I am using the 801BT which is the "Big Toe" version I found on the OCZ support forum. They basically increased the NB voltage to get 4-dimms working. Only difference from the official Asus 801. This was when running four sticks of RAM was a problem. All the Bios' after this seemed to not fix anything I really needed.Quote:
Originally Posted by Radical_53
That's interesting ;) Just the file I always wanted to try out. More NB voltage as the only change is stupid, Tony had some quite nice files for the DFI NF4 back in the days.
I didn't have this high reported voltage with older bios versions, maybe I should flash back and see what happens. Just in case it's really that high :shocked:
Or does someone know where the voltage measurement point is to use a multimeter directly?
my NB temperature:48 SB:58 NB core voltage 1,45
how much is he for who?
I see now why the temp for SPP and MCP are lower. They just cut in half on the version you are running.Quote:
Originally Posted by Radical_53
Please see this link from EU forums:
http://www.lavalys.com/forum/index.php?showtopic=3092
Interesting. But also nice to read the quote from Nvidia that the chip doesn't have a sensor on its own ;) So the temperature reading is pretty obsolete.
Well, I still think this is a very interesting board. Just today I happened to see the somewhat huge effect of the ddr controller / slot voltages. Different clocks, different voltages needed. Fascinating impact on stability :)
I had my RAM tested at ~1100MHz before and the voltages were set accordingly. Now it only ran at ~1000MHz and I had some slight stability problems (not in benchmarks, but overnight stuff).
I lowered the Vdimm from 2.075V to 2.050V which was some success, still not enough. Then I changed the other 3 voltages from -10mv/-20mv/-20mv to -10mv/+0mv/+0mv. Instead of re-booting after 5 mins of OCCT RAM test, the system now runs for over an hour and is still running without an error (I had different stages between, freezes and reboots after 15-40mins were all there).
Yes, I still like this board too. I have a full set of waterblocks for each of two systems with this mobo. I better like it for awhile.
Definitely, yes ;) I really think this is something, somehow fascinating. Some boards I just hate because of their behaviour, but this one... I really like that.
I hope my E8500 arrives today, let's start all over again :P
I like this board because of that too, you have to treat it just right otherwise it will bite your head clear off.
It makes it more fun, and gives you a much bigger sense of accomplishment :p:
I'm just waiting for a newer bios update, the 1002 screwed up my board, and couldn't manage to get FSB to stick after I fixed the bios. so 0801 until I get the balls enough to install new bios again.
Yes, a new bios would be really nice. I "need" to run 1002 in order to let the board detect the E8500 that luckily arrived today.
I'm playing around with it right now, some settings seemingly need to be changed quite a bit to get this stable. I'm really curious where it'll lead me this time :D
Quick question about this board. Is it as picky about memory as the Striker Extreme? I've got the EK block set that fits both the striker and this, but I've fried enough strikers to be done with it.
I have used different OCZ 1GB and 2 GB chips with no problem. Needs a little voltage increase to 1.9 to 2.0V.
I like OCZ due to their great return program. Every year or so a chip will die and they send a new one in no time as long as it fails memtest.
I mainly meant if it's anything like the striker where it won't boot/post with a lot of memory modules.Quote:
Originally Posted by Snyxxx
I was using OCZ modules too, only OCZ, and never had a problem with this board. It even ran 4 sticks 1T up to pretty decent speeds.
This is one of the best boards ever when it comes to ram compatibility. Because of an rma issue early on, that ended up not being the board anyhow, I now have two of these and for some reason I am now the family/neighborhood component tester, lol. I set up a basic test bench for live linux distro's and hardware.Quote:
Originally Posted by crash5s
There is not one brand that I have not been able to get running stable in this board. The only issue ever for me was that rma screwup and that 1002 bios that will not save settings for nothing.
As I said earlier, it will save ;) If it doesn't, just set a ridiculously high FSB, it'll update the "ROMSIP" (anyone knows what that is exactly), OC protection comes in and then you can set your desired FSB.
I'm still testing this Wolfdale CPU I got for myself. It ain't easy at all :D I had to lower a couple of voltages I ran almost all the time unchanged. If I'm lucky, my RAM will be doing a little better than it did the last week with my older E6600.
So far, no easy challenge for me... most stability programs don't give me errors, they just make the machine reboot or give a bluescreen... perfect for diagnostics :D
Well if you want to share some settings radical, I'm running an e8400 at the following
1.250V core
2.075V memory
1.45V northbridge
the rest at auto
This is with spread spectrum disabled, C1E on, EIST on, memory at 5:6 444MHz 4-4-4-12.
