*Semi-Official* San Diego OC Results thread

[GuitarDaddy]

Originally posted by: Painman
Prime's been crunching away for 16:25 now at 2805/1.52. Max load temp 51. This speed seems pretty easy for this CPU, add 55 MHz and things get hairy. I may try walking the HTT up one step at a time and see what's actually between 255 and 260... a steep slope, or a brick wall?

Looks like our chips are pretty much identical, if so, its a steep slope. I can get SuperPI 1m runs from 2805-2822, but SuperPI 32m and prime will error on anything higher than 2805. The max vcore on my board gives 1.57-1.58 per CPUZ, The guys with the DFI
boards can only seem to get another 50-70mhz with vcore over 1.6 (some as high a 1.8), and their temps go way over 60c.

 

[Xed]

Got my 3700+ today, its a 0515 cabge too. Will see what I can do with it tomorrow when my dfi board arrives

 

[Painman]

Originally posted by: GuitarDaddy
Looks like our chips are pretty much identical, if so, its a steep slope. I can get SuperPI 1m runs from 2805-2822, but SuperPI 32m and prime will error on anything higher than 2805. The max vcore on my board gives 1.57-1.58 per CPUZ, The guys with the DFI
boards can only seem to get another 50-70mhz with vcore over 1.6 (some as high a 1.8), and their temps go way over 60c.

I got 2816 to Prime for 2 1/2 hours before it failed on the same 1.52 volts, so the slope starts there. I guess that makes up for my mistaken estimate of 2860 for something doable on reasonable core, I made an educated guess that 1.52 would power 2805 and I was right on the money

I'll see if there's any more to be squeezed out of the chip as a "max" OC but I don't think I'll feed it more than another .05 volt. Whatever that extra MHz might be, I don't imagine it'll be enough to justify the extra stress and heat on the chip for day-to-day. 2.8 on 1.52 looks like a sweet spot, now I have to bring RAM back up and hope nothing drastic has to happen to keep the SD happy.

Originally posted by: Xed
Got my 3700+ today, its a 0515 cabge too. Will see what I can do with it tomorrow when my dfi board arrives

Good luck :thumbsup:

 

[lytalbayre]

Could be helpful to someone. Asus has a new Beta BIOS out for the A8N-E (Bios 1004.002)

I installed it... Doesn't seem to have much affect other than a placebo (I now think my system is a little more stable)

Also, tested 275mhz memory with 2T timings and 10xcpu multiplier with 1.45 v to the cpu. This is actually stable...

Interesting find is that the score on 3dmark 03 with 2.75Ghz cpu/275mhz memory with 2.5-4-4-8/2T is equivalent to 2.8ghz cpu/230mhz memory with 2.5-3-3-6/1T memory... not sure which setting I should keep.

I'm gonna try to oc with 2T timings to get my memory and chip around 2.85ghz/285mhz... I'll let you all nkow how it goes.

Thanks

 

[Painman]

Originally posted by: Xed
Got my 3700+ today, its a 0515 cabge too. Will see what I can do with it tomorrow when my dfi board arrives

I just remembered something that might be important for you - I was running BIOS 310 with my DFI, and with the SD installed, my Command per Clock (1T/2T) option vanished from the BIOS menu. I flashed up to 414-3 beta BIOS and it returned.

 

[bobov]

Originally posted by: GuitarDaddy

Originally posted by: Painman
Prime's been crunching away for 16:25 now at 2805/1.52. Max load temp 51. This speed seems pretty easy for this CPU, add 55 MHz and things get hairy. I may try walking the HTT up one step at a time and see what's actually between 255 and 260... a steep slope, or a brick wall?

Looks like our chips are pretty much identical, if so, its a steep slope. I can get SuperPI 1m runs from 2805-2822, but SuperPI 32m and prime will error on anything higher than 2805. The max vcore on my board gives 1.57-1.58 per CPUZ, The guys with the DFI
boards can only seem to get another 50-70mhz with vcore over 1.6 (some as high a 1.8), and their temps go way over 60c.

I too have a 3700+ SD box version from MonarchComputer, CABGE 0515SPMW. I can do 2.6Ghz (260x10) at 1.40v prime95 stable, 2.75Ghz (250x11) at 1.475v prime95 stable (run 23 hours without an error). 2805Mhz (255x11) at 1.500v prime95 failed after 1 hour. Havn't try higher vcore. I guess my SD should stable at 2805Mhz with 1.525v or 1.55v vcore.

So it does look like our chips are pretty much identical with a limit around 2805-2860Mhz or Air.

A64 3700+ SD @ 250x11 (2750Mhz) 1.475v | XP-120/Panaflo M1BX run at 1400rpm | A8N-E 1004.001 | Gigabyte X800XL silentpipe | Audigy 2 ZS

 

[Xed]

Finally got everything hooked up, time to start playing

Also plan to see if I can manage to get 1t timings with 4x512 sticks

 

[Dadofamunky]

Originally posted by: Xed
Finally got everything hooked up, time to start playing

Also plan to see if I can manage to get 1t timings with 4x512 sticks

Very unlikely, but it might also depend on your mobo.

 

[Xed]

I've got the dfi ut nf4 sli-dr and the ram is ocz plat r2, we'll see

 

[Xed]

While I need to run longer stability tests, so far things look promising.

