it may just be my boards here, but 2 are doing exact the same. I set 1.67V on the 4-16 and I get 1.8V at the dimm slot, i tested with 4-25 and 1.8V gets me 1.89 so its still not working as it should be...i shot a video to show Jimmy so we wait to see what he says.
4-25 is a good clocker though and is worth a shot if you want to play ;)
I used Normal Mode and left all the strength settings at default. The SPD of my mem says it should be 1x for all stregths but that did not post.Quote:
Originally Posted by Rammsteiner
I had CS/ODT Pin Fine Delay On i'm unsure about Bank Swizzle mode.
I had used 2.3V for the mem and also used 75ns tRFC.
I tried to install debian 64bit today at 2.8GHz/1.325V and it froze at those settings. Our cpu's seem to be very identical in behaviour.
The bios allows an offset for the cpu sensor in the range -8°C - +8°C. I use Auto (means no offset is used). The sensor reads ~5°C lower temps in idle and up to ~8°C higher temps under load. Everest display the sensors labeld correct if i use an DFI bios with an Sapphire bios it show's cpu temp as aux and PWM temp as cpu. I tried the 9850e in the M3A with the same cooler and get similar readings there +-1°-2°C. After a few trials with TIM I got the temps down about 5°C so all my testings run below 60°C cpu.Quote:
Originally Posted by KTE
Last time i tried that java benchmark in sandra i got an "out off memory" error, I used the jre coming with oo-2.3.1 also tried the latest sun jdk with the same result.
Yeah, for you this is a wise decision. I had no memory timing and voltage problems so i had to try it out. With my up to 2.2V rated mem i get very good readings now. It seems to be a different thing with 2GB modules rated for lower voltages.Quote:
Originally Posted by Brother Esau
All the freezes i get now do not seem to be related to the 425 bios. It was just a cooincidence. I had not done wise testings at 2.9GHz before. That 2.8GHz setting however was rock solid while i used the sapphire 415 bios so the next thing i will try is this bios again.
Thank you for the awardflash switches you posted in the official thread. They are all required to switch between dfi and sapphire.
Achim, Is that you in the Avatar?
It's howie, the jerk from the "FALL GUY". You remember? :ROTF:
http://eightiesclub.tripod.com/e1fd4330.jpg
I've got 4-25 in now and it destablized like I've not seen before. Apps are now crashing which I think is a voltage issue. The problem I'm seeing with setting the VID is that what I set is not what I get. 1.28 In bios and I'm seeing 1.248 in CPUz and 1.3 in AOD. I think 1.25 is what it is running based on the temperature.Quote:
Originally Posted by Tony
Like I said I am not dedicating anymore testing and time until they get this Voltage table on these Bios sorted out with it constantly overvolting:shakes:
I flashed the Sapphire 415 bios and i'm still having the freezing issues at 2.9GHz. So i decided to play with ref HT.
CPU: 12.5x225 = 2812.5MHz at 1.3625V
NB: 10x225 = 2250MHz at 1.3375V
MEM: 533*225/200 = ~600MHz at 2.12V
EDIT: Settings reflect my current stable setup and i'll modify it if it turns out unstable.Code:>HT Link Control:
HT Link Width......................... - 16 bit
HT Frequency.......................... - 1.8GHZ MHz
IH Flow Control Mode.................. - Disabled
HT Link Tristate...................... - Disabled
2x LCLK Mode.......................... - Disabled
Unit Id Clumping...................... - Disabled
>CPU Feature:
CPU DID.............................. - NA
CPU FID.............................. - NA
CPU-NB DID........................... - 1
CPU-NB FID........................... - 06
AMD CPU stepping..................... - P0
>DRAM Configuration:
MCT Memory Timing.................... - Manual
2T Mode.............................. - Auto
Cas Latency Time..................... - 5
Pre Charge Delay (tRAS).............. - 24
DRAM RAS# Precharge.................. - 7
DRAN RAS# to CAS# Delay.............. - 7
Bank to Bank Cmd (TRRD).............. - Auto
TRC.................................. - 31T
TRTP................................. - Auto
Write Recovery (TWR)................. - Auto
Write-Read Command (TWTR)............ - Auto
TRFC0................................ - 105ns
TRFC1................................ - 105ns
TRFC2................................ - 105ns
TRFC3................................ - 105ns
Trdrd................................ - Auto
TrwtT0............................... - Auto
Twrrd................................ - Auto
Twrwr................................ - Auto
CS/ODT Pin Fine Delay................ - Enabled
Bank Swizzle Mode.................... - Enabled
