BD7 Boot, RAM & OC'ing Probs

Wow I finally just visited this forum section and I think I've been posting in the wrong sections here at O/C Forums. So please mods, forgive me one more time for posting in this section also. If you must delete my duplicate threads, please delete all of them but this one. This one seems to have me covered the best.

Before I paste my previous post, I gotta say I noticed I have some of the similar cold booting problems with my BD7 Raid as has been mentioned here. I just thought it was part or due to the problem I am listing below. I also have problems doing a warm boot, which I haven't seen referenced here yet. Sometimes I have to restart 3 to 5 times before it will go through my BIOS. Most times it just stops with a black screen at startup, showing nothing, and not beeping. I ALWAYS have to cut off the compter at the power supply and then back on, as the reset button and off/on button on the front panel has no effect. Anyway PLEASE read the rest of my problem, concerning my OC'ing below. It is pretty good for consideration, considering the other posts I read here tonight. BTW my BD7 Raid, in the BIOS has core voltage options all the way up to 1.95v. I have mine set at 1.9v and in PC Health and Hardware Dr. it shows around 1.83v. Anyhow read the below:

Stuck at 118 OC; Why?
I hope I'm not out of line by reposting here, but I couldn't make any headway in the memory section. If I'm posting wrong here, I'll understand the mods moving or deleting my thread. Anyways here it is again.

Hi All,
I have a problem concerning oc’ing my ram that I can’t figure out at all! I’ve done a fair amount of oc’ing, but I’ve never came across a problem like this. Some of you here will remember me from December, when I built this system. Batboy will I know.

I have a P4 1.7 ghz on a Abit BD7 Raid mobo. I also had the same P4 on a BL7 Raid before the BD7 Raid I have now. So the problem also ties in to that mobo.

With the BL7 I used 712 meg (3x256 sticks) of Corsair SDRAM PC133 CAS2 ram. I could overclock this ram to 127 fsb, giving my cpu 2159 mhz. BUT I could never start the ram from 3:4 BIOS setting, which would have made the fsb 133 to start with. To get the 127 fsb, I had to start my ram at a 1:1 setting, starting my fsb at 100 mhz. When I had to run the ram at 133 fsb, it limited my overall oc to about 118 fbs on top of the 133, giving me a total oc of about 151 fsb. I ran this setting because the memory bandwith was greatly increased to about 1200 in 2002 Sandra benchmarks. CAS settings at 151 fsb was 3-3-3, without precharging. I could run these settings at 2-2-2 with precharging, starting at 100 fsb.

About a month ago I went to a Abit BD7 Raid DDR mobo, of which I had hoped I could increase my fbs, getting both higher cpu oc’ing as well as higher bandwith. But it was not to be the case. First I used some Crucial 2100 DDR ram I had. I Had run this ram before at 159 with no problems. However once again, with this Crucial ram I was limited to practically the exact same oc’ing numbers/mhz as I was with the Corsair SDRAM. With a 100/200 fsb I could oc the ram about 27/127 mhz, or about 154/254 mhz. Or if I started at 133/266, I could oc my ram about 18 mhz ontop of the 133/266 settings for DDR. I kept my BIOS CAS settings the same for all ram oc’ing here, at 2.5-7-3-3.

This past week I got some *supposedly* real good oc’ing ram. I got 2x256 sticks of PC3000 OCZ with copper heatsinks. This ram *supposed* to oc to 200 mhz! To make a long story short, I am still stuck with the same oc’ing numbers as I was with my Corsair and Crucial ram. Matter of fact this OCZ ram may bit a bit more finicky. I can ONLY run this ram with the following BIOS settings. 2.5-6-3-3. Anything else, regardless to oc’ing or not will not run With this ram. I can start my fsb at 100 and oc my cpu to about 127. or I can start my fsb at 133, and oc my cpu about 18 mhz on top of the 133, giving me a Total fsb of, again, 151 mhz…..

My big question or problem here is; I know my cpu will oc to 2159 with no problems, but it will only do it if I start from a 100 fsb. My ram *supposed* to go much higher than this, but it won’t if I start at 133 fsb! I have also tried turning my ram voltage up, but that has only helped marginally, if at all. I have also tried the *FIXED PCI* settings, and I still can’t get any better results! So what is the problem with my ram and setup here? I am at a loss!

Thanks for reading such a long post!
Mike

welcome to the forums. i was, am having similar problems with the mobo. i have kingmax ram, and someone else on the forum is running it at 175 fsb, with a gigibyte-8IRXP, which is coming to me this week, but i can only currently run my fsb at 127 . i am using a beta bios right now, and i am using the speed settings. i noticed that my northbridge was getting real hott, and i put a Corb on it yesterday, and while the temps still read hott, i can now run stable at 2550. if you get to 2159 with your cpu, that is killer, considering that the willo's dont overclock very well. i bought the abit board for the fix pci option, but i am beginning to doubt the overclockibility of this board. i have read someone else running a 155 fsb on this board, but i cant acheive it . if you find out any tips for this board, let me know, but as of right now, i am disappointed with this board.

