Overclock.net › Forums › AMD › AMD - General › The Socket 939 Appreciation Club and Knowledgebase [Official]
New Posts  All Forums:Forum Nav:

The Socket 939 Appreciation Club and Knowledgebase [Official] - Page 2124

post #21231 of 28692
Quote:
Originally Posted by BlackOmega View Post
Which test did it fail? Usually if it fails the L1 cache test, you need more voltage. If it fails L2 cache or FPU, it's usually because of temps. BTW, the FPU test WILL get it the hottest, be sure to keep aclose eye on temps.
yes it fails L1. what I don't understand is why it fails there, yet I can run prime for hours with no errors and the computer works "seemingly perfect"? on top of that...prime runs the cpu @ 100% and s&m was only running it @ 75%
I'm thoroughly confused about that.
Overclock_Machine
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX-60 DFI LP nF4 SLI-DR Expert 4890 BH-5 
OSMonitorKeyboardPower
Tiny7 23" awesome Corsair TX-850 
Case
None 
  hide details  
Reply
Overclock_Machine
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX-60 DFI LP nF4 SLI-DR Expert 4890 BH-5 
OSMonitorKeyboardPower
Tiny7 23" awesome Corsair TX-850 
Case
None 
  hide details  
Reply
post #21232 of 28692
Quote:
Originally Posted by BlackOmega View Post
At any rate, I swapped the board and RAM out for the other Ultra-D. Passed 3 hours of Prime @ stock, now I'm testing it @ 2640 @ 1.344v. So far it's 1/2 an hour stable at this speed/voltage. I think I might see how high she'll really go, stabily.
I'll also post up a mini CPU cooler comparison when I get back from the store.
post #21233 of 28692
Quote:
Originally Posted by ny_driver View Post
yes it fails L1. what I don't understand is why it fails there, yet I can run prime for hours with no errors and the computer works "seemingly perfect"? on top of that...prime runs the cpu @ 100% and s&m was only running it @ 75%
I'm thoroughly confused about that.
Ny_driver, the testing that S&M performs is more strenuous than Prime95. Txtmstrjoe lost a set of a memory to this specific stress tester. It's quite intensive. I'm perfectly fine with Prime95/OCCT...everyone has their own testing methods. How often though will you be running an application that stresses the processor like S&M does? I'm assuming not often especially if you just browse the internetz and check email. Don't worry about being 100% stable, it's never going to happen regardless

Good luck
Workstation
(13 items)
 
  
CPUMotherboardGraphicsRAM
Athlon II X4 620 Asus M4A785-M Asus 5770 G.Skill 8GB DDR2-800 
Hard DriveOSMonitorPower
WD3200AAJS Windows 7 Pro Acer AL2216W, Samsung P2770 Corsair 550VX 
Case
Tagan Black Pearl 
  hide details  
Reply
Workstation
(13 items)
 
  
CPUMotherboardGraphicsRAM
Athlon II X4 620 Asus M4A785-M Asus 5770 G.Skill 8GB DDR2-800 
Hard DriveOSMonitorPower
WD3200AAJS Windows 7 Pro Acer AL2216W, Samsung P2770 Corsair 550VX 
Case
Tagan Black Pearl 
  hide details  
Reply
post #21234 of 28692
Quote:
Originally Posted by ny_driver View Post
yes it fails L1. what I don't understand is why it fails there, yet I can run prime for hours with no errors and the computer works "seemingly perfect"? on top of that...prime runs the cpu @ 100% and s&m was only running it @ 75%
I'm thoroughly confused about that.
How many hours can it run prime? And which primetest did you run?

As thlnk3r stated, no programs really stress the CPU as much as S&M does. When you run the test, to get the "highest quality" results, it's best to have it set to maximum load.
I have somewhat revised the way I test like this: For initial overclock test (CPU only) I run S&M short duration @ 100%. If it passes that, then I run Prime95 for an hour. If it passes that I move on to my next OC step. If I'm at the final OC, I let prime95 run for a minimum of 8 hours. My sig rig I let run for 16.
Something kind of neat I've noticed with S&M: when you first start S&M and the loading dialog box appears, if the voltage is too low for your CPU frequency the green loading status indicator won't show up. It'll just stay blank as S&M loads up. It's actually very consistent.

