3700 San Diego vs. 3200 Venice: Round 2!

[ben805]

Per Stephen's request, I'd completed the reruns of San Diego vs. Venice in different methodology. The previous comparison at higher speed can be found HERE




Test Configuration:

CPU1: AMD 3700 San Diego
CPU2: AMD 3200 Venice
Cooling: XP90, 90mm Enermax Fan and AS5.
RAM: GSKILL PC4800 512LA 2x256MB DDR560 @2.5-4-3-7 1T
Hard Drives: 2 x 36GB WD Raptors (RAID-0)
Motherboard: DFI LP Ultra-D
BIOS: 414-3
Video Card: MSI 6600GT
PSU: PCP&C 510 SLI
Sound Card: EMU 0404
OS: Ghost image of Windows XP Pro SP1 (fresh install with drivers only)
Target CPU speed: Both CPU at 2800 Mhz (280 x10 @1.66v)


Note: A fresh Windows XP Pro SP1 Ghost image was being use for this shootout purpose, the operating system has nothing but drivers and benchmarking softwares installed. All benchmarks were being executed in the exact same order and manner, every single test was being carry out at the exact same clock speed with the exact same memory timing, each and every single test was follow by a reboot to ensure a fresh start for the next test. Once the 3700 SD cpu has completed the task, it was then swapped out with the 3200 Venice. The BIOS setting was being left as is(exactly the same way as with the 3700 SD), once the AS5 was applied on the IHS, the exact same heatsink/fan was installed, the exact same fresh Windows XP Pro SP1 Ghost image was being restored onto the same RAID0 array, and finally follow by the exact same task again.






Disclaimer:

Warning: Overclocking is dangerous. You could easily damage your hardware components or entire system if you do not know exactly what you are doing. I am NOT responsible for what you choose to do to your own system. If you choose to overclock your system, you do so at your own risk. I am NOT affiliate with any hardware/software vendor or manufacturer in any form or way, the result that I've conducted do not represent any other (and all) Venice or San Diego, please do not rely your hardware purchase decision base on my review, as I am NOT responsible for any dud/dog/pos/sucky cpu that you might end up getting.

[ben805]

3DMark2001 SE: Same as the previous shootout, this 3200 Venice scored slightly higher than the 3700 San Diego

PCMark04: The table has turned....yuppie! and this time around it favors the San Diego in overall score, but please compare the cpu, memory, graphic, and hdd.

[ben805]

Sandra2005: you'd be surprised, hang tight and see for yourself!

SuperPI 1M and CPUMark99: guess who's the winner now?!

[ben805]

3DMark03: Are you sure you're ready for this?!?

Sciencemark 2.0 Primordia: whatever this benchmark is...I don't have a clue what those numbers mean, but a member requested to crunch it away.

[Ubermann]

You better show the taskbar to people LOL
I guess the San Diego really hates something in your computer.

[ben805]

Sciencemark 2.0 Cipher Bench & Mol Dyn: Again, don't know what these mean, but nevertheless it looks like we've got a winner.

Toast: I use this one to test the maximum under load temperature, ran the Venice for 8min straight, only 2.31min for San Diego....eeekkk...bbq anyone??

[ben805]

You better show the taskbar to people LOL

I don't know who the winner is, don't have any fancy 3D graphs to put all these scores together so if anybody care to put a little chart for all these results, please do so. Man...all these tests have got my head spinning, it was the longest 7 hours I've ever had in my life, I really don't know what to think or say anymore, all I know is... this San Diego is NOT done yet, the Mushkin Redline XP4000 will be here this afternoon so I'll see what else is left in him...another project for another day. As for the Venice...my brother wants it back so it will go back to him later on. I better go get some sleep now, have to wake up in about 4 hours!

[Sucka]

Interesting.

For what it's worth, i have a 3500+ Venice and a 3700+ San Diego, which are similar in all respects and i'm getting about the same overclockability from both of them. The SD is in fact showing about that 1-5% difference i assumed it would show compared to the Venice. I would tend to agree that the San Diego really isn't worth the extra money over a Venice, at least that's what i'm seeing with my own results. I to ran tests on the exact same computer to come up with this conclusion. I will say though, my SD enjoys voltage much more than my Venice. Both can max bench at right around 3300MHz under single phase.

[ben805]

Interesting.

For what it's worth, i have a 3500+ Venice and a 3700+ San Diego, which are similar in all respects and i'm getting about the same overclockability from both of them. The SD is in fact showing about that 1-5% difference i assumed it would show compared to the Venice. I would tend to agree that the San Diego really isn't worth the extra money over a Venice, at least that's what i'm seeing with my own results. I to ran tests on the exact same computer to come up with this conclusion. I will say though, my SD enjoys voltage much more than my Venice. Both can max bench at right around 3300MHz under single phase.

