Thuban and 4 sticks of RAM @ 1600 @ 1T ?

[ga1ve1an]

First select blend, then select "custom" and change the amount of ram used to be whatever you want...
As for how it will show error, a thread will fail with something akin to "ROUNDING ERROR: Rounding was 0.x, expected less than 0.4".

Didnt know about those settings or never paid attention to it. Did get a rounding error on first run. Ok have a stable run going right now.. 11 Minutes in. I had to give the NB one notch more of voltage to stabilize it. Didnt have to change anything else. Going solid.



Just a word of experience.. If you get memory errors dont immediately think it is your memory. I have noticed many times that if the nb doesnt have enough voltage it can give you memory errors. Stablize your NB first and you may be surprised what you get with your memory.

[ga1ve1an]

The prime i stopped at about 57 minutes no errors. I will let it run longer at night since I have some work to do.

Here is a quick run through with 3dmark11 with the new settings. ONly difference between this score and the previous one I posted is the higher memory at 1920 with 9-10-9-28 1t and NB 3120mhz compared to 1600mhz 7-8-8-24 1t and 3000mhz nb. You can see the most effected is the physics score which jumped quite a bit. Almost 500 points. I would have to say that most of that is due to the NB. Ok hope that helped...

Old settings

New settings

[Mad-Marty]

Finally my memory arrived.

I've running them at 16000 8-9-8-24 1T now.
Had to bump the vNB to 1.35 V, because on 1.3 V everything seemed ok ... but when copying large amounts to an usb storage,
it somehow hung up, it just stopped in the middle and did not transfer any more data.

Strange symptom ....

TRef is still 200 Mhz for me, since i need to go to 10-11-10-30 on 1866 Mhz ... i dont think this is better than 8-9-8 @1600.

[ga1ve1an]

Finally my memory arrived.

I've running them at 16000 8-9-8-24 1T now.
Had to bump the vNB to 1.35 V, because on 1.3 V everything seemed ok ... but when copying large amounts to an usb storage,
it somehow hung up, it just stopped in the middle and did not transfer any more data.

Strange symptom ....

TRef is still 200 Mhz for me, since i need to go to 10-11-10-30 on 1866 Mhz ... i dont think this is better than 8-9-8 @1600.

Best way to test that would be to to run a program like 3dmark11 with the two different memory setttings.. Check the physics score. So much traffic is passing through the NB from the CPU and the memory it is a good stress test. You will see the most dramatic change there if there is one as you saw with mine. If there is no real change then stay with which is most stable.

[Mad-Marty]

unfortunately i cant get it working stable at 1T, at least not with vNB <= 1,35 V. :-(
Games and Prime95 work without flaw, but usb copy craps out ....

at 1.35 vNB its much better than at 1,3 vNB, but sometimes still hangs.

1,375 vNB is too much, might damage the L3 right? :-/

[ga1ve1an]

unfortunately i cant get it working stable at 1T, at least not with vNB <= 1,35 V. :-(
Games and Prime95 work without flaw, but usb copy craps out ....

at 1.35 vNB its much better than at 1,3 vNB, but sometimes still hangs.

1,375 vNB is too much, might damage the L3 right? :-/

From what I understand 1.35v is about the max most I have read said to run it at. Sounds strange that you are getting this with just a USB copy. How big is the file you are transfering and is it using USB 3? Doesnt make sense.. This is more than likely a memory timings problem. Have you tried to loosen your memory timings .. something may be getting corrupt in the transfer. If you are doing prime95 fine, which is an exaggerated stress of the system in a way that most users never use their system, something is not right.

[AlleyViper]

... but when copying large amounts to an usb storage,
it somehow hung up, it just stopped in the middle and did not transfer any more data.

On my SB750 this can happen independently of OC, or cpu_nb settings. On each hub some ports are prone to fail when transferring large amounts of data and others are 100% OK. This hang is easier to trigger when connecting pen drive/hdd to a monitor hub, and then to an MB port (either on the back or internal connectors)--by this way I know which ports to fully trust. Input peripherals don't seem affected on the flaky ports.
I've seen more reports of this very same problem on other SB750 boards.

[BeepBeep2]

unfortunately i cant get it working stable at 1T, at least not with vNB <= 1,35 V. :-(
Games and Prime95 work without flaw, but usb copy craps out ....

at 1.35 vNB its much better than at 1,3 vNB, but sometimes still hangs.

1,375 vNB is too much, might damage the L3 right? :-/

IMO it would be fine if that is your real voltage, no way to tell though without DMM.

Main problem is temperatures. You can also try upping HT/SB Voltage, maybe it will help.

[Mad-Marty]

How big is the file you are transfering and is it using USB 3? Doesnt make sense.. This is more than likely a memory timings problem. Have you tried to loosen your memory timings .. something may be getting corrupt in the transfer. If you are doing prime95 fine, which is an exaggerated stress of the system in a way that most users never use their system, something is not right.

A few GB (25) over USB 2.0 HDD@Monitor Hub.
It stops at around 4 GB Copied. Going to 2T removes the Problem.


@AlleyViper:
Ok, i think i should try that with non Monitor hub, or at least plug the hub to another backplate port ...

Maybe i should try it with looser timings, but going from 8-9-8-24 to say 9-9-9-30 probably kills the same performance than 1T -> 2T i guess?

[zhadoom]

Testing a new Sabertooth 990fx and a 960T unlocked to 6 core.

-Memory DDR 1600 GEIL Enhanced Corsa kit with 04 4GB modules ( 16GB total )
-Timmings 8-9-9 28 1T at 1600 MHz ( 800 MHz real )
-NB clock 2600 MHz 1.15v
-CPU clock 3000 MHz 1.32v turbo disabled ( some weak unlocked core )

Looks stable until now.

[AlleyViper]

@AlleyViper:
Ok, i think i should try that with non Monitor hub, or at least plug the hub to another backplate port ...

If you are using a hub monitor too on a SB750, it makes the occurrence of the very same problem I was facing more likely. Might have to do with weaker signal on some ports added to the extra USB cable length to the monitor which makes that condition worse. Usually it only happened to me when writing some GBs of data to a usb drive/hdd to let it have time to fail. I didn't even completely pin this problem until after more than a year of use.

You might try to keep it plugged on the same port stack on the backplate, but swapping for the upper/lower port (still keeping it under firewire/lan). If you change stack, it's probable that one of them still shows the same problem.
IIRC, on my board I have 6 ports on the back plate, and 2 are prone for it on separate stacks. I currently run my monitor hub cable from a 2 port expansion on a PCI bay that's connected to an internal USB port, and even there, large transfers will only work 100% on one of the ports.

[Mad-Marty]

Strange .. upping vRAM to 1.68 v ... and now it works on all usb ports.
Which was not the case some days back.

I'm almost sure i did check more RAM voltage ...

Now running 4x4 Patriot AMD Performance Edition @ 8-9-8-24 1T :-) :-) :-)