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

5000+: Can't Hit 3.2GHz

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

Caviman2201

Member
Joined
Dec 30, 2003
Location
MD
I have a 5000+ brissy that I can't get to touch 3.2GHz...

I have it stable at 240x13 @ 1.4v but can't get to 247. I tested 247x7 at a slight increase in chipset voltage and it worked just fine. I also have the memory running a 2:3 divider so its not the memory. I've tried all the way up to 1.5v and can't even get it to boot to Windows. Temps at [email protected] are topping out at 44C...

I know there's no tangible difference between 3120MHz and 3211MHz, but I just wanna go to 3.2GHz

Any suggestions?
 
Last edited:
You might want to try a burn in trick. Stay with your current oc, but up the vcore and leave it at 1.5 and up your chipset another bump or two. Run it like this for a few days, then try to get to 3.2ghz. Sometimes this works and lets you get past that seemingly invisible wall, sometimes it dont. Its worth a shot, and as long as your temps are fine you should be good.
 
What are the maximum volts I should throw at my NB and vCore? I was thinking about staying under 1.5 w/ CPU, but not sure about my NB.

My NB is passively cooled:
321_12_full.jpg
 
since your chip is 65nm i would stay under 1.6vcore if your temps are good, and for the chipset, as long as you have good case cooling and they dont get to hot, up that voltage to what you think is safe.
 
I'm thinking I really shouldn't have to go any higher on the chipset voltage because i was able to run 247x7 without any problems at all at 1.2v (up from 1.15). The thing that worries me about going too high on volts is that the MOSFET (I think) heatsink is getting very hot even at just 1.5v. I'm talking about the other black heatsink right behind all the I/O ports
 
What are your thoughts on trying something like 320x10 and running a 2:3 memory divider so the mem would run at ~213MHz @ 5-5-5-15? I know the RAM will do 230MHz @ 5-5-5-15 w/ 2.0v
 
Well if the sinks get hot put a fan blowing on em if you are that worried about the heat and up that chipset voltage more man at least 1.35 or higher. It will not hurt you to try different timings, RAM speed, and cpu multipliers. The point of burn in isnt to run it like that for 24/7 opperation, but to help you get over a hump.
 
I will try the burn-in as a last resort. Folks over at the Biostar forum are telling me I'm going about it all wrong and that to OC a brissy on this board, its best to attack it with a high reference clock and lower the multiplier. I just don't know what kind of chipset voltage is considered maximum... I tried booting with the following setup and it did not work.

FSB: 300
Mult: 10x
HT Multi: 3x
Mem Div: 2:3 (266)
Mem Timings: 5-5-5-15
vCORE: 1.45
vChipset: 1.3
vDIMM: 2.0 (memory was running at a 2:3 divider (300x.6666 = 200)
 
You want to ease into your clock, 5-10 mhz at a time checking for stability and temps as you go. Jumping straight to 300 fsb if you have been running a lower bus probably won't post.

My clock on my TA770 A2+ is:
Vcore 1.462v
HT Volt Reg 1.25v (default)
NB/SB Volt Reg 1.15v (default)
HT Link Width 16bit
HT Link Freq Auto
CPU Clock 260
Cool & Quiet disabled
NPT FID (multi) 12.5

Don't worry too much about the HT being over 1000 as that was mostly 939 and 90nm chips. You should be good up to 1400 as I've seen clocks run stably with that HTT. Mine runs at 1300.

Keep an eye on your temps as I am running a TRUE and my temps may be lower than yours.
 

Attachments

  • untitled.JPG
    untitled.JPG
    42.9 KB · Views: 196
Really man you are just trying to find the max speed you can get out of your chip and since you have a BE just do multiplyer overclocking then fine tune with the mem dividers and try to keep that RAM at stock speeds and what not. If you know what your RAM can do then feel free to tweak a little with it but try not to since you want to keep that out of the equation till you find out how high you can get that chip. Once you know the max of the chip, then you will want to mess with upping the RAM speed as doing so usually yeilds more performance if you get it high enough it can be like an effective extra 100mhz oc.

Burn ins are usually a last resort, or when you hit a wall that really puzzles you. I had the same with my chip, it hit a wall at 2.65ghz, but i burned it in at 2.64ghz with 1.7vcore and 1.6v chipset for a day or two, then i was able to get over that hump to 2.7ghz and higher.
 
I don't have a BE - thats the problem...

I feel like mine is slamming into a wall right at the 3.2GHz barrier... I've tried all kinds of combinations of FSBxMulti...

I'm about ready to call this one... a 520MHz OC isn't too bad... I got that along with tighter timings on RAM than I paid for... I can't complain.
 
You're right about your clock not being bad. I don't think being a BE means anything except the multi is unlocked.They might be better binned, I don't know. I do know no two chips will necessarily clock the same. Before this BE I ran 2 different x2 3800+ in the old Abit board I had, and one was definately better than the other. I am running a lower multi than default to get the speed up on my Ballistix.
 
Yeah, I guess nobody can really complain about a 20% overclock... now I'm gonna try to get my RAM and 8800GT running as fast as possible...

Do you guys think a high FSB speed on the RAM is more important than a mediocre FSB speed but tight timings?
 
Test it and find out. I havent played with the DDR2 stuff, but i do know that the older athlons with ddr mem controllers loved low latency vs higher speed, but with ddr2 im not so sure. Just test and benchmark and find out.
 
I haven't seen much boost with my Ballistix much over 1000-1100 MHz. Tighter timings close to 1000 gave me the most benefit with my 3800. I need to upgrade my case to improve the airflow before I push this BE and get my ram up to the speed I want.
 
Having the same problem, im stable at 3.0 right now, as soon as i try 3.2 i run prime 95 for not even 10 seconds and it errors
 
your ram could be holding you back i had to go down on fsb and up the multiplier, and i had to go up on the nouth's v and i got it to 3.2 at 1.4v
 
Unfortunately I don't have that option - to up the multi.

Thing is, about the RAM, I'm running a divider of 5:6 when I try to go up to 247FSB which takes the memory down to 205MHz. I have been running 231x13 for a while now with a 1:1 divider so I KNOW the memory will do at least 231MHz.
 
Back