Overclock 64bit 3000XP

jessy

Anyone here overclocked a 64bit 3000XP 512K L2 catch. just wondering whats the max you got out of it.the Max stable speed i can get is 2.24.

Azza

Your overclock is what to be expected of the AMD 64 3000. I assume your using the socket 754 130nm newcastle chip. You might want to consider switching to socket 939 and the new whinchester chip AMD 64 3000. Clocks from 1.8ghz to 2.62 on crap air cooling cheap too

jessy

thanks for the advice azza, but i was thinking about this one
AMD Clawhammer 64bit 3700XP 1mb L2 cache.

done suppose any has overclocked this one yet. if so let me know.

gline

I think i only got an xtra 200mhz out of my old athlon 64 on stock cooling, they arent that ood or overclocking. Yeh ive heard that the socket 939's are really cool, id luv to see the athlon fx53 939's, imagine what i could do to that, mmmmmmmmmm

Azza

I would not recommend the 3700 is way too expensive. And single channel. Getting any sport of 939 90nm chip will get you overclocks near 2.6 which is what you probably get with the 3700 overclocked. Alas you will only have half the cache but your memory will be running in dual will compensate abit. Plus you be saving god know how much cash.

Jammer

got an extra 440mhz outta my A64 3200+ !!

BKtje

hey arent that ood or overclocking.

I beg to differ. 32% overclock on mine (clawhammer and phase change but still). Dunno how far this would go on air but with a decent air cooler ive seen people get pretty good overclocks from em.

Also im my experience (comparing my rig with CC's at same cpu speed) the dual channel showed no extra performance in gaming. Dunno about other apps.

EDIT: Btw if its a 64bit cpu its no longer an XP

Woden

i don't know i don't think 32% on a phase cooling set up to be a particurly good but thats a matter of opinion.

the only time i've looked at the 64bit processors and gone thats gonna be nice for overclocking is the new winchesters

BKtje

i don't know i don't think 32% on a phase cooling set up to be a particurly good but thats a matter of opinion.

Well its no p4 but amd chips are more powerful per clock cycle (usually) so a 32% (aka 630mhz) overclock would probably be over 1ghz on a p4 (tho thats just me guessing).

These might not overclock as well as a P4 but for a relatively immature chip i think its pretty good.

Azza

I calculated it at 100mhz AMD = 150mhz Intel. So probably 950mhz overclock which is good

Again the reputation that AMD 64 are not great overclockers is not exactly fair. Most people get over 200mhz overclocks out of them which equals about 350mhz intel. The P4 are only marginally better overclockers the norm looking around 400mhz on air cooling. You can give or take another 100mhz to each side.

Seem on phase changing Intels overclocking potential improves more than amd. However AMD are still waiting for a 64bit OS. Probably in the end pan out 6 of 1 half a dozen of the other.

BloodBath

I calculated it at 100mhz AMD = 150mhz Intel. So probably 950mhz overclock which is good

Again the reputation that AMD 64 are not great overclockers is not exactly fair. Most people get over 200mhz overclocks out of them which equals about 350mhz intel. The P4 are only marginally better overclockers the norm looking around 400mhz on air cooling. You can give or take another 100mhz to each side.

Seem on phase changing Intels overclocking potential improves more than amd. However AMD are still waiting for a 64bit OS. Probably in the end pan out 6 of 1 half a dozen of the other.

Depends on the intel chip. Most 2.8ghz prescott's will do at least 3.6-3.8ghz on air cooling. That's 800-1ghz overclock. Dubdvd's 3.2 goes to 4.1ghz on air cooling so i'd say 400mhz is a bit of an under estimate. Mine only does 400mhz unfortunetly but the psu is causing some of that trouble and the fact that it's the older C0 stepping. The new D0's are much better.


BloodBath

astrofool

there's also the problem now that intel have cancelled their 4ghz p4, meaning basically that when overclocked to that speed, the chips won't be error free (kind of like what happened with the p3 1.13ghz if anyone remembers).

Azza

Again the reputation that AMD 64 are not great overclockers is not exactly fair. Most people get over 200mhz overclocks out of them which equals about 350mhz intel

Opps bad math on my part that should equal 300mhz intel.

Dubdvd's 3.2 goes to 4.1ghz on air cooling so i'd say 400mhz is a bit of an under estimate

Yes I have seen awesome overclocks on air but its pretty rare to achieve more than 500+ on air intel wise not saying it can't be done taking average agains. From what I have seen 800mhz to 1 Ghz is not the norm. I have heard of AMD 64 going from 2.2 to 2.7 on air which was a one off ...thats a 750mhz overclock intel wise but again its rare to hear above 350mhz.