Quote:
Originally Posted by Hueristic View Post
Thanks Hue, that very slight overclock wound up being 3 hours stable. Then @ 2760 @ 1.344v she was 2+ hours stable. Now I've got her sitting at 2903MHz @ 1.42v (was too tired to mess with the voltage properly, I just upped the "special control" to 110% ), at any rate, that OC is 5 hours 11 min stable. I think that I'll need ~1.5v to get 3000MHz stable. @ 1.42v it booted and even primed for about 5minutes before 1 core failed.
At any rate,I'm pretty happy with it's performance. The Xiggy DK keeping hercool is also doing an excellent job. @ 1.44v (vcore jump under load) it keeps both cores under 60C, according to smart guardian the CPU is at 43/44C stupid thing.

I'll try dropping the voltage later, then I'll mess with OCing those Corsairs to see what they can do. Once all is said and done, I'll load up Everest ultimate and run some benchmarks.
Quote:
Originally Posted by thlnk3r View Post
Ny_driver, the testing that S&M performs is more strenuous than Prime95. Txtmstrjoe lost a set of a memory to this specific stress tester. It's quite intensive. I'm perfectly fine with Prime95/OCCT...everyone has their own testing methods. How often though will you be running an application that stresses the processor like S&M does? I'm assuming not often especially if you just browse the internetz and check email. Don't worry about being 100% stable, it's never going to happen regardless

Good luck
I personally like S&M for checking out the CPU performance and thermal threshold. As usually the FPU test gets it the hottest, I've noticed that the FPU test on average gets the CPU aboutn 5C hotter than prime95 blend test.
But when testing overall system stability, can't really beat Prime95.
I was thinking of trying out linpack as I've seen a lot of others (mainly intel people) using it.
Black Box
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 3770k Asus P8Z77-Vpro Evga 780 Classified Crucial Ballistix 
Hard DriveHard DriveOptical DriveCooling
A-Data 128GB SSD Samsung Spinpoint F3 1TB Pioneer BDR-207DBK Corsair H80 
OSMonitorKeyboardPower
Win 7 pro 64 Hannspree 25" 1080p LCD 2ms Razer Lycosa PC Power & Cooling 760w 
CaseMouseMouse PadAudio
Corsair 550D MX 518 Allsop Creative XFi Xtreme music 
  hide details  
Reply
Black Box
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 3770k Asus P8Z77-Vpro Evga 780 Classified Crucial Ballistix 
Hard DriveHard DriveOptical DriveCooling
A-Data 128GB SSD Samsung Spinpoint F3 1TB Pioneer BDR-207DBK Corsair H80 
OSMonitorKeyboardPower
Win 7 pro 64 Hannspree 25" 1080p LCD 2ms Razer Lycosa PC Power & Cooling 760w 
CaseMouseMouse PadAudio
Corsair 550D MX 518 Allsop Creative XFi Xtreme music 
  hide details  
Reply
post #21235 of 28692
Quote:
Originally Posted by BlackOmega View Post
How many hours can it run prime? And which primetest did you run?
I was running the 1st test-small FFTs(maximum FPU stress, data fits in L2 cache, RAM not tested much) for 5 hours with no errors and I shut it off. Now that I look closer at it the 2nd test is probably better for the CPU test. What does FPU mean anyways(floating point unit), and what is the purpose of the 1st test the one I was running?
Edited by ny_driver - 12/30/09 at 11:17am
Overclock_Machine
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX-60 DFI LP nF4 SLI-DR Expert 4890 BH-5 
OSMonitorKeyboardPower
Tiny7 23" awesome Corsair TX-850 
Case
None 
  hide details  
Reply
Overclock_Machine
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX-60 DFI LP nF4 SLI-DR Expert 4890 BH-5 
OSMonitorKeyboardPower
Tiny7 23" awesome Corsair TX-850 
Case
None 
  hide details  
Reply
post #21236 of 28692
Quote:
Originally Posted by ny_driver View Post
I was running the 1st test-small FFTs(maximum FPU stress, data fits in L2 cache, RAM not tested much) for 5 hours with no errors and I shut it off. Now that I look closer at it the 2nd test is probably better for the CPU test. What does FPU mean anyways(floating point unit), and what is the purpose of the 1st test the one I was running?
Ny_driver, FPU is floating point unit. This part of the processor processes all the basic calculations (ie. adding, subtraction, division ect).

