DC Vault 2

My ATI HD4870s @ 800Mhz average at ~50M/s and the HD4890s @ 900Mhz average ~55M/s.
It takes 2 hours per (71 bit level) TF to finish and I receive about 4.3Ghz days credit.
: 24 hours/2 hours = 12 TFs (71 bit level)
: 12 TFs (71 bit level) x 4.3Ghz days = 51.6Ghz days
So thats about ~51.6Ghz days I can get from a decent HD48xx each day.

When I get done with PSP PRPNet in about a month I'll put all 6 HD48xx cards I have on it for a month or so and see where we end up.:D
 
Razor, thats pretty good, im getting about 1 per hour on a TI560 but im also sharing with Muon1 that has 7 threads of my 2600k. I dont feel like shorting the muon1 processes any more. I have another nvidia 460 that i might try to do a LL and see how that comes out. If we get enough ATI cards that dont have enough projects to work on, we can make major gains on GIMPS
 
I am trying to get this installed for my 5850 but I am having quite a bit of trouble for some reason. I am trying to install it on a Windows x64 bit system but I am having problems compiling the solution. Is it supposed to be compiled in C++ or in App KernelAnalyzer? Neither seem to compile the .sln.

Here is my C++ error when I open up .sln

Code: 
C:\Users\rappr\Desktop\mfakto-0.08 - src\mfakto-0.08 - src\mfaktoVS10.vcxproj : warning  : Platform 'x64' referenced in the project file 'mfakto' cannot be found. Please make sure you have it installed under '%VCTargetsPath%\Platforms\x64'.
 
Last edited:
My bad, I gave link to source not to executable. Link is fixed now.
mfakto version 0.08 for ATI
Windows 32-bit and 64-bit
Linux

"mfakto.exe" is already compiled so you won't need to even install the ATI SDK APP for compiling.
Extract mfakto-0.08 - Win.zip in a directory and make a worktodo.txt file in that directory and copy/paste and save your Manual Testing/Assignments (be sure to select Trial Factoring for preferred work type) into worktodo.txt and run the mfakto.exe from a command prompt that has path to mfakto.exe or make a shortcut to run.
When mfakto.exe starts it will do a short self test, if it passes it will start working on the TF ranges.
After all ranges are complete, return the results.txt file to Manual Testing/Results.
After results.txt has been uploaded, delete it and it will be recreated when the next batch of TFs are run.

Edit: I am so very sorry mmmmmdonuts for all the unnecessary work I've made you go through.
 
Last edited:
No problem Razor. I just got it up and running. My 5850 @780 MHz (default is 725) is running at 135 M/s and it looks like a unit (71 bit level) will take about an hour to do. It is keeping the GPU at 87% at the moment.

I got to leave for work so I won't be able to do more testing till later but it looks like a beast. ~ 100 Ghz days it looks like.
 
just started a P1 LL test using cudalucas on my 460 with the amd 4 core. not sure of how long it will take, its showing about 2:37min for 10000 iteration. good thing is its using about 85%gpu and 1%cpu. will look around for something that will give eta,etc.
 
Way to go chaps. It's looking like we're going to make pretty healthy progress on GIMPS :)

I wasn't able to find any stats on Free-DC so I've been using this site.

Might give PSP a try at the weekend if it's easy to set up.
 
Last edited:
Woot! Grabbed us a spot in yoyo for 15 vault points! More to come fellas. Hopefully we'll move up soon in einstein as well.
 
Good job on finding the GIMPS stat tracker, phoenicis. It's annoying that Free-DC doesn't track GIMPS.

Great work on yoyo, DooKey.
 
Wow sounds like we could really do some catching up in GIMPS with some GPU crunching. Thanks guys for researching and testing this out.

Nice crunching on yoyo DooKey!

Only a few days left until we pass some Muon teams! Crunch On!
 
Hey guys, just wanted to pop in to let everyone know that I'm still alive, just had a few real-life issues to deal with. Unfortunately, right in the middle of all that, BOTH of my boxes went down, and I had to scrape together old parts just to get a single functional computer with my old dual core e2160, nothing that could actually do any kind of crunching. I'm sorta back up with one quad, but 10 minutes after firing up OGR, core #1 is already at about 68*C (2, 3 & 4 are all at 62* or lower), so I'm not sure if my heat sink didn't get seated correctly, my case airflow is messed up after pulling and installing new stuff, or what other thing is up. Unfortunately, we're having to rely on savings a bit each month to just pay the bills until next summer, when I'll finally have the budget to go nuts on an i7 hex or something good! So I'll be here lurking, and living vicariously through all your crunching!
 
