View Full Version : odd 2.6c problem on IS7-E
chip will post as high as 3510 with only 1.525 however when i get into windows and folding the highest i can get is 3250mhz with 1.75v
any ideas? its being cooled by a slk900u with 80mm ~45cfm fan. temps are ok. load 3250mhz 1.75v is about 51c (mostly due to the high voltage).
this just seems rather odd that it will post so high with such low voltage yet isn't stable with high vcore until you knock the mhz down about 250.
a little more info:
i just checked my voltages. i'm using an Antec 350 on the rig. voltages are kinda low.
3.3 = 3.26
5 = 4.97
12 = 11.67
all of this in bios idle. kinda crummy. so much for antec making a decent psu. :rolleyes:
also this board seems to be limiting my cpu vcore. i have it set to 1.7 atm and idle is 1.65 load takes it down to 1.62-1.63. thats pretty sad. :(
bchur83
03-02-04, 09:22 PM
I have a 2.6C also, and it will also post and load windows up to 3.6 @ 1.6vcore, but it it only stable at 3.25 @ any voltage greater than 1.6v. Even at 1.675v, it is not stable at a higher speed. I think that many of the 2.6C's are not the greatest overclockers. This is on the rig in my sig.
thats so weird. this huge discrepency in clockspeed is baffling.
i've tried up to 1.9v and the highest i can get is 3300mhz. load temp about 54c. its mostly stable, but havent let folding run in case i were to lose work units.
i might swap out this psu for my 400w sparkle i have in another case waiting for my 2.8e.
bchur83
03-02-04, 09:41 PM
I don't know why it refuses to be stable at higher speeds and vcore. I have ruled everything else out except my PSU. That could be my problem, since it was a cheap Allied 450W from newegg. I would think that Antec would be a good PSU for overclocking, but who knows.
PiSan314159
03-02-04, 09:51 PM
And I thought it was just me. I've been able to post at 3.6, but everything is horribly unstable over 250 FSB. I just figured it was a case of n00by and let it slide.
Originally posted by PiSan314159
And I thought it was just me. I've been able to post at 3.6, but everything is horribly unstable over 250 FSB. I just figured it was a case of n00by and let it slide.
not a case of the nooby as far as i can tell :)
Are you guys able to run 3D??? Is it only unstable at CPU tests?
I just picked up a 2.60C that does 277 fsb at 1.525 volts, Memtest clean (L333Axx Malay). That's 3.6 gig and I am at my target. I haven't even bothered to take her higher. Strange thing is it can clear both 3DMarks and all my 3D stress tests, but kicks out of Aquamark. I get a 38 sec SuperPI 1M, but when I run the 16M test it locks. Never seen this happen before. Usually you get the rounding error or some error box, but a freeze??? It passes Sandra Arith/Multi, and PCMark2002 fine. 'Guess I just need to bump her up to 1.55 volts.
The thing benches all my 3D so I got no complaints. Another one I have (China) does 3.5 gig at 1.68 volts, stable in everything.
bchur83
03-02-04, 11:16 PM
Is that prime stable. That is what I am basing as a stability test right now.
Is that prime stable. That is what I am basing as a stability test right now.
Why don't you try benching the 3DMarks to see if they are 3D stable? Takes less time than Prime for an hour :). I think they probably will be 3D stable up to maybe 3.4 gig.
Well these CPUs are not guaranteed stable 'cept for 2.60 gig + maybe some headroom, say to 3 gig. They are presumably defective in some way above that or they would be put in 3.0-3.4Cs.
oddly enough i just finished about a 1 hour session of ut2k4 demo at 254fsb with 1.75v no problems at all. i started up folding after disposing of the game and it immediately locked up..
wtf? :confused:
lbbo2002
03-03-04, 03:11 AM
I have 480 TT Antec PSU (2) and a Fox 550. So I know it’s not the PSU but the CPU. The IS7 under volts .025 - .03 but I have compensated for this.
I couldn't get the 2.6C stable (was a Q Mayla) at greater than 3.3 under any conditions.
