Guide For Running bigadv On Windows Natively (A3 Core) {Massive F@H Points!}

[PantherX]

I have been writing this guide on the Official Folding Forum. The most up-to-date guide will be available here. I have removed the data collection section here for privacy reasons.

Originally Posted by PantherX

• Introduction:

Finally after a long wait, bigadv/extra-large WUs are now available on Windows Natively: (compiled from different posts in this forum)

More info here. Please note that it hasn't been updated for Windows yet, but it can give you a general idea about the bigadv WUs. An interesting update is:

The maximum bonus factor of 10x has been removed: <Click>

• Running bigadv WUs:

1) A system with 8 or more Physical Cores. (MUST READ NOTE #1 & #2)

2) Download the latest F@H SMP2 BETA Client (v6.30): <Click>

3) Install the F@H SMP2 Client by following this guide: <Click>

4) If you want to take advantage of the Bonus Points (highly recommended), make sure that your Passkey/Username is already qualified.

5) Run the F@H SMP2 Client with these flags before it downloads the WU:

• -verbosity 9 -bigadv -smp

6) Windows OS 32/64 bit.


Notes:

1) Those of you running Intel Core i7-800 Series CPU or Intel Core i7-900 Series CPU will notice that in the Task Manager and Device Manager, 8/12 CPUs will be shown and may assume that your system is capable of running bigadv. This isn't the case. The fact of the matter is that these CPUs are Quad/Hexa Cores with HT (Hyper-Threading) thus you have 4/6 Physical Cores with 8/12 Threads while Standford recommends at least 8 Physical Cores. The only reason you can obtain these bigadv WUs is because of a loophole(confirm?) in the detection mechanism of the Client which reports 8/12 Cores to the Servers. HOWEVER, if you have an Intel Core i7 system that is Overclocked and STABLE (Read Troubleshooting below), you can try running the bigadv WUs and see if your system can fold within the Preferred Deadline (Read Minimum CPU Speed For Each CPU Model below).

2) The minimum amount of RAM is 0.50 GB/Core which turns out to be 4 GB. The recommended is 0.75 GB/Core which equates to 6 GB. The more-than-enough is 1 GB/Core which results in 8 GB. Note that 32-bit OS can't use RAM sizes above 4 GB. (This is the Official requirement for Linux and OSX. There isn't any Official word for Windows yet but RAM Usage for F@H Client is generally <2 GB)

3) A report has been made that a fast CPU with slow RAM can be overall slower then a slow CPU with fast RAM. This may be due to the fact that the FahCore_a3 uses more RAM to process the bigadv WU then the standard WU. (Further testing is required before it can be confirmed so please experiment with it and report your findings here.)

4) If you are already running the latest F@H SMP2 Client and have been qualified for the Bonus Points, use the -oneunit flag. After the F@H SMP2 Client exits, replace -advemthods flag (if you have it) with -bigadv flag. Do not use both as -advmethods may override -bigadv.

5) If you forget to use the -smp flag, your F@H SMP2 Client will download a Classic WU which will use only one core/thread regardless of how many you have.

6) If you forget to use the -bigadv flag but are using the -smp flag, your F@H SMP2 Client will download a normal a3 WU.

7) If you are using -smp X instead of -smp then it is recommended that you use a non-prime number for the X value as there is a known issue with scaling when using prime numbers. PG is aware of it and will hopefully provide a fix.

8) New users should first obtain the passkey so that they can take advantage of the Bonus Points. I recommend that they qualify for the Bonus Points by first folding normal a3 WUs and once you become eligible for the Bonus Points, you can try folding bigadv WUs.

9) Donors with slow internet connection should take the upload time of the WU Result (~100 MB) into consideration when estimating Bonus Points. If you fail to upload the WU Result before the Preferred Deadline, you will not get Bonus Points even if you have finished processing the WU before the Preferred Deadline. Availability of Standford Servers isn't guaranteed thus any delay will have a negative impact on your Bonus Points and Pande Group isn't responsible.

