Folding Benchmark CD

[notfred]

I've seen some setups where even though it is disabled in the BIOS, Linux still manages to find the network card and so the PCI one is moved to a higher interface number. This could be the issue here.

[Usacomp2k3]

I've seen some setups where even though it is disabled in the BIOS, Linux still manages to find the network card and so the PCI one is moved to a higher interface number. This could be the issue here.

Yeah, quite annoying, in fact.

[Damage]

notfred, that did it. New CD rev works like a charm on the Gigabyte P35 board. Thanks.

[Ragnar Dan]

Sometimes, I wonder about that Ragnar Dan...

I seem to have memory of this [a Linux finding a disabled device] actually having happened to me, so why I'd not think of it, I can't figure. Though Ragnar does need to pick up about a dozen hours of sleep this weekend, I doubt that's the reason. Just plain stupid, I think it must be. And that 2nd person thing, he'd better get in touch with a professional if that doesn't subside soon.

[Flying Fox]

Sometimes, I wonder about that Ragnar Dan...

I seem to have memory of this [a Linux finding a disabled device] actually having happened to me, so why I'd not think of it, I can't figure. Though Ragnar does need to pick up about a dozen hours of sleep this weekend, I doubt that's the reason. Just plain stupid, I think it must be. And that 2nd person thing, he'd better get in touch with a professional if that doesn't subside soon.

I think I recall seeing Windows doing that too, with the device disabled in the BIOS but it still gives me the annoying "!".

[Ragnar Dan]

And he also for some reason changed "3rd person" to "2nd person". He worries me, that guy. *ahem*.

Yeah, I've seen Windows do that too if I remember correctly, but as long as it's only detected but not treated as a running device, I figure that's reasonable. Unless it actually lets you enable the device from the Device Manager instead of just being informational, which would be bad. But Windows has a lot of hardware detection to do, considering how many different device makers there are that the average user expects to work without any trouble, so it finding hardware doesn't seem that extraordinary.

[jeffry55]

Hi Notfred. I got error messages when I tried to use your SMP64 folding CD today.

The first error was "VMWare not found"

The second was "fah5 not found" in a particular directory that I can't remember.

Is it just a bad download on my part? Should I burn another CD or did I download the wrong image?

It did find all four cores of my Q6600, which was nice to see.

[notfred]

"VMWare not found" isn't an error message, it just means that it should run NTP to keep the clock in sync. If it's running under VMWare or other virtualisation environment like VirtualPC then it should not run NTP as the VM runs in stops and starts and that really messes with NTP. VMWare and VirtualPC both keep the clock in sync with the host OS by default, so that is not a problem as long as the host is running NTP or similar to keep clock sync.

The "fah5 not found" is more of a problem - that usually means that it failed to download it from the Stanford site, often for some networking reason. If you are on a P35 based board then you are running in to the same thing Damage did. I only updated the benchmark CD not the diskless folding CDs. I had hoped to release a new version by now, but I'm busy debugging some issues in syslinux detection of 64bit (doesn't work on USB disks, but does work on netboot....)

[jeffry55]

Thanks NotFred! I recall seeing a message about using NTP to sync the clock.

I guess I can download Ubuntu and run the Linux SMP fah software.

I hope you don't get bugged by all of your debugging. That just wouldn't be right.

[oakleeman]

I just found the Folding Benchmark CD tonight when I was searching for some sort of CPU testing live cd. I'm trying to use the cd on a MSI K7D Master-L motherboard with two AMD MP2400+ cpus and 1.5GB ECC ram. When I try to boot to the benchmark cd it shows the loading linux message and then simply reboots and then repeats the process.

I am able to run memtest86 on the system and boot into the Ubuntu 6 live cd. Once Ubuntu booted I downloaded the FAH.exe and am running two instances of it (different machine IDs) without incident so far.

I am able to boot to the cd on a VMWare virtual appliance so I know the cd is good.

Ideas?

[notfred]

That sounds bizarre, if it shows the Loading Linux message then it should keep going and you should see a bunch of messages as the kernel starts up - I'm not really sure what is going on there.

[eitje]

sounds like the same kind of thing that was happening to my VIA mITX boards, back when I was trying to test them out.

[Damage]

A couple of questions for you, notfred, about the new CD generator.

First, how does one produce a benchmark result with this strange, new concoction?

Second, do you all think on SMP instance per four cores is the right formula? Or two? Why?

[Damage]

Ok, getting the benchmark to run is straightforward enough. Never mind that one.

[Damage]

...aaaand, it looks like the benchmark doesn't actually use the SMP kernel, so I suppose the number of instances/core isn't a relevant setting.

[notfred]

There should be 4 settings, 2 normal folding with UniProcessor and SMP kernels, and 2 for benchmarks with UniProcessor and SMP kernels.

At boot time, make sure you type "benchmark64" to get the SMP kernel. It doesn't benchmark with SMP WUs yet as Stanford are still playing around with the the SMP setup, points vary dramatically etc.

[notfred]

Let me just correct what I said, I menat 32bit and 64bit rather than Uniprocessor and SMP kernels. All the kernels are SMP enabled. Sorry, doing too many things at once...