A few issues with Notfred/VMware Player

I hate making new threads cuz I'm sure this has been covered before, but I failed to find it, so here goes:

Here's my VM setup: Win7 64-bit, i7 @ 3.36 ATM, 6GB RAM, running one VM with 4 threads/cores configured for it.

Issue 1: For some reason, after running the VM for a while, it starts killing my internet connection by maxing out my upload ISP cap constantly. I can see this because I have a custom firmware on my Linksys router (DD-WRT) that shows all LAN, WLAN, and WAN bandwidth usage. My ISPs upload cap is 512 Kb/s and my WAN outbound traffic stays there until I close or disconnect my VM, then immediately goes down to 0 outbound traffic. This is not happening while uploading a completed WU because FahMon shows it's still in the middle of working on a WU. It's also not CPU usage because priority on it is "Idle" and I can ping my router with 1ms times constantly (as was the issue with someone else that recently had this issue). Why is this and is there anywhere I can cap the VMs connection usage? I don't see the option.

Issue 2: Notfred is only using 2 or 3 cores sometimes, about 70% of the time it seems to use all 4 cores/threads as I've assigned to it through the folding.vmx file in Wordpad and also through WinAFC. Is this normal? Also, what's really the point in using WinAFC to set the affinity to specific cores/threads? As long as you set the priority low in the task manager, I see no problem with letting it distribute the load across all 8 threads. Care to enlighten me? Here's my config:

folding.vmx config: "numvcpus = "4""
WinAFC affinityinput.txt config: "*\vmware-vmx.exe := ALL [assign=4,priority=idle]"

I'm seeing around 4150k PPD from 4 threads @ 3.36 GHz. That's about on-par, right?

Issue 3: About once a day, the VM stops and gives me some errors. I've gotten two different ones so far, I keep forgetting to get a screenshot of it, but one of them I just got about 30 mins ago. It was giving me an error that pretty much said "not enough memory" even though I have "1024" configured for the VM in the folding.vmx file and I have no less than 3GBs of physical RAM left at all times. I wish I had gotten the exact error for this, but will post back with it if it happens again soon.

Edit: One of the errors says
"attempt to access beyond end of device
hda1: rw=1, want=2096304, limit=2088387"

But it seems to still be folding after this error, so I don't know if I should worry about it.

Thanks for any help you all can provide .

For issue 2, AFC is usefull when you're fully loading 100% of your threads, in you case you're using 50% of your CPU, it doesn't do much for you unless your using a few applications at the same time, which would make the VM wait for a while.
Yeah it's normal for the VM to constantly switch between cores ,and 4k ppd is normal.
What you should do for max ppd is let it run on all 8 threads with idle priority, it shouldn't disturb the other tasks. You should get 7-8k ppd or a bit less if you're heavily using the computer.

For issue 3, have you tried running memtest?

** VM Console is showing "attempt to access beyond end of device"
This is due to a number of different things. Easiest way to correct this is to reboot the VM. Upon reboot, the VM frees up any unused memory and most of the time will clear up this message
I have this error on a vm but no problem.

T4rd said:

Edit: One of the errors says
"attempt to access beyond end of device
hda1: rw=1, want=2096304, limit=2088387"

But it seems to still be folding after this error, so I don't know if I should worry about it.

Click to expand...

This last one that you mentioned happens on both of the NotFred runs that I have. Don't know why, but I do know that it keeps on folding without any issue for me (so far). I've not really had the time to pay that much attention to it, so as long as it's behaving itself I'm not gonna worry too much.

Issue 2: Notfred is only using 2 or 3 cores sometimes, about 70% of the time it seems to use all 4 cores/threads as I've assigned to it through the folding.vmx file in Wordpad and also through WinAFC. Is this normal? Also, what's really the point in using WinAFC to set the affinity to specific cores/threads? As long as you set the priority low in the task manager, I see no problem with letting it distribute the load across all 8 threads. Care to enlighten me? Here's my config:

Click to expand...

I don't use WinAFC, and don't see the point in using it if you have cores assigned in the .vmx...Also, you will occasionally get a work unit that uses an older core instead of the version 2.10 core, and they are not optimized for multi-core machines...thus, it will not utilize 4 cores.

I'm seeing around 4150k PPD from 4 threads @ 3.36 GHz. That's about on-par, right?

Click to expand...

Yes, you could also create another notfreds vm to utilize the other 4 cores...you would probably average around 9k ppd.

As far as the notfreds errors go...as long as it doesn't say anything about segfaults, or say anything about the core shutting down, you're fine.

For some reason, after running the VM for a while, it starts killing my internet connection by maxing out my upload ISP cap constantly.

Click to expand...

Are you sure it's not uploading a previously completed work unit?...I can't think of any reason that notfreds would require bandwidth other than for downloading cores or uploading completed units....Keep in mind, every time you fire up the notfreds from a "shut down" state, it has to download the core again.

Hope this helps!

Ok, here's another error I got just while typing this out. This actually stopped the VM and I think it's the first time I've seen this specific one:

"Fahcore_a1.exe[7123]: segfault at 7f3a0a616b58 ip 00000000005ba9e4 sp 0000000041601f90 error 4 in FahCore_a1.exe[400000+36200]"

That just a bad WU or something?

404 - File not found said:

For issue 2, AFC is usefull when you're fully loading 100% of your threads, in you case you're using 50% of your CPU, it doesn't do much for you unless your using a few applications at the same time, which would make the VM wait for a while.
Yeah it's normal for the VM to constantly switch between cores ,and 4k ppd is normal.
What you should do for max ppd is let it run on all 8 threads with idle priority, it shouldn't disturb the other tasks. You should get 7-8k ppd or a bit less if you're heavily using the computer.

