I cannot believe that I have to start tweaking core voltages and other parameters to get E@H to work on an off the shelf processor/mobo combination!
I don't think that's going to help much. Only testing with another CPU will give you a sturdy outcome. See "What Do You Do If The Test Fails?" at the bottom of this page for more help on what you noticed during testing. Sorry to say, but it looks to me like you have damaged hardware, we're just not sure yet what exactly.
At least the passing of the small FFTs shows it may not be the FPU on your CPU, but still leaves the integer part of the CPU in question, as well as the RAM and the motherboard.
If you have another CPU that fits on the motherboard, run with that one first. Just to take the motherboard/Northbridge/RAM out of the equation.
P.S: Question for moderators reading this, should we stay in this thread for his hardware debugging, or do you want to move the last couple of posts to its own thread, for us to continue there?
P.S: Question for moderators reading this, should we stay in this thread for his hardware debugging, or do you want to move the last couple of posts to its own thread, for us to continue there?
Best to go to its own thread since this problem is nothing to do with "Validate errors". It'll be a bit of work but I'll shift everything related to this.
EDIT: Everything is now shifted to this new thread so please continue your discussions here.
Swappeed the AMD X6 1090T processor with a new one and still have the same problems. Additionally I tested my system with only one RAM stick at a time and verified one more time that I have computing issues regardless of the RAM configuration. So it's looking like the MSI 890FXA-GD70 mobo might be the focus of the problems. So far I've been unable to isolate any mobo configuration settings that will resolve my E@H computing issues. A merry frickin' Christmas this is!!!!!
There are only 10 kind of people in the world, those that understand binary and those that don't!
From the description, it looks like a problem with the first memory bank. Something you cannot repair, but for with a new motherboard. Although for the time being you may be able to overcome it by using banks2 and further. If that's possible, not all mobo's will allow that.
Changed the memory "Command Rate (CR)" timing from 1T to 2T and every E@H task since this morning 12/26 has generated VALID results. There might be light at the end of the tunnel. I'll know more after a few more days or weeks.
PRIME95 torture testing with the large FFT and Blended options and CR=2T doubled the time to failure to a little over an hour vs CR=1T results - but if the E@H tasks are successfully calculating then I think I'm going to be happy.
I'm guessing that I might not have purchased the optimum RAM for my system is why this tweak was needed or the mobo "auto" setting didn't do a good job of configuring the CR value?
There are only 10 kind of people in the world, those that understand binary and those that don't!
Nice to have debug LEDs on the motherboard (behind the PCI slot, if you wonder where). You really need to be a nerd to understand those. ;-)
But seeing how all 3rd party tests show the Command Rate to be 1T, you may want to check the rest of the timings (as shown on the various tests) against what you have.
Click the Review tab, then go for the "Read more" buttons on the three reviews.
RE: I cannot believe that I
)
I don't think that's going to help much. Only testing with another CPU will give you a sturdy outcome. See "What Do You Do If The Test Fails?" at the bottom of this page for more help on what you noticed during testing. Sorry to say, but it looks to me like you have damaged hardware, we're just not sure yet what exactly.
At least the passing of the small FFTs shows it may not be the FPU on your CPU, but still leaves the integer part of the CPU in question, as well as the RAM and the motherboard.
If you have another CPU that fits on the motherboard, run with that one first. Just to take the motherboard/Northbridge/RAM out of the equation.
P.S: Question for moderators reading this, should we stay in this thread for his hardware debugging, or do you want to move the last couple of posts to its own thread, for us to continue there?
RE: P.S: Question for
)
Best to go to its own thread since this problem is nothing to do with "Validate errors". It'll be a bit of work but I'll shift everything related to this.
EDIT: Everything is now shifted to this new thread so please continue your discussions here.
Cheers,
Gary.
Swappeed the AMD X6 1090T
)
Swappeed the AMD X6 1090T processor with a new one and still have the same problems. Additionally I tested my system with only one RAM stick at a time and verified one more time that I have computing issues regardless of the RAM configuration. So it's looking like the MSI 890FXA-GD70 mobo might be the focus of the problems. So far I've been unable to isolate any mobo configuration settings that will resolve my E@H computing issues. A merry frickin' Christmas this is!!!!!
There are only 10 kind of people in the world, those that understand binary and those that don't!
From the description, it
)
From the description, it looks like a problem with the first memory bank. Something you cannot repair, but for with a new motherboard. Although for the time being you may be able to overcome it by using banks2 and further. If that's possible, not all mobo's will allow that.
Changed the memory "Command
)
Changed the memory "Command Rate (CR)" timing from 1T to 2T and every E@H task since this morning 12/26 has generated VALID results. There might be light at the end of the tunnel. I'll know more after a few more days or weeks.
PRIME95 torture testing with the large FFT and Blended options and CR=2T doubled the time to failure to a little over an hour vs CR=1T results - but if the E@H tasks are successfully calculating then I think I'm going to be happy.
I'm guessing that I might not have purchased the optimum RAM for my system is why this tweak was needed or the mobo "auto" setting didn't do a good job of configuring the CR value?
There are only 10 kind of people in the world, those that understand binary and those that don't!
What kind of memory do you
)
What kind of memory do you have?
On what brand and model motherboard? (Checked BIOS upgrades?)
Those are 'standard' questions, of course. ;-)
You may have fixed it --cheaply-- with just setting up your memory correctly.
MSI 890FXA-GD70 Bios 1.7, AMD
)
MSI 890FXA-GD70 Bios 1.7, AMD X6 1090T 3.2Mhz, Win7 Home Premium 64bit
OCZ 8GB (2x4GB) DDR3 PC10666 1333Mhz, PN - OCZ3G1333LV4G, CL 9-9-9-20-33
Antec EA-750 PSU 750W, +12V~56A, +5V~25A, +3.3V~25A
nVidia GeForce 8800GT 512MB PCI-E
WD Caviar WD5000KS-00MNB0 500GB SATA 3.0 (3 disks)
Sony DRU-840A DVD-RW
There are only 10 kind of people in the world, those that understand binary and those that don't!
Nice to have debug LEDs on
)
Nice to have debug LEDs on the motherboard (behind the PCI slot, if you wonder where). You really need to be a nerd to understand those. ;-)
But seeing how all 3rd party tests show the Command Rate to be 1T, you may want to check the rest of the timings (as shown on the various tests) against what you have.
Click the Review tab, then go for the "Read more" buttons on the three reviews.