101.2% on phase change :)

[JCviggen]

I picked up a DLT3C (1.5V default) 1700+ today, XPMW

I had tried a few SPMW and RPMW DLT3C's before, but they all HATED voltage... 1.8 max. However, this XPMW is a whole different story... it just doesnt stop going :)

Here is what happened in my Prometeia :)

at 1.95 Vcore, I got to this

http://users.pandora.be/JCviggen/CPUIDXPMW.JPG

Super Pi (lower FSB because I needed dual channel memory for a good score, maxing out at 233 MHz 3-2-2-2.5)

http://users.pandora.be/JCviggen/superpi36.JPG


At 1.80 volts it reached the same overclock as the SPMW and RPMW ones, by the way.

http://users.pandora.be/JCviggen/XP1700XPMW2.JPG

I have to note, that the only reason I cant pass 3 GHz right now is the multiplier and max FSB on my board.... I cant get past 235 MHz FSB no matter what I try, and this CPU wont go past 12.5x without modding (and I'm reluctant to mod a chip THIS good)
I have roughly 100 MHz left in it I think.... 3050 MHz would have been sweet... but the multiplier just isnt allowing me right now... I hope to try a better board soon to crack 3 gigs with phase change again, this time with a 1700+ (already reached 3006 with a 2100+)

This chip is simply awesome... it cost me 60 euro, and I would say its worth it :)

[palee72]

no pics... pretty little red 'x's tho..

[JCviggen]

Originally posted by palee72
no pics... pretty little red 'x's tho..

they're all on my screen :confused:

[Anderoid]

Sweet man! Good work!!

Iam getting a 08 SPMW tomorow and hope to reach atleast 2400 MHz with my volcano 7+.

[BrainStorm]

JC, congrats! Nice chip. :)

btw, all the pics show for me.

Any idea if/when Xtreme may actually come back to life?

[JCviggen]

Dont get me started on Xtreme :(

should have been up DAYS ago and I dont know why it is not :(

nice chip : I think so :) cheers

[alexhuzar]

I wonder why AMD just doesen't sell the chips factory overclocked. This *miracle chip* can put out 2.3+ GHz easily with simple air cooling at ~1.6 volts and a lot more with a bit more voltage. AMD would have such an edge over Intel in these matters!!!

[Lithan]

The reason these chips do so well is they are made with the higher clocked processors. So technically, the 2800+,2700+ etc are factory overclocked 1700+'s. It's cheaper to make 500 of one kind of chip then 100 of this kind 200 of this kind and so forth. But they can't ONLY make topend chips or else intel would take over the value market (Probably amd's most important market). So they make 500 2800+ and test xxx amount to 2800+ speeds, and wire them and ship them out, then test xxx amount to 2700+ speeds, etc.

At least that's my take on the process.

[ScorpionNL]

wow, I just bought a 1700 DLT3C as well, to see how far it will go in comparison with my DUT3C, and guess what, it's the same stepping as you have :D

[Anderoid]

As me or JCviggen :)?

[BrantUnger]

The statement about 2700+s 'basically' being overclocked 1700s isn't quite accurate.

In each processor revision they add more transistors, and the Instructions per Cycle is upgraded, this is why an authentic 3000+ runs better than a 1700+ overclocked to a 3000+


Performance = IPC * mhz

I believe the 3000+ has somewhere around 20000 IPC while the 1700 would have more like 15000. Dont quote me on those numbers they are off. By a quite a bit, (2000 give or take maybe). In either case, the 3000 is better.


Keep that in mind next time you buy a CPU!


EDIT: Grammar, spelling.

[krairach]

1700+ DLT3C runs out of stock.
At excaliberpc?
In Holland too?

[JCviggen]

lol ...... a 3000+ has no IPC advantage over lower clocked models (with the same amound of cache, like the 2500+)
Clock a 2500+ to the same level as the 3000+ and it will have 100.00000000000% the exact same performance.

