AMD A64 4000+ San Diego

Post by **Apoptosis** » Wed May 04, 2005 8:41 pm

This afternoon Legit Reviews Forum member wickedld9 came over armed with his brand new AMD Athlon 64 4000+ "San Diego" processor. He was even nice enough to let us throw some water cooling on it and run some tests on it!

Here is a retail 4000+ on the DFI SLI-D motherboard running Corsair's PC-4400C25 memory on DFI's 427-3 Beta BIOS.

Stock Everything:

Best Stable Overclock:

We ran both Corsair's Samsung TCCD based memory and OCZ's Winbond UTT memory on the board and tried a numerous voltages, but was only able to get up to 2.91GHz with some stability.

We were able to run Comanche 4 at 640 x 480 resolution and came up with this for the results:

Both At 2.2GHz (11x200)
3500+ Venice: 72.82
4000+ San Diego: 76.46

Both At 2.9GHz (10x290)
3500+ Venice: 95.46
4000+ San Diego: 99.50

This was the closest Legit Reviews has ever been to breaking 100FPS on Comanche 4! The extra cache is used up on games, but doesn't make much difference on benchmarks like Sciencemark, Cinebench, Everest, or Sandra.

I'm sure wickedld9 will keep us inforemed how his CPU does after it has been broken in!

infinitevalence · Post by **infinitevalence** » Wed May 04, 2005 9:30 pm

wow that would be awsome to hit 3.0 on that sandi, its looking more and more like 2.8-2.9 is the limit on both venice and san diego. now i just need to see if i can swing the $$ for either.

Post by **Illuminati** » Wed May 04, 2005 10:11 pm

Ya, I was there too... it was fun watching Nate and wicked changing all the little settings on the DFI BIOS trying to get the most out of the CPU. After watching some of the set backs they encountered in the few hours I was there, I'd say they walked away with some success getting the Sandi to 2.9!

eva2000 · Post by **eva2000** » Wed May 04, 2005 11:33 pm

yeah i just bought the same exact stepping/batch 4000+ san diego as well should have it next week

why not up vcore to 1.6v and see how far it goes ? what water cooling ya using ?

427-3 bios never seen that before ?

Post by **Apoptosis** » Wed May 04, 2005 11:53 pm

427 is the latest series of BIOS's from DFI... They are said to work best with the new San Diego CPU's.

We are using Corsair's HydroCool 200ex and was at 37C when running 2.9GHz at 1.6V... We did hit 1.6, but it didn't give anymore head room. We then ran out of time, but I'm sure Brian will O/C is more later at his pad.

eva2000 · Post by **eva2000** » Wed May 04, 2005 11:57 pm

can you ask when dfi will be releasing 427 bios

-mogwai · Post by **-mogwai** » Thu May 05, 2005 12:39 am

what's the difference between san diego, clawhammer, and sledgehammer?

infinitevalence · Post by **infinitevalence** » Thu May 05, 2005 1:40 am

sledgehammer == optron, 1mb cache, 130nm
clawhammer == skt 754, 4000+, FX 53,55, 1mb cache, 130nm
san diego == 3700+, 4000+, FX 55, 57, 1mb cache, 90nm

-mogwai · Post by **-mogwai** » Thu May 05, 2005 3:07 am

ah. excellent. i'm learning so much! thank you guys for bearing with me. i used to know tons about stuff back in the day (nforce2, athlonxp, p4, 865pe, 875p, etc), but then technology decided to go into hyperdrive and i couldn't keep up anymore.

Bwall · Post by **Bwall** » Thu May 05, 2005 8:14 am

infinitevalence wrote:wow that would be awsome to hit 3.0 on that sandi, its looking more and more like 2.8-2.9 is the limit on both venice and san diego. now i just need to see if i can swing the $$ for either.

3.0 on this one without high end water cooling or below ambient cooling is probably a pipe dream. I think you should hang on to your winny just a few more months and let them refine the process a little more and see if they are getting 3GHz out of them.
I'm disapointed with the overclocking of the CPU but overall I'm pretty happy with the it. I actually played some games with it last night and it screams as it doesn't have quite the dips in frames that I saw with Winny and Newcastle.

I'll keep posting up new air results as long as you guys care to read them. I'll be busy with other things this weekend so it will be Monday before I can dig into it much. My new Lian Li will be here today and it will be a big help for getting good clocks on air.

And I'm going to figure out how to break 100fps on Commanche.

infinitevalence · Post by **infinitevalence** » Thu May 05, 2005 10:34 am

wicked your thinking the same way i am, my guess is that if i ever get off my lazy ass and put my cpu under decent water i can hit 2.7 maybe even 2.8. Heat is deffinatly a limiting factor for me now as i run in the 50's at load. Its looking like pushing the venice/san diego core to 3ghz takes more than i have for cooling and would therefor not net me any real performance improvement over my current system.

eva2000 · Post by **eva2000** » Mon May 09, 2005 7:27 pm

mine arrived too

infinitevalence · Post by **infinitevalence** » Mon May 09, 2005 10:01 pm

cool man pop the cherry and lets see some OC action

Bwall · Post by **Bwall** » Tue May 10, 2005 12:13 pm

Just a quick update as I'm trying to sort out the cooling with this new case.
I've been using my UTT too long, I tried to use my Mushkin TCCD's Sunday and just don't have the touch with them.

