*Semi-Official* San Diego OC Results thread

[GuitarDaddy]

Originally posted by: UzairH

GuitarDaddy, don't take this wrongly , I respect you and all (thanks again for the help you've given me in questions I've posted in the past ), but I am almost tempted to call BS to your bechmarking methods, this is so weird. Waiting for tests from other guys here, as well as review websites. But thanks to GuitarDaddy for these extensive tests :thumbsup:

Weird!
And oh, the OC results are so MUCH better than what we've seen with Venice. Glad I didn't order one

I'm still not quite beleiving the game numbers myself. One thing I considered is both benches were run using the 1007 bios, which was terrible for O/C with the Winchester but seems fine with the SD, even though I had no stability issues while testing the Winnie at 2200mhz, its possible it wasn't playing well with the 1007 bios. But then again the other numbers don't support this.

The HL2 bench is very definitive in that it runs to completion and calculates and displays the avg FPS, so there couldn't be anything different in the way I ran it. The X2 bench is a bit more subjective, because its just a looping demo and you have to run a frame capture program like FRAPS and stop and start the benchmarking. I was very careful to start and stop the benchmarking in the same spots and felt somewhat comfortable with the results since they were so close to the HL2 results.

But I too am struggling to explain the 17-18% increase I,m also having a hard time with the memory bandwidth. How does SD beat the Winnies by 9% with the same ram, same timings, and 20mhz less clock?????

*Update*
2822mhz errored in prime after 4 1/2 hours. I dropped to 2800mhz and restarted it, I stopped it this morning before I went to work and it had run just over 6 hours with no errors, looks like I will be able to hold at 2800 or just a bit higher
My load temp after six hours of prime was 53c
:thumbsup:

 

[UzairH]

*** CALLING ALL FOLKS WITH SAN DIEGOS AS WELL AS CLAWS/WINNIES/VENICES TO BENCHMARK AND COMPARE AT SAME CLOCKS***

Hey GuitarDaddy why don'tcha edit the topic title and OP to ask people for benchmark results also? This would be good to verify the incredible gaming and bandwidth results you've obtained, and in other games/apps as well such as Doom 3, Quake 3 (still a good CPU test - mostly CPU limited these days), 3DMark, PCMark, etc.

If the dual-cores have same core as the San Diego, then the 2.4GHz 1MB 4800+ is one HELLUVA powerful chip. OC'ed to 2.6GHz it should own the FX-55 even. And at a slightly higher cost. Or you could just ge the 2.2GHz 1MB 4400+ and OC to 2.6 and rock! (but with some monster cooling lol)

 

[Dadofamunky]

Originally posted by: GuitarDaddy

But I too am struggling to explain the 17-18% increase I,m also having a hard time with the memory bandwidth. How does SD beat the Winnies by 9% with the same ram, same timings, and 20mhz less clock?????

*Update*
2822mhz errored in prime after 4 1/2 hours. I dropped to 2800mhz and restarted it, I stopped it this morning before I went to work and it had run just over 6 hours with no errors, looks like I will be able to hold at 2800 or just a bit higher
My load temp after six hours of prime was 53c
:thumbsup:

Hah. I am VERY excited. Thanks for the quick report. I'm taking my time getting mine installed. I flashed my BIOs and that was all I had time for. I may wait until my new XP-90c comes in, as I haven't been happy with the performance of my existing HSF. However, I don't know if I can bring myself to be that patient.

Our CPUs didn't even physically exist until less than three weeks ago - mine has an 0515 date on it. We are on the leading edge here!

 

[xenolith]

Originally posted by: Zebo

i guess you really do pay for what you get in this case

Indeed. I just can't understand the diff. I mean same manufacturing process between V and SD but more than 200mhz differential so far in the overclcoks!!!???!!!



Edit: Here's that thread GD was talking about!!!

SD seems the chip to have... all into and over FX57 levels (2800Mhz)

It's been my understanding that the San Diego uses SSOI (strained silicon on insulator) and the Venice uses DSL+SOI (dual stress liner silicon on insulator).

