My little quest

THe chip AMD socket 754 Mobile 3400+
Mobo DFI UT 250 GB
cooling vapochill LS

I've had this chp for a few months now buying it frm newegg, as a new rig for my wife. Also something to hold me over till newer systems arive. She used it for a while with a stock HSF from another A64 i had last year and was Happy for the most part except for an issue with playing java games on it. So being a nice guy i give her my system so she can play games without issues (cus i just could not figure why java was screwy with it) Now I have it and play for 2 weeks running conservativly at 250x10 on the old HSF. Seems to run hot at these speeds, and figure betteer cooling is needed. OK then order up an xp90C and wait. Well since i choose ground shipping and live far from SVC it's gonna be a long wait, and figure if I have to break it down for the new HSF, i might as well see what she can do with my vapo (which just sits idle till I wanna play like now) So i set it up and start playing around. It seems to go well, And i tooling around at 2800mhz. doing well, priming and all i come across gautams SS of 2700 on air. With a lesser chip. and think what i'm doing sucks with ultra fast cooling, and much more expensive chip. Then to make me feel worse I see this thread And say damn, thats same stepping and week as mine. Why cant i get chips like that one. I say, and start on this kick to see what kind of screenie I can get on just the vapo. So the first thing i do is stop prime, and just bump the volts up a tad and go. Well i get to play around at 2900 for a few, and things aint lookin good. cant get crap to run and start looking for answers. I end up finding tweaks here and there to get 2900 going better and i'm much happier now and push further. After a little more toil 3gig falls.w00t! Actualy stable enough to play and bench. running some pi runs 3d01 sandra benchies without crashing. So now happy i can finnaly play around at 3gig i decide to try for a max screenshot. and figure to try for 3.1 pft no problem. go some more I say, and have to start using clockgen , cus booting above 3.1 aint gonna happen. and finnaly get this screenie. which can be verified here http://valid.x86-secret.com/show_oc?id=11512 All in all nice for a 754 chip, and within 100MHZ of the top screenie ever for 754 (done by kunak I believe) Either way tho it's something that i never thought i would see without an FX, and or a tripple cascade.

awesome

I did tell you you'd see 3.2 GHz out of this puppy, didn't I? Nice work, glad to see you've come across a proc you can be happy with, at least for a day or two.

You make me wanting to pour some LN2 on my Winnie which can run Sandra CPU at 2.94 GHz and run SuperPI 1M at 2.9 GHz on air, to see what it could potentially do.

what stepping? once i get my replacement (vmodded) board I have two chips to max out

200 more mhz and you'd have a 3400 that actually ran at that speed.... good job!

crimedog said:

what stepping? once i get my replacement (vmodded) board I have two chips to max out

Click to expand...

Mine is Winchester 3000+ CBBHD 0447.

Overclocking sandbox for A64 939 systems with Winchester and beyond

not you! *slap* that's a great chip though

also digits, what was your ram at for the super_pi ss? just wanna compare, best i got was a little faster than that. why is the 3d01 so slow!?!!

CAA2c is stepping on it, and 3d run was a stock 9700np (locked clocks ATM) cus my 9800 is dead ;(
SuperPI was done at250x10 and bh5 1:1 222 with some extra timmings from a64tweak, but did it with 3d open and a few other windows. I can do pi at 3100, but scores realy suck, cus i have to use 11x multi. that drops rams to like 225 with 166 divider, and the 180 divider isnt very stable. I'll get a shot of a 3100 run today, and try a 3d run at that speed to, but no garauntees on 3d.

hitechjb1: i'd love to see what ur chip could do frozen. you could no doubt hit a 3.4 screenie.

CAA2C is associated with the FX-55 which is believed to be the first 130 nm A64 fabricated in a form of strained silicon called dual stress liner (DSL). Many of those FX-55 can do close to 3 GHz on air/water and 3+ GHz under sub-zero. Dual stress liner technology is believed to be used in the coming 90nm Venice and San Diego.