Note that the voltage can scale on it's own to 1.45v when set to auto.
1.45v boots her to at least 2.6GHz. I'm able to play CS:S and all other games for hours and run all 3dmarks at this speed. It is Prime stable for at least 5 hours. I don't have the video card quite to the level I had at the LAN. Still sorting out cooling issues as mentioned above and don't want to cook her.

http://service.futuremark.com/compare?3dm05=840273
http://service.futuremark.com/compare?2k3=3719715
I should have more tomorrow.

eva2000 · Post by **eva2000** » Wed May 11, 2005 2:27 am

not having much luck with my sandy it seems http://www.dfi-street.com/forum/showthread.php?t=11402

any pointers ?

Post by **Apoptosis** » Wed May 11, 2005 4:03 am

Eva,

When wickedld9 brought his CPU over i didn't have much luck either... 2.9Ghz as you can tell. Looks like our 2.9GHz stable on water is a good frequency to hit after all!

I was running 4/27-3 BIOS, but the newest is 5/09-3.

Just like on my Venice I was able to better overclock with the TCCD memory than the UTT or Winbond/Hynix modules. I was hitting 2.8 with UTT and then 2.9 on some late week code Samsung TCCD w/ Brainpower PCB's.

Something interesting i noted during testing is that our San Diego cpu didn't like more than 1.55Vcore. Even with the VID @ 1.425 and the SVID @ 113% a higher overclock wasn't possible. I took it almost up to 1.7V trying to figure the darn thing out.

eva2000 · Post by **eva2000** » Wed May 11, 2005 8:18 am

thanks for the info maybe i need to get my hands on the 427-3/509-3 latter bios heh

seems i am unable to achieve what OPB achieved with this cpu (2850mhz at 1.45v air), but mine handles more vcore fine so far

Bwall · Post by **Bwall** » Wed May 11, 2005 8:54 am

Update:
Spent most of the night re-wiring the case for better airflow and trying to hide the wiring. There is still more work to be done but my temps are way down...it also helps that we now have the A/C on. ;) I finally have similar load temps I was seeing with my Winchester.

I tried to boot 241x11 @1.45v but it went into a reboot loop before loading. Simply added 4% (1.508v) and started up fine. I didn't have time to try a lower voltage or different multi, I just wanted to test the cooling mostly and it was already 1am. I stopped Prime after 15 minutes just to test my cooling and then went through 3dmark01 fine, and left it running with a few tasks overnight and was still going strong this morning.
I'm going to put off the rest of the wiring job until tomorrow and try to get some overclocking done tonight.

On my previous post I forgot to mention that I was running 260x10.

As far as the case goes, this V1200b Plus is really nice to work with. The front 120mm fan and blower fan come with molex connectors, I converted them to 3 pin and hooked them up to a fan controller. I've also moved my 86cfm Panaflo from CPU duty to pulling air from the front of the case. CPU is now being cooled by a 115cfm Panaflo. There is a ton of room inside of this case, which is the reason I bought it..in case I decided to go H2O.
The hard drive bays need more airflow in my opinion. The front 120mm blows straight on them but I have 4 drives and the first 2 eat pretty much all of the airflow, and the fan has a filter limiting the airflow. I had a 60mm Delta 5k rpm screamer laying around so I hooked it up to the controller as well and zip tied it to the bottom of the case (it got lots of holes for pulling in cool air) so that it's blowing on the bottom of the middle drives, and deflecting air into the PSU.
I'm probably a little overboard when it comes to cooling, if it feels warm it needs a fan.

Until tomorrow..........

Bwall · Post by **Bwall** » Wed May 11, 2005 9:17 am

Apoptosis wrote:Eva,

When wickedld9 brought his CPU over i didn't have much luck either... 2.9Ghz as you can tell. Looks like our 2.9GHz stable on water is a good frequency to hit after all!

I was running 4/27-3 BIOS, but the newest is 5/09-3.

Just like on my Venice I was able to better overclock with the TCCD memory than the UTT or Winbond/Hynix modules. I was hitting 2.8 with UTT and then 2.9 on some late week code Samsung TCCD w/ Brainpower PCB's.

Something interesting i noted during testing is that our San Diego cpu didn't like more than 1.55Vcore. Even with the VID @ 1.425 and the SVID @ 113% a higher overclock wasn't possible. I took it almost up to 1.7V trying to figure the darn thing out.

aahhhh..I thought the Comanche results were with UTT...I smell 100+fps comanche coming up tonight!!!

eva, That looks to be just about the same as mine as far as voltage requirements. Try setting the memory on a divider and using some different multipliers. I've seen some strange things..
At 225*12 1.45v it won't boot but 245*11, 270*12, and 300*9 all boot without issue. I'll test this out some more tonight.

Bwall · Post by **Bwall** » Wed May 11, 2005 9:37 am

Sorry to keep posting, I just noticed something strange. Last week I was able to boot 245x11 1.45v but last night could not get 241x11. I'm pretty sure I'm using tighter timings now but nothing that the memory hasn't run in the past. I guess that's something else for me to play with tonight.