The expectation has been the former will have better yields than the later, and therefore have better o/c headroom. Maybe this is bearing out... ?

edit: spell check is my friend, lol.

 

[Painman]

Wow. Very interesting benches and a 2.8 OC to boot. I wasn't even considering picking up SD because of the cost, and not believing the extra cache would justify it. I hope some more SD owners report in with similar news. I'm pretty sure I'm gonna sell my Venice regardless, if SD is consistently good I may just have to grab one...

 

[Duvie]

Damn....for me the cost of the 3700+ to get only a 2.8ghz is not enough.....I will stay with the 2.66ghz I have now on a 140 dollar chip.....

3ghz must have been a pipe dream.....

 

[explosean]

Here's something that may be of interest

high speeds

 

[GuitarDaddy]

Pretty nice SuperPI 32m

26 minutes 29 seconds

*edit*
Interesting note: I downloaded and tried the SSE3 patched "Prescott" version of SuperPI and I get worse scores than with the SSE2 patch? I guess they will have to release an SSE3 patched "San Diego" version. Must be something different about SSE3 for Intel vs SSE3 for AMD

 

[Dadofamunky]

The brain swap is done.

Well, so far things are looking pretty good..... VERY preliminary results have my 3700+ at 2.607 GHz, 11x237. No memory divider on 4x512MB stix of PC4000 RAM, but memory defaults to 2T. An 18% OC is not enough but it's a good burn-in point. I flashed up to the "Official" 1.5 BIOs and it sees the chip fine but it takes away the 1T/2T Command Rate BIOS setting. Needless to say, I'll be doing a lot more testing and benchmarking (beginning with an overnight Prime run, hopefully - so far this thing seems to be behaving very well - my 0451 3000+ wouldn't boot Windows at this speed), flashing to a newer BIOs and slowly raising the HTT. When I get my XP-90c HSF I'll be swapping that in....

45C idle speed in Windows. For now I'll keep my sig as it was until I'm closer to being finished.

 

[rise]

san diegos only run 1T anyway, right?

these are looking like some badass chips. i'm not regretting cancelling my 3200+ on sunday

 

[GuitarDaddy]

Mine will run 1T or 2T

 

[wfn]

so whats's the max overclock on air on can expect with these? given proper mem / xp-120

 

[GuitarDaddy]

From my results and the others I've seen 2.8ghz is fairly easy. I've seen one 2.9ghz so far. The best I can get prime stable so far is 2805mhz, I can run most benches at 2830mhz but prime fails after several hours

 

[wfn]

600mhz oc... not bad but i wouldn't go as far as to call it spectacular.

 

[arcenite]

I am going to keep my venice until amd comes out with something that can overclock to 3ghz

 

[GuitarDaddy]

I guess the insane overclocks that were seen by some of the eary Winchester 3000+'s got everyones expectations up too high

All I've seen for the last couple days is tons of Wha Wha Wha, my venice sucks it will only do 2.7, and so many people expecting 3gz on air from 1.8-2.2 chips. Personally I,m more than happy with 2.8ghz

 

[wfn]

i hear ya.

 

[Shimmishim]

Originally posted by: arcenite
I am going to keep my venice until amd comes out with something that can overclock to 3ghz

then get the FX55... 3 ghz has been reported on air

 

[Dadofamunky]

Originally posted by: rise4310
san diegos only run 1T anyway, right?

these are looking like some badass chips. i'm not regretting cancelling my 3200+ on sunday

Nope, not on the Neo2 Plat - defaults to 2T and no way you can move to 1T

 

[Dadofamunky]

Originally posted by: GuitarDaddy
From my results and the others I've seen 2.8ghz is fairly easy. I've seen one 2.9ghz so far. The best I can get prime stable so far is 2805mhz, I can run most benches at 2830mhz but prime fails after several hours

Not much luck here so far.

BIOS at 1.6 or 1.8 - makes no difference (current 1.8)
VCore set to 1.55 +5%, shows at 1.504 in CPU-Z
12V rail at 12.38V in CoreCenter
Memory set to 2.7V
AGP set to 1.60V
RAM: 200MHz 3-4-4-10-2T (OCZ PC4000 here)
HTT: 11x245 2695 GHz on air, fails Prime immediately, 3DMarks fail

The only really good thing is suddenly I'm getting CPU temps of 26-28C idle. Never seen it that low with any CPU.

