Neo2 Platinum & 3200+ Venice OC

[Stokes]

I haven't really checked how much air that fan is actually moving. It's staying at a constant 1600RPM and if i remember right its in the range of 36CFM which isnt the greatest. I'd rather keep this fan for a screamer and sacrifice a little for the quietness.

Prime95 Just ran stable for 13 hours on these settings until I stopped it myself. Should I now go and try the tests for 5 min as I increase FSB to see what I can get more?

Yea I know 33% is a great overclock but i want more! lol (within reason as not to blow my computer up)

I basically went up to 1.56-58V when I failed after 3 hours on the 1.52V, and get this when I did that, the test failed after 8 minutes, which did no make any sense to me.

I also read in the other Venice 3200 thread that possibly lowering the fsb:htt down to 2x that might give a better overclock? But will that hurt system performance?

 

[bjc112]

Originally posted by: Stokes
I haven't really checked how much air that fan is actually moving. It's staying at a constant 1600RPM and if i remember right its in the range of 36CFM which isnt the greatest. I'd rather keep this fan for a screamer and sacrifice a little for the quietness.

Prime95 Just ran stable for 13 hours on these settings until I stopped it myself. Should I now go and try the tests for 5 min as I increase FSB to see what I can get more?

Yea I know 33% is a great overclock but i want more! lol (within reason as not to blow my computer up)

I basically went up to 1.56-58V when I failed after 3 hours on the 1.52V, and get this when I did that, the test failed after 8 minutes, which did no make any sense to me.

I also read in the other Venice 3200 thread that possibly lowering the fsb:htt down to 2x that might give a better overclock? But will that hurt system performance?

AS long as the LDT multi x HTT is above 500(1000) there is almost ZERO peformance hit..

 

[Stokes]

Originally posted by: bjc112

Originally posted by: Stokes
I haven't really checked how much air that fan is actually moving. It's staying at a constant 1600RPM and if i remember right its in the range of 36CFM which isnt the greatest. I'd rather keep this fan for a screamer and sacrifice a little for the quietness.

Prime95 Just ran stable for 13 hours on these settings until I stopped it myself. Should I now go and try the tests for 5 min as I increase FSB to see what I can get more?

Yea I know 33% is a great overclock but i want more! lol (within reason as not to blow my computer up)

I basically went up to 1.56-58V when I failed after 3 hours on the 1.52V, and get this when I did that, the test failed after 8 minutes, which did no make any sense to me.

I also read in the other Venice 3200 thread that possibly lowering the fsb:htt down to 2x that might give a better overclock? But will that hurt system performance?

AS long as the LDT multi x HTT is above 500(1000) there is almost ZERO peformance hit..

But will that make a difference in stability for overclock?

 

[Ike0069]

You can lower the voltage back to 1.52 and try the 2XHTT and see if that is stable.
Or just try upping the FSB and see if you can get 270 (or higher) at the current voltage.
Remeber you can always raise the FSB while lowering the CPU multiplier. 300X9 would be 2.7 GHz, and then lower LDT to 2x which would give you 600 HTT. That might help with stability.
It's all trial and error at this point because every system is different.

 

[Stokes]

After further testing:

When I put the voltage down to 1.56V and ran at 2.65Ghz, the rest ran for no longer then 2 minutes without failing. I tried going the 9x multiplier route with 2x htt and slowing moving up the FSB, right when I hit 280 FSB, the computer will not boot and goes back to 200FSB. It appears that 1.65V for 2.65Ghz is my best overclock producing 31C idle 56C load.

I probably just have to leave it at that and work on RAM now, unless there is something I am missing.

 

[Ike0069]

I can't think of anything you're missing. Try 270 FSB with everything else left as is and see if you can get that stable.
Then its time to work on the RAM which IMO can be a PITA because of all the variables. You have FSB and all the multipliers, plus you have to tweak all the individual timings. Good luck.

 

[Stokes]

Update on the memory:

I upped the voltage all the way to 2.8V
Frequency is 133
1T
2-3-3-5

Is what the computer is giving me on auto. For some reason the third value if I try to select it myself only start with 5 and go up.

I tried going up to 166mhz giving me DDR333 I believe, and when I did that it booted up just fine. I went into bios and bios was frozen. Windows would not load either.

Currently I'm finding stable overclock at: 2.67Ghz 267x10 with 1.63V
HTT: 2x

What am I doing wrong here?

 

[Ike0069]

First, the third number on the BIOS RAM timings for a Neo2 is the number you have listed as 5. The BIOS goes 1,2,4,3 when comparing to your 2-3-3-5 setup.

Now if you set it to 166 with the FSB at 267, your RAM is not at 333, but actually DDR 445. Yes, your RAM is DDR 500, but at CAS 2.5 I believe (correct me if I'm mistaken here); so you'll need to loosen the timings more, even though you are still below the RAM's rated speed.
Try going 2-4-4-8 (2-4-8-4 in BIOS) and see if that boots.