Since it sucked so bad (compared to 2.4 ,2.8, 3.0 chips) I just moved it out the door and on to a 3.0 priming at 3.7 (247) w/mem at 1:1. If I could find a 2.8 that would do 260-265 that would be comparable to the 2.4 I have.
I think it’s the chips, sucks I know as its is not what you wanted to hear.
Any one with an 04 2.6 doing any better?
You are getting about the right speed. Average overclocking of 2.6C is in 3.3GHZ range. A little higher or lower are both normal. If it's stable in UT2k4, then it should be ok in most cases. Unless you want this system folding only. If so just lower the FSB a few MHZ, it can't be that much different.
My 2.6C used to be able to 100% stable at 258FSB with nothing could crash it at that speed not even Prime95. Maybe it's getting dusty, or maybe it's getting old, now this standard is down to 256FSB, which nothing can crash it, but I still can play games at a little higher FSB. Yours aren't that much different.
There is nothing to worry about. After all you have your all powerful Athlon 64 for gaming.
P.S. I see you grabbed a 2.8E to play with as well. How is the resault?
and yes it is that much different. i can be at 252 and fold perfectly. move to 255 and it immediately locks up. there is no inbetween with this chip. odd. not really important since its not my main rig i don't really care how it oc's. plays ut2k4 demo with my 9600pro stock very smooth with most details all the way up.
???? = getting it thursday/friday :)
The_man27
03-03-04, 01:06 PM
Since we all seem to have the same problem (me included), I found it weird that we all have ABIT motherboards (except #18)...
Check out this link and tell me what you think: http://forum.abit-usa.com/showthread.php?threadid=13877
There is huge debate going over at ABIT forums. They are talking about the inability to go over the 250FSB mark, and the debate is quite lengthy...
PiSan314159
03-03-04, 01:53 PM
I'd rather it be my chip than my mobo.
Besides, I have no trouble posting, it's just unstable at that speed.
Does newegg allow returns based on dissatisfaction?
Since we all seem to have the same problem (me included), I found it weird that we all have ABIT motherboards (except #18)...
I don't think it's the Abit boards, it's the CPU. I have tried pretty much all the boards from Abit, Asus, DFI, and Epox and I can tell you I have the highest regard for Abit boards when it comes to extreme fsb in the 300 fsb area. One of my 2.40C MO steps does 302 on the IS7 (yet to test higher), and another 300 on the AI7 - on air. I have not had to use ClockGen or off fixed PCI bus. A third is doing 299.6 on water on an IC7. The CPU can make your board look like a champ or a chump.
I fixed the problem on my 2.60C :D. Just needed a bump in VCORE. I was getting greedy with default voltage. 1.536 actual was enough to clear the 16M SuperPI test and Aquamark at 277 fsb (3.6 gig). On the CPU database here, I am #6 I think and one of the highest (if not the highest) on air, and I haven't even tried for more. On top of that, I took the chip out and found out I hadn't used enough AS3 and the center of the CPU was not making contact with my Alpha!
I have tried three other 2.60Cs and they did 3.3 at 1.75 volts (China), 3.50 but at 1.75 volts (Malay), and 3.5 at 1.68 (China). I got rid of the first two. That's it for me with the 2.60Cs and 2.40Cs.
My 1 GB ADATA PC4000 max out around 275-280, 1:1 so I wanted a 2.60C which maxes out in this area at around 3.6 gig. That's good for 3500-3700 unbuffered, albeit at 3-4-4-7 or 2.5-4-4-7. Makes for one sweet rig, though. The 38 sec SuperPI 1M time on this 2.60C at 277 is the same I get at 300 fsb on a 2.40C. Reason why is I run 277, 5:4 at 2-3-2-5 on the IS7 with SR1 enabled, which nets me almost 3400 unbuffered! I can barely do this with my XMS3500 at the 2.8 VDIMM limit of the board.
PiSan314159
03-03-04, 04:57 PM
My voltage is already at 6.5 (6.25 actual). How high did you push yours to get it further? I'm a little paranoid about the S.N.D.S and I'm trying to prevent from destroying my chip.