Hope that helps
Workstation
(13 items)
 
  
CPUMotherboardGraphicsRAM
Athlon II X4 620 Asus M4A785-M Asus 5770 G.Skill 8GB DDR2-800 
Hard DriveOSMonitorPower
WD3200AAJS Windows 7 Pro Acer AL2216W, Samsung P2770 Corsair 550VX 
Case
Tagan Black Pearl 
  hide details  
Reply
Workstation
(13 items)
 
  
CPUMotherboardGraphicsRAM
Athlon II X4 620 Asus M4A785-M Asus 5770 G.Skill 8GB DDR2-800 
Hard DriveOSMonitorPower
WD3200AAJS Windows 7 Pro Acer AL2216W, Samsung P2770 Corsair 550VX 
Case
Tagan Black Pearl 
  hide details  
Reply
post #21237 of 28692
Quote:
Originally Posted by ny_driver View Post
I was running the 1st test-small FFTs(maximum FPU stress, data fits in L2 cache, RAM not tested much) for 5 hours with no errors and I shut it off. Now that I look closer at it the 2nd test is probably better for the CPU test. What does FPU mean anyways(floating point unit), and what is the purpose of the 1st test the one I was running?
The small FFT test is similar to S&M, except it's not as intense (IMO). THe large FFT test more of the memory and the CPU to a much lesser extent. The blend test tests large amounts of both. That's why I use it for overall system stability. I've had CPU OC's pass small FFT's and S&M yet fail the blend test.
The blend test not only tests the CPU and memory it also tests the integrated memory controller (IMC).

Try running the prime blend test and see what happens.

PS: You never did say which S&M test your CPU failed.
Black Box
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 3770k Asus P8Z77-Vpro Evga 780 Classified Crucial Ballistix 
Hard DriveHard DriveOptical DriveCooling
A-Data 128GB SSD Samsung Spinpoint F3 1TB Pioneer BDR-207DBK Corsair H80 
OSMonitorKeyboardPower
Win 7 pro 64 Hannspree 25" 1080p LCD 2ms Razer Lycosa PC Power & Cooling 760w 
CaseMouseMouse PadAudio
Corsair 550D MX 518 Allsop Creative XFi Xtreme music 
  hide details  
Reply
Black Box
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 3770k Asus P8Z77-Vpro Evga 780 Classified Crucial Ballistix 
Hard DriveHard DriveOptical DriveCooling
A-Data 128GB SSD Samsung Spinpoint F3 1TB Pioneer BDR-207DBK Corsair H80 
OSMonitorKeyboardPower
Win 7 pro 64 Hannspree 25" 1080p LCD 2ms Razer Lycosa PC Power & Cooling 760w 
CaseMouseMouse PadAudio
Corsair 550D MX 518 Allsop Creative XFi Xtreme music 
  hide details  
Reply
post #21238 of 28692
Quote:
Originally Posted by BlackOmega View Post
The small FFT test is similar to S&M, except it's not as intense (IMO). THe large FFT test more of the memory and the CPU to a much lesser extent. The blend test tests large amounts of both. That's why I use it for overall system stability. I've had CPU OC's pass small FFT's and S&M yet fail the blend test.
The blend test not only tests the CPU and memory it also tests the integrated memory controller (IMC).

Try running the prime blend test and see what happens.