For issue 3, have you tried running memtest?

** VM Console is showing "attempt to access beyond end of device"
This is due to a number of different things. Easiest way to correct this is to reboot the VM. Upon reboot, the VM frees up any unused memory and most of the time will clear up this message
I have this error on a vm but no problem.

Click to expand...

I would run all 8 threads on it, but when I max this thing out, temps get too high for comfort for me. Just with 4 threads maxed it gets up to almost 80°C. With all 8 threads maxed it goes to around 90°C. Also, yeah, I usually just restart the VM too and it's fine after those errors. It's just annoying because I can't babysit this thing all day from work. If I remote into my PC, my GPU clients mess up and stop receiving WUs (known issue with GPU client, right?).

Axdrenalin said:

This last one that you mentioned happens on both of the NotFred runs that I have. Don't know why, but I do know that it keeps on folding without any issue for me (so far). I've not really had the time to pay that much attention to it, so as long as it's behaving itself I'm not gonna worry too much.

Click to expand...

Yeah, I just noticed it happens on all three of my boxen that run notfred at home. So I guess it's not a big deal. Thanks .

newxperiment said:

I don't use WinAFC, and don't see the point in using it if you have cores assigned in the .vmx...Also, you will occasionally get a work unit that uses an older core instead of the version 2.10 core, and they are not optimized for multi-core machines...thus, it will not utilize 4 cores.

Are you sure it's not uploading a previously completed work unit?...I can't think of any reason that notfreds would require bandwidth other than for downloading cores or uploading completed units....Keep in mind, every time you fire up the notfreds from a "shut down" state, it has to download the core again.

Hope this helps!

Click to expand...

Well what's weird about it is that it sometimes constantly changes in the middle of a WU from 2 to 3 to 4 cores utilized. So I don't think it's specific to the WU, but I dunno, heh, that's why I'm here.

I wouldn't think it'd still be uploading a WU when it's already 40% into the next one an hour or two later. But even if it is, is there any way to cap the VMs bandwidth usage? I would like to keep it from affecting everything else on my network (like my 360, PS3 and laptops) by killing my connection.

Thanks a lot guys.

When I was troubleshooting my own segfault errors, I found it to be an issue with my ram...I relaxed the cas timing, and the segfaults went away...

As far as remote desktop screwing up the gpu clients, yeah this is a known issue...To avoid it I use logmein.com (free)...After using it I actually prefer it over microsoft's RDC.

I'm at a loss on the bandwidth issue though...hopefully someone else has an answer...

Thanks, Newexp. I'll try the logmein. I haven't messed with the latencies on my RAM though, just have it OC'd to 1600 MHz (which is what it's rated for).

Ok, now I have a WU that refuses to use more than 2 cores and it's not even maxing those 2 cores out, they're staying at around 70-90% for some reason. It's a P2665 1920 point core so it's taking forever to get through . Anyone seen this?

Maybe the ram needs more voltage? Have you tried memtest?

T4rd said:

Thanks, Newexp. I'll try the logmein. I haven't messed with the latencies on my RAM though, just have it OC'd to 1600 MHz (which is what it's rated for).

Ok, now I have a WU that refuses to use more than 2 cores and it's not even maxing those 2 cores out, they're staying at around 70-90% for some reason. It's a P2665 1920 point core so it's taking forever to get through . Anyone seen this?

Click to expand...

I did, it's slow!
Project : 2665
Core : SMP Gromacs
Frames : 100
Credit : 1920

-- Opteron 8346HE 1.8ghz --
Min. Time / Frame : 23mn 21s - 1184.07 ppd
Avg. Time / Frame : 24mn 14s - 1140.91 ppd

404 - File not found said:

Maybe the ram needs more voltage? Have you tried memtest?



I did, it's slow!
Project : 2665
Core : SMP Gromacs
Frames : 100
Credit : 1920

-- Opteron 8346HE 1.8ghz --
Min. Time / Frame : 23mn 21s - 1184.07 ppd
Avg. Time / Frame : 24mn 14s - 1140.91 ppd

Click to expand...

Yeah, it's on auto voltage. I think my P6T Deluxe is supposed to adjust voltage on the fly depending on your OC, but I'll check it out. Thanks.

Yeah, that's garbage. I'm getting about 2100 PPD on it. Which seems about right given I usually get around 4300 PPD when all 4 threads are utilized and only 2 threads are here at ~80%. Gheyness, it's gonna take another 9 hours to get through this WU. Hopefully I don't get another one.

404 - File not found said:

I did, it's slow!
Project : 2665
Core : SMP Gromacs
Frames : 100
Credit : 1920

-- Opteron 8346HE 1.8ghz --
Min. Time / Frame : 23mn 21s - 1184.07 ppd
Avg. Time / Frame : 24mn 14s - 1140.91 ppd

Click to expand...

You're running an a1 unit. That would explain the drop in performance.

Just finished a WU on my VM and checked the log and saw this:

[08:51:02] Completed 245000 out of 250000 steps (98%)
[08:57:47] Completed 247500 out of 250000 steps (99%)
[09:04:24] Completed 250000 out of 250000 steps (100%)
[09:04:25] CoreStatus = 89 (137)
[09:04:25] Client-core communications error: ERROR 0x89
[09:04:25] Deleting current work unit & continuing...

Click to expand...

WTF!? Did it seriously delete the WU just over a communications error!? This was a 1920 pointer!