Auto Tweak Performance............... - Enabled
Optimal Performance Mode............. - Enabled
>ODC Control Mode:
DRAM Drivers Weak Mode.............. - Normal
Output Drivers Comp................. - Normal
>Channel A:
CPU On-die Termination.............. - 300 ohms +/- 20%
DQS Drive Strength.................. - 1.00x
Data Drive Strength................. - 1.00x
MEMCLK Drive Strength............... - 1.00x
Addr/Cmd Drive Strength............. - 1.50x
CS/ODT Drive Strength............... - 1.50x
CKE Drive Strenght.................. - 1.50x
>Channel B:
CPU On-die Termination.............. - 300 ohms +/- 20%
DQS Drive Strength.................. - 1.00x
Data Drive Strength................. - 1.00x
MEMCLK Drive Strength............... - 1.00x
Addr/Cmd Drive Strength............. - 1.50x
CS/ODT Drive Strength............... - 1.50x
CKE Drive Strenght.................. - 1.50x
>PWM Settings:
PWM Control.......................... - Enabled
CPU VID Control...................... - 1.36250v
CPU Vid Special Add.................. - Auto
CPU NB Vlotage Control............... - 1.33750v
CPU NB Vid Special Add............... - Auto
DRAM Voltage Control................. - 2.12v
SB PLL 1.2V Voltage.................. - 1.20v
NB Core Voltage...................... - 1.250v
NB PCI-E Voltage..................... - 1.120v
NB HT Voltage........................ - 1.270v
Genie BIOS Settings:
Timing Mode.......................... - Manual
Memory Clock value or Limit.......... - DDR 1066
PCIE 3,4 State....................... - Auto
PCIE 2 State......................... - x2
CPU TLB Cache Fix.................... - Disabled
AMD K8 Cool & Quiet control.......... - Disabled
CPU Clock Ratio...................... - Auto
Spread Spectrum...................... - Disabled
CPU Host/HTT Clock................... - 225mhz
CPU Host/HTT Boot Up Clock........... - 225mhz
CPU Host/HTT Adjustment Gap.......... - 5mhz
PCIE Clock........................... - 100mhz
w00t,
240Mhz HTT is pretty much stable. I ran Prime95 for 14 hours and 30 minutes. 2.76Ghz CPU, 2.4Ghz NB, 2.16Ghz HT and DDR960.
Now testing timings since my CPU REALLY has an 'anywhere near or over 2.8Ghz fobia'. Which sucks, but nothing to do about.
Been testing with timings and subtimings. Although some subtiming makes one core crash (1st time it was 3rd, 2nd time it was 2nd). But with just 4-4-4 and TWR (or however it's called) at 4 forced trough BIOS it seem stable thus far.
Still I stand my point, I would have prefered higher CPU clocks though. I think it's still strange 2.76Ghz is stable and at any Vcore around 2.8Ghz and over is a no go.
Justapost has more or less same problem, looks like we indeed got a crap stepping:(.
And oh yes, Im wondering... How do we check whether Vnb is actually being applied or not? There's no way, AFAIK, to check Vnb so Ive been wondering. ATM I use Special Add option to be sure.
Lol, thank you for the update rammsteiner. It made me realize that im on 2250MHz HT speed atm. ;)
Still trying to install 64bit debian here. It worked once with above settings but froze on the second try. Trying to stabilize it with higher CPU and NB VID's atm.
In 1066mode 225 seems to be the stable limit here. 227 boots but freezes soon.
I use relaxed timings and try to find the lowest voltage for the mem at this frequency. With the 3-07 bios i got 4-5-5-15 timings up to ~580MHz at 2.3V during first tests.
Nice!
I think using 4 DIMM's is ruining some RAM clocking for me at this point. Although I dont mind, Im very close to settings I used on X2.
Some timings applied via Memset make the system BSOD while I can boot up fine if it's done by BIOS. And some BIOS timings aren't being applied in real:confused:.
My NB seems to be clocking fine then, I just dont get the poor CPU results really. How much I'd love to understand, I just dont. I can imagine too many Volts, too high temps or the wall of a stepping (like ~3.5~3.6Ghz on F3 Windsor) would stop a CPU clocking more at a certain point. But we all know and see people hitting nice clocks with B3 Agena, so there's no wall. But Voltage scaling is just... not there. I only added a little bit of Vcore to make sure tighter RAM settings wouldnt make the CPU crap over.
Im actually starting to wonder if it's possible that with a certain batch a bug occures between BIOS and CPU? Like Vcore isnt being applied. On the otherhand I think it is since temps rise with more Vcore.
[EDIT]
http://en.expreview.com/2008/04/24/t...helf-tomorrow/
I demand DFI releases this for AMD! And switch mounting holes for the heatsink so RAM coolers actually can be used without interfering with the TRUE.