Again with these Winchester chips going from 1.80 to 2.62 thats a 45% (which translating to intel terms is 1200mhz overclock) on average air cooling looks like AMD have caught up in overclocking terms. Incidently AMD are claming these chips have no extra tweaks but according to several independent reviews there out preforming there old 130nm by a small margin at the same clock speed. Phase cooling has seen them hit 3.0Ghz

BloodBath

there's also the problem now that intel have cancelled their 4ghz p4, meaning basically that when overclocked to that speed, the chips won't be error free (kind of like what happened with the p3 1.13ghz if anyone remembers).

Your wrong astrofool. People test their max overclocks for what is stable using the likes of prime 95, running the test for a good 12 hours or so is usually enough to determine whether the chip is stable and error free. Dub's runs stable past 4ghz and I have seen many other machines run stable and error free past 4ghz.

Yes I have seen awesome overclocks on air but its pretty rare to achieve more than 500+ on air intel wise not saying it can't be done taking average agains. From what I have seen 800mhz to 1 Ghz is not the norm. I have heard of AMD 64 going from 2.2 to 2.7 on air which was a one off ...thats a 750mhz overclock intel wise but again its rare to hear above 350mhz.

Your right in a way but only if people buy the high speed chips. It's the same with the amd's. You could buy a 3800 but you wouldn't get much of an overclock out of it. I'm talking about the lower speed D0 steppings especially on the new socket. They all seem to max out around the same level. The 2.8's will usually get 3.6-3.8ghz while the 3.2's will normally get at least 3.8ghz.


BloodBath

astrofool

if you can find it read up on the p3 1.13ghz debacle. Basically Intel released it, then had to withdraw it after it failed some high load compilations with a cpu error, even though other chips could be overclocked to beyond that level, the chips were never stable enough to resale. Having Intel cancel their 4ghz p4 is basically the same problem all over again.

BloodBath

Ffs astrofool. I know plenty of people who have chips past 4ghz with no stability problems what so ever. Instead of comparing it to a problem that happened with a P3 read up on the facts.

It's not like the chips magically lose stability past 4ghz. I'm sure it's probably because intel couldn't get a high enough percentage of them to clock to 4ghz on a low voltage so they decided to scrap it all together.

The people who get them to 4ghz and beyong push the voltage to 1.55v and sometimes higher which is a decent bit above stock and requires a decent air cooler.


BloodBath

Azza

Bloodbath is right. For what I heard it was nothing to do with the architecture behind the Prescott that prevented from being sold at a stock speed of 4.0ghz. The reason I have heard is that Intel did not want to go far behind AMD in terms of releasing a a dual core cpu's so they abonded the 4.0ghz to concentrate on the new smithfield core. There due out now just a month or 2 after AMD's rather than almost a year behind. Alas for hyper-threading intel fans its being dumped for the dual cores.
They also will not be running the new 1066fsb but will be sticking with a 800mhz bus.

On the Plus side for Intel these are gonna be for desktop while AMD dual cores will be on server first. Its like the situation with amd 64bit chip technology Intel will have it on the desktop first but will haev next to nothing to take advantage of it. The 3 chips will be 2 .09nm prescotts x20 (2.8ghz)
x30(3.0ghz) and x40 (3.2ghz). They will also have 64bit extensions.

BloodBath

Cool about time they took things a different direction. It's still going to take a while before they are much use though. All programs and games will have to be programmed to take advantage of the 2 cores unless they come up with some clever way of doing it.


BloodBath

astrofool

right, before you say "ffs", find out a little about what happened with the p3 1.13. For Intel to have diverted a team from doing 4ghz means that they were looking at manufacturing tweaks in order to do so, and means that the current prescott core cannot be verified for sale at 4ghz.

Yes it will run at 4ghz and run 99% of things properly, but some things won't run like they would at stock OTHERWISE intel would have released a 4ghz model already.

capiche ?

BloodBath

No not capiche. Your wrong.


BloodBath

astrofool

no u're wrong (how mature an approach !)

gline

Intel must have had a damn good reason to pull out from producing of 4 ghz chips, because it would have cost them millions to stop that

gline

Azza wrote:

I calculated it at 100mhz AMD = 150mhz Intel. So probably 950mhz overclock which is good

Again the reputation that AMD 64 are not great overclockers is not exactly fair. Most people get over 200mhz overclocks out of them which equals about 350mhz intel. The P4 are only marginally better overclockers the norm looking around 400mhz on air cooling. You can give or take another 100mhz to each side.

Seem on phase changing Intels overclocking potential improves more than amd. However AMD are still waiting for a 64bit OS. Probably in the end pan out 6 of 1 half a dozen of the other.