6.5? you mean 1.65?
i've pushed my chip all the way up to 1.9v which is too high really, but i wanted to see what would happen.. unfortunately.. nothing gained. :(
lbbo2002
03-03-04, 08:34 PM
Same here I max voltage@1.9 with nothing gained but heat.
hi,
have yours tryed setting the last two gat's to dis ?
me too, have first much trouble with high fsb.
reading larva's postings and yepp. THX
with gat setting to
auto-auto-auto-dis-dis
i can run my 2.4C m0 up to 300fsb on water.
setting the vagp to 1.60v.
later you can the first auto set to turbo, pat is now fully enabled.
but over around 260fsb pat is to hard for running, so i stay with auto. (ram @ 5:4 = ddr480)
There is no generic profile for GAT settings: it depends on both the ram and the board. In general, you need to disable the last two settings.
For example, the IS7 and AI7 will run GAT settings like SR-Enh-A-D-D with BH-5. This enables partial PAT and boosts unbuffered memory scores. However the IC7 and IC7-MaxIII do not like this and you have to run A-A-A-D-D. I have NOT been able to run any GAT settings but AUTO on the Abit 875P boards with BH-5. Now this applies to my two 512 MB XMS3500. They are not shabby as they are capable of 250, 1:1, 2-2-2-5 in DC given the right volts (Epox boards).
For OCZ PC3700 Golds Rev. 1.0, same thing as BH-5, except I can only run them 1:1 at 2-3-3-6. Damn sticks will not run async on Abit boards but have no problem on Asus boards.
I find with Hynix PC4000, I can run A-A-A-A-A at 1:1 but I have to change to A-A-A-D-D for 5:4.
You really have to do your own experimenting. Memtest will give you a rough dial-in, but as always, the proof is in the pudding and make sure you bench the 3DMarks to check stability.
i will try it with the last 2 disabled.
lbbo2002
03-04-04, 03:41 AM
Clevor
What abit boards you running the OCZ 3700 on? The IS7 with my rev 1 3700 Gold El's run the mem dividers just fine. Have you had any issues running 1:1 above 446DDR and 3d graphics corruption?
As for the GAT settings I always have the last to GATs disabled and it was not use, it just stopped. Notda, end of story.
Originally posted by Maxvla
not a case of the nooby as far as i can tell :)
erm, I'm not sure the evidence exists to support that conclusion :D
A 350W power supply (any that I have seen) is overwhelmed by the 12V draw generated by operating a P4 well in excess of 3GHz. It's not that it cannot work, but you cannot expect it to and should expect it to limit the oc.
As well, if you've been running the last two GAT fields on A that is problematic in nearly all cases.
Originally posted by larva
erm, I'm not sure the evidence exists to support that conclusion :D
A 350W power supply (any that I have seen) is overwhelmed by the 12V draw generated by operating a P4 well in excess of 3GHz. It's not that it cannot work, but you cannot expect it to and should expect it to limit the oc.
As well, if you've been running the last two GAT fields on A that is problematic in nearly all cases.
with a sparkle 350w i was able to run my 2.4b @ 3.5ghz no problem. this psu doesn't seem to be near the quality, so it could be limiting me. still haven't tried with the last 2 gat's off. just haven't had time yet.
just tried with last 2 off and no gain at all. still stuck at 255fsb as my max sandra benchable speed. complete stability is still 250. :(
What abit boards you running the OCZ 3700 on? The IS7 with my rev 1 3700 Gold El's run the mem dividers just fine.
Up until what fsb? Believe me, I have flogged them all: IS7, IS7-E, AI7, and IC7. The IS7s are Rev. 1.2. I have not tried much below 280 asynch, but last time I tried as low as 250 on the AI7 I still could not boot.
From Bigtoe's comments he's made on these sticks, there may be two sub-versions of Rev. 1.0, an early type and late. I have both and notice a difference in the labels. I've only been flogging the early type, I think. 1:1 works fine to 241, 2-3-3-6 at 2.8 VDIMM, but boards won't boot at 5:4.
What does the label on yours say? Does it say 2-3-3-7 with a long s/n? I've only tested the version which states no CAS settings.
As for the GAT settings I always have the last to GATS disabled and it was not use, it just stopped. Notda, end of story.