Hang in there Jathanis and take care of you and your family first. We will still be here and awaiting your return when you can. Your still a very important part of this team and I'll not let you forget it.
 
phoenicis said:
There are settings you can play with within the config file but best leave as is until one of us gets brave.
Click to expand...
If you download and install GPU Caps Viewer and check the OpenCL tab and see how many Compute Units your video card has, then you can change the value of NumStreams=x in the mfakto.ini file to the same number.
Taken from the mfakto.ini file -
Code: 
# Set the number of CUDA streams / data sets used by mfaktc.
# NumStreams must be >= 1. In this case mfaktc can process one stream /
# data set on the GPU while the GPU can preprocess the other one. When
# NumStreams is >= 2 than the time needed to upload (CPU->GPU transfer)
# the data sets can be hidden (if the hardware supports it (most GPUs are
# supporting this)).
# On Linux systems 2 or 3 seems a good numbers. There are comments that
# Windows systems need a greater number of streams.
# A greater number increases the memory consumed by mfaktc (host and GPU
# memory). The current limit for the number of streams is 10!
 
NumStreams=5
My Compute units were listed as 10 and after changing the NumStreams=10 my avg. rate went up 2-3M/s.
I'm not exactly positive if Compute Units and NumStreams are the same meaning but it did help on my cards.
 
Last edited:
Lost a spot in RNA World and GPUGRID. I'd do some grid, but I don't have any NV cards anymore.
 
OMG! - phoenicis just pounded 2814.297 Ghz-days into Gimps today!:eek:
/me tucks tail between legs and goes pee in the corner.
borgsmile.gif
 
Got home, uploaded the results files and was a bit shocked by the credits too. Between the four of us we’ve netted a princely 158 Vault points on GIMPS over the last day :)

It gets worse. I’ve set up a second instance for each card sending GPU usage to 99%. A 570 is now getting 400M/s, another 350M/s (x4 PCI-E slot) and the 560Tis 265M/s each. This is beginning to feel like discharging firearms at fish in a hollow cylindrical container :D

Razor_FX_II said:
My Compute units were listed as 10 and after changing the NumStreams=10 my avg. rate went up 2-3M/s.
Click to expand...

Thanks Razor. Gave NumStreams=8 a bash on the 560s (8 compute units) and it slowed them marginally. I then gave it another go now that things are GPU limited and it seems to have lifted the speed slightly.
 
i tried to set it to 8 also no change at all. i might try setting up a second instance on my ti560 , i dont think my 460 can handle it since it is at 85% already. my prob is going to be cpu, i have muon set to 7 threads leaving 1 thread for gpu gimps
 
Good job fellas! I threw my 2 6950s and 2 6970s at Primegrid to hold ground there.
 
Changed my numstreams = 10 on my 5850 (says it has 18 but max 10) and saw a slight bump but nothing too significant. Right now I am toying with running a single instance vs a double instance in GIMPS. I get about 135 M/s on a single instance and then between 150-160 M/s while running a double instance.

I don't know about you guys on using nVidia cards but I am getting slaughtered with the CPU. I got an X6 just under 3.7 GHz and 1 instance pegs all six cores around 30-40% and 2 instances around 75%.

Also threw my server back on RNA World to churn away at that for a while so we don't lose to much ground.

Keep up the good work.
 
mmmmmdonuts said:
I don't know about you guys on using nVidia cards but I am getting slaughtered with the CPU. I got an X6 just under 3.7 GHz and 1 instance pegs all six cores around 30-40% and 2 instances around 75%.
Click to expand...
Each of my HD48xx video cards when running Gimps use 2 full cores of a QuadCore.
Computers with 2 video cards use the whole processor.
 
started a second instance of gimps on the ti560 now, both instances running around 146M/s, each one is using 13% cpu, (one thread worth), with the other 6 threads on the 2600k doing Muon1. must be a Nvdia thing.
 
So phoenicis is about to top my lifetime output in GIMPS in a couple days. GPUs look like they'll make as much difference there as they have in other projects. I have a bunch of GPUs that I can fire up when it gets colder.
 
Well we gained 26 more points today in GIMPs. We should be making some great progress this week between GIMPs and 3 teams falling in Muon1. Not sure if we will overtake the #7 spot or not but we are inching closer.

Also anyone looking at getting a new toy soon aka BD or SB-E? I am looking at possibly upgrading to one of these but I am not sure what I am going to do or willing to spend at the moment. The 6 core SB-E looks tempting but so does Ivy Bridge and I am really hoping BD turns out to be a winner. I need to keep warm this winter:)
 
Last edited:
Razor_FX_II said:
We are getting a boost in NFS by Majestic-12's server.
http://stats.free-dc.org/stats.php?page=team&proj=nfs&team=613
Click to expand...
Thanks to the folks over at Majestic12 for lending a helping hand!

