• 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.

low PPD on i7

N

nitrobass24

[H]ard|DCer of the Month - December 2009
2FA
Joined
Apr 7, 2006
Messages
10,477
I just installed the new VMware Workstation 7 RC.

And I have installed two instances for Notfreds using 4 cores each.

But im only getting about 2800PPD from each.

What do think it is?
Anyone here running the new vmware products on i7?
 
Still using 6.5.something here. What has changed other than the VMware version? Are you using the same Notfreds files as before? Did the new version rename any processes that WinAFC manages?
 
jimbowdang said:
Still using 6.5.something here. What has changed other than the VMware version? Are you using the same Notfreds files as before? Did the new version rename any processes that WinAFC manages?
Click to expand...

Well the process name is still vmware-vmx.exe but for some reason WinAFC couldnt see it.

I am now using Bill2's Process Manager to do the same thing, but my PPD is shit.
If I turn one of the VMs off then I can pull around 4700 on a single 4 core instance which is more inline of where it should be.
But turning the second one on to get 100% Utilization, the performance just goes down the toilet.
 
Dunno. I just got a whole bunch of massive (really long protiens...) WU.

Way slower, I'll tell you.
 
That's odd, still sounds like the VMs are fighting each other for resources. Maybe it's just a bug in the release candidate. I'd just keep using the version that was working, maybe check out 7.0 once the final version is released.
 
If you're running at stock, then 4700 ppd is really the max you are going to get with one client turned on. I'll give you some numbers from my i7, at 3.6GHz:

1 4-core instance: 6200ppd

1 4-core instance + GPU (on XP, so GPU takes up a full core): ~5900ppd if the GPU client is running on one of the same cores as the VM, otherwise 6200ppd

2 4-core instances: 4500ppd each, for about 9250ppd total (this is actually quite close to what running 1 8-core SMP WU on native Linux gives me, around 9500ppd...I imagine that the overhead with running Windows as well as 2 other OSes takes up the rest)

2 4-core instances + GPU: average of about 8400ppd total, not including GPU.

This is still much better than running 4 2-core VMs, that gave a total of around 6000ppd over all 4 clients, and memory usage now is much lower than it used to be (I used to get lockups, freezes, and lag all the time because I had no memory left)

That's odd, still sounds like the VMs are fighting each other for resources. Maybe it's just a bug in the release candidate. I'd just keep using the version that was working, maybe check out 7.0 once the final version is released.
Click to expand...
That is what is supposed to happen. It's the reason why 4 cores + 2-thread hyperthreading is inferior to 8 native cores, because each virtual/real core unit has to share the same resources in terms of cache, bandwidth, etc. If Intel doubled cache sizes and bandwidth, then you would see better scaling from hyperthreading, but they haven't, so you don't. In any case, 50% increase (which is what I'm getting) is pretty impressive for hyperthreading.
 
So are you saying that ~5200PPD with 2x 4core VMs @ stock and 3GPUs is normal?
 
So are you saying that ~5200PPD with 2x 4core VMs @ stock and 3GPUs is normal?
Click to expand...
If the CPU is at stock clocks, then yes. I would expect around about those numbers. If those numbers include the GPUs, man, you have a serious problem on your hands.
 
Mr. Pedantic said:
If the CPU is at stock clocks, then yes. I would expect around about those numbers. If those numbers include the GPUs, man, you have a serious problem on your hands.
Click to expand...

no with the GPUs im getting around 27k-30kPPD depending on GPU WU
 
The best way to fold on a Core i7 is to have a dedicated, non-VM, Linux install and run it with the -smp 8 flag. You also want to overclock it to at least 3.6GHz You should see 9000 to 10K PPD like this. This is how I ran my Core i7 rig earlier this year. Running two instances of -smp 4 only slowed things down for me. If you must run in a VM environment because you can't get rid of Windows, Mr. P's numbers are pretty close to accurate.


 
The best way to fold on a Core i7 is to have a dedicated, non-VM, Linux install and run it with the -smp 8 flag. You also want to overclock it to at least 3.6GHz You should see 9000 to 10K PPD like this. This is how I ran my Core i7 rig earlier this year. Running two instances of -smp 4 only slowed things down for me. If you must run in a VM environment because you can't get rid of Windows, Mr. P's numbers are pretty close to accurate.
Click to expand...
And then you can run bigadv WUs as well, which if you leave the machine alone will give you maybe double or triple your ppd over just running SMP -4.
 
You must log in or register to reply here.
Back
Top