Overclock.net › Forums › Overclockers Care › Overclock.net Folding@Home Team › Folding@Home Guides and Tutorials › [Windows 7/8] Complete Client v7.3.6 Guide
New Posts  All Forums:Forum Nav:

[Windows 7/8] Complete Client v7.3.6 Guide - Page 44

post #431 of 745
Quote:
Originally Posted by scorpiontsi View Post

Yea I have come to the conclusion that I cant run -bigadv with a 4 core non hyper threading cpu currently. So many older posts I think I was probably reading from a few years ago.. anyhow I am running the advanced ones for video card and overall not sure its worth the effort. Seems without quite a bit of hassle I cant fold on both. Right now folding on both yields virtually the same as just folding on the 2500k OC @ 5ghz. The video card alone pulls slightly lower ppd. With the HP server folding going on it seems less than worthwhile to let this rig fold. We will see ...


My 2500K is at 4.7 GHz and can pull anywhere from 18,183 to 26,715 PPD depending on WU using the standard SMP 4 setup. (allowed large memory units and allowed -advmethods getting 11021, 11040, 11060, and 11070 WUs with the 11040 being the max ppd and the 11021 being the min PPD).

 

My GTX 580 (clocked at 900 core / 1800 shader / 2106 memory) is getting a rock-solid consistent 20,007 PPD. I'm running both clients simultaneously.

 

I'm not sure what sort of PPD I could get if I folded only on the 2500K and I think that the extra 300 MHz you have will yield a significant points increase over my clock speed.

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #432 of 745
Quote:
Originally Posted by shad0wfax View Post



My 2500K is at 4.7 GHz and can pull anywhere from 18,183 to 26,715 PPD depending on WU using the standard SMP 4 setup. (allowed large memory units and allowed -advmethods getting 11021, 11040, 11060, and 11070 WUs with the 11040 being the max ppd and the 11021 being the min PPD).

My GTX 580 (clocked at 900 core / 1800 shader / 2106 memory) is getting a rock-solid consistent 20,007 PPD. I'm running both clients simultaneously.

I'm not sure what sort of PPD I could get if I folded only on the 2500K and I think that the extra 300 MHz you have will yield a significant points increase over my clock speed.

It's really wierd I see my numbers change a lot. What other client are you using? When I installed gpu client it told me that my video card(HD6950) was not supported. Right now Im running on just the v7 client. What's the command to enable larger memory units? I had to change to smp3 to get the two to work semi-effectively on v7 client. I did activate bigadv last night just to see if I can complete them in time. Dont think I have recieved a bigadv work unit yet though. Some might be due to the fact that I didnt put in a passkey my first 15-20 WUs hehe..
    
CPUMotherboardGraphicsRAM
i5-2500k Asus Maximus gene-z/gen3 Gigabyte 6950 HD OC Corsair Vengeance 16 gig 8x2 CL7 
Hard DriveCoolingOSMonitor
Adata 120gig Sata 3 AS510s3 Noctua 2011 SE NP-D14  Microsoft Windows 7 (Home Premium) 64bit 55 inch Samsung 3d Tv 
MonitorPowerCaseMouse
27 inch Gateway  OCZ ZT Series 750W Fully-Modular 80PLUS Bronze  Rosewill Thor V2 Logitech Performance Gaming Mouse (wireless) 
Mouse PadAudio
rocketfish Onkyo tx-nr1008 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500k Asus Maximus gene-z/gen3 Gigabyte 6950 HD OC Corsair Vengeance 16 gig 8x2 CL7 
Hard DriveCoolingOSMonitor
Adata 120gig Sata 3 AS510s3 Noctua 2011 SE NP-D14  Microsoft Windows 7 (Home Premium) 64bit 55 inch Samsung 3d Tv 
MonitorPowerCaseMouse
27 inch Gateway  OCZ ZT Series 750W Fully-Modular 80PLUS Bronze  Rosewill Thor V2 Logitech Performance Gaming Mouse (wireless) 
Mouse PadAudio
rocketfish Onkyo tx-nr1008 
  hide details  
Reply
post #433 of 745
Quote:
Originally Posted by scorpiontsi View Post

Quote:
Originally Posted by shad0wfax View Post



My 2500K is at 4.7 GHz and can pull anywhere from 18,183 to 26,715 PPD depending on WU using the standard SMP 4 setup. (allowed large memory units and allowed -advmethods getting 11021, 11040, 11060, and 11070 WUs with the 11040 being the max ppd and the 11021 being the min PPD).