• Preliminary Results:

Please use this format to submit the data:

• Project #:
  Average TPF:
  CPU Model:
  CPU Speed:
  # of CPU sockets:
  # of cores/threads available:
  # of threads used by fahCore_a3:
  RAM installed:
  RAM used (F@H):
  OS:
  Dedicated:
  Username:

or this one:

• Project #---CPU Model---CPU Speed---RAM Used (F@H)---SMP X---TPF---OS---32/64---Dedicated---Username

Explanation Of The Terms:

• Project #: -> The Project # available to Windows Users. So far there are Projects 2684 & 2685
  Average TPF: -> The Average Time Per Frame for this Project. You can use HFM.NET to get this data.
  CPU Model: -> What is the Processor's Model that you are using?
  CPU Speed: -> What speed is your Processor running at?
  # of CPU sockets: -> Most motherboards have 1 CPU Socket. The exceptions are server motherboards and EVGA SR-2.
  # of cores/threads available: -> The number of Physical Cores followed by the number of Threads available.
  # of threads used by fahCore_a3: -> How many cores/threads have you dedicated to F@H SMP2 Client?
  RAM installed: -> How much RAM is installed in your system?
  RAM used (F@H): -> How much RAM F@H SMP2 Client is using when folding. Memory value of FahCore_a3.exe can be found in the Task Manager under the Processes Tab.
  OS: -> What version of Windows are you using and is it 32 or 64 bit?
  Dedicated: -> Is you system folding 24/7 or do you use it for other purposes?
  Username: -> In case you need to be contacted about your submitted data.

Collected Data:

Project #---CPU Model---CPU Speed---RAM Used (F@H)---SMP X---TPF---OS---32/64---Dedicated---Username

REMOVED

Minimum CPU Speed For Each CPU Model:

Please note that this is obtained from the above data and isn't a complete or official list. I will be only using values from Project 2684 instead of Project 2685 and the reason is that the TPF of Project 2685 is lesser than that of Project 2684. If a CPU at a particular speed barely makes the Preferred Deadline for Project 2685, it won't be able to make it for Project 2684 as its TPF is significantly more. However, if a CPU at a particular speed will make the Preferred Deadline for Project 2684, it can easily make it for Project 2685. For future Projects, I will be using the one with the Maximum TPF to ensure that a system can finish all bigadv WUs and not only WUs from a particular bigadv Project.

• Core i7-860 - 3.36 @ -smp 8 OR 3.80 @ -smp 7
  Core i7-860S -
  Core i7-870 -
  Core i7-875K - 4.00 @ -smp 8
  Core i7-880 -
  Core i7-920 - 3.80 @ -smp 8
  Core i7-930 - 3.52 @ -smp 8
  Core i7-940 -
  Core i7-950 - 3.77 @ -smp 8
  Core i7-960 -
  Core i7-965 Extreme Edition -
  Core i7-975 Extreme Edition -
  Core i7-980X - 3.78 @ -smp 12 OR 3.34 @ -smp 8

• Observations:

1) Please remember that all these values are calculated from Project 2684 & 2685. Future Projects may decrease/increase these requirements.

2) The current Projects 2684, 2685, 2686, 2692 (modified Project 2682 as it caused problems) have a Preferred Deadline of 4 days which would mean that if you want to get Bonus Points, your TPF should be less than 57:36 otherwise you will exceed the Preferred Deadline and will only get Base Credits which is 8955.

3) This category of WUs is still classified as trial: <Click>

4) Due to donor requests, normal SMP2 WUs (Project 6701 & 6702 plus future ones) will have a new method for calculating their Preferred Deadline and Final Deadline. HOWEVER, this method doesn't apply to bigadv WUs so if you can't meet the Preferred Deadline, please switch over to normal SMP2 WUs. (Details)

5) When reading the above table, you may notice that some machine are not dedicated. I would like to clarify that these machines do run the F@H SMP2 Client 24/7 but are also used for personal usage in such a manner that it doesn't have a significant negative impact on the F@H SMP2 Client and are able to return the WU Result before the Preferred Deadline.

7) Users with Unknown fields, please post the missing data so that I can update your record(s).

