A64 OC Data

Status
Not open for further replies.
My OC

Case- Antec SOHW1040 (2x120mm top exhaust fans)
Mobo- Asus K8n (non deluxe)
CPU- Athlon64 Socket 754 CAAOC 0443SPMW Clocked at 2.5Ghz
Memory- Micron PC3200 Cas 2.5 3-3-3-6 at 250FSB 1:1 Ratio
Power- Enermax True 465watt dual fan
HDD's- 2X120gb WD1200JB Specials in RAID-0
Audio- Audigy 2 ZS Plat
Video- MSI 6800GT flashed to Ultra @ 420/1160
http://service.futuremark.com/compare?3dm05=455246
http://service.futuremark.com/compare?2k1=8339451
showimage.php
 
Might as well add to the fun:

3000+ Winchester, with an Asus A8V:

On stock air - 285fsb (2565Mhz) Prime stable at 1.55v

Anything more and the system won't load windows.

I can push it to 290fsb (1.50v is fine, haven't tested lower) on my Vapochill XEII, but games are not stable for more than a couple seconds. Again, Prime95 is completely stable. Any higher and the system either won't POST or won't load windows.


The motherboard is definitely holding back the chip. Setting HTT down to 3x, or using a lower memory divider (or even a lower CPU multiplier) does nothing.


... thinking of getting an FX-55 instead :p
 
oc3eh.jpg



HTT at 4x
HTT at 300
Multiplier at 9x
Vcore at 1.6V
Ram Divider at 100
Ram Timings: 2.5,3,3,10


My ram sucks but the cpu and motherboard rocks!


As for the temps. Well I have no clue lol. Everything is reporting around 30C and if you add 10 degrees due to the incorrect temp reporting bug its 40C idle. im amazed.
 
holy shit dude rock on man!!!

awesome overclock

but like you said the ram sucks especially since your running at 10z mhz

take the ram and run it at 133 or 166 or 200 then overclock. since your running the ram clear down at 100,,, all that overclocking your doing is for nothing, you ram is running like molasiss on a cold winter day, and its making your overclock not mean a thing.

running your ram the way you are actually makes performance worse than if you did it the other way.
 
actually the ram is at 150mhz.. which isn't too far off from it's rated speed.
you board isn't able to do 338mhz htt is it? if so, set the cpu multi to 8 and the ram to 100mhz. it'll bump the ram speed up to 168mhz.. though 18mhz isn't a crazy amount, it'll help out the ram bandwidth a bit (though 150mhz isn't gonna absolutly kill the performance.. just slow it down like 5-10%)
 
OC.png


First time experience with A64, any tips?

AMD64 3000+ @ 2500ghz (1.52v) 250x10 1:1
DFI Lanparty UT 250gb
Zalman 7000cu
Geil 512mb PC4000 Ultra Series 2.5 7-4-4 (2.6v)
 
Navi -

thats a good overclock and unfortunately about all you'll get out of that 3000+ CG core.

definately a good overclock for the clawhammer.

i have 2 cg's and they wont go past 2.55ghz so they sit happily at 2.5 any higher gets unstable.

and water cooling isnt even needed since the temperature dont rise THAT MUCH on these. ther eis a noticeable temp. difference but a good HSF takes care of that.

I just got my lanparty last week and its a piece of dog crap.
i can run 2.5ghz but the system 90% of the time either wont post or it wont boot past the raid bios screen.
i cant use the nvidia sata either, the system is totally unusable cuz its unstable. and thats even when i use ports 3/4 cuz sata 1 and 2 wont handle overclock past around 230. but thats cuz port 1/2 isnt locked. so when you overclock the sata bus speed moves too. port 3/4 is locked. but wow,, its the slowest sata i have ever seen.
i have to use my promise TX2 S150 pci card in order to get a stable overclock.

so am rma'ing the piece of poop and getting a MSI K8N NEO Plat. sent to me instead. so hopefully it will do better.
 
quite nice navi. you can probably get an extra 100-200mhz out of the core if you bump the voltage up to 1.6 or 1.65. unless you feel like making a bunch of heat, i wouldn't go too much over this because it really doesn't raise the maximum clock too much.
as an example... my 3200 can do the following:
2ghz with 1.3v
2.2ghz with 1.45v
2.3ghz with 1.57v
2.35 with 1.65v
2.4 with 1.78v

but since you're already at 2.5, you've got a head start on me ;)
 
eclipse those cg clawhammers wont go over 2.5ghz no matter what voltage you use.

even phase change cooling wont let it,. its just the limit of the chip. and is the limit for almost all cg clawhammers.
those who can go over 2.5 are rare. ive seen tons of 3400+ cpu's not go over 2.5ghz

and if your gunna push it up to 1.65v then boy be carefull cuz thats as high as you want to every go. those clawhammers from everything i been told and the research i been dont max out in the voltage at 1.7 to go higher is stupid you'll fry it. so 1.65 is about as high as you wanna push it.

unless of course his is NOT a mobile but i thought all the CG clawhammers were mobiles.
 
