A8N-SLI Premium motherboard + AMD FX-57 Wont Clock

Associate
Joined
1 Apr 2006
Posts
117
Location
Central Scotland
Hello,
I have an A8N-SLI Premium motherboard with 1 gig (2x512) of OCZ EL DDR PC-3200 Dual Channel Platinum Rev 2 memory and an AMD FX-57 processor

everything is run at stock including the memory timings ( 2-2-2-5 1T ) manufacturer recommended timings

Not fully understanding the multiplier stuff on this board and not being an overclocker much anyway I thought the Ai Booster was the answer as it would take my chip up to a nice 3 gig and pressumably get some extra out of the memory

But whenever I set the AI Booster to 3% it stumbles along for a few hours or days then locks
at 5% it fumbles along for about 20 mins and anything higher wont even boot fully !

I was told by somebody in here to just shove the multiplier up to 15x from 14x to reach 3 gig but it wont ware it it will accept 14.5 x 200 = 2914Mhz
but thats it :(
so now i'm wondering about going the other way and dropping multiplier and increasing the chip BUS/FSB
I know this will affect memory timings but i'm thinking 218x13 = 2834 Mhz which doesnt overclock the chip but does clock the memory which may seem strange until i mention this machine is used solely for climate modeling and the Memory and it's timings seem to play a bigger role than overclocking the chip a little
The reason I picked 218 for the FSB ( I know it's called something else weird ) was that according to reviews the best setting for my memory is 436 at those timings
much higher fequancies and you have to loosen the timings which defeats the purpose
theres something else new in the mix chipset frequancies have to be kept below 1000Mhz and there is an automatic setting ( default ) for that
now i have been led to believe two differant things
1. this can be pushed to a massive 1600Mhz without the systen crashing which would mean it would accept 218 as thats only = 1090 Mhz
2. this must be kept below 1000Mhz by dropping the chipset to a value of x4
it seems these two peices of advice are wildly contridictory so what to believe ?
one other thing about the motherboard I've read is the mem Frequancy is controlled by the FSB no matter what you set it at in the memory timings section ?
whats the point of that if it's true ?
if it's not true i could just change the frquancy in the memory timings to 436 and be done surely but seems not :(

hope this doesnt seem like a massive amount to ask
Seems the FX-57 although a fast chip is not renowned for overclocking even amongst other AMD chips although 3 gig would have been nice :(

hope somebody can shed some light
best regards
Ian
 
up the multi to 15 and increase the vcore a little bit

would be handy to know what core the FX57 has as the san diego clocks further than the clawhammer
 
I dont think there was ever a clawhammer FX57, it was released after the introduction of 90nm. It should clock well, make sure you use whole multi's too. You might need to increase the vcore slightly to get it stable, I'd avoid using AI booster and clock using the bios.
 
What was wrong with the other thread? I already answered half your questions :confused:

1. this can be pushed to a massive 1600Mhz without the systen crashing which would mean it would accept 218 as thats only = 1090 Mhz
2. this must be kept below 1000Mhz by dropping the chipset to a value of x4
it seems these two peices of advice are wildly contridictory so what to believe ?

Dude, it depends on the board, there is no contradiction in that advice. Generally you should keep it below 1000 Mhz to keep the system stable and since there are no performance hits from lowering it significantly easier to play safe than to not. Some specific boards, such as the A8N32-SLI have reports of hitting massive frequencies (like the 1600 you mention) without any issues, but as I already said, it depends, and the 1000 is good general advice for all boards.


one other thing about the motherboard I've read is the mem Frequancy is controlled by the FSB no matter what you set it at in the memory timings section ?
whats the point of that if it's true ?
if it's not true i could just change the frquancy in the memory timings to 436 and be done surely but seems not

Memory is tied to the FSB since having them in synch is generally the best performance, like I said in the other thread your gaining so little (if anything) by clocking it to 436 there isn't much point, especially as the timings your using are already really fast at default frequencies.

You should just leave the FSB/memory alone and up the CPU mulitplier, you might not get much more out of that chip, as again I said in the other thread San Diegos tend to top out around 3 Ghz on air and especially when using stock CPU voltage which I think you are. Dividers are available to lower the memory clock but you don't need them unless your upping the FSB.
 
