• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Nice Chip, trying to get the best out of it. 2.66b (@3.46ghz, 1.8v)

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

RyanSG

Registered
Joined
Nov 19, 2002
System at max stable:
P4 2.66b (SL6EH) @ 3.46ghz (173fsb) 1.8volts (reading 1.79)
Giga-Byte GA-8PE667 mobo
Samsung 512mb DDR @ 346mhz, 2.5-7-3-3


I'm quite happy with how well its done so far but I have run into a couple of problems.

I need to raise the voltage to 1.8 to get the chip up this high and I've heard anything over 1.75 might be bad, can anyone confirm?

Idle diode temp reads 38-40, full load around 45.

I tried lowering my memory timings to 2-6-2-2 at 346mhz memory speed but it wouldn't post. This is odd because I left the Dram/Cpu ratio at 2.5 all the way up until 3.2Ghz (which means the memory was at 400mhz stable) at which point i could overclock no more till i lowered the ratio down to 2.0

Can anyone tell me which of the 4 timings is most beneficial or whether the timings are independant of each other in terms of overclocking.

Also assuming I can't improve the timings any is it better to run at 3.2ghz with 400mhz memory, or 3.4ghz with 340mhz memory (highest stable settings at 1.75)?
 
Also curious if anyone has done better then 3.46 with a 2.66 and aironly cooling.
 
No, 3.46 is with the memory set at 2-6-3-2 which the the best it will run stable at. It won't do any better then 3.46 with the default settings of 2.5-6-3-3 tho.
 
Ryan,
I curious about how far you got with stock voltage... and if you have any notes regarding how far you got with each voltage increase, that would be good info as well.

I'm seriously considering ordering a 2.66 tomorrow, I'd like to hit 3ghz but I hope not to jack the voltage up much, if at all.
 
Vcore: MaxStable Bus/Mhz(actual)

1.525: 155/3103
1.65: 160/3207
1.75: 168/3364 (3.4 appears stable @1.75 on benchmarks (sandra's burn in utility and 3dmark2001) but weird things happen like music skips once in awhile and i can't soft reboot)
1.8: 173/3460 (only ran briefly due to fear of hurting chip. Benched Sandra's cpu/mem tests fine, 3dmark 2001 also fine)
1.85: 176/3520 (same as 1.8)
 
With that much voltage you are just asking to fry the chip. I'd go no higher than 1.65 if you want to keep it long.
 
I'm pretty sure 1.75 is ok. I know for instance thats the maximum safe voltage listed in Intel's documentation, i.e. if you go over that you void your warranty. If it fried at 1.75 I would be rather surprised and get a replacement from Intel anyway ;)

However its quite true from everything I've read on these boards and elsewhere that anything over 1.75 involves a real danger of burning the chip out and leaving the equivalent of internal scorch marks that Intel might not replace your chip.
 
Hmm........

From what i got from SNDS is that people push their CPUs past 1.75 with extreme cooling or crappy cooling in the first place. This leads me to a conclusion that if you keep your CPU at 1.75 and a decent "average" temp, you should be fine.

Good Luck

Kam
 
RyanSG said:
If it fried at 1.75 I would be rather surprised and get a replacement from Intel anyway ;)

This is encouraging unethical and illegal behavior. We members of
O.E.M of overclockers.com try to discourage such behavior because it is wrong and illegal as well as driving prices up. Please refrain from encuraging illegal activity. If you want to participate in illegal activities that is your business :rolleyes: but please do not ask our other members to do so.
Thank you
 
hawk is right. kamilkluczewski and RyanSG you should really both look into P4 SNDS a little more. even i know that 1.75 is a no go with p4's on air or water - its just too hazardous.

And RyanSG you should know better, dont rma a processor that you killed because you didnt listen to us. thats plain wrong and you know it.
 
I didn't say RMA as in to the distributor. I said return under warranty. Unless I misread Intel's documentation, Intel seemed to say the warranty was void over 1.75, not at it. Granted, overclocking the clock rate probably voids the warranty anyway so Intel wouldn't take it back, but if you are talking strictly about voltage I don't see how setting it at 1.75 and returning it if it dies is wrong, its within spec limits.