There's been quite a few bit of Winchester OC data flowing thru, could anyone that's kept an eye on it tell me what's the highest vcore most go for on them? I know Winchesters generally don't benefit as much from more voltage but I'm curious nonetheless.

I seem to have hit a wall with my 3000 Winchester, currently at 2.3GHz and it just won't do 2.4GHz unless I juice 1.65v thru it (even then it locked up after 6 hours of Prime, and it was getting hotter than I'd like in this PR climate). I've tried just about everything to get it to 2.4GHz, the only thing I haven't tried is setting the HT multiplier to 4x instead of 3x...

Some people have actually reported more instability at 3x but at 4x it'd be slightly over 1,000 (1,068MHz for 2.4GHz or 267MHz FSB)? I've seen a few cases where it was actually more stable that way but I dunno really.

It's silky smooth stable at 2.3GHz / 1.5v though, been running a very thorough stress test on it overnight. Prime95 hasn't been making as much progress as when I run it standalone (I'm also running rthdribl and a looped MP3), think I might close the other apps for the last leg of the test so Prime can take up more cycles.
 
v_lestat said:
eclipse those cg clawhammers wont go over 2.5ghz no matter what voltage you use.

even phase change cooling wont let it,. its just the limit of the chip. and is the limit for almost all cg clawhammers.
those who can go over 2.5 are rare. ive seen tons of 3400+ cpu's not go over 2.5ghz
Click to expand...
you are a victim of the missing chip. the 3700+
it gets the yeilds that don't quite make the fx cut for some reason or another, be it a faultly memory controller or it doesn't scale quite as high as amd would like. sadly, i don't see too many people with these, but i haven't seen one that won't do 2.6ghz easily.
also, you're thinking of the chips before week 40 or there abouts. the newer clawhammers are far better than the older ones.
and as a side note, i can almost get my 3200+ to 2.6ghz. it's just not very stable.. i can make 2.5ghz game stable, but not prime
proof. though i stress now that it's not very stable, so makes the second screen kinda pointless
 
http://www.13thstepdesign.com/benches.jpg

3000+ @ 2.6, 289FSB, CPU/11 Divider
Completely Stock
Asus A8N-SLI Deluxe 1.02, 1003.500 BIOS (Though it was just fine on the ship Bios, 1002.something)
RAM is mediocre though, Corsair ValueSelect - surprised it handles 236mhz to be honest.
Maybe with a decent TCCD chip I can run at 1:1, which would be nice.
Don't remember what week this chip is, will look at it later.
 
the word on the winnies as voltage doesn't really help them. For me I can run at 2.65 ghz 100% prime stable on 1.45 V's upping it more doesnt help, and dropping to 1.4 V's is pretty stable at 2.65 ghz. I can prime for about 20 mins at 2.7 ghz. This is stock cooling, full load in runs about 48 C, and 30 C idle
 
(cf)Eclipse said:
you are a victim of the missing chip. the 3700+
it gets the yeilds that don't quite make the fx cut for some reason or another, be it a faultly memory controller or it doesn't scale quite as high as amd would like. sadly, i don't see too many people with these, but i haven't seen one that won't do 2.6ghz easily.
also, you're thinking of the chips before week 40 or there abouts. the newer clawhammers are far better than the older ones.
and as a side note, i can almost get my 3200+ to 2.6ghz. it's just not very stable.. i can make 2.5ghz game stable, but not prime
proof. though i stress now that it's not very stable, so makes the second screen kinda pointless
Click to expand...
week 32 or older clawhammers (it seems they're rejected fx-55 cores) can easily do 2.8GHz on air, and most could do even 2.9-3GHz on water, which is a "meh" overclock for an FX. (I've seen FX that do over 3GHz on stock cooling).
 
you mean week32 and newer.
and that was precisely the point i was trying to make.. however the fx-55 is different than the rest of the hammers. it has strained silicon. i question what they do with all the ones that don't make spec, because technically it's the only cpu with SS...
too bad there isn't an easy way to find out if a normal a64 is SS or not.
 
CBID can identify the new revision of ClawHamemrs, in the checksum thingy. I forget what the code is, but it's different from the new CH.
 
it switches from 2.5 (what i have) to 2.6 (the new ones that oc really well).. someone needs to tell us the code that their fx-55 is.
centvalny? can we have a cbid screen? hehe :D
 
(cf)Eclipse said:
it switches from 2.5 (what i have) to 2.6 (the new ones that oc really well).. someone needs to tell us the code that their fx-55 is.
centvalny? can we have a cbid screen? hehe :D
Click to expand...
Heres with week42 chip
cbid.jpg
 
hmm, you have sSOI, i'm guessing the extra s makes the difference. and that voltage isn't right is it? i'd be scared if you can really do 2.7 with 1.13v :p

here's what mine looks like
cbid.png
 
Oh, yes I remember now - the Version should be 2.6 for the new ClawHammers. I wonder what's up with the 1.6 NewCastles, though? Maybe they're actually Winchesters?
 
yeah, but i was wondering how much they changed with adding the strained silicon. the version is still 2.6, but the string gets changed from SOI to sSOI..
 