Hello,
Sorry it's the San Diago core didnt realize that made a differance
and i tried 15x but no go exactly how much of a voltage increase are we talking here as i'm getting 2.9 gig at stock voltage 1.39 volts put it to 1.45 when I aised the multiplier to 15 and still no go
should i be putting much more voltage into it as i've heard its absolute max is 1.55 and thats trying to get to 3.5 gig with some weird cooling ( i'm on air cooling artic freezer pro 64 )

could you also tell me if i up the multi and get that to work how i up the memory speed if its not done in the memory timings section ?

thanks for your speedy response
Ian
 
Hi Tetras,
Sorry I didn't mean to offend it's just it looked more like my post should be in here than the other section and from what I can gather San Diago's are taken above 3 gig on air all the time 222x14 = 3.1 gig being an example

I I'm just trying to squeeze as much memory performance as chip performance out of this as crunching climate models rellies hugely on memory performance i got a huge performance increase just changing the memory to 1T from 2T something I havent got by adding 1 gig to the chip speed :(

best regards
Ian
 
You should easily be able to run 1.55v for air providing your temps are cool enough. Can you tell us everything it says in the bios. Also what slots are the memory in and have you ran memtest? What are your temps by the way?

As for the ht rubbish it has no affect - best comparison is 3 cars on 3 lane motorway. Make motorway 5 lane and the 3 cars still have a lane each - no speed diff. Try and keep it between 800 and 1000 so if fsb is 200; ht=5. Between 200 and 250 ht=4 and 250+ ht=3. You wont need any lower ht settings as fsb on asus boards doesnt go high enough.

What bios you using as well?
 
Hello,
Everything it says in the Bios ? :eek:

so far after trying to have 200 x 15 to get 3 gig and failing I'm now trying 14 x216 and failing it will however run at 13.5 x 216 to give me 2915Mhz but I've been told not to use a half multiplication factors why i don't know as it works
the HT is at x4 according to cpuz the link ht is sitting at 680 Mhz no where near 800 to 1000, I've turned the mem voltage up to 2.9 volts warranteed maximum and increased the chipset voltage just because i could :)
i had the CPU voltage at 1.5 volts to try the 3gig test but seeing as i cant get it I've lowered it again
I havent run memtest, never heard of it but i run a climate model through it and that stresses the system more than any test I've came across unless memtest tells you something spacific I don't know about
and the memory is sitting in the second dual channel rails as in the first pair of rails they would run but not give me 1T timing no matter what I tried
they now sit at 2-2-2-5 at 1T at 216 Mhz quite happily now
as far as bios version i couldn't tell you but what i can tell you is I can't update it as the ASUS site is crap and wont let me :mad:
if Ive forgotten anything please let me know thats all in the Bios I can think of you may need to know but theres a heck of a lot in their now
best regards
Ian
 
If you load Cpuz and go into the mainboard tab, it will tell you what bios version you have. If you're trying to update the bios directly from the net using asusupdate, don't bother, you'll never connect to the server! Best bet is to download the newest bios manually from the ASUS site & then choose the 'update from file' option. This will then flash the bios whilst in windows. It's not a good idea to have you PC overclocked when you do this however.

Alternatively, you could put the bios on a floppy and do it in dos but it's slightly more complicated this way.
 
Hi,
Under stepping it just says " 1 " ?
the bios is " revision 1009 "

wondered why i couldnt get the online bios update to work at all why do they bother if you cant connect to the server ?

about the voltage thing surely it wont need more than 1.5 volts to get it to 3 gig ?

thanks for your help guys
Ian
 
1 isn't the stepping, he means the 2nd code line under the model code, starts with KAB, CAB or something like that. An example would be CACJE 0548 FPAW.
 
from what you have said, it doesn't sound like you have much experience in clocking chips. Personally if I was in the same boat as you, I'd not be tinkering with a £300+ processor with a good chance of toasting it, unless of course you can afford to fork out for a new processor/mboard. There are many overclocking guides, find and do a little research, as you have a processor which is unlocked (multiplier) it makes overclocking much easier, hence why these chips always cost a premium.