Clock a 1700+ to a 2800+ Tbred and its the same story. "less" good silicon always ends up in lower chips, rather than be thrown away. a 1700+ will almost never go as high as a 2800+ you buy.... but there is always the chance, since this binning isnt 100% accurate, to save money.

krairach, the DLT3C's are relatively rare. They had only DUT's at the store where I got my previous DLT's. All 0309xxxx crap. Nothing good.
Only in one store they had 3 left, one of them this XPMW :)
Its impossible to predict if AMD is going to make more of them in the next weeks. they may, or they may as well not.

[BrantUnger]

I think you better go find two .pdf from AMD and compare the 2.

It's not possible for IPCs to not be upgraded when they upgrade processors. They change pipeline length, etc etc...pipeline has direct corolation between Frequency and how many Instructions per cycle. I know this because the intel's have larger pipeline (somewhere like 20) for higher speed and lower IPC, while AMDs have shorter pipeline, for lower speed, higher IPC. (Which is the point of quantispeed architecture)

EDIT: Well couldn't find any AMD authenticated stuff, but heres a reputable site to start.

http://www.cyberbreakcafe.com/jpj_computers/cpuvscpu.htm

[JCviggen]

lol DUDE ... please... I dont want to seem like an arrogant ******* but you dont know what you're talking about.

Every AMD Athlon Thoroughbred-B chip has the exact same internals as a higher clocked model. (note : Barton is different than Tbred, obviously). Down to the exact transistor, a 1700+ = a 2600+ for example.

Also, if what you are saying would be true, the actual performance of every AMD CPU would be the same. higher clock = lower IPC, totalling the same total "work"

I am sorry, but you are wrong...

edit , checked that site you linked to. quote :


AMD Athlon XP 2700+ 2166.66 , IPC = 9 , instructions per cycle 19,500.

AMD Athlon XP 2000+ 1666.66 IPC= 9 , instruciotns per cycle 15,000


This site is sort of stupid... as "instructions per cycle" they just do clock speed x IPC ... they should have said "instructions per second" or something. Even then they're wrong, because it would have to be multiplier by MEGA Hz not just Hz

[BrantUnger]

Ok...This has nothing to do with overclocking. When they introduce a completely NEW chip....they increase the IPC.Ok? This is not because of speed...This is because of Architecture.

Architecture is a physical thing...overclocking is not.


If you get an Athlon XP 1700+ it has less transistors (among other changes) than an Athlon XP 1800+ will have. Why do you think they introduce new chips? If they didnt make enhancements to architecture then introducing chips would be WORTHLESS. Because everyone would overclock!

Is this making sense? I mean you dont have to sit here and flame me, go read a .pdf like I had orginally said, the 'specifications' .pdf with very technical data will tell you all you need to know.


Edit: And yea...I know they have the definition of operations per cycle wrong. It shouldnt say 'second'

[Kunaak]

you two need to chill out alittle.

Nice Oc on that XP 1700, I have 3 of them on thier way to me right now, hopefully one is worth a damn, the last 2 I had did less then 2.6 ghz, but they were only DUT3C's.
I haven't tried a new XP 1700 in about a month, as I been focusing on the Xp 2100's. (6 and not one stable past 2.8 ghz yet)

it would be pretty funny if this new XP 1700's kill my XP 2100's.

;)

[JCviggen]

Originally posted by BrantUnger
Ok...This has nothing to do with overclocking. When they introduce a completely NEW chip....they increase the IPC.Ok? This is not because of speed...This is because of Architecture.

Architecture is a physical thing...overclocking is not.


If you get an Athlon XP 1700+ it has less transistors (among other changes) than an Athlon XP 1800+ will have. Why do you think they introduce new chips? If they didnt make enhancements to architecture then introducing chips would be WORTHLESS. Because everyone would overclock!

Is this making sense? I mean you dont have to sit here and flame me, go read a .pdf like I had orginally said, the 'specifications' .pdf with very technical data will tell you all you need to know.


My friend I dont want to flame you at all... sorry if I seemed like that.

but you are... wrong.

The Thoroughbread-B is an architecture. If you will say a 3000+ has a different architecture, then yes. One is a barton, the other a thoroughbread. There is no architectural difference between 1700+, 2000+ and 2700+ THoroughbred-B's .... they are tested to see how high they can go, and this depends on the purity of the wafer, and how well the lithography was done. some parts of your wafer will be rated higher than others....