Looks like it might be a pretty good week. Looks like you guys are making some good progress in GIMPS, I see DooKey is on pace to gain 4 spots in yoyo this week and we should get 3 teams in Muon this week as well. Props to all you guys running Muon - 5.5M points today alone has to be a daily record for us! :eek: Nice work!

mmmmmdonuts said:
Also anyone looking at getting a new toy soon aka BD or SB-E? I am looking at possibly upgrading to one of these but I am not sure what I am going to do or willing to spend at the moment. The 6 core SB-E looks tempting but so does Ivy Bridge and I am really hoping BD turns out to be a winner. I need to keep warm this winter:)
Click to expand...

I'm waiting on some reviews but as long as there are no major problems with the new AMD 8 core chips I'll probably try to grab one assuming they have a competitive price/performance ratio. I've been wanting to build a Windows Home Server box so that might work nicely. :)
 
Razor_FX_II said:
Hang in there Jathanis and take care of you and your family first. We will still be here and awaiting your return when you can. Your still a very important part of this team and I'll not let you forget it.
Click to expand...

Thanks - managed to get OGR-27 running nice and stable on my q9550, so at least I can limp along with that for the next few months... :D
 
Could any of the [H]ard DC Commandos put to use a good working Asus P5N-D Intel LGA775/DDR2/SATA RAID/ATX/NVIDIA's nForce 750i SLI chipset motherboard with northbridge fan, backplate and manual.
Good mild overclocker and has the latest bios flashed, if needed I can burn all the latest drivers on disk as well.
Best part is all make someone a great deal on it - Free and I pay the shipping (USA).
I'm building another pfSense router in a miniATX case with the processor and memory and this board will be extra when I'm done.
I'll have it out and ready for shipping this Friday if a DC Commando can put it to use.
 
picked up a spot on muon1, 138 points , with 4 more spots upcoming. pour it on, with the muon1 and gimps charge, we should pass Rechenkraft.net for 7th soon.
 
With the Majestic 12 Boost in NFS of nearly 1 million points we passed a team for 15th spot in that project which brought about 25 points into our DC-Vault score yesterday.
Its really cool to see a project willing to give back to there participating teams.
 
On the same token as Razor, could anyone put to use a Radeon 4850 512MB? (Razor, I know you want another:) ). I would just ask you maybe cover the cost to ship it to you and that it be put to use with for the [H]ard Commandos. I would use it myself but my electric bill hates me.
 
mmmmmdonuts said:
On the same token as Razor, could anyone put to use a Radeon 4850 512MB? (Razor, I know you want another:) ). I would just ask you maybe cover the cost to ship it to you and that it be put to use with for the [H]ard Commandos. I would use it myself but my electric bill hates me.
Click to expand...
Yeah, that would fit into the farm nicely. I can move the 4850 from my HTPC into my wifes case with it's match and put yours into the HTPC. Thats what I had in mind to do anyway but hadn't ordered a card to put in the HTPC. All my equipment is on crunching as it possibly can be, during the summer I have to be selective with video cards running but during the winter all of them run full blast to heat the house and cut down on natural gas bills. I'm use to paying $250-$300 per month every month of the year for my electric bill and another 4850 isn't going to even be noticeable on my bill.
I'll PM you my info and let me know how much and how you want me to get money to you for shipping and any extra to cover your costs.
Thanks a bunch...need a motherboard?:)
 
My rear view mirror is now slowly (quickly?! Only 15 days left!) filling with Phoenicis on Muon1...
http://stats.free-dc.org/stats.php?page=user&proj=dpad&name=%5BH%5DBriliu

This graph is kind of terrifying from my point of view:
http://stats.free-dc.org/charts/usercumulteam.php?proj=dpad&name=%5BH%5DBriliu

I can't wait to see the slope you have when you show up on my project user cumulative score graph! What do you even have crunching on this? I remember reading about some 2 socket or 4 socket beast, was that you?
 
You want to know what Phoenicis has crunching on Muon1? I want to know what Phoenicis has crunching on PSP PRPnet now. He's everywhere!
Holy cow, I need to sell my house and do some upgrading.:eek:
 
Razor, When I read your first sentence of your post I thought:

"Oh great, the next sentence is going to be "You can't handle what Phoenicis is crunching with on Muon!"

Thank you for not doing that ;)
 
Briliu said:
"You can't handle what Phoenicis is crunching with on Muon!"
Click to expand...

Oh damn. That would have been awesome. Im going to start using that line at least once a week in the DC subforum. :D

IIRC he has 2 SR2 systems(and possibly others). So yeah some serious crunching power!
 
It's nice to have a big hitter like phoenicis helping out, even if I do get mowed here and there. :D
 
Phoenicis, I have done some testing with PSP PRP First Pass (8100) & Second Pass (8101) work and I come up with that over time that I produces more points with First Pass tests.
Please let me know if you experience something different.
 
You must log in or register to reply here.
Back
Top