Project Piledriver 64

Godbox Folder

  • Intel

    Votes: 5 12.2%

  • AMD

    Votes: 32 78.0%

  • GPU Farm

    Votes: 3 7.3%

  • Screw that!

    Votes: 1 2.4%
  • Total voters
    41
  • Poll closed .
quickz said:
In my estimation, if you want the power draw of a 4p 6100 @ 3.0GHz to be around 600W when folding, the vcore should be set to as low as ~1.000V.
I think this is impossible, but it might be possible for 4p 6200 (with very small possibility).
Click to expand...

Which is why I stated that for power consumption like that, the chips would have to be at stock clocks.
 
Linden said:
OK, I see your point: If it's at 3GHz, then yeah, one must assume it has a nice voltage boost. :)
Click to expand...

a very hefty voltage boost...

I was running around ~950w+load depending on the wu.... 8101s were averaging 975w.
6901 was 925w...

but I am curious to see how much lower I will be with 4x 2ghz ES chips... I had those running at 1.25v instead of 1.375 that the 1.7s required for 3ghz...
 
Purchased a matching set of 6274 ES from a reputable source who pretested the cpus with my board, so hopefully, smoothness...troubleshooting a 4P seems like a PITA.
 
Take these, you shall need them.

steampunkgoggels.jpg
 
Pulled the trigger after I read some random article about the Titan supercomputer, figure the Department of Energy has got 18688 Opteron 6274 nodes, I can has one or two.


magnifying-glass.jpeg
 
Almost all parts are in, just waiting on the CPUs and some Elegant Musky mounts for testing from AXm77 . I did some preliminary testing with my i7 rig in Windows 8 and in Xubuntu 12.10, the difference on a basic SMP core is ridiculous. Almost 50% more points in Linux.

P8056 in Windows TPF 1:38 | PPD ~8600
P8055 in Linux TPF 1:22 | PPD ~16000

When I get the rig built I will do some side by side comparisons from Windows Server 2012.
 
Last edited:
RushiMP said:
Almost all parts are in, just waiting on the CPUs and some Elegant Musky mounts for testing from AXm77 . I did some preliminary testing with my i7 rig in Windows 8 and in Xubuntu 12.10, the difference on a basic SMP core is ridiculous. Almost 50% more points in Linux.

P8056 in Windows TPF 1:38 | PPD ~8600
P8055 in Linux TPF 1:22 | PPD ~1600

When I get the rig built I will do some side by side comparisons from Windows Server 2012.
Click to expand...

I could not leave it alone and was curious about the differences so I did some side by side by "virtual side" comparisons:

Core i7 Desktop Rig in my Sig.

Native Windows 8 x64:
Project 8056
TPF: ~1:38
PPD: ~8600

Ubuntu 12.10 x64
Project 8055
TPF: ~1:23
PPD: ~15900

Hyper-V Ubuntu x64
Project: 8055
TPF: 1:24
PPD: ~15600

So I get about >98% Native Linux performance in Hyper-V, not bad. It would probably run faster if I closed the client window and just let it run as a service. The linux version of F@H is definitely better optimized.
 
Cpu's came in last night. Everything installed. She boots to Live OS. My USB wireless card was acting up so I have to move this beast from the garage to my office for stability testing.
 
Its on a test bench in my garage. I was hoping to finish testing it out there and then move it inside; however, my wireless network card crapped out on me. I hope to have it up and folding by tomorrow.

What program can you use to test 64 cores and associated memory for stability? I used to use Prime and Linpack (LinX), but afraid they might not scale up well.
 
Can you post your final system specs? I was thinking of investing for a 4P platform as well.
 
Here is a good start:
http://hardforum.com/showthread.php?t=1652906

I will post it as soon as I know everything works, especially in Linux.

Basics:
Motherboard:
http://www.supermicro.com/aplus/motherboard/opteron6000/sr56x0/h8qgl-if_.cfm
CPU: Opteron 6100/6200/6300 series processors
Memory: http://www.newegg.com/Product/Product.aspx?Item=N82E16820148488
PS: Coolermaster UCP 1100
Heatsink: Supermicro, thermaltake, Dynatron, Musky modified Coolermaster 212, or water.
Drive: Any
Case: Tell me when you find one...
 
Last edited:
RushiMP said:
What program can you use to test 64 cores and associated memory for stability? I used to use Prime and Linpack (LinX), but afraid they might not scale up well.
Click to expand...

Best bet is to fire up f@h and let it rip on a unit, not many programs will push a machine as hard, there used to be a f@h benchmark program that would be good for testing but i don't know what happened to it.
 
So the "it never fucking happens, don't worry about it" happened to me. Latest Supermicro BIOS just for fun, not so fun now. Attempted Super.rom recovery, no go. FUBAR.
 
pull the plug and battery and let sit for a couple hours....if not already done so
 
Last edited:
Was the flash successful? == did it return to command line?

super.rom recovery takes _noticeable_ amount of time. Give it at least three to five minutes. Using USB
drive with activity LED helps in that department as well. Also, make sure to repeatedly press the key
sequence (don't remember if it's Home or Ctrl+Home) like crazy for quite a while (for 10-15s since
power-on -- preferrably on PS/2 keyboard).
 
I completed the flash and returned to the command line. The last time I had a bad flash was like on a Pentium 60 Mhz, I have clearly come full circle. :rolleyes:

I left the battery out overnight and disconnected the power supply connections.
I created a USB key with super.rom on it.

Bootup, it does not even read the flash drive. I am thinking RMA.:mad:

I have the motherboard box, but man I wish I would have kept that shipping box, what a weird size.
 
Did you trigger super.rom recovery correctly? (IIRC there's some kind of keystroke involved [?])

What ROM did you flash? (exact name)

Perhaps flash process coincided with something else...

I'd suggest going through usual process of troubleshooting:
1) Try different PSU, if that doesn't help
2) Remove all CPUs but CPU1 (retain its memory); if that doesn't help
3) Replace or otherwise reseat CPU1
 
SUPER.ROM

Ctrl + Home is the trigger key

I get no beeps, no activity. Wiredzone is closed on the weekend, I will try giving them a call on monday to see if I can RMA or if I have to deal with Supermicro directly. They a couple of hours drive from here and I may be able to swing a trip by there to exchange.

Wouldn't that be some shit luck to have a confounding event. I really think its the board, the other stuff was working perfectly until I rebooted after the flash.
 
I do not have any retail chips available.

I can try those steps over the weekend before I try and RMA.

I found this guy on Ebay, it will fit SWTX with minimal modification and can probably be used to retrofit a case if someone wanted.

img20121110155914.jpg
 
You must log in or register to reply here.
Back
Top