So look at your website .... and take the 2700+ total work per second. divide by clock speed and do the same for 2000+
Do you get the exact same number, yes or no?

[BrantUnger]

Aye, I know. I am done posting, I dont really care who's right or wrong. I'll just leave the post. Sorry.

[JCviggen]

Originally posted by BrantUnger
Aye, I know. I am done posting, I dont really care who's right or wrong. I'll just leave the post. Sorry.


:(

Still... look it up m8
I have no interest in being "right", but you got a wrong idea somewhere so just straighten that out..

cheers :)

[deathstar13]

i see where both of ya comen from.
on amd's papers i do believe they say a high cpu has more transisters.
BUT as being an ocer i dont see how that would be true far as i know as a xp2100 and a 2800 tbred oced to the same mhz have the same results.
the reason being this happens is the 2100 has the same amount as the 2800 as it was binned down so thats why it will oc so well.
so actually it does have the same amount of transisters.but just not in theory.

jcV btw great ocing dude. i gotta buy a cpu in a week and im having a hard time finding what i want.but ill be looking for what you just posted tho.

"edit" btw any news on when es.org will be back up?

[sirtoby]

All chips of the same model are exactly the same, it's only when they change model the number of transistors will increase.

i.e. going from Thunderbird to Athlon XP obviously meant an increase in transistors amongst other things.

Now, what is the difference between these chips? None at all..

How do they determine which chips is a low end and which is a high end then you might ask?

They are all produced in the same way, then they are tested and labeled to whatever speed they run stable on.

Obviously they use the stock AMD fan for the testing, and when the processor reaches a certain temp (or is becoming unstable) they move it down a couple of 100 Mhz and label it as whatever speed that is.

Now, this is the very core to o/cing. This is why we are able to overclock. If the chips would be maxed out at stock we obviously wouldn't be able to overclock the chips. Because of better cooling and additional voltage we are able to squeeze those extra MHz out of the CPUs.

[jdmcnudgent]

nice chip you got there bud,;)

[AcEmAsTr]

what exactly is prometeia?

[JCviggen]

Originally posted by AcEmAsTr
what exactly is prometeia?


google is your friend ;)

but try www.chip-con.com anyway...

[pctuner@amd]

Actually their right a higher rated procesoor will yeild higher clocks than a overclocked on my 2600/333 at 2535 had much higher benchmarks in pcmark and sandra than my 2100b has clocked to the same speed on the same mobo and ram everything the same just the 2100 produces more heat and does yeild lower benchmarks..

[jonspd]

Nice

Now mod the chip and let's see 3.5 LOL

[Warlord2]

VERY nice overclock, hard to believe a cheap little proccessor can beat the snot out of a $200+ dollar one. Do you have a accurate temp reading of the core? Im just curius because I playing with phase-change also and would like to know what it take for such a awesome overclock:D

[JCviggen]

sadly the EPoX 8RDA board uses a socket thermistor.... heated by all kinds of outside heat sources :(
Its in the +10C range under load... actual on-die temperature is about -15/-10C .. but there's no way to tell :(

[Warlord2]

Im in the same boat as you, I got a probe but it doesnt display below 0c and I too have the 8RDA. I wish they would of had the 8RDA say on-die temps:( Im going to have to buy another display for the probe that reads negatives:p

[Emericana]

JC you continue to impress me good work :D

where did you get this stuff directly from AMD? you live in belgum so isnt that near where amd produces their chips or somethin (i aint good at geography)..... lets just hope for the ppl here in the USA that excal has the XPMW soon!

[JCviggen]

Although Dresden is relatively near, the things still get assembled in Malaysia... so hardly matters where they make the cores...

New cores are at least as fast in the US as they are down here a few hundred Km from Dresden.

[Y.E]

you know what an overclocked 1700 is not the same as a 2800. the 2800 will score a little higher.... but you also have to take into acount that u can overclock the 2800. if the 1700 was the same as a 2800 then the overclock max of a 1700= overclock max of 2800, meaning u could not go past the speed of a 2800. But if u overclock a 2800 u will have a lot more mhz(total not increase) than the overclocked 1700

[OC Detective]

Originally posted by JCviggen
Although Dresden is relatively near, the things still get assembled in Malaysia... so hardly matters where they make the cores...