Running at 2805mhz 11x255 (1.45v, 1.43 actual) with a 3/4 divider (191mhz, 4x512) at 1.5-2-2-7 2t (haven't tried 1t yet)

No errors in a few full passes of memtest and prime has been going for an hour, will see if it can go overnight

 

[GuitarDaddy]

I found a tweak that allows me to utilize ram much better with the San Diego

I had previously not be able to acheive even a modest ram overclock while running the CPU at 2.8ghz. But I have found that if I use A64 tweaker to disable the mem clk for banks 2 and 3 (the empty banks, I have 2 x 512 in banks 0 and 1) that I can now run ram at 235mhz 2-3-2-7 while running CPU at 2.8ghz My Sandra bandwidth is now back up around 7000, and I got a new SuperPI 32m personal best of 26m 17s

This tweak also seems to bring the SD's memory bandwith scores back in line with my Winchester and eliminates the 300+ point difference at the same clocks. I,m guessing it will also help those who are having high temps overclocking the SD's on air.

 

[lytalbayre]

What's that tweak to disable mem clk??? Link??

With 2T and timings of 2.5-4-3-8 my max memory spped is 281mhz. I'm running this 1:1 with the cpu resulting in a stable cpu clock of 2.818 ghz all with vdimm of 2.75 and vcore of 1.5.

This so far gives me better benchmark results than my second and third best OCs with this chip:

2.8Ghz & 230mhz 2.5-3-3-6/1T
2.8Ghz & 200mhz 2-2-2-5/1T

 

[GuitarDaddy]

A64 tweaker is a free utility that you can download, Google for it.

When you open the application, in the bottom right corner it shows the mem clk option for each of the 4 banks 0,1,2,3. There is a drop down box for each with the options (enable,disable), they are all enabled by default. Disable the banks that you aren't using.

If your not sure which banks you have populated, run Sandra memory bandwidth benchmark and look at the report details it shows you which banks your memory is in.

 

[Dadofamunky]

Originally posted by: GuitarDaddy
A64 tweaker is a free utility that you can download, Google for it.

When you open the application, in the bottom right corner it shows the mem clk option for each of the 4 banks 0,1,2,3. There is a drop down box for each with the options (enable,disable), they are all enabled by default. Disable the banks that you aren't using.

If your not sure which banks you have populated, run Sandra memory bandwidth benchmark and look at the report details it shows you which banks your memory is in.

Oh, I get it, this would not apply to me. Great find though!

 

[Painman]

The DFI board + SD combo isn't playing nice with my RAM. At all. Errors during every stage of Memtest at all combinations of settings I've tried so far. Before I delve much further into the maelstrom of DFI BIOS settings or try to exchange this RAM, I may try the SD and Ballistix together on my Neo4 - I still have it kicking around...

 

[Xed]

Just a few more notes

I can run 9x300 1:1 at 1.43v, so I'm not having the fsb probs others are.

At 1.63v I can get into windows at 3ghz but it's not stable =(

2.86 (260x11) was stable at 1.48v with mid 30's idle temp so I think I'll just keep it there

best unpatched superpi 2m was 1min 7.328s , 32m 26min 16.653s

 

[Dadofamunky]

Originally posted by: Xed
Just a few more notes

I can run 9x300 1:1 at 1.43v, so I'm not having the fsb probs others are.

At 1.63v I can get into windows at 3ghz but it's not stable =(

2.86 (260x11) was stable at 1.48v with mid 30's idle temp so I think I'll just keep it there

best unpatched superpi 2m was 1min 7.328s , 32m 26min 16.653s

That's awesome. Do you have the DFI board? If so, which BIOS version? Details, please. Your temps and voltages are excellent at those speeds.

Hey, I'm still playing around with mine but kinda wish I'd held out for the DFI NF4 now. Oh well, I'm definitely not in the mood to throw more money at this thing much less rip the guts out again.... I just hope MSI gets its SH*t together and starts releasing some better BIOSes soon...

 

[Xed]

I have the dfi-ultra d (non sli) and 4x512 ocz plat r2

Those results were done on the 4.14-1 bios. I flashed to the 5.10-2 bios today and it seems to have made an improvement. Testing stability still but appears to be stable at 2805mhz on 1.43v now with a 33c idle temp

 

[IEC]

Originally posted by: Painman

Originally posted by: GuitarDaddy
Look what I found !!!

CABGE? Sounds like it oughta be boiled with some corned beef, celery 'n carrots

Ahahaha! Nice one!

 

[sonnywaters]

Hey Painman,

Did you get that SanDiego to work with your Neo4? I currently have a Winchester, ( on a Neo4-F ) but of course, it's not workin' right. Just wanted to know before I drop that chunck of change and it not work too.

Thanks, John

 

[sonnywaters]

I went ahead and ordered a 'Diego last night. I was hoping to get some feedback about them working on the MSI boards, but... I'll let you know.

 

[geezman]

FYI, I just built a new system with an MSI Neo4\SLI and a SD 3700+ this past weekend. I am happy to report that all is working great!. I'm loving my new system. I haven't tried OC'ing yet, but will after I get everything installed and setup.

 

[GuitarDaddy]

geezman, did the 3700+ work with the bios that came on the board? or did you have to update the bios?

 

[Painman]

I'll provide my brief MSI intel report, the SD booted up fine on the Neo4 board with BIOS 1.2, I promptly flashed to BIOS 1.3 but no 1T/2T command rate option was available. 1.41 beta BIOS corrects this.

Unfortunately the setup refused to POST at 255 HTT, even with RAM dividers and slack timings so I stuffed the DFI back in. I didn't do much testing to see what max HTT would POST with the Neo4, I put the DFI back in as I'm just going to work on the RAM issues it has and try to make it happy.

 

[geezman]

GuitarDaddy, It worked right out of the box. I believe the BIOS was dated 12/30 or sometime in December.