In my experience you can boot, but 3D will be unstable. This is quite evident regardless of the fsb, even down at 250. I have not found any increase in bandwidth with it enabled anyway.
Originally posted by Maxvla
with a sparkle 350w i was able to run my 2.4b @ 3.5ghz no problem. this psu doesn't seem to be near the quality, so it could be limiting me. still haven't tried with the last 2 gat's off. just haven't had time yet.
My sparkle 350 sagged all the way to 11.43V on the 12V line at 3.42GHz and 1.71V. It ran stably, but the supply was obviously overloaded. I had to run the 48cfm fan in wide open to keep it stable. The Fortron units are indeed better than the SL series Antecs, but in any event I recommend a TP430 or the 530 Fortron as a practical minimum for P4 users expecting to operate in excesss of 3GHz.
Do try disabling those GAT settings though, it almost always produces a usable gain in stability.
Methodx
03-04-04, 04:58 PM
I am in the exact same situation. I got up to 3.5ghz posting but it wouldn't get stable until I got down to 250 fsb, just like you guys! And that is with the system in my sig.
Originally posted by larva
My sparkle 350 sagged all the way to 11.43V on the 12V line at 3.42GHz and 1.71V. It ran stably, but the supply was obviously overloaded. I had to run the 48cfm fan in wide open to keep it stable. The Fortron units are indeed better than the SL series Antecs, but in any event I recommend a TP430 or the 530 Fortron as a practical minimum for P4 users expecting to operate in excesss of 3GHz.
I have better psu's but they are all in rigs atm. i don't feel like spending 70 bucks for a 530 just to (possibly) run a little higher clockspeed on an unimportant computer. 3250 is fine.
Do try disabling those GAT settings though, it almost always produces a usable gain in stability.
Originally posted by Maxvla
just tried with last 2 off and no gain at all. still stuck at 255fsb as my max sandra benchable speed. complete stability is still 250. :(
PiSan314159
03-04-04, 09:11 PM
6.5? you mean 1.65?
Ah, my n00b is showing.:-/
All this evidence seems to point to the chip. I'm thinking about ripping out my old 1.7 to see if anything else is lowering my oc.
However the 1.7 was bought with a dell so I don't know much about it. It was purchased September 2001 so it may not even have a 533 FSB. Would it be possible to stick this sucker into my setup?
Originally posted by Maxvla
I have better psu's but they are all in rigs atm. i don't feel like spending 70 bucks for a 530 just to (possibly) run a little higher clockspeed on an unimportant computer. 3250 is fine.
I hear ya, and there is no guarantee a better supply would alleviate the problem. It's just the general trend towards bigger supplies that generally becomes mandatory in the 3.2-3.4GHz range. If the cpu is just poor no amount of ram or power supply thrown at will change its fundamental nature.
Originally posted by larva
I hear ya, and there is no guarantee a better supply would alleviate the problem. It's just the general trend towards bigger supplies that generally becomes mandatory in the 3.2-3.4GHz range. If the cpu is just poor no amount of ram or power supply thrown at will change its fundamental nature.
agreed. the trend in this thread has shown this chip to hard lock at 250fsb so i'm inclined to think its the chip and not the supporting hardware.
Methodx
03-05-04, 11:30 AM
Yeah, a better power supply will not help. I know because I've tried.
I hady a crappy Powmax 400w that wasn't providing enough power to the rails, so I thought this was my problem.
I went out and got a Fortron 500w with adjustable rails and I'm still stuck at 250 mhz!
Maybe its just the chip?
Originally posted by Maxvla
just tried with last 2 off and no gain at all. still stuck at 255fsb as my max sandra benchable speed. complete stability is still 250. :(
mmm, with an p4m (here to test on my hands) i have same prob as yours.
setting 245 fsb in bios an check in win with setfsb the stability.
after this i set the found stable-fsb in bios and noch luck on win-load.
i tryed some ram/gat settings... nothing.
so i have finally succes with bios setting 245fsb and setfsb with commandline in autostart.
... try this.
with my 2.4c m0 i have not this problem. :nut:
so possibly the is7-e have an "stability-hole"...?
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.