My GTX 580 (clocked at 900 core / 1800 shader / 2106 memory) is getting a rock-solid consistent 20,007 PPD. I'm running both clients simultaneously.

I'm not sure what sort of PPD I could get if I folded only on the 2500K and I think that the extra 300 MHz you have will yield a significant points increase over my clock speed.

It's really wierd I see my numbers change a lot. What other client are you using? When I installed gpu client it told me that my video card(HD6950) was not supported. Right now Im running on just the v7 client. What's the command to enable larger memory units? I had to change to smp3 to get the two to work semi-effectively on v7 client. I did activate bigadv last night just to see if I can complete them in time. Dont think I have recieved a bigadv work unit yet though. Some might be due to the fact that I didnt put in a passkey my first 15-20 WUs hehe..

 

My statements were more applicable to a nvidia user, which brought you into the confusion as well. I'm sorry for that. In my defense, I did post that I was running a 580, but I didn't state the disparities between the way Folding handles radeon vs Fermi and that may have hurt your PPD. Again, I apologize for the confusion and poor response that post.

 

However, many settings still apply:

 

If you've activated -bigadv on your SMP client and you aren't actually using a Linux distro with a "core spoof" to increase your core count to, I believe it's 16 cores now rather than 12, you're never going to get assigned one of the -bigadv WUs. You mentioned that you're on 3 cores of SMP, so you'll never see a -bigadv on the SMP.

 

There is a larger memory unit setting in the configurations but as near as I can tell it doesn't actually do anything with the SMP or GPU cores; it seems to be a vestigal setting from previous core versions for lower end equipment where it made a difference. Now, it seems as if the larger memory size flags don't have any effect on the updated cores.

 

As I made clear in my edited (and very bold-faced edit at that) you will probably need to leave the core lock enabled (it's enabled by default) and force your SMP client to use cores 2, 3, and 4 while your GPU client (and system idle and whatever else) uses core1.  (Assuming you're calling your cores 1, 2, 3, and 4 here even though they're normally called cores 0, 1, 2, and 3 in computing circles.) The core affinity lock will prevent the GPU client from switching to one of the SMP use cores and vice versa. Disabling it, as I did, will hurt your performance significantly.

 

You will want to put the GPU client priority to "low" (which is higher than the default idle) and leave the SMP client priority on "idle" (which is lower priority than the GPU) and that applies to both Fermi and Radeon cards.

 

The -advmethods GPU WUs aren't available to anyone except for Fermi users and those are the WUs that are optimized to stress the GPU heavily and put almost no load on the CPU at all, which is why the Fermi users are able to get good performance out of both GPU and SMP clients with all threads available for SMP. This will not apply to you, unfortunately. However, for Fermi users, disabling the core lock if you're running the GPU on -advmethods seems to increase performance, in my experience.

 

I'm sorry for the confusion. I've done some strike-through and bold/italic editing of my post that contributed to the confusion so that it won't trip up any other AMD GPU users.

 

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #434 of 745
Quote:
Originally Posted by shad0wfax View Post


My statements were more applicable to a nvidia user, which brought you into the confusion as well. I'm sorry for that. In my defense, I did post that I was running a 580, but I didn't state the disparities between the way Folding handles radeon vs Fermi and that may have hurt your PPD. Again, I apologize for the confusion and poor response that post.

However, many settings still apply:

If you've activated -bigadv on your SMP client and you aren't actually using a Linux distro with a "core spoof" to increase your core count to, I believe it's 16 cores now rather than 12, you're never going to get assigned one of the -bigadv WUs. You mentioned that you're on 3 cores of SMP, so you'll never see a -bigadv on the SMP.

There is a larger memory unit setting in the configurations but as near as I can tell it doesn't actually do anything with the SMP or GPU cores; it seems to be a vestigal setting from previous core versions for lower end equipment where it made a difference. Now, it seems as if the larger memory size flags don't have any effect on the updated cores.

As I made clear in my edited (and very bold-faced edit at that) you will probably need to leave the core lock enabled (it's enabled by default) and force your SMP client to use cores 2, 3, and 4 while your GPU client (and system idle and whatever else) uses core1.  (Assuming you're calling your cores 1, 2, 3, and 4 here even though they're normally called cores 0, 1, 2, and 3 in computing circles.) The core affinity lock will prevent the GPU client from switching to one of the SMP use cores and vice versa. Disabling it, as I did, will hurt your performance significantly.

You will want to put the GPU client priority to "low" (which is higher than the default idle) and leave the SMP client priority on "idle" (which is lower priority than the GPU) and that applies to both Fermi and Radeon cards.