New cores are at least as fast in the US as they are down here a few hundred Km from Dresden.
It doesnt make any difference where they are assembled either - unfortunately for me:( What matters is where their distribution centres are as I dont see the latest stuff until about 2 - 3 months after all you guys!
Great overclock by the way - simply awesome.

[ZeroWing29]

JC...very nice work!

Obviously I don't have a Prometeia so I won't be getting what you're getting...but what week is it. 0308? I might pick one up to stick in my watercooling system and NF7-S rev 1.2.

I want it if it tolerates voltage well :)

[OC Detective]

Originally posted by BrantUnger
The statement about 2700+s 'basically' being overclocked 1700s isn't quite accurate.

In each processor revision they add more transistors, and the Instructions per Cycle is upgraded, this is why an authentic 3000+ runs better than a 1700+ overclocked to a 3000+


Performance = IPC * mhz

I believe the 3000+ has somewhere around 20000 IPC while the 1700 would have more like 15000. Dont quote me on those numbers they are off. By a quite a bit, (2000 give or take maybe). In either case, the 3000 is better.


Keep that in mind next time you buy a CPU!


EDIT: Grammar, spelling.
I think the problem here is you are comparing a Barton (XP3000) to a Tbred and yes the Barton does have a different amount of transisitors - however all 266 Tbreds have exactly the same number. I think the 2 .pdf files you are referring to are the model 8 XP Data Sheet (Tbred) and model 10 XP Data Sheet (Barton)

[dreammmatt]

how does week 10 perform?

[Y.E]

i still dont think that an oc 1700 is the same as 2700 tbred

[dropadrop]

Originally posted by Y.E
i still dont think that an oc 1700 is the same as 2700 tbred

If they have the same fsb and multiplier then they will give the same excact results. (with a slight error margin in one way or the other)

We just tested it with two identical systems based on 8rda+ and ati 9100. One had a 1700+, and the other had a 2100+. Both had the same kind of powersupply, ram, and our corporate win2k. At the same speed, the results are identical.

The diferance starts to show when you push it as high as you can, the 2100+ will overclock alot higher then the 1700+. Still, as long as the multi and fsb are the same, the results are identical.

[JCviggen]

Originally posted by ZeroWing29
JC...very nice work!

Obviously I don't have a Prometeia so I won't be getting what you're getting...but what week is it. 0308? I might pick one up to stick in my watercooling system and NF7-S rev 1.2.

I want it if it tolerates voltage well :)



My friend,

Its indeed week 8, but I have had several week 8 chips (SPMW and RPMW) and they HATE voltage. This XPMW likes it though. I have long been saying that the MOST IMPORTANT piece of info on the core is actually the 4-digit code after the date (3-digit in the MPM '41xxxxxxxxxx' case)

[dreammmatt]

ok, that guy said that the 1700+ running at 2.5Ghz will last like a month before it starts wearing out!?! is this true!?! and he also said that at 2.5GHz it will not perform as well as a 2100+ @ 2.5GHz, how big a difference will i notice? I am about to buy one of these...so...any info is appreciated

[ZeroWing29]

Originally posted by JCviggen




My friend,

Its indeed week 8, but I have had several week 8 chips (SPMW and RPMW) and they HATE voltage. This XPMW likes it though. I have long been saying that the MOST IMPORTANT piece of info on the core is actually the 4-digit code after the date (3-digit in the MPM '41xxxxxxxxxx' case)

Thanks JC, I was curious because ExcaliberPC has the 0308 XPMWs in stock...a DLT3C T-Bred B that loves voltage sounds like the miracle chip to me. Perfect under a watercooler (or in your case a Prometeia) :D

Looks like that is my next chip, I should order today before they run out :D

[JCviggen]

I hope they all like voltage as mine does! :)

I got a 2100+ MPM '4' once and it loved voltage... I got it to 3 GHz at 2.22V. Last week I tried a chip from the exact same batch and it didnt like over 2 volts still :( I think you need to get a bit lucky, but for the price of a 1700+ XPMW, you'd be crazy not to get one :)

[ZeroWing29]

Originally posted by JCviggen
I hope they all like voltage as mine does! :)

I got a 2100+ MPM '4' once and it loved voltage... I got it to 3 GHz at 2.22V. Last week I tried a chip from the exact same batch and it didnt like over 2 volts still :( I think you need to get a bit lucky, but for the price of a 1700+ XPMW, you'd be crazy not to get one :)

You better hope they do...else I'm putting your head on a stick!! :D

Nah J/K but it's worth a shot. Gotta be faster than my 1600+ AGOIA Y :)

[JCviggen]

even if you get an XPMW which hates voltage, I am fairly confident they can all do 2.5 GHz on air.

