The CALL to ARMS. What's your contribution?

[SNM]

Of course, it's perfectly possible to remove your battery from the laptop when it's plugged in.

[Gerbil Jedidiah]

I don't fold on them all the time, and I do work off the socket a bit so the batteries have a chance to discharge (no folding on battery power is set)... I'm not worried.

[CampinCarl]

Well, I finally got around to installing the latest drivers and the GPU2 client from nvidia's website (it was part of that package thingy). Hopefully it should finish it's first WU in another hour or two. Debating whether or not to toss in a standard client as well.

[Gerbil Jedidiah]

Well, I finally got around to installing the latest drivers and the GPU2 client from nvidia's website (it was part of that package thingy). Hopefully it should finish it's first WU in another hour or two. Debating whether or not to toss in a standard client as well.

It's definitely worth throwing in a standard client. You may even be able to get the SMP client running, but I'd make sure it doesn't detract from your GPU2 power.

[CampinCarl]

Well, I've decided to play it safe and just run an extra single-core client. F@HMon predicts a total PPD for them just shy of 2000 (1996.59). Not much, but something. Too bad Dell doesn't support the CUDA drivers for the 8600MGT yet.

[jeffry55]

2000 ppd is good!!

[Gerbil Jedidiah]

IIRC you need to enable a flag in the vmx file. Try this and see if it helps after you enable the BIOS option.

Code: Select all

Linux Tip: How to Tell if Your Processor Supports VT

Virtualization Technology (VT) is a set of enhancements to newer processors that improve performance for running a virtual machine by offloading some of the work to the new cpu extensions. Both AMD and Intel have processors that support this technology, but how do you tell if your system can handle it?

It's quite simple: We'll need to take a peek inside the /proc/cpuinfo file and look at the flags section for one of two values, vmx or svm.

    * vmx - (intel)
    * svm - (amd)

You can use grep to quickly see if either value exists in the file by running the following command:

    egrep '(vmx|svm)' /proc/cpuinfo

If your system supports VT, then you'll see vmx or svm in the list of flags. My system has two processors, so there are two separate sections:

    flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall nx lm constant_tsc pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr lahf_lm
    flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall nx lm constant_tsc pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr lahf_lm

VT technology can still be disabled in your computer's BIOS, however, so you'll want to check there to make sure that it hasn't been disabled. The flags in cpuinfo simply mean that your processor supports it.

From my research, VT is required in order to run 64-bit guests under the free VMware server for linux… so it would logically follow that if you can do so, VT is enabled.

32-bit VT is not enabled by default under VMware server. If you want to enable it, you need to add the following line to your *.vmx file for your virtual machine:

    monitor_control.vt32 = TRUE

VMware does not recommend that you use VT for 32-bit guests, because they say it will actually hurt performance.

Uh, wth is egrep? Unix command? How would I run that? I can't even get the VM environment to come up. I am so lost. I think I have to find this flag and enable it somewhere, but they just say the /proc/cpuinfo folder... Uh, what about the stuff before the /proc??? Me. Confused.

[Flying Fox]

IIRC you need to enable a flag in the vmx file. Try this and see if it helps after you enable the BIOS option.

Uh, wth is egrep? Unix command? How would I run that? I can't even get the VM environment to come up. I am so lost. I think I have to find this flag and enable it somewhere, but they just say the /proc/cpuinfo folder... Uh, what about the stuff before the /proc??? Me. Confused.

In the .vmx file, add this line:

Code: Select all

monitor_control.vt32 = TRUE

[Gerbil Jedidiah]

Line added. Now I get this:

[Flying Fox]

More googling.

Add this one too:

Code: Select all

monitor_control.restrict_backdoor = TRUE

Remember to restart the VM. The BIOS option has been enabled right?

[Gerbil Jedidiah]

More googling.

Add this one too:

Code: Select all

monitor_control.restrict_backdoor = TRUE

Remember to restart the VM. The BIOS option has been enabled right?

I restarted the VM, removed and reinstalled the VM, double checked VT is enabled in the BIOS... all that.

[Flying Fox]

More googling.

Add this one too:

Code: Select all

monitor_control.restrict_backdoor = TRUE

Remember to restart the VM. The BIOS option has been enabled right?

I restarted the VM, removed and reinstalled the VM, double checked VT is enabled in the BIOS... all that.

With both lines in the .vmx file still nothing?

Failing that, I run out of ideas. VT is supposed to be supported with your CPU, but somehow it did not work. Or you may have to hack VMware Server to run on that thing.

[Gerbil Jedidiah]

Thanks FF. I may give it some more efforts over the weekend.

We haven't had anyone answer the CALL to ARMs lately?

[kitsura]

Watch this space. I might get a dual proc G5 up and running soon.

[Ttocs]

Some early results...

MSI HD 4870 X2, which is factory overclocked to 780MHz core...


Running on a e6700@3.4

I'm going to have to watch heat closely, though... the card barely fits in my case.

[Gerbil Jedidiah]

Watch this space. I might get a dual proc G5 up and running soon.

Now that would be most excellent

@Ttocs - What's your PPD?

[pikaporeon]

Watch this space. I might get a dual proc G5 up and running soon.