PS: You never did say which S&M test your CPU failed.
Cache 1 level test #2 I think is what it was. I'm going to run the prime blend test now until I get errors.
Overclock_Machine
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX-60 DFI LP nF4 SLI-DR Expert 4890 BH-5 
OSMonitorKeyboardPower
Tiny7 23" awesome Corsair TX-850 
Case
None 
  hide details  
Reply
Overclock_Machine
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX-60 DFI LP nF4 SLI-DR Expert 4890 BH-5 
OSMonitorKeyboardPower
Tiny7 23" awesome Corsair TX-850 
Case
None 
  hide details  
Reply
post #21239 of 28692
L1 cache failure is typically due to a lack of voltage. I'm betting that you'll fail the blend test relatively quickly.
Black Box
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 3770k Asus P8Z77-Vpro Evga 780 Classified Crucial Ballistix 
Hard DriveHard DriveOptical DriveCooling
A-Data 128GB SSD Samsung Spinpoint F3 1TB Pioneer BDR-207DBK Corsair H80 
OSMonitorKeyboardPower
Win 7 pro 64 Hannspree 25" 1080p LCD 2ms Razer Lycosa PC Power & Cooling 760w 
CaseMouseMouse PadAudio
Corsair 550D MX 518 Allsop Creative XFi Xtreme music 
  hide details  
Reply
Black Box
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 3770k Asus P8Z77-Vpro Evga 780 Classified Crucial Ballistix 
Hard DriveHard DriveOptical DriveCooling
A-Data 128GB SSD Samsung Spinpoint F3 1TB Pioneer BDR-207DBK Corsair H80 
OSMonitorKeyboardPower
Win 7 pro 64 Hannspree 25" 1080p LCD 2ms Razer Lycosa PC Power & Cooling 760w 
CaseMouseMouse PadAudio
Corsair 550D MX 518 Allsop Creative XFi Xtreme music 
  hide details  
Reply
post #21240 of 28692
***? yesterday I ran blend test for 45 minutes with no errors, and today I get 1 error in 1 minute. I think since the fire that I get slightly different voltage when I boot up, other than that the system runs fine.
In case you missed the fire story....on Christmas eve I was sticking the multimeter in there to test the cpu voltage on the mosfets and I think I must have caused a short somehow because the system shut down. I fired it right back up and then immediately saw sparks followed by a tiny flame which I mushed out with the tip of my finger. Anyways after the fire and after turning the computer back on and seeing that it was working without catching on fire...I tested the cpu voltage on the each mosfet and I noticed that the one closest to the fire was reading .10 lower than the other two. Now lately I have been noticing that according to HWMonitor, CPUZ, etc.. it appears that my voltage fluctuates approx. .20. for example right now according to those programs it's at 1.47 roughly and last night it was at 1.49 roughly as I didn't test with meter. I will right now 1 sec....I get 1.50 on the meter and 1.48 in windows. Maybe there is a problem with the voltage being properly regulated. I'd like to get another Neo2 Platinum cheap, or maybe something 939 and PCI-E that overclocks well with a good video card and cheap.

Oh yeah the picture of what caught on fire(not my mobo in this picture) same one just not mine..... http://img709.imageshack.us/img709/2761/firedamage.png it looks like it's still intact, but I don't really even know what it is.
EDIT: I just checked in bios and it said 1.47, but the meter said 1.52. I upped it from 1.35+8.3% to 1.35+10% and now the meter says 1.55 and in windows it says 1.50. It doesn't really seem that consistant. The voltage settings in the bios are whacked anyways..I went through and tested them with the meter, and from 1.40 through 1.55 at each of the percentages(+0,+3.3,+5,+8.3 and+10) I get the same readings (@ each voltage the readings are 1.37, 1.42, 1.47, 1.52, and 1.56)

re-EDIT: cache 1 passed now. cache 2 failed(47c max)??................... primes torture test only made it 12 minutes this time. (fatal rounding error) I don't know what is going on, I'm telling you I ran the torture test for 45 minutes yesterday. maybe my 3700+ has just had enough overclocking...I got it used almost 3 years ago from a guy here named GeekMan that OC'd it.
Edited by ny_driver - 12/30/09 at 1:24pm
Overclock_Machine
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX-60 DFI LP nF4 SLI-DR Expert 4890 BH-5 
OSMonitorKeyboardPower
Tiny7 23" awesome Corsair TX-850 
Case
None 
  hide details  
Reply
Overclock_Machine
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX-60 DFI LP nF4 SLI-DR Expert 4890 BH-5 
OSMonitorKeyboardPower
Tiny7 23" awesome Corsair TX-850 
Case
None 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD - General
Overclock.net › Forums › AMD › AMD - General › The Socket 939 Appreciation Club and Knowledgebase [Official]