[Anderoid]

Got my 0308 SPMW stable at 2.4 GHz now with 1.725 V. Problem is heat as stated before, Iam getting 46 c under full load and that temp is read from my 8RDA+ so add a certain amount of degrees to that.

Still I think 2.5 might be possible with some luck.

[zindane18]

I dont think an overclocked 1700 to a 2800 per say is the same as an org 2800. If this were the case then why cant you overclock a lot of 1700 past 2500 for that matter....Are you telling me that AMD just makes a bunch of chips and test to see how fast they run, and then label them when they get to thier fastest overclock.....I think not......IMHO

[Y.E]

Hi all,
ok first is said 2700 not 2100 but thats what i mean they are the same if they have the same settings but u can obviously take the 2700 to more than the 1700 which means that they cant be exactly the same chip the 2700 is better becasue it can go more i mean after crazy overclock the 2700 totall mhz will be more than the 1700 total mhz which means the chips differ......

if i run my 1800+ at 145X11.5=1667 which is the same as the 2000+ but according to sandra my multimedia benchmark is less than the 2000+'s so then an overclocked chip is not the same as a nonoverclocked chip(they might be the same benchmark if u do it rite but the higher up one will still be better)

[only my opinion do not think im trying to say anyone is wrong just stating my opinion]

[Y.E]

exactly what ive been tryin to say zindane18 thank u

[ZeroWing29]

Originally posted by Y.E
Hi all,
ok first is said 2700 not 2100 but thats what i mean they are the same if they have the same settings but u can obviously take the 2700 to more than the 1700 which means that they cant be exactly the same chip the 2700 is better becasue it can go more i mean after crazy overclock the 2700 totall mhz will be more than the 1700 total mhz which means the chips differ......

if i run my 1800+ at 145X11.5=1667 which is the same as the 2000+ but according to sandra my multimedia benchmark is less than the 2000+'s so then an overclocked chip is not the same as a nonoverclocked chip(they might be the same benchmark if u do it rite but the higher up one will still be better)

[only my opinion do not think im trying to say anyone is wrong just stating my opinion]

That is because your system is not optimized, it has nothing to do with it being "slower"

And no, a lot of times you can't take the 2700+ to more than the 1700+, read up about it and you'll find out.

Bottom line? On the same platform, a 1700+ @ 2.5Ghz will perform just as well as a 2700+ @ 2.5Ghz, period.

[Y.E]

scrap the crap about my 1800+ its not the same comparison wasnt thinkin but does anyone know why that is tho?

[Y.E]

haha ok fine what ever it doesnt really matter all that matters is that they overclock great

[ZeroWing29]

Originally posted by Y.E
scrap the crap about my 1800+ its not the same comparison wasnt thinkin but does anyone know why that is tho?
It's because you're on a different platform, with a different motherboard and different RAM and a different amount of system resources being taken up.

Your chip is actually FASTER than a 2000+ if anything, because you have a higher FSB :)

[Y.E]

maybe but it scores excatly the same as the 2000+ in all other benchmarks except this one which is alot off
mine 9200/9700
2000+ 9900/9900

[ZeroWing29]

Close out all other applications that you possibly can, make sure DC services such as Folding or SETI are shut off.

I will however guarantee that if you had a true XP2000+ in your computer it would not be any faster than what you have now :)

[Y.E]

ok thanx

[zindane18]

OK please explain to me everyone.....Read my first my post and tell me....If all of AMD Tbred are the same exact chip just different muti etc. then how can amd take a 1700 chip and run it at 2800 on STOCK VOLTAGE??? If anyone here can take a 1700 and run it above 2.3 ghz on stock volt. then your deserve a medal!! Trust me...not all tbred are exactly the same via model numbers.....not meaning though that a 1700 cant outperform a 2800....thats a whole different issue....please try to understand what Im saying...