I'd be really curious as to how that'd be in PPD

[Gerbil Jedidiah]

I have been running two single core clients on a laptop... Same one I failed to get the SMP client to run on... Well, it got so hot today running the clients that the power adaptor would just about burn you and the laptop itself was giving off the smell of burning plastic! While it did this all day (since I was not here to stop it) I felt bad for the poor thing and shut it down. Maybe I should just run one client on it to keep it below the melting zone.

[HurgyMcGurgyGurg]

Well I've finally got the first stage of my contribution up. 2 console clients on one laptop and another 2 console clients on another. Not much but if I kept them on 24/7 I would be getting close to 650 ppd, (According to fahmon) To be honest I know they will probably be on for only 12 hours a day though, I don't like the whole psychological issue of leaving it on while I sleep. So it should be around 300-400 ppd. Not much but its something to tide me over until I get a new desktop which should definitely help maybe 10x my current ppd.

[kitsura]

v6.20 for some reason can't run on the G5 and keeps crashing immediately. I reverted to version 5 of the software and could only run non-SMP mode. I guess support for the PPC client is really sucky now.

[Gerbil Jedidiah]

v6.20 for some reason can't run on the G5 and keeps crashing immediately. I reverted to version 5 of the software and could only run non-SMP mode. I guess support for the PPC client is really sucky now.

That sucks. All that wasted power. Hopefully someone will have a brilliant idea *looks around the room expectantly*

[Ttocs]

I have no idea what my PPD is, I'd have to do the calculations.

I shut down my PCs at night, as well, so it goes down somewhat.

Unfortunately, I'm having some problem where it thinks my GPU isn't supported on the second client. Strange...

[Flying Fox]

I have no idea what my PPD is, I'd have to do the calculations.

I shut down my PCs at night, as well, so it goes down somewhat.

Unfortunately, I'm having some problem where it thinks my GPU isn't supported on the second client. Strange...

FahMon/FahSpy will do the calculations for you.

[Ttocs]

FahMon reports ~1480 per instance, on project 4727.

Turns out, however, that the F@H client doesn't properly detect the second X2 core. That might explain a few things...
I'll probably wind up waiting years for them to fix that, if at all. Unfortunate.

[just brew it!]

FahMon reports ~1480 per instance, on project 4727.

Turns out, however, that the F@H client doesn't properly detect the second X2 core. That might explain a few things...
I'll probably wind up waiting years for them to fix that, if at all. Unfortunate.

It may be a limitation of ATI's current GPU computing framework, and not something specific to the folding core. (I don't actually know, I'm just wondering...)

[Pegasus]

FahMon reports ~1480 per instance, on project 4727.

Turns out, however, that the F@H client doesn't properly detect the second X2 core. That might explain a few things...
I'll probably wind up waiting years for them to fix that, if at all. Unfortunate.

Those numbers don't seem far off from a single 4800. Why don't you think it detects the second core? Unless crossfire can't be disabled.
But if it can't see the 2nd core then both clients you are running are sharing the one core they can see. Why run 2 then? If you try just running 1 then you should get the same PPD as both of them added together now if it really can't see the 2nd core.

[Ttocs]

I wasn't aware of the problem when I made the screenshot.

Apparently, a lot of it is discussed starting at about halfway/at the end of page 1 here: http://foldingforum.org/viewtopic.php?f ... &sk=t&sd=a

I can run two at once and get no slowdown right now (PPD doesn't change when I close one of my clients), because unless it's one of the new "big" workunits, I'm only using about 320 of the 800 stream processors. That's part of the reason why nVidia is beating AMD/ATI on F@H right now - fewer shaders, clocked higher.

JBI, it likely is an ATI Driver (or CAL.. or whatever their variant of CUDA is) addressing issue. I'm not sure how responsive ATI is with these issues. We'll see!

[Pegasus]

Apparently, a lot of it is discussed starting at about halfway/at the end of page 1 here: http://foldingforum.org/viewtopic.php?f ... &sk=t&sd=a

I'll check it out. thanks

I can run two at once and get no slowdown right now (PPD doesn't change when I close one of my clients), because unless it's one of the new "big" workunits, I'm only using about 320 of the 800 stream processors. That's part of the reason why nVidia is beating AMD/ATI on F@H right now - fewer shaders, clocked higher.

I ran another GPU2 client on my 4850 to see if it would use the unused processors. (The core speed is OCed) PPD was split between the two. From 1800PPD on 1 client to 900 and 900 PPD on 2 clients. I'm not seeing how I could use the extra processors on 1 core by running multiple clients.

[pikaporeon]

my WUs havn't successfully sent for like four days now

[41.60l23]

Hey guys,

I quit folding for the summer (my AC couldn't quite keep up with it), but now that it's getting cooler, I'm going to start getting back into the game. I have a few machines to add:

P4 w/ 2600Pro GPU2
Celeron 2.66 w/ Linux Client
X2 5400 @3.0 w/ 8800 GTS (GPU2 client, I guess)

The first two are working fine, and I'll have them up soon. However, the third one (at a different house) is having some problems. Every time I start up the client, it gives the error "Error code 0 - Could not download core". This happens no matter the client I use. I have no firewall software, and the Windows firewall is disabled. As far as I know, there is nothing that should be blocking it on my router, either. Doesn't the client just use port 80 as a HTTP browser would? There's nothing helpful on the Stanford forum.