venice on ice 2850mhz 1.38v -40Â°C

saaya · 03-25-2005, 02:15 AM

first of all BIG thanks to s7e9h3n for making this preview possible!
thanks a lot for bringing those chips to XS and letting us freeze and torture them

and of course big thanks to our trusty member chunkylover77 for letting us play with his 3500+!
thx a lot guys!

after last weeks results of a 3800+ venice from s7e9h3n here
pc ice now received the 3500+ venice from to s7e9h3n to play with it using his phase change skills

and it looks like the low temp bug is gone!!!

i will keep adding more results here step by step

Quote:

EDIT: bad news! opp cant get it to work under his cascade, seems there is still a low temp bug :S

now lets hope E4 changes that, so far i only heard that E4 had changes made in the temp probe , thats why it needs a bios update to work on almost all boards.
i really hope the low temp problems dont have to do with the manufacturing process!

phase change cooled system specs:

AMD 3500+ 2.2ghz dafault vcore 1.4v 90nm Rev E3 Venice core @-40Â°C
pc ice single stage phase change cooling
DFI NF4 ultra D bios 3/14
2x512mb G-SKILL PC4800FF @ddr660 2.5-3-3-6 1T
SAPPHIRE X850XTPE 688 / 644
OCZ POWERSTREAM 420w

here are the first results at stock speed (11x200):

................cpu-z............... ......sandra arythmetic....... ........sandra memory........ ...........3dmark 2001......... .........3dmark 2003.......... ..........3dmark 2005.........

click on the thumbs for a larger image

the bios undervolts, bump to 1.4v real gives 2850mhz:

.........sandra memory........ ........sandra memory........ ......sandra arythmetic....... ...........3dmark 2001......... ...........3dmark 2001.........

click on the thumbs for a larger image

2950mhz with still the same vcore (1.4v):

.........sandra memory........ ......sandra arythmetic....... ............superPI 1M.......... .............the cooling..........

click on the thumbs for a larger image

3ghz and still default vcore (1.4v) :

........sandra memory........ ........sandra memory........ ...........3dmark 2001......... ............aquamark3.......... ............aquamark3..........

click on the thumbs for a larger image

3100mhz and still default vcore!!! (1.4v) :

..............cpu-z.............. ..............cpu-z..............

click on the thumbs for a larger image

320fsb seems to be no problem for this memory controller :

................cpu-z............... ........sandra memory........ ........sandra memory........

click on the thumbs for a larger image

watercooled system specs:

AMD 3500+ 2.2ghz dafault vcore 1.4v 90nm Rev E3 Venice core
asetek waterchill
DFI NF4 ultra D bios 3/14
2x512mb G-SKILL PC4800FF @ddr660 2.5-3-3-6 1T
SAPPHIRE X850XTPE
OCZ POWERSTREAM 420w

stock 11x200 and 11x220 2420mhz :

...............cpu-z................ ...............cpu-z................ ........sandra memory........ ......sandra arythmetic....... .............waterblock........... ...........idle temps............

click on the thumbs for a larger image

11x240 2640mhz :

...............cpu-z................ .......sandra arythmetic...... ........sandra memory........

click on the thumbs for a larger image

11x260 2860mhz and 11x258 2838mhz :

...............cpu-z................ .......sandra arythmetic...... ........sandra memory........ .......sandra memory......... ...........3dmark 2001......... ...........3dmark 2001........

click on the thumbs for a larger image

memory controller oddities :

..............sandra............... .............sandra...............

click on the thumbs for a larger image

update:
cpu-z 1.27.2 or the latest CBID couldnt recognize the cpu either, dave will try to use cpu-z 1.27.2 to verify a shot though

update:
the superpi version he used was not sse3 patched, but he will test superpi with and without the sse3 patch to see what the difference is and how much sse3 helps those chips to perform

update:
the bios incompability issues you have all heard about are about the E4 stepping (san diego 1024kb L2 cache 90nm) not the E3 (venice 512kb L2 cache 90nm) so you dont need a bios update for the venice chips that will be released on april 4th

update:
we will add watercooled and aircooled results later

update:
please stop pming pc ice! hes very busy atm pushing this baby even further

update:
the screenshot indicates he is running 2GB of memory, this is a bug!
however, he will test with 4 DS 512mb sticks and post results

update:
bad news, pc ice hurt his back has the water and air results on a raid array he has problems accessing now. he will post the results asap though.
lets all wish dave to get well again soon

update:
more than stock vcore barely seems to do anything to this chip

lets hope that its just because this is an early chip and that future chips will scale better with more vcore. and 2.8ghz with stock vcore on water is not a bad result by any means

update:
pc ice tried it again under his cascade and the chip didnt boot at -90Â°C either, so theres def a low temp bug. :/

_

Alec · 03-25-2005, 02:26 AM

woah! Thats a sweet overclock man!

crank those volts up a bit and lets see 3GHz!

Awsome 3dmark 2k1 scores too.

