• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Linux VM for large work units?

A

AMD_Gamer

Fully [H]
Joined
Jan 20, 2002
Messages
18,287
Is there a guide to run those big work units in a linux VM? do you need a certain CPU? I have a Xeon X3440 on my ESXi host i might run while I am not using the host for other stuff.

How much PPD could I get?
 
you wont be able to run bigbeta on that CPU (atleast not without tricks) since its ment for 12 threads and above
 
AMD_Gamer said:
Is there a guide to run those big work units in a linux VM? do you need a certain CPU? I have a Xeon X3440 on my ESXi host i might run while I am not using the host for other stuff.

How much PPD could I get?
Click to expand...
the absolute minimun is i5 2500k OC to like 4.6 ....:eek:

with VM tricks ofcourse...;)
 
Oh man, here we go again...

To answer the OP; yes there is a guide. Yes, you need a certain CPU or multi-proc systems which meet the 12 thread minimum. No, that CPU shouldn't do large WU's. Your PPD for SMP WU's should be around 30,000.
 
Jester14 said:
Oh man, here we go again...

To answer the OP; yes there is a guide. Yes, you need a certain CPU or multi-proc systems which meet the 12 thread minimum. No, that CPU shouldn't do large WU's. Your PPD for SMP WU's should be around 30,000.
Click to expand...

How do you get a VM to use the cpu properly?
 
AMD_Gamer said:
How do you get a VM to use the cpu properly?
Click to expand...

I'm not following the question. I am guessing that since you are running an ESXi box, you probably know more about virtual machines than must of us. The key points are to set up your VM with 8 cores, install a Linux OS (Ubuntu is as good as any, but use your favorite), probably install BFS for your kernel, and you should be set. Is there something more specific you are asking?

FWIW, I wouldn't even try bigbeta with that proc. If you had a 2600K, maybe. You can run normal bigadv with it without issue, though. Another thing to consider is, does the box need resources available for actual use? If so, sucking up all 8 threads may not be a great idea. I assume ESXi lets you prioritize VMs, but I don't know enough about it.
 
musky said:
FWIW, I wouldn't even try bigbeta with that proc. If you had a 2600K, maybe. You can run normal bigadv with it without issue, though. Another thing to consider is, does the box need resources available for actual use? If so, sucking up all 8 threads may not be a great idea. I assume ESXi lets you prioritize VMs, but I don't know enough about it.
Click to expand...

One issue with assigning 8 VCPUs to a guest in ESXi is that for anything going on in the guest, the hypervisor will wait for 8 threads to be idle to give cycles for that guest.

So if you have 8 cores, one 8VCPU guest and three lightly busy 1VCPU guests, then you could find that the host box is mostly idle and that the 8VCPU guest is waiting for CPU >50% of the time.

It would be very apparent if this were happening, though.
 
And its also running on 1/3 the recommended core/thread count .... Thuban on big adv was one thing (3/4 the thread count "wanted"), but running big beta on anything under 12 cores is dam irresponsible (not to mention puts you bloody close to the time frame IIRC.....anyhow enough of this :D

nicfolder said:
but it works...:D
Click to expand...
 
You can run smp inside a vm... your cpu is not powerful enough for bigadv...
 
Ok thanks for the help everyone but let me clear this up.

I have an ESXi host with an X3440 and 8GB RAM that I would like to get folding. Is there a way to run it inside a VM to get the best PPD possible?

I remember years ago there was a folding boot CD. That would be awesome to use when I am not using the machine for VMware.


I used to be really into this folding stuff I don't know why I got out of it. MY GTX460 has been chugging along though.
 
Is that proc going to cut it on some of the chewy (aka 2684) WU?

If the system is doing much of anything else I'm not sure if you will hit the deadlines on all the -bigadv
 
Sounds like the OP isnt goign to have 100% uptime, so no probably not ....

running just -smp is the best bet for PPD .

Kendrak said:
Is that proc going to cut it on some of the chewy (aka 2684) WU?

If the system is doing much of anything else I'm not sure if you will hit the deadlines on all the -bigadv
Click to expand...
 
AMD_Gamer said:
So just run the regular smp client in a vm? just a normal 1 vCPU setup?
Click to expand...

Yeah. If you use the web configure page don't check Enable Bigadv. Use SMP auto-startup instead.

Otherwise do the manual configure with ./FAH6 -configonly and start it with:
./FAH6 -smp
 
AMD_Gamer said:
Is this the current guide http://hardforum.com/showthread.php?t=1511766

Now i just need to figure out the vCPU thing?:confused:
Click to expand...

8 VCPUs will work... just know that *any* scheduler time on the guest will need all 8 host threads to process it.

So if you have one 1VCPU guest running at 100%, one 1VCPU guest at 1%, and one 8VCPU guest running F@H.... that 8VCPU guest might be waiting on CPU half of the time. The hypervisor will alternate between scheduling the 8VCPU guest and the two 1VCPU guests. So effectively you will have 6 cores empty 50% of the time.

Good read:
http://blog.peacon.co.uk/understanding-the-vcpu/

Basically VCPU doesn't scale vertically like you'd think it would. There are gotchas.

I'm not aware of any tuning parameters that affect this behavior...might ask in the Virt subforum.
 
You must log in or register to reply here.
Back
Top