3700+ overclocking :(

Dup

Dup

Dup

Soldato
Joined
10 Mar 2006
Posts
11,313
Location
East Lancs
I posted a thread and a follow up elsewhere but no luck, but you guys seem to have more indepth knowledge so I'll quote my posts here. Any help would be really great!

I've been messing with OCing my 3700+ all day but I'm not having any luck in any CPU stress tests.

The CPU will run stock 2.2Ghz, 200fsb, 5x HTT just fine and will frime for a few hours. My memory will complete one full run of tests in memtest86 at stock timings.

As for overclocking, I cannot get my CPU to pass even 15 mins of Prime 95 at all. It will however complete 1M of superpi upto 2.6Ghz

At the moment I have got it to 2.6 Ghz with the following settings:

235x11
1.525v vcore
1.8v DDR
1.5v Chipset
1.2v HT Voltage
HTT 4x
2.5-3-3-7 RAM timings.

I have seen the 3700+ San Diego OC much higher and I have seen my ram go much higher on stock tighter timings.

At the moment everything seems fine, apart from the stress testing. I can run games ok and temps go to 40C full load, idles at 32 minimum. I don't see any instability, yet prime fails almost instantly.

It might not be the best thing to rely on, but OCCT etc do the same. Superpi has thrown up some errors with tighter timings and lower volts etc but it will do !m at this present time.

Do I have a bad chip? Its a CABGE stepping. I have also used a ram 166 divider which runs the ram well below its rated spec, which rules out the RAM for me.

Any ideas?
Click to expand...
Well, have tried everything. cheers for the help but nothing is working

I set my ram timings to 3-4-4-8 and prime starts then stops immediately.

I've tried using auto vcore and auto HT Multiplier, no joy

I have run the mad mod mike demo, I have run sandra burn in once and played FEAR on full settings for an hour and had no stability issues. I do realise this does not mean my PC IS stable but then can I trust a few pieces of software?

I'm thinking my CPU has a bad memory controller, or just isn't upto it. I might start over and find the max values for everything and piece it all together again.
Click to expand...

EDIT - Specs might help!

Antec Neopower 480
Asus A8N SLi Premium
A64 3700+ Sandy (sat under a Thermalright XP-90C and Panaflo 120mmx35mm fan)
2GB (2x1GB) Corsair PC3500LL Pro (2-3-2-6)
XFX GeForce 7900GTX 512Mb Extreme Edition (675/1700 core+mem)
2x 120GB Maxtor Diamondmax 9+
 
Last edited:
If everything else, including a full overnight run of metest, superpi, fear etc are stable, then its quite likely it might simply be a bug in prime95, what version do you have as I seem to remember an issue with prime a while back when torture testing amd/newer cpus.
 
Yeah, I'll have to check my prime version. I have brought up errors in super pi but I am running tight timings on the RAM to casue that. 1M runs fine, upwards of that can poke up some errors.

I have a thought that it might be my PSU. The Neopower is rated 15A on the first 12V rail and 18 on the other. I might try removing everything I possibly can and retry prime with the latest version.

My list of upgrades is going through the roof now. If it is the PSU I will need to spend £100 on one of those, I want the Swiftech Apex Ultra kit at £150 and I want a £430 NEX 20" widescreen monitor, I also want a dual core opty at about £250. All that on a 12.5k a year job... my hobbies are too much for my bank balance!
 
A few things,
- The CABGE stepping is iirc one of the bad ones that requires more volts to run stable.
- I need to run my memory 1:1 with my SD 3700+ anything else is unstable
- use HTT 3x (might not help but speed difference is minimal)
 
Yeah, I meant 2.8v.

I'm intent on running 1:1 on the RAM in any eventuality, but I've been trying dividers for the sakes of getting prime to run stable for at least an hour.

My mobo will only allow me to go to 1.55v vcore.

Just been home and tested with everything unplugged and just running the bare system, didn't work.

I have now set my timings to 2.5-3-3-6 and changed some other timings, get the vcore to max 1.55 and set DDR voltage down to 2.6.

Prime ran for 2 mins.

Unfortunately the cat got in the open case and half unplugged my power cable to my boot hdd and now I get NTLDR is missing, whoops! Shame the damn thing didn't get sucked into my 120x35 panaflo over the CPU... Hopefully I can repair the boot sector or the sata cable itself is loose too, but back at work now :(
 
Cheers Dutch guy! That works great, I also found another tool which is supposed to be a little more demanding.

I've been running it for 30 mins now with my new settings and everything has gone smoothly. It's called S&M. You can down load it here:

S&M 1.7.6

It has pushed my temps from 27C cold boot to 43C load. I'm at 2.6 GHz at 1.55V.

What I changed was the vcore to max 1.55v and the HTT to 3x instead of 4x.

Now I will try lowering the VCore to a happy place and begin to push the MHz again until I run into problems, then I think I'll be happy!
 
Dup said:
I have now set my timings to 2.5-3-3-6 and changed some other timings, get the vcore to max 1.55 and set DDR voltage down to 2.6.
Click to expand...

You could probably drop your timings to 2.5-3-2-5 without any problems. Spoken to a few people with the same RAM and they have all been able to:)

Up to 230MHz i can get 2-3-2-5 and from 231MHz to 254MHz I can get 2.5-3-2-5. Again, most other people have been getting similar results:)
 
Yeah, at this speed I can't boot unless I'm at 2.5 Cas. I'll not drop them yet, I'm going to push to 2.8 at 1:1. That'll be 255x11, 512Mhz on the RAM. if not, 2.7 will do me fine, 2.8 is just a milestone.

Been running SP2004 for 20 mins now without hickup. I think it was the HTT.

Thanks for all the help guys. Nice to have a friendly croud to geek out with.
 
You must log in or register to reply here.
Back
Top Bottom