Palermo 2600+ OC

[sluggard]

anyone else OC'ed it? stepping CBBID 0503. DFI nF3 and aircooled with venus12.









almost same spi time but lower cpuclock, vcore and memtimings. OC's like winchester

[shuRe]

haha that beats my a64 3200+ @ 2.6 :s

well done!

[c7775]

omg now that is some good overclocking m8, what voltage you tried max?

[cadaveca]

i have the same stepping but 0504 week apew. my board sucks tho. we will see what happens in the end tho.

[sluggard]

omg now that is some good overclocking m8, what voltage you tried max?

1.8v was the maximum stable. above that windows simply freezed. the cpu doesnt tolerate that high voltage with a plain aircooling i guess

[Nano2k]

I got 2320Mhz on this machine with the Sempron 2600. The mobo (Chaintech VNF3-250) is holding me back, it only detects half of the Ram on the sticks and always been a bad overclocker.

Nice o'clock Sluggard!

[$a1Ty]

are there going to be any s939 semprons?

[cadaveca]

they should scale well, i think. i maxed out @ 2.25 ghz, was prime stable(hours), but not 3d. same settings got the same results on a 3400+, so the board is the culprit. Even so, max load temps with prime95 were 41c under water...the lack of enabled cache makes for a huge difference in heat in comparison to their winchester cousins. 1.45v for a 650mhz overclock is good for me...but i cannot wait until i get a better board for this cpu!(besides the fact that 754 will be alive for quite a bit, and should get dualcores too!)

[sluggard]

i think ill try one or 2 out tomorrow @ -100c

then you definately need a DFI board for that, if you're going for 2600+ cause the multi is locked @8x my board reached the limit@428MHz HTT with newcastle, tho theoretically you can get at least 3,4GHz before mobo sets the limits default clock for 3000+ and 3100+ sempy is 9*200 which is a bit better.

[Malachia]

how much for a 2600??? i think i will get one for my DFI 754...

[cadaveca]

about 100 Canadian. i paid 105 with tax or something. not bad for 2.9ghz...

[Malachia]

yes but molti is really low

my dfi do max 350mhz benchable...

[sluggard]

they should scale well, i think.

they do indeed. did a little research

voltage needed for certain speed increment:


also performed superpi 4M on these previous voltage/MHz:


almost linear scaling, MHz increase transfer ca. 1:1 into performance numbers not bad


memtimings on different setups with TCCD were following:

200MHz CL1.5-2-2-0 1T
250MHz CL2.5-3-3-0 1T
300MHz CL2.5-3-3-0 1T
325MHz CL2.5-4-3-0 1T

yes but molti is really low

my dfi do max 350mhz benchable...

i'm sure your board also does 400+ HTT for sure, just needs some fine tuning

[i found nemo]

damn, i'm gettin' the 1.8ghz version (3000+) hope mine does half this good good job buddy

[reject]

spi doesnt do much with cache, try a 2001 and you wont go too well
nice oc but

[ent]

I also wanted to buy such Sempy, but in the end I spent my money on 939 + Winnie setup .

[sluggard]

spi doesnt do much with cache, try a 2001 and you wont go too well
nice oc but

well yes, superpi is plain mem+cpu+mobo subsystem benchmark. for 2001 you have to eliminate the gfx bottleneck (but anyway i've got gf2 atm )

[i found nemo]

wondering howcome in one cpu-z screen shot it shows fsb and bus but no voltage, then in another one shows voltage and fsb but no bus

not pointin' fingers just a bit curious

[cadaveca]

different cpu-z versions. one is .27(no voltage) the other is .28

[i found nemo]

eh, makes sense

[sluggard]

i used version 1.28 every time. the validation screen only doesnt show the voltage and it uses the 1.27 template while generating the screenshot it says v1.28 in the second row too. no idea why it doesent show the voltage, it just uses the data it gets from the validation file