Hey jd,
From what I've read, and have understood in this forum, the Abit mobos are actually defective, or at least crippled by either programming in the BIOS or actual hardware. I had spent the better part of three months tweaking and wondering what my problem was. I have asked several times in other places, but have never really gotten any answers. Even in Abit's newsgroup, I have not gotten the first single reply to my questions! I am afraid we are stuck with half or porrly performing mobos here, and I seriously doubt that this will be corrected. I read in one post here that Abit has been aware of the problem, yet they have not released any official BIOS upgrades for our mobos since December.

What gets me as mad or madder than anything else, is all these reviews that showed how high they oc'ed their Abit BL and BD 7's! Personally I have never trusted reviewers much, and I know I now don't believe the first thing they have wrote about these boards, concerning their claims of they oc'ed them to the 170 range. Well for that matter I don't know if I believe anything else they wrote about the boards either!

I think your mention of the northbridge being hot is pretty much normal. I have read and seen myself, that they can get easily get hot enough to burn you. And yeah I think I got lucky with my Willy. But my brother has the same setup and cpu as I and he can do the same. Although we got the last stepping series that was available for the Willy, and that made oc'ing these cpus the easiest of all the Willy's.

Cheers,
Mike Lamb

that really sucks, but why can those people with the th7-II's get up so high? i wonder.

I've had my BD7 upto 166mhz so far, i can probably go higher as the cpu is still only on 1.625v, so they do overclock well.

Anonicle said:

I've had my BD7 upto 166mhz so far, i can probably go higher as the cpu is still only on 1.625v, so they do overclock well.

Click to expand...

you got a lucky board, could not go any higher than 127 on mine, bummer.

Anonicle said:

I've had my BD7 upto 166mhz so far, i can probably go higher as the cpu is still only on 1.625v, so they do overclock well.

Click to expand...

You no doubt had your memory set for DDR200 or 1:1 because the BD7 has a slight misrepresentation of the memory speed when you set it to DDR266 or 3:4. When you set it to DDR266 with the CPU at 100 then the memory is running at 133mhz. No problem there. But as you increase the FSB the memory speed doen't increase the same as far as mhz goes. Say you go from 100mhz FSB to 120mhz. You would think that the memory speed would then be 153/306mhz right? Wrong. The memory speed is really 160/320mhz. If you set you FSB 10 125mhz then your memory moves to 166/333mhz. If you set your FSB to 133mhz your memory will be running at somewhere around 185/370mhz.

This little oddity may account for why some can hit the higher FSB setting with no problems while others can't get anywhere near them. It's all in what speed you have the memory set to. So if you want a high FSB then make sure you keep your memory at the same speed.

I'm aware of the memory thing, i have some Samsung PC2700, its good for over 200 fsb, but yes, running at 166 i only have the ram on 1:1, i dont think the ram could handle running at over 220!

Anonicle said:

I'm aware of the memory thing, i have some Samsung PC2700, its good for over 200 fsb, but yes, running at 166 i only have the ram on 1:1, i dont think the ram could handle running at over 220!

Click to expand...

I was just passing that on for others who may not have known it. It may help them is seeing why they can't get higher with their FSB settings.

Just would like to share my personal experience of my BD7raid and 1.6a northwood. First off, owning the BD7raid is like in a love-hate relationship or playing russian roulette. Just the other day, I absolutely love it when it booted up on the very first try (especially enlightening after getting back from a long day work).
Apparently there's another issue with my BD7raid besides the boot problems. In WinME, it can't resume from a standby state..the monitor remain blank however I might jerk/press the mouse or tap on the keyboard. Haven't tried it on XP or 2K. Anyone else have this issue?
I'm currently using the beta bios(dated 1/11/2002) with HPT 2.31 bios. Although it didn't solve the #@^%$@* boot problems, I did notice marginal improvement while cold booting(or it might be me just hallucinating). Also memory settings seem to be better: now at 2,6,3,3 @ 158FSB, 1:1 configuration on my Kingmax DDR333 @ 2.6volt. Yep that's right,...CAS 2 @ 158 bus..
Allow me to share my experience/ordeal on how I've hit 2536MHz ROCK STABLE with the 1.6a woodie:

I'd actually purchased the CPU(week 51/Malaysia) three weeks earlier than the rest of the components; BD7raid, 2x 256MB Kingmax DDR333, AVC Sunflower, Arctic Silver3, generic casing + 350Watt generic P4 PSU. The rest of the parts are transfered from my old setup; SBLive, MSI Geforce2MX, Aopen DVDROM 16/48x, Teac CDRW12-10-32, 46GB Maxtor ATA100. The long wait to setup the entire system was due to the lack of decent HSF sold here in Malaysia for the P4. Even as I write this, there's no Thermaltake Volcano 7+, Swiftech, Alpha P4 solutions available here yet