8) Links to interesting Threads in this Forum:
A) Hardware Tweaks For Running bigadv WUs: <Click>
B) Frame Times on BigAdv A3 Projects: <Click>
C) Project 2692 Feedback: <Click>

• Troubleshooting The F@H SMP2 Client: (bigadv Only)

If for some reason, you can't run bigadv WUs on a system that is capable of running it, you may want to read this:

1) To ensure system stability, F@H developers have their own stress tool called Stress CPU which is highly recommended to run on an Overclocked system to ensure that they are F@H Stable. Personally, I find that Prime95 is less stressful then Stress CPU while IntelBurnTest configured for Maximum with 5/10 iterations is more stressful then Stress CPU. I would recommend that first time users for IntelBurnTest start with Standard Mode and select the appropriate number of threads and set the iterations to no less than 5 and no higher than 20. Start the application. During this burn-in period, you should monitor the temperatures of the CPU (if the temperature rises above your comfort level, you can abort the test). Once it completes this run successfully, select the Higher Mode and repeat. You can also select the Maximum Mode once the Higher mode has finished a successful run. Do note that each time you select a different Mode, the time taken to complete the test will vary. If any errors are found, you may want to reduce the Overclock and try again.Remember that Overclocking may damage your hardware and will void your warranty. Do it at your own risk and no-one other than you is responsible for any damage that may occur when using any stress software or when Overclocking your system.

2) If you want to experiment with your Overclock, you can "capture a WU" and let your system fold it. Do remember that you will not get any points for it. You can capture a WU by copying the entire SMP2 directory while the WU is being folded. You will have to reconfigure the Client and set this option (in advance configuration):

Set it to "Yes". You can then hold this WU indefinitely and use it to test your system. In the Work folder, you can keep only wudata_0X.dat and wuinfo_0X.dat files and delete the rest so it will fold from the start. (This may vary from version to version and I tested it with v6.30). I also choose to block the Client from internet connection to ensure that it won't connect to the Servers.

3) If you TPF is more than 57:36 and your system isn't dedicated, you may want to use -smp N before switching over to normal WUs. The way I test what value of N in -smp N is better for my system is like this:
Step 1: I first run the SMP2 Client only with -smp flag
Step 2: I note the TPF for the WU to fold from 0% -> 1% by looking in the Console Window (Value 1)
Step 3: Terminate the SMP2 Client and delete all the files in the Work folder except the wudata_0X.dat and wuinfo_0X.dat files so that I have "reset" my WU to start from 0%
Step 4: Run the GPU Client(s) and the SMP2 Client with -smp flag
Step 5: I again note the TPF from 0% to 1% (Value 2)
Step 6: I compare Value 2 to Value 1. If there is a significant difference, then I lower the N value by 1 in -smp N (i.e in your case, it will be -smp 11) and redo from Step 3 -> Step 5 until I get a Value 2 that is close to Value 1.


If I have made a mistake, please tell me so that I can correct it.

Thanks to:
DonMarkoni = Suggested that I elaborate on Intel Core i7 Series which will help new users and who also found an interesting relationship between slow RAM on fast CPUs and fast RAM on slow CPUs.
7im = How to "capture a WU" and keep it indefinitely.
Plus anyone else that I may have forgotten.

[PantherX]

I have successfully installed v6.30 without any problems and am folding a bigadv WU. I deleted the v6.29 files and made a new client.cfg rather than using the old one. I was able to get my hands on G.Skill Trident 1600 MHz 1 GB and got 2 sets (4 GB total) and captured a live WU to test my system Initially I decided to do an extensive benchmark with -smp 6, -smp 7 and -smp 8 but I wasn't able to do so as I was too busy However, I was able to test it at some important settings and the results are below:

