there, there's probably dozens (if not hundreds) of a given PIII or P4 chip.

Actually, not necessarily contributing, but potentially available now and later.

 

[Assimilator1]

Hmm, maybe, but I bet that a lot of PIII rigs have long since been scrapped, at least in the developed world.

I still have a PIII Tualatin laptop though , but I don't crunch with it.

 

[VirtualLarry]

I would think that running a PIII or P4 rig to do DC on would be a very "un-green" thing to do.

Edit: Because you can buy a decent Core2Duo refurb box for under $200 (some under $150, geeks.com had some E8400 3.0Ghz dual-core boxes for $119.99), and they are so much more power-efficient.

In the interest of conserving energy, I would think that many DC projects would discourage use of those older, power-hungry, slow, systems.

 

[Turbonium]

I would think that running a PIII or P4 rig to do DC on would be a very "un-green" thing to do.

Edit: Because you can buy a decent Core2Duo refurb box for under $200 (some under $150, geeks.com had some E8400 3.0Ghz dual-core boxes for $119.99), and they are so much more power-efficient.

In the interest of conserving energy, I would think that many DC projects would discourage use of those older, power-hungry, slow, systems.

~2.6 GHz Core 2 Quad at 100% load
vs.
3.4 GHz P4 Northwood at 100% load

What is the power usage difference? Literally just power, not cost of power, or PPW.

EDIT: According to this, the power draw of the two CPUs are virtually identical (within 5W of each other). Is that accurate?

 

[VirtualLarry]

I don't know the exact power usage, but the Core2Quad can put out nearly 8x the speed of results, compared to the P4, for similar power usage. Why wouldn't DC projects want to promote more efficient use of power?

 

[Turbonium]

I don't know the exact power usage, but the Core2Quad can put out nearly 8x the speed of results, compared to the P4, for similar power usage. Why wouldn't DC projects want to promote more efficient use of power?

I'm asking this more out of curiosity than anything. I'm a bit of a hardware enthusiast, old and new.

 

[piasabird]

You cant always go by simple Processor speed. Just going from a P4 to a Core 2 duo is like night and day. Advances in processor designs and Silicone density can result in significant speed increases. A pentium 630 is plenty faster than A P4.

 

[Turbonium]

You cant always go by simple Processor speed. Just going from a P4 to a Core 2 duo is like night and day. Advances in processor designs and Silicone density can result in significant speed increases. A pentium 630 is plenty faster than A P4.

Trust me, I'm well aware of this. I was just curious as to the power draw difference, though I probably should have started a new thread in CPUs.

 

[Centauri]

^I think you rubbed a lot of people the wrong way when you asked what is the most wasteful possible system you could run 24/7.

Your wasted power affects my power bill. Your wasted power affects the vitality of my power grid. Your wasted power affects the quality of my air.

 

[EagleKeeper]

I have a AMD Athlon 1700 running XP that still churns out units as long as grandkids leave it on.

Fan is a "leaf blower" and wife does not like it running all the time. I need to try to replace the OEM HP fan with a quieter on.

 

[EagleKeeper]

^I think you rubbed a lot of people the wrong way when you asked what is the most wasteful possible system you could run 24/7.

Your wasted power affects my power bill. Your wasted power affects the vitality of my power grid. Your wasted power affects the quality of my air.

I expect that you waste power just as well but in different ways.

 

[EagleKeeper]

I would think that running a PIII or P4 rig to do DC on would be a very "un-green" thing to do.

Edit: Because you can buy a decent Core2Duo refurb box for under $200 (some under $150, geeks.com had some E8400 3.0Ghz dual-core boxes for $119.99), and they are so much more power-efficient.

In the interest of conserving energy, I would think that many DC projects would discourage use of those older, power-hungry, slow, systems.

One still had to "justify" the purchase and setup of the HW and SW.

I take my hand me down systems and turn them loose on DC.

To go out and spend $$ to upgrade a through away box can be considered a waste.

 

[Turbonium]

^I think you rubbed a lot of people the wrong way when you asked what is the most wasteful possible system you could run 24/7.

Your wasted power affects my power bill. Your wasted power affects the vitality of my power grid. Your wasted power affects the quality of my air.

lol

 

[Centauri]

I expect that you waste power just as well but in different ways.

Don't assume.

 

[EagleKeeper]

Don't assume.

I am not assuming - I am stating

People waste energy for their conviences

 

[Assimilator1]

~2.6 GHz Core 2 Quad at 100% load
vs.
3.4 GHz P4 Northwood at 100% load

What is the power usage difference? Literally just power, not cost of power, or PPW.

EDIT: According to this, the power draw of the two CPUs are virtually identical (within 5W of each other). Is that accurate?

Don't know but C2Q rig running DC (on CPU only) draws ~207w from the wall running F@H.