You just remind me of that, at the end of each task...I brought the clock back up to 318 x9 @1.71v then ran the Toast, my San Diego was running dangerously hot at 70~71'C at around 2min, I'd shut off the system immediately....didn't feel like frying the damn thing. At the end of Venice task, I'd also ran the Toast....never get to see the result as it BSOD and rebooted after few seconds.

[Ubermann]

Try running test at 2.6ghz or something, in case heat is making problems for San Diego.

[ben805]

Try running test at 2.6ghz or something, in case heat is making problems for San Diego.

No more test between these two cpu anymore....I'm done, finished, completed!! people can believe in whatever they want.

[eva2000]

Sandra2005: you'd be surprised, hang tight and see for yourself!

SuperPI 1M and CPUMark99: guess who's the winner now?!

those superpi 1Ms look slow you sure some settings aren't right i.e. 4000+ sandy @12x234 2.5-3-3-7 1T



but with 1x eVGA 6800U PCI-E at stock 425/1100

[andL64]

lol 1M 2800mhz san diego 31 sec. wit DDR 560 thats weird i get with my newcsatle core @ 2750 mhz better scores

and the 3dmark01 score is pretty never the san diego will be slower than a venice i m sure.


this (p)review is only weird

seEn

[Ubermann]

The conclusion should not be that Venice is faster, it should be what problem is it that makes the San Diego perform bad.
Need to be solved i think else ppl will be sitting with San Diegos thats not performing 100% and they dont know it.

[googles]

outstanding job ben.

looking at those figures and barring any defects/problems with your SD cpu, id definitely have to go with a venice.

your results would definitely force some to put on their thinking caps and reduce the superficial/impulsive debunking.

i really wonder what useful programs there are that would cast doubts buying a 512k venice over 1mb diego.

thanks for the shootout.

[LoKi2k]

great job. i think that was a lot of work

thank for that!

[Ubermann]

I ordered a Venice, but there is still problems with this tests.
You cant accept that the Venice is faster just cuz its faster for him.

[Mastakilla]

kewl work man!!!

thx 4 the results

it seems they both have about the same speed to me (a few marks difference in 3dmark happens all the time)

the strange part is:
ur PWMIC, toghetter with ur cpu temp, of the san diego is muuuuuch higher then the venice..?!

so i guess it has to do something with clockgen of ur san diego?

[s7e9h3n]

Great job Ben. Notice that people actually are commenting on the results of your tests rather than crapping on how you ran the tests? . Hey your work's not over. Get some rest tonight and get ready for the onslaught of questions tomorwow

[s7e9h3n]

Oh BTW, for comparison's sake, here's what my 3800+ Venice did @ 280x10 a couple of months ago.......

[AMDnewbie]

SD is definitely sucking more amp and create much more heat than venice as from the results.If thats the case than the X2 512K would be much more appealing than the 1MB cache.Run more cooler and maybe can clock higher

[agenda2005]

Thanks for running those sciencemark benchmarks. It appears that your 3700+ SD have a very high L2 cache latency. Primordia just gain only 1sec for SD over Venice after 319Sec run! Wow!

I will conclude that the 1MB cache is not worth it for A64 arcitecture. I reckon someone in Prime95 website claiming that one should go for clock speed over cache when it comes to A64, because the on-die memory controller has render any extra cache worthless.
I made a mistake, I should have gone for 3800+ Over 3700+ for just $40 more and 200MHz higher clock and damn the useless 512KB cache.

[Zebo]

LOL. Got cheap? Tests like this is why I always say buy the lowest rated chip in iteration and turn it into $1000 chip. Pyrite into gold is where it's at.

You kinda over spent gettin the 3200 brudda

[Ref]

Great testing, thank you

now waiting for FX-56 with 512 kB L2 cache

[ewitte]

lol 1M 2800mhz san diego 31 sec. wit DDR 560 thats weird i get with my newcsatle core @ 2750 mhz better scores

and the 3dmark01 score is pretty never the san diego will be slower than a venice i m sure.


this (p)review is only weird

seEn

Superpi tends to care more about latency. If I leave the memory timings at default I'd probably get about 31s at 2.8Ghz as well. With them tweaked its 29s. Thats 2.5-3-3-7-1T (10,12 and mostly 3's). It could easily be 27s or even 26s if I could pull 2-2-2-5-1T.

Eric