Sen's Ultimate Winchester Thread

[thousandLies]

well i can get into windows with 260 with 3-4-4-10-2t 2.85v....unstable

 

[man_utd]

OC550: Please help us confirm for you that you do have a winchester
GCPUID
http://addict3d.org/index.php?page=downloadfile&ID=3110

Please download and post a screenshot.

 

[Sentential]

sent, how do you know if its your ram or your mobo that wont go over 280fsb? even if it wont, just use the 10x multi, dont waste time rmaing a mobo that was only guaranteed to do 200fsb. My mobo did 300fsb, this is the highest I tested it so far but CPU-z says I got a newcastle and dual channel wont post. I can get into windows at 2.6GHz at 1.4(overvolts to 1.45) but more volts does not help me overclock higher, it only stabilizes 2.6GHz. I think 2.54GHz at 1.4v was stable. This newcastle sure does not like volts!

Ill figures something out.... most likely Ill ebay this on monday once I order my replacement. Either way, this mobo is on its way out.

 

[thousandLies]

so guys, should i pick up the ocz ddr booster and try my luck with 3v+?

or should i try new ram

 

[man_utd]

If it's bh-5 yes. However, I have never seen bh-5 stable with cas3

 

[bass]

so guys, should i pick up the ocz ddr booster and try my luck with 3v+?

or should i try new ram

Get a booster mate, with mine my bh5 is good for 260+ 2/5/2/2 with 3.5v up it's ***, not a problem if you have a fan cooling your dimms

 

[G0DZILLA]

So what kind of MOBO and ram are you going to get now ?

 

[cmitchellfly]

Well, with the wonders of my own luck, I have successfully screwed myself.
I can get into Windows with 9x / 285 with stock voltage, but Prime 95 will error instantly, as it will do all the way to 1.60V. I'm down to 9x / 277, for a whopping 2497MHz. Doh. Hopefully it will prime at this, if it doesnt I'll cry. My memory is stable at 285 at 2.5-3-3-8 (if my memory (no pun) serves me right), but cant go much higher. Suckage.

Ben, What ram are you using?

 

[cmitchellfly]

No....Im gonna beat up on this sorry sack-of-**** until I sell it. This is just the way I am on my parts Im sure Ill be happy with CPU #2, which will probly happen shortly after Xmas.

Sen, I know how you feel....and the best I could get with my 3200 Win was 2.4g Prime stable (with my OCZ 3200 EL Plat r2). I almost sent my Neo2 back but I bought a 3500 Win instead. I was obsessed with getting at least 2.6g stable.

 

[Sentential]

Im not 99.9% positive that my mobo is to blame. Not even 2 days ago did I have 2.38 stable on default vcore. Now it fails instantally even with 10%??

Pushed the voltage down to 0% and primed at 2.32....ironically it was stable for 3 hours.

Tried 2.52 again @ 10%....stable for a 1/2 hour.......Is it me or does this not make ANY sense at all. Im blaming hte mobo, especially since it wont even boot over 290 under ANY configuration, when Osirus is at 314 stable

 

[Optimus Prime]

my msi craps out at 310... so I would say if you are sure that your board cant handle at least 300 it is bunk. As for prime, I think we need to stop using prime here. No offence but it is just gonna drive you crazy. with an a64.

And long and the short of it. I have 2 of the gigglebytes here for testing so I will through my chip in em and see what happens. If the results are good I will post back tonite and let you know.

 

[Sentential]

my msi craps out at 310... so I would say if you are sure that your board cant handle at least 300 it is bunk. As for prime, I think we need to stop using prime here. No offence but it is just gonna drive you crazy. with an a64.

And long and the short of it. I have 2 of the gigglebytes here for testing so I will through my chip in em and see what happens. If the results are good I will post back tonite and let you know.

I dont think you understand.....it physically will not POST at 279 with 1T and 289 with 2T. Thats REALLLY low.

 

[Osirus]

Well when you get a new board, I do wish you all the luck with it. I hope you get a good one.

 

[Joe Camel]

I dont think you understand.....it physically will not POST at 279 with 1T and 289 with 2T. Thats REALLLY low.

ya, that is low...my AV8 ran P95 @ 300x9 for 3.75 hours lastnight, i had to stop it when i went to bed

Sen, how does it "not POST"...blinking cursor in the upper left or does windows start then restart or what/how?

odd things for me: my mem will run 270 (2.5,3,3,10) but with this 300x9 i have mem @ "333" so it runs @ 240...so you would think it could do the same timmings?...NOPE! needs to be: 3,3,3,10 or windows wont load (file repair or something message)

 

[CandymanCan]

My b oard is totally wierd, if i use diff ratios like 166-133 when i set my HTT up really high even tho my memory runs much slower then it normally does i get memory error's in memtest 86, thisi s beyond 300HTT when it happens. I dunno why but my board also doesnt like running past 260HTT using the 3xHTT multiplyer, so i have to revert to 4x cuss it gets stable again. Its like my board refuses to run @ lower speeds but loves it when i push it really high.