I'm gonna throttle it back some until I get a stable OC.

Any firther ideas? Hell, I'd love to get 2.8G outta this puppy, but I don't know if my board can handle it.

 

[anandtechrocks]

So that is only 1 out of 2 people from these forums with these high results. I think everyone is overblowing the SD, one guy gets lucky with a great chip and everyone want to jump on.

I don't mean to be pestimestic, but I think we will see the same overall overclocking results from the SD's as the Vience's.

 

[UzairH]

Hoi GuitarDaddy include a poll for the oc's also in the OP, like Duvie's older poll for Winchesters. Sandy's are rare now but in a few weeks we'll be seeing a lot more of them. Let this be the *official* thread for SD overclocking and results.

 

[Duvie]

Originally posted by: Dadofamunky

Originally posted by: GuitarDaddy
From my results and the others I've seen 2.8ghz is fairly easy. I've seen one 2.9ghz so far. The best I can get prime stable so far is 2805mhz, I can run most benches at 2830mhz but prime fails after several hours

Not much luck here so far.

BIOS at 1.6 or 1.8 - makes no difference (current 1.8)
VCore set to 1.55 +5%, shows at 1.504 in CPU-Z
12V rail at 12.38V in CoreCenter
Memory set to 2.7V
AGP set to 1.60V
RAM: 200MHz 3-4-4-10-2T (OCZ PC4000 here)
HTT: 11x245 2695 GHz on air, fails Prime immediately, 3DMarks fail

The only really good thing is suddenly I'm getting CPU temps of 26-28C idle. Never seen it that low with any CPU.

I'm gonna throttle it back some until I get a stable OC.

Any firther ideas? Hell, I'd love to get 2.8G outta this puppy, but I don't know if my board can handle it.

Both the 1.62 and 1.8v have a temp reporting bug for me and show temps at like 14-16c idle and only get up to 38-40c under load.....I went back to 1.41mod to see and see enough it goes back to 32c idle and 51c load....

 

[Dadofamunky]

Originally posted by: Duvie
Both the 1.62 and 1.8v have a temp reporting bug for me and show temps at like 14-16c idle and only get up to 38-40c under load.....I went back to 1.41mod to see and see enough it goes back to 32c idle and 51c load....

OMG that's right, you'd mentioned that. I knew that must be too good to be true.

I have managed to get some better behavior out of this thing at 2.75 GHz.... Fixed a video card setting in the BIOS (I forget which one) and I reached a personal best of 13,337 with 3DMark03, 5580 with 3dMark05, and a 5418 with PCMark04. The synthetic benchies (Sandra and Everest) show huge losses in memory performance, but with my graphics results starting to show up this well it's hard to take those seriously....

Prime fails in less than one minute. I will not be happy until that runs reliably. Still have some tweaking to do but I think I'm getting closer. I'm going to try a divider next. I found the 1.5 official BIOs to be really unstable beyond stock speeds. Running BIOS 1.8 now. I really wish MSI woudl get its poop together in that area.

 

[Zebo]

Originally posted by: GuitarDaddy
I guess the insane overclocks that were seen by some of the eary Winchester 3000+'s got everyones expectations up too high

All I've seen for the last couple days is tons of Wha Wha Wha, my venice sucks it will only do 2.7, and so many people expecting 3gz on air from 1.8-2.2 chips. Personally I,m more than happy with 2.8ghz

There was justification for the optimistic outlook. Mainly the process is good for 3Ghz on defaut 1.4 vcore. Just look at the FX-57 which is 2.8 on 1.4V!!! add to the fact 5/5 reviewers thus far have all gotten 2.8 minimum, some default Vcore leads to optimism.

Anyway my feeling is they are doing something to the chips or speed binning. If they speed binning, which looks more likly because some have now gotten 2.8 on or about default core, then it's back to luck and stepping, no automatics.