I think amds are just as good for overclocking, probably better. The only thing holding amd's back is Voltage, with a decent volt mod on ur board with a phase change kit u can get massive oc's on amd's, The fx51 can get 3ghz and over with a board thats voltmodded.

astrofool

you can read a few articles on the P3 1.13ghz here:

http://www.tomshardware.com/cpu/2000.html, from august. Interesting bit of processor history. The chip was removed from sale soon after this. (it only ever shipped a couple thousand, Dell's mostly if I remember correctly). Only when tualatin came out did the p3 go to 1.13 properly (a complete die/process revision).

secret_squirrel

astrofool wrote:

right, before you say "ffs", find out a little about what happened with the p3 1.13. For Intel to have diverted a team from doing 4ghz means that they were looking at manufacturing tweaks in order to do so, and means that the current prescott core cannot be verified for sale at 4ghz.

Yes it will run at 4ghz and run 99% of things properly, but some things won't run like they would at stock OTHERWISE intel would have released a 4ghz model already.

capiche ?

Sounds like wishful thinking to me...whats far more likely is that intel couldnt get economically viable yields for 4+Ghz prescotts - without some manufacturing tweaks - which would require a team better used on the dual core prescott or the dual core pentium M projects.

StRiKeR

astrofool has a point, and how many of us uses everything that are out there? but hey intel does indeed has to test every single applications that are out there to make sure they works fine with it! if one major or comment professional app fail for any reason, surely they not gonna release it! cos intel are not only aiming at users like gamers, also professional! and they not gonna release a gaming cpu that runs at 4GHz but they gonna say hey this cpu not being release for such app so dont complain if it crashes!
no they cant do that!
if they find one test fail, they wont release it which is the right thing to do, but we never gonna know exactly why!

btw did you guys read all of the toms article?? remember seeing this part that was stated??

"Interestingly, stress tests as Prime95 or CPUburn under Windows98 would not get my 1.13 GHz processor to fail on the VC820."

does that mean anything to you??

Woden

it would be unreasonable to believe that intel cannot make a 4ghz chip. a number of chips would surely pass all the q.c for 4ghz. just looking at the fact that will ship 3.8ghz chips in number would mean a number would get to 4Ghz. a different question though would be is it feasible for them to release a 4ghz and the answer is obviously no.

secret_squirrel

StRiKeR wrote:

but hey intel does indeed has to test every single applications that are out there to make sure they works fine with it!

Major assumption there - I'd lay odds that its not true. I'd be surprised if the number of programs tested get much out of the ten's.

They are far more likely to have custom apps that stress particular parts of the core.

Any one else remember the calculation bug in early pentiums?

gline

secret_squirrel wrote:

Major assumption there - I'd lay odds that its not true. I'd be surprised if the number of programs tested get much out of the ten's.

They are far more likely to have custom apps that stress particular parts of the core.

Any one else remember the calculation bug in early pentiums?

I'd have to agree, they would probably have special ways of testing a cpu, rather than using other companies software which has its own bugs

StRiKeR

I was just trying to make an example, if theres anything wrong at all, they not gonna release, they dont want to have the same thing happens again!

thats one thing that has proved even if it passes Prime95 it doesnt really mean its a perfect overclock! it works fine for us cos we not using anything that may crashes or hang!

I dont feel the needs of overclocking my rig apart from doing some bhenchmarks to see what it can go up to without crashes, but I'm happy with the stock speed!

at the end the day, all those CPU are vary from one to another, even if they could be the same chip being sold at different speed!

I've read some review and benchmarks that was showing the Athlon64 3800 was running very close to a FX speed, dont remember exactly but atleast they can clock to the FX speed! and cheaper, a good bit cheaper! I still have a Athlon64 3000 rig also, running at stock speed but can clock to nearly 2.4GHz on watercooling!(oh I bought 2 set of watercooling before I even got this P4, just as well)

gline

StRiKeR wrote:

I was just trying to make an example, if theres anything wrong at all, they not gonna release, they dont want to have the same thing happens again!

thats one thing that has proved even if it passes Prime95 it doesnt really mean its a perfect overclock! it works fine for us cos we not using anything that may crashes or hang!

I dont feel the needs of overclocking my rig apart from doing some bhenchmarks to see what it can go up to without crashes, but I'm happy with the stock speed!

at the end the day, all those CPU are vary from one to another, even if they could be the same chip being sold at different speed!

I've read some review and benchmarks that was showing the Athlon64 3800 was running very close to a FX speed, dont remember exactly but atleast they can clock to the FX speed! and cheaper, a good bit cheaper! I still have a Athlon64 3000 rig also, running at stock speed but can clock to nearly 2.4GHz on watercooling!(oh I bought 2 set of watercooling before I even got this P4, just as well)

"I dont feel the needs of overclocking my rig apart " - Do you not want more speed for your money???