Zindane

[ZeroWing29]

Originally posted by zindane18
OK please explain to me everyone.....Read my first my post and tell me....If all of AMD Tbred are the same exact chip just different muti etc. then how can amd take a 1700 chip and run it at 2800 on STOCK VOLTAGE??? If anyone here can take a 1700 and run it above 2.3 ghz on stock volt. then your deserve a medal!! Trust me...not all tbred are exactly the same via model numbers.....not meaning though that a 1700 cant outperform a 2800....thats a whole different issue....please try to understand what Im saying...

Zindane

No, all AMD chips are not the exact same chip, they are the exact same architecture. Some yield higher clocks than others.

Let me explain...AMD does not actually test every chip that comes off of their production lines. They do know that chips towards the center of the wafer will run at 2800 or 2700 or 2600 so they mark them accordingly, but some chips they do not know HOW high they go...thus they rate them 2100+ or 1700+ to be safe.

Some of them OC to 2800 speeds on default voltage, and some do not

Bottom line is...a 1700+ at 166 x 13.5 will be the EXACT SAME as a 2800 @ 166 x 13.5 (a 2800+ T-Bred that is)

Maybe it won't run at the exact same voltage...or maybe it will...but performance wise it will be the same.

[zindane18]

OK thanks for explaining that Zero, I just think some people get a flase since that they can buy a 1700 or 2100 and overclock up to a 2700 or 2800 and really its all a crap shoot, sometimes you can and sometime you cant...

Thx

Zindane

[ZeroWing29]

Originally posted by zindane18
OK thanks for explaining that Zero, I just think some people get a flase since that they can buy a 1700 or 2100 and overclock up to a 2700 or 2800 and really its all a crap shoot, sometimes you can and sometime you cant...

Thx

Zindane

Well, it may be a crapshoot, but there's a pretty big gun and not a whole lot of crap.

If you go to Excaliber and get the guaranteed steppings, it is almost a guarantee that you'll hit super-high speeds :)

[cozmo_d]

Awsome Chip Awsome Voltage

Has anyone noticed a trend of Lower Voltages & Higher FSB

:) Thats a real good thing

Real Good

I thought I had a sweet Chip

Thats super Sweet

Rox ON!!!

[v8440]

Another thing to throw in here that people seem to be missing-let's say for an example that you have a 2100 and a 2800. Let's then say that you overclock both as far as they'll go, and you find that the 2800 reaches a higher speed in terms of raw MHz. THAT DOESN'T NECESSARILY MEAN THAT IT WILL OUTPERFORM THE 2100 AT THE SAME CLOCK SPEED! Yes, there is obviously some difference between the two chips, otherwise they'd have the same maximum overclock. But, the differences that affect max overclock don't necessarily change performance at a given speed. The signal/noise ratio in some critical part of the core may be better in the 2800, thus allowing it to run faster and/or hotter before thermal noise obliterates the signal. Does that mean it will outperform the 2100 at a speed that they both can run? No. Does that mean that it has more transistors, or some other architectural change? No.

To put what I'm saying another way, the fact that one chip can hit a different speed than another doesn't mean that the transistor count is any different, even if they are rated at different speeds. That MAY be the case, but to assume so simply because one reaches a higher speed than the other is false logic; you have no way of knowing WHY one chip can go faster than the other, so you certainly can't logically say "well, this chip can be clocked faster than the other, therefore the benchmark scores will be different at the same speed."

[ZeroWing29]

Originally posted by v8440
Another thing to throw in here that people seem to be missing-let's say for an example that you have a 2100 and a 2800. Let's then say that you overclock both as far as they'll go, and you find that the 2800 reaches a higher speed in terms of raw MHz. THAT DOESN'T NECESSARILY MEAN THAT IT WILL OUTPERFORM THE 2100 AT THE SAME CLOCK SPEED! Yes, there is obviously some difference between the two chips, otherwise they'd have the same maximum overclock. But, the differences that affect max overclock don't necessarily change performance at a given speed. The signal/noise ratio in some critical part of the core may be better in the 2800, thus allowing it to run faster and/or hotter before thermal noise obliterates the signal. Does that mean it will outperform the 2100 at a speed that they both can run? No. Does that mean that it has more transistors, or some other architectural change? No.