IntelBurnTest @ 2048 for 10 iterations With CPU Speed 3906

RAM Speed 1488 @ 8-8-8-24-2T
Latency 5.38 ns
VCore 1.36875
Time 68 Seconds
~~~~~~~~~~~~~~~~~~~~~~~~~~~
RAM Speed 1488 @ 8-8-8-21-2T
Latency 5.38 ns
VCore 1.36875
Time 55 Seconds
~~~~~~~~~~~~~~~~~~~~~~~~~~~
RAM Speed 1488 @ 7-7-7-21-2T
Latency 4.70 ns
VCore 1.36875
Time 55 Seconds
~~~~~~~~~~~~~~~~~~~~~~~~~~~
RAM Speed 1488 @ 7-7-7-18-2T
Latency 4.70 ns
VCore 1.36875
Time 55/54 Seconds
~~~~~~~~~~~~~~~~~~~~~~~~~~~
RAM Speed 1488 @ 6-7-6-18-2T
Latency 4.03 ns
VCore 1.36875
Time 55 Seconds
~~~~~~~~~~~~~~~~~~~~~~~~~~~
I processed the same WU from 0% to 1% and calculated the TPF:

CPU Speed 3801
VCore Auto
RAM Speed 1448 @ 7-8-7-22-1T
Latency 4.83 ns
TPF 0% -> 1% 52:02
~~~~~~~~~~~~~~~~~~~~~~~~~~~
CPU Speed 3801
VCore Auto
RAM Speed 1448 @ 6-7-6-18-2T
Latency 4.14 ns
TPF 0% -> 1% 51:39
~~~~~~~~~~~~~~~~~~~~~~~~~~~
CPU Speed 3822
VCore 1.34375
RAM Speed 1456 @ 6-7-6-18-2T
Latency 4.12 ns
TPF 0% -> 1% 51:18
~~~~~~~~~~~~~~~~~~~~~~~~~~~
CPU Speed 3906
VCore 1.36875
RAM Speed 1488 @ 7-7-7-18-2T
Latency 4.70 ns
TPF 0% -> 1% 50:16
~~~~~~~~~~~~~~~~~~~~~~~~~~~
CPU Speed 3906
VCore 1.36875
RAM Speed 1488 @ 6-7-6-18-2T
Latency 4.03 ns
TPF 0% -> 1% 50:32
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Currently I am running at the bold setting as this gave me the lowest TPF and the RAM timings are not too tight (I think) to cause any real threat. However, I will continue to monitor the system and will make necessary adjustments. Happy folding

[iancook221188]

that good what sort of ppd does a TPF of 50:18 get you??

[PantherX]

Here is from HFM.NET:

Min. Time / Frame : 00:48:32 - 18,214.90 PPD
Avg. Time / Frame : 00:50:22 - 17,229.48 PPD
Cur. Time / Frame : 00:51:14 - 16,775.33 PPD
R3F. Time / Frame : 00:51:11 - 16,791.72 PPD
All Time / Frame : 00:51:14 - 16,775.33 PPD
Eff. Time / Frame : 00:52:06 - 16,496.28 PPD

This is from P 2684. If I come across P 2685, the PPD is way higher:

Min. Time / Frame : 00:39:20 - 24,965.82 PPD
Avg. Time / Frame : 00:39:47 - 24,543.42 PPD

However, I had only one WU from this Project Hopefully, I will be getting more of them in the near future.

[iancook221188]

i like the look of those 2685 i mite have to go down the bigadv root soon helpfully the end is near for the 2684 project

[iancook221188]

Project #:2685
Average TPF:28:32
CPU Model:i7 970
CPU Speed:3.33
# of CPU sockets:1
# of cores/threads available:12
# of threads used by fahCore_a3:11
RAM installed:6GB
RAM used (F@H):2.82
OS:win 7
Dedicated:no
Username:iancook221188

hey panther can we have a sticky with all the link to the guide we have to help people get set up so they are easy to find as well

[Ghost15]

Hope this helps with your data collection.

Project #: 2684
Average TPF: 31:50 - 34,289.7 PPD
CPU Model: Core i7 - 980x
CPU Speed: 4137 MHz
# of CPU sockets: 1
# of cores/threads available: 12
# of threads used by fahCore_a3: 12
RAM installed: 6GB @ 2004MHz 9-9-9-30
RAM used (F@H): 1.4GB
OS: Windows 7 x64 Ultimate
Dedicated: No
Username: Allen_Waldrop

------------------------------------------

Project #: 2686
Average TPF: 00:21:51 - 59,939.7 PPD
CPU Model: Core i7 - 980x
CPU Speed: 4137 MHz
# of CPU sockets: 1
# of cores/threads available: 12
# of threads used by fahCore_a3: 12
RAM installed: 6GB @ 2004MHz 9-9-9-30
RAM used (F@H): 1.14GB
OS: Windows 7 x64 Ultimate
Dedicated: No
Username: Allen_Waldrop

-------------------------

Some tips I've learned. Running a Hex core, Don't do GPU Folding on that machine. For some reasons Hex core systems are greatly effected (TPF) by GPU Folding on the same machine. Much more than a Quad Core.

In the machine above, Folding on a 280 netted 8-9k PPD, but effected the SMP TPF by 2.5 Minutes, loosing 10-11k PPD

The Same card runing in a Quad i7 only effected TPD by 1 Minute at most.

[Mr.Bigtime]

wow man..when ever i read those GUIDES, i start to loose myself

But the thing is, since iam using a FERMI now, my FOLDING process increased massively ..unbelievable.