BloodBath

no u're wrong (how mature an approach !)

Maybe not mature but it's true.

you can read a few articles on the P3 1.13ghz here:

http://www.tomshardware.com/cpu/2000.html, from august. Interesting bit of processor history. The chip was removed from sale soon after this. (it only ever shipped a couple thousand, Dell's mostly if I remember correctly). Only when tualatin came out did the p3 go to 1.13 properly (a complete die/process revision).

Has no relevance to this situation at all.

Sounds like wishful thinking to me...whats far more likely is that intel couldnt get economically viable yields for 4+Ghz prescotts - without some manufacturing tweaks - which would require a team better used on the dual core prescott or the dual core pentium M projects.

Exactly.

it would be unreasonable to believe that intel cannot make a 4ghz chip. a number of chips would surely pass all the q.c for 4ghz. just looking at the fact that will ship 3.8ghz chips in number would mean a number would get to 4Ghz. a different question though would be is it feasible for them to release a 4ghz and the answer is obviously no.

And I guarentee you practically every single one of those 3.8ghz chips will do at least 4ghz with no voltage modification, maybe some will need a slight increase and they will be perfectly stable. I'd expect at least 4.2ghz out of them tbh.


BloodBath

StRiKeR

gline wrote:

"I dont feel the needs of overclocking my rig apart " - Do you not want more speed for your money???

I'm not too bother, If I want more speed I buy a faster chip, I can afford it!

astrofool

I guess Intel are too idiotic to release a 4ghz p4 then, seeing as every chip goes to that speed anyway.

Or maybe they want AMD to have all the speed records.

StRiKeR

theres no harm in holding off for a while for whatever reason they may want to do that I guess!

secret_squirrel

Since the fastest A64 is only running at 2.6 or 2.8Ghz I think Intels speed record is safe for a bit.

Its become pretty meaningless at this stage anyway. Especially since the pentium M is argueably a better processor than the p4 anyway.

All the major manufacturers have to move away from pure Ghz ratings since next years dual core processors are likely to run significantly slower (in Ghz) than the processors they replace.

Its just sad that pride on intels part wont go down the 'notional Ghz' route like AMD with their 3800+ etc, but would rather opt for a numbering system that is completely alien and incomprehensible to the majority of users.

BKtje

Since the fastest A64 is only running at 2.6 or 2.8Ghz I think Intels speed record is safe for a bit.

Fx goes a fair bit above that m8.

secret_squirrel

err...no it doesnt - stock speed for the new FX-55 is only 2.6Ghz. Anything thats overclocked doesnt count since its not an official product.

Max overclocks of any chip is a completely different issue to the one I was posting about - namely Intel's 'Ghz is Good' ex-policy.

Azza

However much a fan I am of the AMD 64 processors the naming convention AMD is too confusing for example the AMD 64 3200 there is three of them (2.0Ghz 1MB Cache, 2.2Ghz 512kb Cache both socket 754 and the socket 939 version at 2.0Ghz 512kb Cache).....how about the socket 754 3400 out preforming the 3500 with its extra half a mb of cache despite not using dual channel. Also its hard to tell the difference between them and the FX chips.
All I know about the FX is its mulitplyer is fully unlocked and I think it makes better use of memory....not sure really.

You might say that people can tell by the core name (newcastle or winchester)but the average Joe Soap aint going know the difference.
But then again perhaps AMD buyers are the sort of people who go out of there way to learn about cpu's otherwise they would of got Intel P4's. By that I mean there people looking for an alturnative to Intel hence done there research (not mean't as a jab at Intel). Not a fan of Intels clock speed is everything approach. But maybe its smart marketing most people have the perception clock speed is important the higher the better so if there higher than there rivals people think they must be better.

Finally what the hell is the AMD 64 3300 doing with 256kb cache....thats crazy!

BKtje

despite not using dual channel. Also its hard to tell the difference between them and the FX chips.

Im not convinvced that dual-channel makes much difference on A64 platforms (due to the memory controller being on the cpu perhaps?)
I know CC's dual channel system was outperformed by system similar to mine at same clock speed (used newcastle instead of clawhammer so l2 cache was identical).

Azza

Well the performance of dual channel memory benefit is minimal 5% increase in memory performance tops. Thats why the extra half a mb of cache on the 3400 socket 754 helped it outpreform the dual channel but only 512kb cache of the similar clocked 3500 on socket 939. Fellow midlaner Pyrogenx runs dual channel on Athlon XP rig and claims he has not noticed any performance increase.

Incidently I see these AMD 64 winchester cpu's are outpreforming there newcastle equivelents that have the same clock speed and cache by 5% despite AMD's claim that they have not been tweaked. Seen it verified by several independent sites now. Perhaps its something to do with a higher hpertransport setting.