The -advmethods GPU WUs aren't available to anyone except for Fermi users and those are the WUs that are optimized to stress the GPU heavily and put almost no load on the CPU at all, which is why the Fermi users are able to get good performance out of both GPU and SMP clients with all threads available for SMP. This will not apply to you, unfortunately. However, for Fermi users, disabling the core lock if you're running the GPU on -advmethods seems to increase performance, in my experience.

I'm sorry for the confusion. I've done some strike-through and bold/italic editing of my post that contributed to the confusion so that it won't trip up any other AMD GPU users.

Hehe I know that there are some big differences with the GPU folding (I should say I have learned). Apparently what I did to get my GPU a bit higher in points was add client type advanced. Supposedly these WUs are better for ATI/AMD based chips. Might be some confusion based on that. I have gotten a few WUs that really kick ... I had one that was at 35k with just smp3. Unfortunately I havent gotten that one again. The gpu does about 12kppd with smp3. That SMP WU was really hot and fast. Was bouncing my temps unlike most WUs. I don't know if it was advanced but I would like to get more. So how difficult is it to set up a linux distro and core spoof? This machine is not a dedicated folding machine so not sure I will go as far as to setup native linux on it.
    
CPUMotherboardGraphicsRAM
i5-2500k Asus Maximus gene-z/gen3 Gigabyte 6950 HD OC Corsair Vengeance 16 gig 8x2 CL7 
Hard DriveCoolingOSMonitor
Adata 120gig Sata 3 AS510s3 Noctua 2011 SE NP-D14  Microsoft Windows 7 (Home Premium) 64bit 55 inch Samsung 3d Tv 
MonitorPowerCaseMouse
27 inch Gateway  OCZ ZT Series 750W Fully-Modular 80PLUS Bronze  Rosewill Thor V2 Logitech Performance Gaming Mouse (wireless) 
Mouse PadAudio
rocketfish Onkyo tx-nr1008 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500k Asus Maximus gene-z/gen3 Gigabyte 6950 HD OC Corsair Vengeance 16 gig 8x2 CL7 
Hard DriveCoolingOSMonitor
Adata 120gig Sata 3 AS510s3 Noctua 2011 SE NP-D14  Microsoft Windows 7 (Home Premium) 64bit 55 inch Samsung 3d Tv 
MonitorPowerCaseMouse
27 inch Gateway  OCZ ZT Series 750W Fully-Modular 80PLUS Bronze  Rosewill Thor V2 Logitech Performance Gaming Mouse (wireless) 
Mouse PadAudio
rocketfish Onkyo tx-nr1008 
  hide details  
Reply
post #435 of 745
Quote:
Originally Posted by scorpiontsi View Post

Hehe I know that there are some big differences with the GPU folding (I should say I have learned). Apparently what I did to get my GPU a bit higher in points was add client type advanced. Supposedly these WUs are better for ATI/AMD based chips. Might be some confusion based on that. I have gotten a few WUs that really kick ... I had one that was at 35k with just smp3. Unfortunately I havent gotten that one again. The gpu does about 12kppd with smp3. That SMP WU was really hot and fast. Was bouncing my temps unlike most WUs. I don't know if it was advanced but I would like to get more. So how difficult is it to set up a linux distro and core spoof? This machine is not a dedicated folding machine so not sure I will go as far as to setup native linux on it.


It's easy to set up in a virtual machine environment but it's not something that will net you any significant point increases unless you're dedicated to purely SMP folding 24/7 with no GPU going and only some light web browsing in the background and as long as you're not wanting to game in parallel with it, you'll net some significant performance gains in the SMP field and even a tiny performance gain in the GPU field. (Thanks juano for correcting my mistake in the below post.) If you're going to do any gaming at all though, the windows client will throttle down and allow you to game while you fold without much lag, but the VM environments will not at all. If that's alright with you, then the linux VMs are the way to go. If that's your cup of tea, you can possibly get your 2500K to crank out -bigadv units if you have a very high overclock. At 4.7 GHz, my machine isn't capable of finishing the -bigadv fast enough to net me bonuses unless I get lucky, as each one is down to the wire. If I do anything on my PC at all, then I lose my bonus and it's pointless and probably hurting the overall folding science as you're taking that WU away from folks like Deebs, who can crank it out in a few hours instead of a couple of days. (EDIT2: As far as the -bigadv goes, that was my experience prior to them restructuring the deadlines and juano's probably right that it's completely impossible on a 2500K now on any clock.)

