OC-ing woes- testing max CPU? venice 3000+

[dopke]

Well I just started testing max CPU this morning and I am now at the point where I'm not quite sure whether to settle at the FSB that I got as the max CPU. Or go for higher-- I thought max CPU should be higher than 260ish? Where should I go next to get higher max CPU? I tried upping Voltage to CPU, didn't help. I went up to 1.3V +123%. Do I need to go higher?

 

The highest FSB I can go w/o failing prime95 or OCCT is 258FSB. The voltage I'm giving the chip is 1.3 Startup, 1.3 CPU VID, and 110% Special Vid.

 

I dropped the memory to 100FSB (1/2) divider to supposedly throw that out of equation. Here's my bios.

 

Amd64 Venice 3000+ 0518DPDW.

 

FSB: 258

LDT/FSB: 2.5

CPU/FSB: x9

CPU Start up: 1.3V

CPU VID: 1.3V

CPU SPECIAL: 110%

LDT Volt: 1.4V

Chipset Volt: 1.7V

DRAM Volt: 2.7V

 

DRAM Freq: 100 (1/2)

CPC: Enable

CAS: 2

TCL: 2

TRAS: 7

TRP: 2

TRC: 7

TRFC: 14

TRRD: 5

TWR: 3

TWTR: 2

TRWT: 5

TREF: 3120

TWCL: 5

DRAM Bank interleave: Enabled

 

DQS Skew Control: Increase Skew

DQS Skew Value: 0

DRAM Drive Str: Level 7

DRAM Data Drive Str: Lvl 4

Max Async Latency: 7.0 Nano sec.

Read Preamble Time: 5.0 Nano sec.

IdleCycle Limit: 256

Dynamic Counter: Disable

R/W Queue Bypass: 16x

Bypass Max: 7x

32 Byte Granularity: Disable(8 Bursts)

[Tecnolover]

I've settled at running mine at 2.61mhz because the jump to a stable 2.7 even though it's capable requires significantly more voltage to get stability at least for my venice 3000+. The small gain in performance from 2.6 to 2.7 isn't worth the additional voltage it needs which in turn will also lower the cpu's lifespan.

 

I run at 2.6 with 1.56v and have sandra/prime cpu stability

 

To just get cpu 2.7 Pi 2m stable i have to bump up to around 1.73v.

 

That is too big of a jump. Not worth it.

[Jimbo Mahoney]

Hey dude,

 

Default voltage for Venice is 1.4v and this board undervolts by 0.025 - 0.05v.

 

I would just use the CPU VID setting and leave the special alone for now.

 

Put both the startup voltage and CPU VID to 1.4 and see how high you can clock.

 

Then when you fail a test or get a BSOD, set both to the next value up, 1.425.

 

Repeat this until the temperatures get too high, but if you aren't using CPU VID special, you can only go to 1.55 volts anyway.

 

I find beyond 1.525 volts, you start to get really very little for a lot of volts increase, as Tecnolover has pointed out.

 

I have always run my system with max LDT and Chipset voltages too.

 

I also always use an LDT multi of 3x.

 

Also, if it helps, here is a table of the Tref values you should be using:

 

http://homepage.ntlworld.com/james.waynewright/Tref.html

 

Obviously, they are based on the memory speed, not the HTT.

[dopke]

Hey dude,

 

Default voltage for Venice is 1.4v and this board undervolts by 0.025 - 0.05v.

 

I would just use the CPU VID setting and leave the special alone for now.

 

Put both the startup voltage and CPU VID to 1.4 and see how high you can clock.

 

Then when you fail a test or get a BSOD, set both to the next value up, 1.425.

 

Repeat this until the temperatures get too high, but if you aren't using CPU VID special, you can only go to 1.55 volts anyway.

 

I find beyond 1.525 volts, you start to get really very little for a lot of volts increase, as Tecnolover has pointed out.

 

I have always run my system with max LDT and Chipset voltages too.