Richie · 03-25-2005, 02:27 AM

Woah! .6ghz improvement with stock volts! nice work man, thats a sexy chip,

really goes to show that 3d mark 2001 is based alot on CPU, as i use to think that 3dmarks were mostly video card related. thats a great improvement,

thanks for pics

Good work

conrad.maranan · 03-25-2005, 02:28 AM

Take note of the 2GB of RAM at 285MHz 2.5-3-3-7.

saaya · 03-25-2005, 02:30 AM

Quote:

Originally Posted by conrad.maranan

Take note of the 2GB of RAM at 285MHz 2.5-3-3-7.

yeah and dave said it could do 330mhz :o
dont know if thats with 2gb though

great work dave! keep pushing that baby

conrad.maranan · 03-25-2005, 02:31 AM

Quote:

Originally Posted by saaya

yeah and dave said it could do 330mhz :o
dont know if thats with 2gb though

I'm pretty sure it is. Those new memory controllers rock!

bachus_anonym · 03-25-2005, 02:31 AM

Nice! I believe CPU-Z showing 2048MB is a bug, right?

also...

Quote:

G-SKILL XXXXXXXX @ddr660 2.5-3-3-6 1T

???

conrad.maranan · 03-25-2005, 02:33 AM

bachus_anonym - It's not a bug. And about the G.Skill, I'll keep my mouth shut and let pc ice do the talking.

bachus_anonym · 03-25-2005, 02:41 AM

Quote:

Originally Posted by conrad.maranan

bachus_anonym - It's not a bug. And about the G.Skill, I'll keep my mouth shut and let pc ice do the talking.

are you sure, Conrad ? - http://www.xtremesystems.org/forums/...&postcount=430

otherwise, it would be awesome

maybe that gskill is FF... let's wait for pc ice...

EDIT: Just looked at Sandra... "Memory used by test - 512MB"

So it's 1GB

conrad.maranan · 03-25-2005, 02:45 AM

You're right. I was just looking at that as well.

googles · 03-25-2005, 02:49 AM

at last a 3500+ venice being tested

looking forward to more results...

saaya · 03-25-2005, 02:50 AM

Quote:

Originally Posted by bachus_anonym

EDIT: Just looked at Sandra... "Memory used by test - 512MB"

So it's 1GB

should have seen this as well

gundamit · 03-25-2005, 03:20 AM

Quote:

Originally Posted by bachus_anonym

EDIT: Just looked at Sandra... "Memory used by test - 512MB"
So it's 1GB

Dang! Too late! I saw the same thing well after I had already crapped my pants! Would still like to know if 4x512 will run 1T. Wouldn't mind if it results in slighty lower OC, but don't want the double whammy of lower HTT and 2T timing.

Evil_Spork · 03-25-2005, 03:40 AM

woot!

*buys a 3200+*

MaxxxRacer · 03-25-2005, 03:55 AM

I just fell out of my chair... stock volts and 2.8ghz.... i think i smell 4ghz on venice!! Dont call me ti4200 but I think we got something here.

Cant wait to get my winnie out of here and grab a venice.

skate2snow · 03-25-2005, 03:57 AM

probably he has 4x512Mb and we dont see lower? Because that would be the first time I see CPUZ showing bad memory size.... If it is a bug, it is probably because of the new Memory Controller?

matt9669 · 03-25-2005, 03:57 AM

Just FYI:

http://www.xtremesystems.org/modules...rder=0&thold=0

It's up on the front page!

skate2snow · 03-25-2005, 03:58 AM

Quote:

Originally Posted by MaxxxRacer

I just fell out of my chair... stock volts and 2.8ghz.... i think i smell 4ghz on venice!! Dont call me ti4200 but I think we got something here.

Cant wait to get my winnie out of here and grab a venice.

Its doing a while GeForce ain`t connected.... you are replacing him? lol

Mastakilla · 03-25-2005, 04:31 AM

in the 3th screenshot (sisoft sandra mem bench) u can clearly see its 2x 512MB

u can also see some of the timings..

does that "2CMD" mean its only running at 2T?

also im wondering why its only using 2 way mem interleaving?
isnt that normally 4 way?

intresting results btw, would love to know how it does at normal temps too though

Th3__W00T_GuY__ · 03-25-2005, 04:38 AM

omg rocks!!!

saaya · 03-25-2005, 04:39 AM

Quote:

Originally Posted by Mastakilla

in the 3th screenshot (sisoft sandra mem bench) u can clearly see its 2x 512MB

u can also see some of the timings..

does that "2CMD" mean its only running at 2T?

also im wondering why its only using 2 way mem interleaving?
isnt that normally 4 way?

intresting results btw, would love to know how it does at normal temps too though

if we have enough time dave will also post watercooling and air cooling results

Master_G · 03-25-2005, 04:55 AM

What's that you say? Cold bug gone?

Hot diggity damn

I'm going to see what my LS can do with one of these babys ASAP. Might have to change my

motherboard though.

G

odb · 03-25-2005, 05:28 AM

thats pretty damn good results so far, any preorders online anywhere yet? heh

cyph3r · 03-25-2005, 05:32 AM

and which is the stepping ?

saaya · 03-25-2005, 05:38 AM

ADA3500DAA4BP
CBBLE 0504EPAW

same as the 3500+ from steven i think