 

If you're not trying to "core spoof" (which is probably a bad thing to call it anyhow) there's almost no advantage to running Linux in a virtual environment and the no nonsense windows console client or the v7 client are probably your best bets.


Edited by shad0wfax - 1/30/12 at 11:12pm
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #436 of 745
Running linux in a VM does bring a significant benefit to CPU folding no matter what sort of WUs you are folding. With a SMP3 (or even a SMP 4) 2500k you will not want to bother trying to fold bigadv, you won't even come close.

The Linux VM guide that I like is this one. it's pretty simple and easy to follow but just make sure to go into your BIOS and turn on virtualization and remove the CPU SMP slot from your v7 client before you start that guide. Once you do start it just be sure to select 3 cores instead of 8 and select your kernel as "SB i7" or "AVX i7" or something like that instead of regular i7 once you get to the web setup stage.

Using that VM I have increased the PPD from a i7 920 at 3.3GHz from ~12K to ~15K PPD with that change alone.
Edited by juano - 1/30/12 at 8:53pm
Main Rig
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 2700k ASUS P8P67 WS Revolution EVGA 980 Ti SC+ Samsung 4x4GB DDR3 1866MHz 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 1TB Samsung Spinpoint F4 2TB Samsung BD Combo Noctua NH-D14 
OSMonitorPowerCase
Windows 10 64 bit Asus PG279Q Kingwin Lazer Platinum 1000W Silverstone Raven RV03 
  hide details  
Reply
Main Rig
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 2700k ASUS P8P67 WS Revolution EVGA 980 Ti SC+ Samsung 4x4GB DDR3 1866MHz 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 1TB Samsung Spinpoint F4 2TB Samsung BD Combo Noctua NH-D14 
OSMonitorPowerCase
Windows 10 64 bit Asus PG279Q Kingwin Lazer Platinum 1000W Silverstone Raven RV03 
  hide details  
Reply
post #437 of 745
Quote:
Originally Posted by juano View Post

Running linux in a VM does bring a significant benefit to CPU folding no matter what sort of WUs you are folding. With a SMP3 (or even a SMP 4) 2500k you will not want to bother trying to fold bigadv, you won't even come close.

The Linux VM guide that I like is this one. it's pretty simple and easy to follow but just make sure to go into your BIOS and turn on virtualization and remove the CPU SMP slot from your v7 client before you start that guide. Once you do start it just be sure to select 3 cores instead of 8 and select your kernel as "SB i7" or "AVX i7" or something like that instead of regular i7 once you get to the web setup stage.

Using that VM I have increased the PPD from a i7 920 at 3.3GHz from ~12K to ~15K PPD with that change alone.


 

My i5-2500K @ 4.7 GHz posts anywhere between 18,043 and 26,986 PPD -SMP 4 in the no-nonsense console client depending on what WU I get and that's with the 580 cruncing -advmethods on the GPU client.

 

The link you posted is excellent and seemed to give me the best results of all of the different distros that I tried.

 

I will try it again, now that I have a very large sampling of SMP WUs to compare to and see if I get different results, but the last time I tried, I saw no significant change in PPD. I will let you know in an hour or two when this WU finishes and I've had a few dozen frames to analyze in the VM environment.

 

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #438 of 745
Quote:
Originally Posted by juano View Post

Running linux in a VM does bring a significant benefit to CPU folding no matter what sort of WUs you are folding. With a SMP3 (or even a SMP 4) 2500k you will not want to bother trying to fold bigadv, you won't even come close.

The Linux VM guide that I like is this one. it's pretty simple and easy to follow but just make sure to go into your BIOS and turn on virtualization and remove the CPU SMP slot from your v7 client before you start that guide. Once you do start it just be sure to select 3 cores instead of 8 and select your kernel as "SB i7" or "AVX i7" or something like that instead of regular i7 once you get to the web setup stage.

Using that VM I have increased the PPD from a i7 920 at 3.3GHz from ~12K to ~15K PPD with that change alone.


Alright, I'll eat my words on my last post. I've just tested my system out using the distro optimized for VirtualBox (because it's native 64 bit) on the link that you posted along with the most recent version of VirtualBox and the most recent kernel update posted on the link that you posted.

 

I happened to get lucky and get a WU that I've crunched enough in Windows to know what my TPF is to the second. (It's a 7506 WU and my windows TPF is 4:21 on it with the GPU running -advmethods).

 

So, I am cranking away on it in Linux and my TPF (with the exact same system configuration) is 3:56.