Brutally honest, even if you get 3.3 on air, you are not going to see a huge performance increase, the work you do sounds like dual core would probably be much better suited, in which case flog the fx57 and buy an X2.

It's very easy to tell people what they should be doing, however in this case, the OP really needs to learn about clocking processors a little more, before blindly following people's advice. I can see his next thread 'OMG system is dead!!!'.

To the OP, we've all been where you are at, it's hard learning from scratch, but you need a little perseverance, if you learn the basics, us ugly lot can then help you do the fine-tuning to get that little extra :P

G luck
 
Hi,
I've been building my own computers 25 years and am fairly knowladgable about the electronic and electrical processes involved in high frequancy data processors, I thought i'd lay it all out in a conventional format instead of lab speak so others with similar problems could follow any advice I was given,
as far as the chips being unlocked I know this and i also know that shifting the multiplier should be the answer but its not and as i wanted to raise the clock of the memory in tune with the chip raising the multiplier wasnt really even if it had worked what i wanted to do
I've tried both ways of overclocking this chip and added up to 0.10 volts and still it wont go above 2950 and remain on full load stable
as you say adding all sorts of voltages into these devices is not a clever idea I have heard of people and seen how to reports by people getting these chips to run stable at 3.1 on air but mines is just not interested so i can only assume the chip is weak or the batch was weak that month and it just wont clock much above 2.9 gig
as you say i only wanted 3 gig anyway so the differance of the 100 meg of cpu power has been small but the advantage of sticking the memory up to 436 meg has been massive in comparrison
don't know where you can get FX-57's for £300 but mine cost £400 a few months back before I decided to start climate modeling and didnt realize the benifits of twin core processors in that spacific application
and i couldn't really bring myself to sell my FX-57 half price now to get a mid range twin core
best regards
Ian
 
Last edited:
IanBoyle said:
Hi,
I've been building my own computers 25 years and am fairly knowladgable about the electronic and electrical processes involved in high frequancy data processors, I thought i'd lay it all out in a conventional format instead of lab speak so others with similar problems could follow any advice I was given,
as far as the chips being unlocked I know this and i also know that shifting the multiplier should be the answer but its not and as i wanted to raise the clock of the memory in tune with the chip raising the multiplier wasnt really even if it had worked what i wanted to do
I've tried both ways of overclocking this chip and added up to 0.10 volts and still it wont go above 2950 and remain on full load stable
as you say adding all sorts of voltages into these devices is not a clever idea I have heard of people and seen how to reports by people getting these chips to run stable at 3.1 on air but mines is just not interested so i can only assume the chip is weak or the batch was weak that month and it just wont clock much above 2.9 gig
as you say i only wanted 3 gig anyway so the differance of the 100 meg of cpu power has been small but the advantage of sticking the memory up to 436 meg has been massive in comparrison
don't know where you can get FX-57's for £300 but mine cost £400 a few months back before I decided to start climate modeling and didnt realize the benifits of twin core processors in that spacific application
and i couldn't really bring myself to sell my FX-57 half price now to get a mid range twin core
best regards
Ian

Well I quoted 300 mainly because that's what you get on 'the bay' on a good day. Also, have you had a look at how much dual core processors are from overclockers themselves : http://www.overclockers.co.uk/acatalog/Socket_939_X2.html

you'd see that for a smidge under 200 you get the 4800, which is easily clockable to prob 2.8 with only using fsb o/cing. As you have given a little more of the game away about how much you know, I personally wouldn't touch the fsb, leave that and the htt set as they were, and try for x15 multi.. As long as you don't go past 1.6 vcore, you should be fine providing there's no load on the core. Just keep adding little voltage each time, make sure you are able to monitor the temps very closely, the only way at 1.6 you'll kill the processor, is if the cooling is inadequate, and if you monitor the temps good enough, this will be avoided..

good luck anyways mate..

PS Gibbo, how do I sign up for commision!!! :D
 
Back
Top Bottom