Strained Silicon-on-Insulator is just as good as a generation die shrink, using regular Silicon substrate.

So, 130nm gate with SSOI is about as good as 90nm gate on plain old Si.
 
that seems a little generic and broad, but i suppose i can be willing to go with that.
the question is what happens if the cpu doesn't make the 2.6ghz spec amd sets? they can't just toss it. i'm kinda doubting that it goes into the 3700+'s.. though amd is really slow with their tech documents, so..
 
Why not? If it can't do 2.6GHz, it could do 2.4GHz... Or, they just bin a certain percentage of all clawhammers for fx55, just based on best performance. They certainly didn't tell me how they do it :p
 
but the fx-55 is made on a different process than the rest of the claw/sledgehammers. *shrug*
i guess i'll try not to worry about it too much. if need be, an email to amd could always humour me :D

oh, and for those who are wondering. the IHS is copper plated with.. something. nickel would be my first guess
 
Why would it be any different? They still use the same lithography masks; you just swap the substrates... It's relatively easy to switch from SOI wafer to SSOI... and I don't see why AMD would have two production lines, one that makes ClawHammers on SOI substrates, and on that makes the same ClawHammers on SSOI... it just wouldn't make sense. Nor upgrading to SSOI and degrading back to SOI (since the lithography processess are a bit different)

edit: it seems s939 isn't worth it right now. You can get a 3500+ Winchester, which would do about 2.8GHz, or you could get a 3700+ that wil do 2.8GHz for the same money, and still have better performance. Untill E0 stepping is available, IMO it's too early to adopt WR's yet. Just like with the C0 CH's.
 
http://www.freewebs.com/spudsdude

spudsdude_oc.JPG


That darn SATA 1 & 2 is what got me at first, couldn't get past 209FSB, put the drives on 3 & 4, AGP to 67, and wooooommm .. lots of goodness.

And yup, the voltage is correct (3.5% over in bios).. haven't done prime test yet, but been in autocad for the past 10 hours, 3dmark bunches of times, seems good so far.
 
well igot 2.4ghz pretty easy. uhm on stock everything. will get some screenshots up if i can.... need a proggie to take em with tho. anyhoo here are the specs

3200+s939 @ 2.4ghz / 1.4 vcore (randomly set it there, not sure if it can be done with less)
1gb ocz regular ram @ 200mhz 2-3-3-10 (i think spd is pc3500 @ 2.5-4-4-7 or some wackness)
mobo: neo2 plati

its nice, speedup is definately felt from the extra 400mhz. around sumemr time i'll have enough saved and ready for a nice, quiet internal water cooling system then we will really see whats up. i definately want 2.6 and i can, i'll get the best ram to work on the mobo... cuz 1:1 will be necessary if i get the WC
 
el rolio said:
well igot 2.4ghz pretty easy. uhm on stock everything. will get some screenshots up if i can.... need a proggie to take em with tho. anyhoo here are the specs

3200+s939 @ 2.4ghz / 1.4 vcore (randomly set it there, not sure if it can be done with less)
1gb ocz regular ram @ 200mhz 2-3-3-10 (i think spd is pc3500 @ 2.5-4-4-7 or some wackness)
mobo: neo2 plati

its nice, speedup is definately felt from the extra 400mhz. around sumemr time i'll have enough saved and ready for a nice, quiet internal water cooling system then we will really see whats up. i definately want 2.6 and i can, i'll get the best ram to work on the mobo... cuz 1:1 will be necessary if i get the WC
Click to expand...


If your getting new ram, might at reccomend the G.Skill PC4400LE. Its proven good up to 300mhz 2.5-4-4-8. Very impressive ram, I love it.
 
you just haven't played around with settings enough would be my guess.. that's still not bad.. approaching the limit of my chip ;)
htl = hypertransport link
 
meh, I just haven't looked into it enough I suppose, don't really understand it enough. I get the basics of it ya know. It just didn't seem stable, crashed a few times and rebooted on me.
 
Uhh, I dunno. Don't have sandra installed right now. I bumped back down to stock anyway. Maybe in like an hour or two I'll do it again and let you know.
 
It's a non-CG revision, and a 3000+. This leads me to beleive that it is one of the early "NewCastle"s, actually, a ClawHammer with half it's cache disabled.
 
Status
Not open for further replies.
Back
Top