 

Better yet, my best TPF on my GTX 580 (@ 900 MHz shader) for the 7622 WU is normally 3:44 and now using the VirtualBox my GPU TPF is 3:43 without changing any settings.

 

Oddly enough, I had to set the execution cap to 99% per core instead of unlimited in the VirtualBox settings to prevent my system from choking miserably and even with a 1% reduction in execution across all cores I'm seeing a 25 second gain in TPF on the CPU and 1 second gain in TPF on the GPU.

 

So the long and short of it is 64 bit VirtualBox + 64 bit Linux distro = 9.6% performance increase for SMP and 0.4% performance increase for GPU.

 

I stand corrected; with the right configuration, the Linux client is better for everything (except for gaming in parallel with it).

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #439 of 745
Quote:
Originally Posted by shad0wfax View Post



Alright, I'll eat my words on my last post. I've just tested my system out using the distro optimized for VirtualBox (because it's native 64 bit) on the link that you posted along with the most recent version of VirtualBox and the most recent kernel update posted on the link that you posted.

I happened to get lucky and get a WU that I've crunched enough in Windows to know what my TPF is to the second. (It's a 7506 WU and my windows TPF is 4:21 on it with the GPU running -advmethods).

So, I am cranking away on it in Linux and my TPF (with the exact same system configuration) is 3:56.

Better yet, my best TPF on my GTX 580 (@ 900 MHz shader) for the 7622 WU is normally 3:44 and now using the VirtualBox my GPU TPF is 3:43 without changing any settings.

Oddly enough, I had to set the execution cap to 99% per core instead of unlimited in the VirtualBox settings to prevent my system from choking miserably and even with a 1% reduction in execution across all cores I'm seeing a 25 second gain in TPF on the CPU and 1 second gain in TPF on the GPU.

So the long and short of it is 64 bit VirtualBox + 64 bit Linux distro = 9.6% performance increase for SMP and 0.4% performance increase for GPU.

I stand corrected; with the right configuration, the Linux client is better for everything (except for gaming in parallel with it).

Nice .. so should I download virtual box or is the VMplayer3.0 in the write up just as capable? I couldnt finish my WU last night before I went to sleep so still havent shut down f@h to set up the virtual linux.
    
CPUMotherboardGraphicsRAM
i5-2500k Asus Maximus gene-z/gen3 Gigabyte 6950 HD OC Corsair Vengeance 16 gig 8x2 CL7 
Hard DriveCoolingOSMonitor
Adata 120gig Sata 3 AS510s3 Noctua 2011 SE NP-D14  Microsoft Windows 7 (Home Premium) 64bit 55 inch Samsung 3d Tv 
MonitorPowerCaseMouse
27 inch Gateway  OCZ ZT Series 750W Fully-Modular 80PLUS Bronze  Rosewill Thor V2 Logitech Performance Gaming Mouse (wireless) 
Mouse PadAudio
rocketfish Onkyo tx-nr1008 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500k Asus Maximus gene-z/gen3 Gigabyte 6950 HD OC Corsair Vengeance 16 gig 8x2 CL7 
Hard DriveCoolingOSMonitor
Adata 120gig Sata 3 AS510s3 Noctua 2011 SE NP-D14  Microsoft Windows 7 (Home Premium) 64bit 55 inch Samsung 3d Tv 
MonitorPowerCaseMouse
27 inch Gateway  OCZ ZT Series 750W Fully-Modular 80PLUS Bronze  Rosewill Thor V2 Logitech Performance Gaming Mouse (wireless) 
Mouse PadAudio
rocketfish Onkyo tx-nr1008 
  hide details  
Reply
post #440 of 745

As an update, after a full night of unattended runs, the Linux client in a VM had a negative impact on my Windows client GPU performance. I saw a 1.3% performance decrease in my GPU client but a 12.17% increase in SMP performance.

 

Since I'm folding in the team competition in the Fermi category, I'll probably stick with the Windows clients. If I were folding in the SMP category in the team competition or not folding in the TC at all, I'd be using a mix and match of 64 bit virtual machine SMP folding and the windows client for Fermi folding.

 

As for which version is better for you, VMWare Player or VirtualBox, it depends on your preferences. I already had a 64 bit VirtualBox set up and ready to go and so that's what I used. VMWare Player is probably a bit more user friendly but it is also 32 bit and not 64 bit. That may make a difference in folding.

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
Overclock.net › Forums › Overclockers Care › Overclock.net Folding@Home Team › Folding@Home Guides and Tutorials › [Windows 7/8] Complete Client v7.3.6 Guide