The joy when I laid my hands on the Sunflower....it's time to kick into 5th gear and see what's there. But the highest I could get was a disappointing 138FSB equivalent to 2208MHz(ram settings at 2,5,2,2 : 1:1ratio). Vcore is at max 1.625 at that time. This setup is by far the WORST in cold/reset reboots. Then after scouring various forums trying to solve this I've finally decided to do the vpin mod trick on the CPU. Instead of tying the wire on the 1.6a, I'd inserted the wire on the socket itself(much easier). Voila, 1.7v...actual 1.62-.1.66v, but it didn't solve the boot problems even at then 138FSB. Frustrated and !@#$ed off, I went straight for 150-170FSB range. Amazingly it even POST at 2720 @ 170FSB but crashed almost instantly. Then I went for 2600 @ 162FSB, it went to windows and was pretty stable...the only test that it failed was PRIME95(I also found that running PRIME95 solely increases the CPU temp by another 3-5C). Last I settled for 2536 @ 158FSB. All this are at 2.5,6,3,3 ram settings.

The detail Softmenu3 settings are:

FSB = 158
FIXPCI = 33

vcore = 1.70 (default after vpin mod)
ram = 2.6

Standard BIOS setting:

Halt on = No errors (usual is All but keyboard)

Advanced Chipset setting:

RAM = 2, 6, 3, 3 (manual)

Mainboard temp = 42-43C, CPU temp = 60-64C @ Max. load PRIME95 torture test

General assumption/experience:
After vpin mod at pin 3 & 4 which makes the default at 1.7v, it's better to leave the the settings as it is, ie..1.70v. Any values higher does not seem to improve the woodie overclockability. Lower settings is a definite NO-NO...boot problems!!

The beta BIOS seem to improve the mem timings and cold boot issue?

Constant burn-in seems to improve the cold boot issue?

Still waiting for the day if ever ABIT will solve the boot problems..

quote:
"I'm aware of the memory thing, i have some Samsung PC2700, its good for over 200 fsb, but yes, running at 166 i only have the ram on 1:1, i dont think the ram could handle running at over 220!
--------------------------------------------------------------------------------
I was just passing that on for others who may not have known it. It may help them is seeing why they can't get higher with their FSB settings."

Well I can go to 127 fsb, before I start locking up, if I set my ram clocks at 100 - 1/1.

However, I set my fsb ram clock to 133 - 4/3, AND then I can only oc my ram to 20mhz MAX. (Usually keep it at 118 so it boots easier.)

Even though I can oc my ram's fsb more at 100 - 1/1, that still is not enough to make up for starting at 133 and then adding another 18 steps on top of that. Still, I know my cpu will easily do 2159 without problems, yet I am stuck at 2040 (max) because my ram won't go higher using the 4/3 - 133 setting. And I have used good ram, that should easily handle more oc'ing than what I am now using.

Mr. Natural, hope you wouldn't mind explaining the quirks of how you figured out the final oc numbers for the ram. Cuz, for example if I am starting at 133 and add 20 mhz/steps to my ram, I am in fact running a total oc for my ram of 153 mhz. So what has Abit done to change this and mislead everybody? OTOH, even if I was actually running 160 or 165 fsb with my ram, I still should be able to go higher with the ram I have.

Thanks,
Mike

What is happening for some odd reason even tho you may have the AGP/PCI clock set to 66/33 the memory clock follows the FSB+PCI setting which means that the standard 1/3 setting of the PCI to FSB of 133mhz then memory will be 4/3 FSB ao as you add the higher FSB speed your PCI 1/3 clock gets added onto that which means if you have a 4/3 setting for 100mhz then your memory is at 133mhz (ie 100mhz*4/3). Now as you add the higher FSB speeds you are still working with the /3 divider of the FSB so that means say you change to 133mhz like you are saying. That means 133*4/3 or 133.33/3*4 which works out to around 178mhz. Now take in to account the variations of the clock itself which can vary by 1/2mhz then what you get is 133.83/3*4 which now works out to about 179mhz. This can vary consideribly from board to board. I've found my clock can vary upwards of .5-.6mhz which can make a big difference in the end clock speeds.

Duh MrNatural,
I guess I'm a little dense today, so don't think I'm getting on you or anything. But exactly how did you come by 178? I just ain't getting it. Are you saying with me running *133*4/3 x 18 or 20 oc'ed fsb, that I am running 178? If that's the case; why can't I go higher? Still, with the fixed pcu/agp, high overclocks still shouldn't affect my hard drives, sound cards, slots, etc... Right?
Like I said, I know my cpu will do 2159 (17x127), and I know my ram *supposed* to be good to at least 190 mhz. Or is the problem what I was explaining in my first post, starting the thread? I'm just lost here.