Quote:
Originally Posted by justapost
Achim......I hope you are not experiencing the corrupted Bios Phenomenon that I had a couple of months ago. Sure sounds like it though! If thats the caase then what was stable proven settings for many hours Prime or otherwise would fail within minutes and no longer be stable. If you are experiencing that let Me Know and I will re give the very long winded procedure thats involved in sorting it out. Let me know if I can help:yepp:
Overall this memory tweaking does not have an big impact on performance but it's fun to play with. In opposite to X2 I recognized Max Async Latency has an big impact on performance. It's value gets optimized with some pattern testings during startup. I recommend you relax MAL if you play with other subtimings.
Debian installed fine here with 1.3625V CPU-VID and 1.3325V NB-VID. I tried to run the phoronix-test-suite and the system froze during a compilation test (build-imagemagick). So I lowered the ref HT to 215MHz to see if it's stable here.
Thank you for the offer Brother Esau,Quote:
Originally Posted by Brother Esau
I verified my 2.8GHz/2.2GHz setup with the lowest voltages i found 1.3125V/1.25V. It primed fine for ~5hrs. But it froze during wise testing. I did not test normal application usage very often during my lowest voltage trials so i think i just discovered that problem now. :)
Can I call you Howie?:D JK.....why don't you put you're Mugshot up instead? That is unless you are on the back of a Milk Carton somewhere (Have You Seen This Man):rofl:
Well, I miss MAL options in both BIOS and Memtest. There's Maximum Read Latency at 54 or so. It freezes at 50~51. But indeed this setting has a big impact, and MAL had a huge impact on X2. That's why I ran RAM 1:1 to get tighter MAL. With a devider I could only get 6~7, with 1:1 4ns was the best I could.
Ill try upping multi from 11.5 to 12 for CPU sometime and check whether Special Add would help better than normal Vcore.
LOL, i start to like this avatar. But it seems i get some of it's karma since i use it.Quote:
Originally Posted by Brother Esau
Howie is some sort of insider joke in the hip-hop scene here and he's and 80's guy. I also like stocius's avatar.
You are right it's MRL on phenoms i mixed that.Quote:
Originally Posted by Rammsteiner
215MHz ref HT passed the phronix-test-suite here I'm testing with 220MHz now (CPU 2750MHz).
EDIT: 220MHz passed - Results
Most likely you'll be able to add another 2Mhz and top out there for CPU speed:(.
I made my way up in small steps 222,224,225MHz and i'm benching under 64bit debian since an hour without problems :confused:Quote:
Originally Posted by Rammsteiner
Just left a watermark on the phoronix-suite with an 430s kernel build time here.
Less than half the time all others got. :) (I'm the only one with a quadcore in there)
225MHz Result
Gonna run memtest now and plan to find the lowest voltages for this setup this week.
EDIT: Ran a full memtest serie. Then booted into winxp and it froze during 3dmark06 install while priming. :( Enough for today.
Thanks.:)Quote:
Originally Posted by justapost
JP, in 425 (which I found out is beta) I saw a warning that NB VID must equal CPU VID. I wonder is this is what you are actually seeing as unstable?
I'm back on 416 and stock. I can run programs that load the system be every now and then when I start a new program or just surfing I get a program crash or a rare reboot. I just got finished building a cad project and running the simulator back and forth between screens and did not see any problems.
CPU VID = 1.292. (CPUX 1.262).
NB VID = AUTO
RAM on 1.9 stock.
All others Volts and settings AUTO.
I'm beginning to think these need 1.275 to 1.325 to run but some of you have seen lower numbers with a DVM. I'm looking for stability for now until we know more of what is going on. I want to believe both the CPU and MOBO are OK just the numbers are not where they need to be.
Oh ffs, this is more scary then bad actually. You seem not to be stabler after all... so do I:eek:. It ran so many things ok! And then, re-installed XP64 and after install I boot up and it freezes... Trying your 220Mhz method now woth a 10xNB multi. RAM at DDR1066...Quote:
Originally Posted by justapost
Im going mad really.
Pheww! that was close, nearly hit the buy AMD Phenom Quad Core 9850 buttton
So what addition does this post make?Quote:
Originally Posted by Final8ty
9850's are AWESOME, but you got top be lucky enough to get least not the stepping in my sig. Didnt see any bad performing other steppings yet.
My 0809EPMW is also a bad clocker. 2.8GHz @ 1.27V is max. 2.85 is unstable even at 1.47V. NB does not like more than 2.2GHz....Quote:
Originally Posted by Rammsteiner
I'll order a new one soon.