amd's rev F3 silicon

[ozzimark]

...is looking mighty fine

i saw this over at newegg, and i just had to get it.
http://www.newegg.com/Product/Produc...82E16819103868

stock speed is actually 2.6ghz, but that's another matter

just got it in today.. with stock voltage, 3.25ghz is relatively close to being stable



bumped the voltage up a bit and found that voltage scaling sucks


but curiously, it seems that 32M fails faster when unstable than sp2004 does

anyway, some more voltage.. and a mighty untweaked OS



gonna see if i can't get 3.4ghz stable. all with watercooling

[Postal Dude]

new silicon is looking good indeed.
Wish my 939 opteron scaled that well!!

[hstuehmeyer2000]

thats a nice cpu, could you imagine if they made single core opty f3's(i know they don't make single core opy's for am2)

[ozzimark]

afaik, the dual core opterons with F3 clock pretty similarly.. just impossible to get ahold of and much more expensive than a single core A64

i've gotta get amd to tell me what exactly they changed with this revision.. it's nuts

[xenolith]

Nice chip.

You lucky AM2 owners finally have something to really cheer about over us who decided to stick it out with 939.

I wish K10 would hurry up and get here already.

[nn_step]

pretty good single core

[mAJORD]

Whoa, nice single core action.

What a perfect gaming CPU for the $!

[s7e9h3n]

but curiously, it seems that 32M fails faster when unstable than sp2004 does
[/img]

Ummmm....maybe because you're running large FFT's @ priority 1 with Orthos?

Looks like I was right afterall with my initial guestimate...

i've gotta get amd to tell me what exactly they changed with this revision.. it's nuts

You're not gonna find out and they're not gonna tell you

[hstuehmeyer2000]

got my f3 today that i paid $90 for, this a suicide run

[ozzimark]

not a bad start for how cheap that thing was. i'm jealous, how did you manage to get a dual core F3 for that little?

Ummmm....maybe because you're running large FFT's @ priority 1 with Orthos?

Looks like I was right afterall with my initial guestimate...



You're not gonna find out and they're not gonna tell you

probably. i'm using a slightly modified version of blend now, seems to work much better, but the transition from stable to reboot is still on the order of like 20-40mhz at lower voltages

and... since when has that stopped me from asking? but yes, i think you were about right with your original guess... though i haven't gone over 1.55v yet.

at least i can do 3ghz stable with 1.225v

[Postal Dude]

Lets hope they can keep this clocks/Volts ratio up when K8l comes out

[ozzimark]

heh.. we'll see. i'll share my theory. no, i have no proof that it is right, it's just a suspicion of mine...

[put on tinfoil hat]the clocks we are seeing with F3 is a result of SiGe.[/remove tinfoil hat]

once we start seeing more mature 65nm with SiGe in it as well, i am fully expecting K10 to clock a bit better...

[hstuehmeyer2000]

i posted it over at the lHl, Eclipse, under hot deals, it was buy it now on ebay, the seller had 5 of them for sale

i just hit 3.2 at the same voltage, and i upped the voltage a little more and it still won't go farther, my motherboard is now the limiting factor

edit for some reason now my motherboard won't do 320x10, so im back at 310x10 and i undervolted it a notch to 1.375

edit, got it to under volt to 1.325 stable

[Lightman]

Simply WOOW! Both of you guys have great chips (or steppings ).

BTW today I played with X2 4000+ 65nm 0702 and I run out of HT on MSI board based on MCP61 (new single chip GF6100). Basically board can't clock from BIOS and in Windows ClockGen can't recognize this combo. My only way of ocing was to use nTune but it is limited to 250HTT (any known reg entry to change that limit??) which gave me 2622MHz@1.312V easily!
I can confirm now what other lucky guys was telling that Brisbane is extremely cool running CPU .

[hstuehmeyer2000]

thats the only thing bad about the 90nm chips is that they don't run as cool as the 65nm

[ozzimark]

thats the only thing bad about the 90nm chips is that they don't run as cool as the 65nm

well... i'm not too sure about that... my 4000+ is a LOT cooler than the 3400+ sempron that came before it, though it may be because of the soldered on IHS

[hstuehmeyer2000]

i also though it was weird that my opteron ran about as cool as my sempron to

it idles about 33-35c and is about 45-47 load

[high5]

ozzi, what mobo are you running?

[hstuehmeyer2000]

i think he is running a dfi 590sli

[s7e9h3n]

i think he is running a dfi 590sli

yes he is

[ozzimark]

yes he is

yup, home to the wonderful smartguardian

i also though it was weird that my opteron ran about as cool as my sempron to

it idles about 33-35c and is about 45-47 load

hm, you have a much larger delta between idle and load than me, but that kinda makes sense, seeing as i'm using a single core.. with 18c air at 1.35v and 3.2ghz, idle is around 21c while load is around 25-27

[s7e9h3n]

hm, you have a much larger delta between idle and load than me, but that kinda makes sense, seeing as i'm using a single core.. with 18c air at 1.35v and 3.2ghz, idle is around 21c while load is around 25-27

Kinda reminds me of my ADV4400....load vs. idle was about 5-6C difference:

[erwinz]

very nice results..

[ozzimark]

Kinda reminds me of my ADV4400....load vs. idle was about 5-6C difference:

nice temps for a dual core, but the mhz is just a smidge lame

[s7e9h3n]

nice temps for a dual core, but the mhz is just a smidge lame

Better?