But it's not getting me very far, I currently have a 370MHz FSB. I was stable at 380MHz for 24 hours but I don't think 10MHz is gonna make that much of a difference in the real world so I dropped it down a little bit just to feel safer. I could not get anything stable beyond 380MHz, even up to 1.35V vore and 1.75V northbridge. I'm scared running the northbridge at 1.75V with stock cooling and a side intake fan blowing down on it. Also, I didn't try every single possible combination of voltages and FSB speeds, but 405, 410, 420 would all crash within 10 minutes of running stability tests. This is even with loose memory timings and dropping the memory into sync mode
If you have some more insight, let me know as I would love to crank this thing above 400MHz.
Well, sure ;)
First thing that seems to have helped me todas is a clean CMOS clear. The big one, with removed power cable, leaving the system resting for at least half an hour without any cord connected to it.
Right now I'm running Prime95, has been running 1.5 hours right now, with the following settings:
421x9,5, RAM "auto" (resulting in 1010MHz RAM speed)
VCore 1,31875V
Vdimm 2,025V (OCZ ReaperX, needed 2,075V with the Quad and 2,050V with the E6600)
1.2V HT 1,25V
Vnb 1,30V
Vsb 1,50V
CPU VTT 1,20V
I do have C1E and EIST enabled also. RAM is running with 5-4-3-15 2T latencies.
Don't ask me why, I think it has to do with the CMOS clear, the system now seems to run so far. The same setting, same options, before the CMOS did result in a freeze with followed bluescreen every time I did something that stressed the RAM, even when they were set below 900MHz.
Also, you should try to jump above 400 directly, 410-415 to get out of the range of the FSB hole. "Within", you need much higher voltages.
From my previous E6600, I had to lower Vnb, lower CPU VTT and even lower Vdimm to get it stable.
Edit: Back to the drawing board... so much for Prime. It ran nearly 2 hours, then I thought everything was ok so far (I used the blend test to check the RAM, 64-bit version of Prime). Then I started SuperPi... and it nearly instantly froze when I hit the button to calculate 32M. Wow.
If you read a little further down in my post I said I tried 405, 410, 420.Quote:
Originally Posted by Radical_53
How the hell are you stable with a Vnb so low??? Mine is probably not even stable that low at stock clocks.
Also in regards to your SuperPi crashing, I was getting very similar errors like that above 380MHz. It would seem stable, like any temp that I could read from any program would be stable and maxed out for 1 hour+, I would think I'm in the clear, then I load up a game and it crashes.
And regards to your RAM voltage, I have had stable runs at 444MHz 4-4-4-12 at 1.95, 2.00, 2.05, and 2.10 but there doesn't seem to be any difference in the RAM temps, and my RAM is rated at 2.10V 400MHz 4-4-4-12, so I figured 2.075 was definitely not high enough to hurt it, why would I bother trying to undervolt when one day I might crash, and never know why. This way I feel a little more comfortable that it wasn't caused by the RAM.
Well, don't ask me, it's my board who should know :D I actually needed to lower Vnb to get it "stable" with the E8500. Same thing with the Vdimm. I really don't know.
It's just too strange that a thing like Prime runs, and the always cried-about SuperPi makes the machine crash...
My RAM seems to be very picky about voltage, it doesn't like not enough just as too much... both things do cause problems.
Really nice settings to be almost stable, looking forward to run an E8400 daily on my revived Striker.... for me CPU VTT was the setting to keep the rig stable with the quads...Quote:
Originally Posted by Radical_53
I wouldn't keep saying I can't save the settings if I hadn't already exhausted everybody's suggestions trying to get my settings saved with the 1002 bios. What I will not and cannot afford to do is risk burning this board or another piece of hardware out in order to test a buggy bios that most likely will not benefit me seeing how I am still running an old school e6600.Quote:
Originally Posted by Radical_53
Well, ok then ;) I thought you might have missed that one. If all those things don't work, well...
@leeghoofd: Yes, I'd love them to be fully stable. I'm fuddling around with all kinds of things, yet I can't seem to make a big difference. I don't see the "line"... no pattern, no real structure that would lead me closer to my goal.
If I were you I would flash to an older BIOS. I don't know much about the older ones because I bought this motherboard with my e8400 and flashed to 1002 as soon as it was available, but 0803, 0801, and 0805 seem to be the most popular for the 65nm processors in that order. 0803 is hard to find as it's no longer available from asus. If you go on the asus forums for this motherboard you should be able to find the proper techniques for downgrading your BIOS.Quote:
Originally Posted by >GSXR<mrbusa
i have them all.Quote:
Originally Posted by panfist
i already have right after i tested the 1002 i flashed back to 0805