Thanks,
Mike

Mike360000 said:

Duh MrNatural,
I guess I'm a little dense today, so don't think I'm getting on you or anything. But exactly how did you come by 178? I just ain't getting it. Are you saying with me running *133*4/3 x 18 or 20 oc'ed fsb, that I am running 178? If that's the case; why can't I go higher? Still, with the fixed pcu/agp, high overclocks still shouldn't affect my hard drives, sound cards, slots, etc... Right?
Like I said, I know my cpu will do 2159 (17x127), and I know my ram *supposed* to be good to at least 190 mhz. Or is the problem what I was explaining in my first post, starting the thread? I'm just lost here.

Thanks,
Mike

Click to expand...

What I'm saying is that the quotion of 4/3 increases as the fsb increases. basically you can say 1*4/3 = 1.333... Now if you figure that for each mhz of FSB then what you would get is this. 133mhz*1.333...= 177.72889. Now you follow what I'm saying? The fact the AGP/PCI may be fixed will have no affect on how the memory will change in speed.

MrNatural,
O!
gotcha!

Thanks,
Mike

Hi, I'd like to add my own experiences to this torrid affair if I may.
So far I have O/C 4 BD7 non Raid boards with the same level of success as the rest of you. All have had good 1.6 Woodies, and all of them have been up to 2.4GHz at 1.5Volts, but all on a hit and miss basis, when there up at 137fsb giving 2.2G's they are rock solid when they boot so the processor is definately not the issue here and the memory most certainly is, the BIOS memory setting are always on DDR200(or 1:1 as it should be stated).
The one thing I've noticed and I think is key here is to see what the BD7 does at the stock 100fsb....on first boot it will use DDR200 memory timing....but on subsequent boots it will make a unilateral decision to add 33Mhz (or the PCI bus speed) to the memory timings giving DDR266 and states it so upon boot. It may also be that instead of +33Mhz that it is using the 4:3 mulitiplier to the memory timings. At around 115MHz fsb it decides to pull back the +33Mhz adder or the change the 4:3 back to 1:1, but not always. And here is the dilema. Depending upon which of the two options it is doing will give an fsb of either 148 or 153Mhz both of which are starting to fall outside of resonable expectations of standard PC2100 memory. Anything higher on the fsb like 120MHz takes you up to 160MHz and a definate no no for most of us, but sometimes the board decides to use standard timing, no adders or multipliers and you end up with "in-spec" memory timings and a good boot, and any utility will show you what you expect to see for the memory fsb, but when we have a no-show boot, we cant tell what its decided to use as there is nothing to show for it!!!! I have tried the new beta 7H Bios and it doesnt help at all, except clear up the memory multiplier ****-up, but it does allow higher voltages without the need for Vpin modding. So I think harrasing Abit for a fix on this is paramount as us mere mortals do seem to be at there mercy once again.

Today I reached 133fsb with mem. settings at 3:4
I'm using the beat bios allowing vcore settings above 1.65
I found that using fix pci is no good for me.
I set fsb 133 , divider 1:4 , cpu dram 3:4
vcore 1.75 and ram voltage to 2.6
The ram is now running at 354mhz , this must be wrong , the ram should now run at 333mhz.
Abit must now get their finger out and fix this.
I'm using corsair xms2700 ram

cjordan61 said:

Today I reached 133fsb with mem. settings at 3:4
I'm using the beat bios allowing vcore settings above 1.65
I found that using fix pci is no good for me.
I set fsb 133 , divider 1:4 , cpu dram 3:4
vcore 1.75 and ram voltage to 2.6
The ram is now running at 354mhz , this must be wrong , the ram should now run at 333mhz.
Abit must now get their finger out and fix this.
I'm using corsair xms2700 ram

Click to expand...

Guess you didn't read my post just 3 before yours huh? I explained it in simple terms. When you set your ratio to 3:4 that means you are clocking your memory at 1.333mhz for every 1mhz of FSB. This means when you set your FSB to 133mhz (ie 133.33) and you memory ratio to 3:4 then you are setting it to 133.33x1.333...mhz=177.78 which in terms of DDR means 354-355mhz. Now you see why it's at that speed?

bd7r mem timing

Thanks for clearing up that. But still , I wished they change that so it's 1fsb = 1mhz mem with 3:4
It just sounds odd like it is now

regards cjordan61

Gigabyte BIOS to cure POST 26 Hang

Anybody noticed that Gigabyte have posted a new BIOS for the 8IRXP Mobo, that cures the following "Fixed setting VID items then power on/off, system sometimes hangs up at POST26h". Sound familiar to anyone....!!! So if Gigabyte can spot it why not Abit..