So some late NewCastle (?)/ClawHammer could possibly be from those strained silicon DSL 939 ClawHammer core configured from dual channel to single channel and with 1/2 L2 disabled. ->

So your 754 3400+ could possibly be from those strained silicon DSL 939 ClawHammer core configured from dual channel to single channel.

Some links about latest silicon technology, Silicon on Insulator (SOI), Strained Silicon (SS), Dual Stress Liner (DSL)

gautam and i were discussing this last night on AIM. It does sound like a good explaination, bbut the 4000+ series is also on this stepping, and knowone seems to test these to see if strained silicon is present in those also. And these new mobiles may be derived from that line as well. THe only thing i cant agree with is that while half the memory controller is killed off (64bit controller on 754 as opposed to 128bit on 939) the L2 cache is still a full 1meg. which is why i went with another mobile

Yes, your 754 3400+ should have the entire 1MB L2.

So some late NewCastle (?)/ClawHammer could possibly be from those strained silicon DSL 939 ClawHammer core configured from dual channel to single channel and with 1/2 L2 disabled.

So your 754 3400+ (CAA2C) could possibly be from those strained silicon DSL 939 ClawHammer core configured from dual channel.


The 4000+ 130 nm 1MB L2 (939 ClawHammer) is possibly from same core as FX-53 which I think may not have strained silicon.

Desktop A64 939 (130 nm SOI)
4000+: ADA4000DEP5AS 1.5V (SH7 CG rev, F7Ah) <- "ClawHammer 939", 1 MB L2, 2.4 GHz x12, 89 W **

A64 FX 939 (130 nm SOI)
FX53: ADAFX53DEP5AS 1.5V (SH7 CG rev, F7Ah) <- "ClawHammer 939", 1 MB L2, 2.4 GHz x12, 89 W

A64 FX 939 (130 nm SOI DSL)
FX55: ADAFX55DEI5AS 1.5V (SH7 CG rev, F7Ah) <- "ClawHammer 939", 1 MB L2, 2.6 GHz x13, 104 W


A64 940, 754, 939 CPU Models, OPN code, rating

WHile you are probably correct. One would need to wonder why they dont reach the speeds of frozen FX cpu's. which can bench at above 3.2gig, while this one can barely do at 3.1, and reports of others are on par with mine.
Does it matter much, not sure, but your point is taken. and comments very appreciated.

here is very crappy compare at 3.1gig the score really sucks, and is worse than above since ram is running much slower. http://service.futuremark.com/compare?2k1=8527279 some sandras, and more pi coming soon.

Question:

Are there any 4000+ ADA4000DEP5AS (939 ClawHammer) with stepping code CAA2C?

Is CAA2C only associated with FX-55 type of core which is believed to have strained silicon.

What is the model code (OPN) of the 754 3400+?

Such as
4000+: ADA4000DEP5AS <- 130 nm SOI
FX53: ADAFX53DEP5AS <- 130 nm SOI

FX55: ADAFX55DEI5AS <- 130 nm SOI strained silicon

The temperature "P" or "I" are different among them. May tell whether the 754 3400+ is strained silicon based or not.

AMN3400BIX5AR is me OPN (corrected)


*edit*
PI at 3100MHZ

sandra at 3100MHZ

seems that dropping drive strength down to 1 helped get it stable enough to do pi at 10x258 with it at 2 the runs would never complete at 258, or sometimes even get started.

Should be AMN3400BIX5AR (with a 5).

The whole discussion started with the CAA2C code which may (or may not) solely associated with the FX-55 type of core. I am not 100% sure at this point.

hitechjb1 said:

Question:

Are there any 4000+ ADA4000DEP5AS (939 ClawHammer) with stepping code CAA2C?

Is CAA2C only associated with FX-55 type of core which is believed to have strained silicon.

Click to expand...

I wasnt sure,if they were the 2C, or ZC but a little digging found a 2C for sale in another forum. (will PM ya a link to pic hitechjb1 as i dont want to kill the guys bandwith,by linking here)
And newegg has a ZC pictured for there 4000+ CPU's

MY mistake. it does have a 5. AMN3400BIX5AR will correct it.

what's the rest of your code man? i have a caa2c 0450dpmw and a caa2c 0450cpmw to play with