Besides I really don't think anything wrong will happen to the chip at 1.75, especially considering the temp is now down to 36-38 Idle and 45 under 100% full load. And remember thats an internal diode reading which reads much hotter then the readings you get off the under-chip sensors you find on AMD chips or older intel chips. From what I've read as long as your chip never gets above 50 on the new p4 readings you are relatively safe.

Intel's Documentation on the SL6EH which I have says the fail-point is 73. Granted that the point at which it probably melts but I'm still way shy of it.

Do you know of any chips actually failing at 1.75 that were running cool? I'd be intrested to know, because Intel does get its specs wrong sometimes. Almost all chips I've seen people write about failing were either running hotter then 50 under load or forcing more juice then 1.75 or both
 
RyanSG said:
I didn't say RMA as in to the distributor. I said return under warranty. Unless I misread Intel's documentation, Intel seemed to say the warranty was void over 1.75, not at it. Granted, overclocking the clock rate probably voids the warranty anyway so Intel wouldn't take it back, but if you are talking strictly about voltage I don't see how setting it at 1.75 and returning it if it dies is wrong, its within spec limits.


Besides I really don't think anything wrong will happen to the chip at 1.75, especially considering the temp is now down to 36-38 Idle and 45 under 100% full load. And remember thats an internal diode reading which reads much hotter then the readings you get off the under-chip sensors you find on AMD chips or older intel chips. From what I've read as long as your chip never gets above 50 on the new p4 readings you are relatively safe.

Intel's Documentation on the SL6EH which I have says the fail-point is 73. Granted that the point at which it probably melts but I'm still way shy of it.

Do you know of any chips actually failing at 1.75 that were running cool? I'd be intrested to know, because Intel does get its specs wrong sometimes. Almost all chips I've seen people write about failing were either running hotter then 50 under load or forcing more juice then 1.75 or both
I'm sure if you read the warranty you will find running the chip over clocked voids the warranty. Why would one run higher voltage and not overclock?
Any way It is unethical to ruin a chip by running it out of spec and sending it back as defective. Regardless if it is the distrubitor or the manufacturer! If the chip is defective that is one thing but running it for all you can squeeze from it then getting a new one every time you fry one, that is wrong. Do you think the chip cpompanies owe you a new chip if you fry it?? This drives up costs for all. We are talking about XXnm sized transistors that are carefully designed to specific voltage threshold. I have not seen P4s running too hot. What fries these chips is voltage. You say 1.75 is ok? Well how sure are you that your Vc readings are being measured correctly?
I would suggest you go study some of the design parameters and critical dimensions and you will understand how over-volting effects the transistors.
I'm not saying this just to hear my self talk these are facts that one need to properly OC a chip without frying it. I have all my pc's overclocked and have never fried one.
 
Well, I don't understand the exact electrical engineering of the chip, no. I leave that up to Intel, and Intel tells me 1.75 is the maxmimum safe volatage so I believe them. If it turns out 1.75 is not safe for the chip then I do think they should take it back and change the specs.

I've never friend any of my other past/present dozen or so chips either.

That aside it turns out I'm running mine at 1.65v (actual1.61-1.64) at 3.2ghz now because I prefer having my memory run at 400mhz rather then being forced to reduce the muliplier to 2.0 and run it at 342.
 
Do you know of any chips actually failing at 1.75 that were running cool? I'd be intrested to know, because Intel does get its specs wrong sometimes. Almost all chips I've seen people write about failing were either running hotter then 50 under load or forcing more juice then 1.75 or both

what about the chips that die with 1.65v or less and vapochiled...

it's time to read the SNDS "sticky", as well as past posts.

yet I think that most chips at that voltage might have died early anyway.

mica
 
Hawk said:
If it fried at 1.75 I would be rather surprised and get a replacement from Intel anyway ;)

This quote is what concerns me. You are giving advice that could lead to unethical behavior and that is not what we here are about, I do beleive we try to overclock with out frying chips and if we do we should understand that once we overclock the chip is ours and the warranty is void.
If you are not willing to go to 1.75 why would suggest to Takumi it's ok?
I think this is going no where. I just wanted you to know that there are some of us here that overclock and understand that it means you VOID the warranty.
I hope you change your thoughts on the honosty of having someone else paying for your hobby.
Thats all I have to say on the subject but here are some thoughts of other members.
:)
 
Back