Off topic here, do you like the change from red/black to greyblue/white ? Should i change the blue or white to something else cuss my eyes were dying after awhile with my black and red sig

 

[Joe Camel]

im a little lost...i understand when we were running non64s and wanted to get the FSB way up to get more MEM bandwidth, but (to me) mem bandwidth comes from: total CPU speed and mem "FSB".

i say this because with CPU: 247(HTT) x 11 = 2717MHz MEM: 400 (247MHz), im getting almost exactly the
same benchmarks as....CPU: 300 (HTT) x 9 = 2700MHz MEM: 333 (240 MHz)... am i missing something?

 

[hitechjb1]

im a little lost...i understand when we were running non64s and wanted to get the FSB way up to get more MEM bandwidth, but (to me) mem bandwidth comes from: total CPU speed and mem "FSB".

i say this because with CPU: 247(HTT) x 11 = 2717MHz MEM: 400 (247MHz), im getting almost exactly the
same benchmarks as....CPU: 300 (HTT) x 9 = 2700MHz MEM: 333 (240 MHz)... am i missing something?

As far as CPU and memory bus frequencies, the two cases are:

First:
CPU = 247 x 11 = 2717 MHz
memory_HTT_ratio = 1:1 (200 setting)
so memory_divider = 11
memory_bus_frequency = 2717 / 11 = 247 MHz

Second:
CPU = 300 x 9 = 2700 MHz
memory_HTT_ratio = 5:6 (166 setting)
so memory_divider = 11
memory_bus_frequency = 2700 / 11 = 245 MHz

So both CPU and memory frequencies are less than 1% in both cases, hence no clear performance measurement difference.

For the second case, if the LDT multiplier is kept the same (say x3), the system HT bus would be much faster, about 20% faster. But HT bus might not have signigicant impact for those benchmark you were measuring.

 

[StriderGT]

Strange BIOS OCing behaviour

Hi everybody, I am new 2 the forum, not new 2 ocing and yes A64s are different beasts...
I have 3000+ Winchester and a Neo2 Platinum 1.3 official bios. I am trying to develop an OCing calculator for A64s on Neo2 Platinum (Propably will be usefull on all similar m/bs) but stumbled across some weird clocking behaviour:

When I set
Memory to 133 (2/3 divider)
HT multi: 4x
CPU Multi: 9x
FSB/HTT: beyond 209 (including usefull 250 :-(
the system crashes during post and reboots in safe bios mode.
All voltage settings were defaults.
Memory timings were: 3-3-8-3-1T with 2x512 DDR400.
All values are theoretically legit 836+ HT bus, 1881+ CPU, 139+ Mem Frequency.
Has Anyone reproduced the same situation?

PS1 The system runs stable (Prime95 etc) with much more agressive settings like FSB:300, Memory 133(2/3) 200Mhz, 8x cpu = 2400
I can even boot in 2700 just for office work though...

PS2 Is the grey DDR Memory Frequency reported by the Cell Menu (when Memory Timing is set to 166 or 133) correct?!? I guess not because it does not coincide with the memory frequency reported by CPU-Z.

PS3 I have allready posted a new thread under the MSI AMD boards topic, but I later found out that more people are reporting strange Winchester clocking behavoiurs in this forum. I suggest that we sum up what is not workable under a new thread (within the theoretically very easily achievable settings) and start investigating what's going on. I think that the MSI K8N Neo2 BIOS is doing much different things than those it reports and its not just a half multiplier case as clearly indicated by my example...

Any help would be appreciated

Clarification:
I am not trying to overclock my CPU. I ve done that with many different ways. As I said I am trying to develop an Overclocking calcualtor for A64 Winchesters. The problem is that some of the valid values: CPU Freq, FSB Freq, HT Freq, Memory Freq produced by some multiplier combinations are not workable in BIOS (as those mentioned in my previous example). I am trying to figure out why these very easily attainable values are unworkable and I suspect BIOS bugs. I would appreciate if someone could repeat the unworkable combinations of my previous example and post his results or way better if he understands what's going on under these very specific situations.

 

[enduro]

Well, with Nforce4 emerging you guys might have more options opening for you. Sen, you can finally ditch that MSI junk board and try out a NF4. Granted it's Asus, but their quality was good about a year ago. Maybe they are on the rebound.

 

[ewalk123]

Ya, and all you have to do is sell your AGP card and get a PCI-e one