 

I also always use an LDT multi of 3x.

 

Also, if it helps, here is a table of the Tref values you should be using:

 

http://homepage.ntlworld.com/james.waynewright/Tref.html

 

Obviously, they are based on the memory speed, not the HTT.

 

I read that the newer processors liked to run with lower actual volts on CPU VID and adding more special, than having a higher CPU VID and lower special. I think it was AG that said it in one of the sticky's?

 

In any case, I just tried these, and BTW I was running x3 LDT already, not x2.5. But after these changes I couldn't boot into windows, I dropped LDT to x2.5 and now I'm in. I'm testing it again at these volts with OCCT right now at same FSB (258) since the CPU volts are actually lower than when I did 1.3V + 110%.

CPU Startup: 1.4V

CPU VID: 1.4V

CPU Special: Auto

LDT Volt: 1.5V (max)

Chip set Volt: 1.8V (max)

 

Well, was in middle of OCCT-ing and computer just restarts. That means there was an error right? I'll up the volts now.

 

Edit: That didn't work at all. Upped startup and regular Vid of CPU voltage up to 1.5V and couldn't get into windows. Either BSOD or locks up.

[dopke]

Went back to original settings for now. Weird thing was that with LDT Voltage and Chipset Voltage, I couldn't get into windows, but if I dropped both of them .1V (my original volts) I could load windows. BTW my original settings I primed for 8 hours.

[willyrybone]

I had to loosen up my ram timings to get mine running stable at those speeds ... you loose some bandwidth, but most seem to think more MHz will be better in the end

 

i am still tweaking to try to get better ram timings. So far i can only prime for four hours. If i get it working i'll update the values in the OCDB

[dopke]

I had to loosen up my ram timings to get mine running stable at those speeds ... you loose some bandwidth, but most seem to think more MHz will be better in the end

 

Well right now I'm just trying to find my max CPU/FSB- and so I dropped the DRAM to 100MHZ (1/2) to take that out of the question right? I can leave the timings that I have above (2-2-2-7) if divider is 1/2 right?

[bread]

Well right now I'm just trying to find my max CPU/FSB- and so I dropped the DRAM to 100MHZ (1/2) to take that out of the question right? I can leave the timings that I have above (2-2-2-7) if divider is 1/2 right?

 

Should be fine. Run a couple of quick passes of memtest test #5 to be 100% sure.

[Randallw23]

My 3000+ venice runs stable full time at 292 with my ram running at somewhere around 236 2-2-2-5. My superPI 32m was like 27m 26s, and 1m was right around 30s. I havent touched it since i got it stable at 292 and its been running pretty much non-stop for about a month. no errors. try to get your cpu voltage around 1.48v thats around what mine is. oh and for sure flash to the 510-2 beta bios if you havent already.

[Jimbo Mahoney]

I read that the newer processors liked to run with lower actual volts on CPU VID and adding more special, than having a higher CPU VID and lower special. I think it was AG that said it in one of the sticky's?

 

Interesting, thanks...

 

Not sure what's going on with your system though.

 

A different BIOS is a good idea, although 310 wasn't THAT much different from 414-3 for me. 414-3 is better for my system though. Not tried the 5 series...

[dopke]

Should be fine. Run a couple of quick passes of memtest test #5 to be 100% sure.

 

Running memtest #8. Have had 3 passes already. Decided to flash to 310P first.

 

It seems like I am missing something? No matter how high I put the voltages, I can't get past 260FSB. I went up to 1.6V and couldn't get through to boot windows.

[Jimbo Mahoney]

I don't think this is the problem, but worth a try:

 

Are you using XP SP2?

 

If so, edit the boot.ini file (on the root of C: )

 

Change this:

 

/noexecute=optin

 

to this:

 

/execute

 

 

This will turn off DEP which seems to be causing SOME low-end venice owners (like me) problems when clocking high.

 

It only seems to help SOME people though...