 

[Stokes]

Ike -

Took your advice tried 2-4-4-8

System failed memtest86 before it started

Changed timings to 2-5-5-8

Failed #2 in a heartbeat.

 

[Ike0069]

Try 2.5-4-4-8 and see if that works. If that still doesn't work, then I'm not sure what is wrong.
Duvie did say you might not be able to get your RAM to run CAS2 at anything above DDR 440, and you are now just slightly above that.
You can also try dropping to 260 FSB (would have RAM at and running DDR 433) and using 2-4-4-8.

 

[Stokes]

2.5-4-4-8 passed with flying colors on all tests.

So basically is that where I should leave it be? Is that my best option?

 

[Duvie]

try cas 2.5-3-2-7...most like the trcd one notch higher but the trp may be fine at 2....

then go cas 2.5-3-3-7 or 8 thru 10

then try cas 2.5-4-3-8 thru 10

these are all minor and wont result in much...this is "fine tune" tweaking...

 

[Duvie]

Did I miss it buit what bios??? I found 1.41 to be extremely stable with TCCD and my winchester...1.62 allowed me slightly bettter timings and 1.8v was really no different then that...I have never used any "official MSI bioses"...always mods...

 

[Ike0069]

He's using 1.8. I don't think he can go back to 1.41 because I doubt it would "see" his Venice core.
Actually, I'm not sure if anything before 1.8 works with the Venice.

 

[Duvie]

Originally posted by: Ike0069
He's using 1.8. I don't think he can go back to 1.41 because I doubt it would "see" his Venice core.
Actually, I'm not sure if anything before 1.8 works with the Venice.

Actually it does work with 1.62 for sure....

 

[Ike0069]

Originally posted by: Stokes
2.5-4-4-8 passed with flying colors on all tests.

So basically is that where I should leave it be? Is that my best option?

If you decide to stay at 2.5, you need to get the RAM speed up to make up for the drop in Latency. You should be able to run at least DDR 500 using 2.5.

Using Gogar's optimzer and 2650/DDR 500 as max speeds you should try:
294 x 9 = 2646
5:6 (166) x 294 = 240.54
3 x 294 = 882

Then you can maybe try and tweak the setting from there.

 

[Stokes]

Well interesting turn of events that just happened. As mentioned before the motherboard doesnt like going above 280FSB and I double checked every single setting from the fsb multiplier to cpu multi to ram etc.

Well I thought I would give it a chance and got it up to 290FSB and then bam, i up to 295FSB and it won't boot. I thought maybe i got it then that happens. So I try lowering the settings and nothing still. I put the settings to default and it wouldn't boot. Finally it booted running like 1.8ghz speed.

And windows went corrupt again so I am ghosting as we speak. I inched up towards 280FSB afterwords and it didn't like that at all. What could be causing this?

 

[Stokes]

Does this motherboard have a defect or is that just simply my max FSB?

 

[Ike0069]

Not sure, but I would expect it to at least reach 300 FSB.

 

[Stokes]

Anyone else have a solution for this problem?

 

[Stokes]

Just realized one thing, but not sure if this is related to the FSB problem.

My ram is in slots 3 and 4 running dual. The tweaking guide for the Neo2 calls for putting them in 1 & 2 which is the nearest to the cpu socket. Will that limit me at all? Thats all my research has come up with thus far.

And then using the +/- key to change the FSB, but thats more so it doesnt hang on you right, or could that be my other problem?

 

[Vegitto]

I'm not really sure. Try it!

 

[Stokes]

I'm at work right now so I'm contemplating what possibly could be holding me back. FSB of 280 is soo low. Even when I remove the cpu and memory out of the picture it won't go past that. And when I have memory and cpu in overclocked it still won't go.

Weird.

 

[Duvie]

Stokes...It may definitely make a difference as 3-4 for me was horriby unstable...

I have hit a max of 353 and most MSI boards have hit 300 a stated...

One could also try bumpingthe vagp in the outside chance it possibly controld the northbridge chipset like othr chipsets in the past...also may try removing NB chipset and scraping off the crap they installed and placing some AS5 or thermal paste of your choice instead...

 

[Stokes]

Originally posted by: Duvie
Stokes...It may definitely make a difference as 3-4 for me was horriby unstable...

I have hit a max of 353 and most MSI boards have hit 300 a stated...

One could also try bumpingthe vagp in the outside chance it possibly controld the northbridge chipset like othr chipsets in the past...also may try removing NB chipset and scraping off the crap they installed and placing some AS5 or thermal paste of your choice instead...

I'm hoping those ram sticks are whats stopping me. I really think I could hit 2.7ghz+ if I used a 9x multiplier.

I was actually going to up the agp voltage when I got back to 1.6V is what the MSI overclocking guide recommends. As far as removing the heatsink/fan on the NB. i've heard it can be difficult and possibly chip the chipset. But I'm pretty careful about that kind of stuff. I got some AS5 sitting around here. How do you suggest removing the thermal compound MSI uses?