To put what I'm saying another way, the fact that one chip can hit a different speed than another doesn't mean that the transistor count is any different, even if they are rated at different speeds. That MAY be the case, but to assume so simply because one reaches a higher speed than the other is false logic; you have no way of knowing WHY one chip can go faster than the other, so you certainly can't logically say "well, this chip can be clocked faster than the other, therefore the benchmark scores will be different at the same speed."

About time someone starts talking sense! :D

[futura2001]

Okay, lets say I have two chips, one is a 2600+ and one is a 1700+
Both of these chips are of the same process. They are made from the same achitecture, they are pretty much the same except for clock speed.
Now the reason for this is that AMD cannot afford to make a new fab plant every time they come out with a new speed chip. Not even intel could afford this. Come on, do you really think that AMD can change the actual chips so that performance is different between overclocked and stock chips? No, If I run a 1700+ at 2600+ it has become a 2600+ in performance. Because the performance enthusiasts are such a small portion of the market, AMD does not have to worry much. What do they do? They make all their chips from the same wafers. Excluding obvious differences such as Duron to T-Bred to Barton. So I have my 1700+ which is pretty much the exact same chip as the 2600+ and it may or may not overclock well. If it can reach 2600+ speeds, good for me. If I were to run benchmarks a hundred times in a static environment, they would benchmark exactly the same. Why do benchmarks differ? Benchmark your system a few times. Be sure to turn off your computer and turn it back on again, maybe even going as far as to remove and install your CPU again each time. Now try to remember exactly what your system was like and what programs were running. Now that we have filtered out all the people with the attention span of a fruitfly on crack, take a look at those results. There is some varience between the results just because of the nature of software. You have windows running in the background, I assume. Windows will not go through the exact same processes in the exact same order every single time. Thus you get a slight varience. The purpose of running benchmarks many times under the same settings after restarting each time is to show you that while there is a varience, the average will be damn close. Don't be an uneducated benchmark junkie. Just because it can do it once doesn't mean it can do it again. Just because there is a miniscule varience doesn't mean one is better. Have fun overclocking, just don't make yourself look unnecessarily stupid.
*Whew*
Futura

[dreammmatt]

anyone got a link on how to unlock a 1700+ t'bred-B? or...do i not needa unlock them seing as i have a nForce2 board...i cant remember....

[altec]

im pretty sure the nforce2 will unlock it but im not totally sure...dont take my word for final say so...:p

[JCviggen]

you dont need to unlock, but you cant get any mults past 12.5

i"m going to do CPU surgery tomorrow and cut some bridges... to get higher mults... wish me luck ;)

[Jupiler]

Nice oc, m8, but I already told you that. ;)
Mine is crunching nicely @ 2460Mhz, 1.825V. Hope it passes Prime95 overnight.

Good luck with unlocking it, hope it goes well. It might be hard to find another great chip like you have now.

[JCviggen]

I think i'll head back into germany on monday anyway.. i'll look around ;) I wont hesitate to buy another one of these if I can!

[dreammmatt]

cool, thanx

[Jupiler]

Originally posted by JCviggen
I think i'll head back into germany on monday anyway.. i'll look around ;) I wont hesitate to buy another one of these if I can!

I don't think I'll be going next week, I've seen enough of it in the last few weeks :D

[JCviggen]

update after multiplier change to 13x :)

http://users.pandora.be/JCviggen/CPUIDXPMW3.JPG

http://users.pandora.be/JCviggen/CPUXPMW.JPG

http://users.pandora.be/JCviggen/MMXXPMW.JPG


This is at 2.00 Vcore... and after this my board cuts out due to over-current. I have removed R678 of course on my 8RDA, which enabled me to run 2.2V on my 2100+ ... but on this chip, a bit past 2 volts it will shut down under load or even when windows starts to load :( Apparently these chips draw more current at the same speed/voltage as the DUT chips... very weird.
It has some more in it, but not until I get a board capable of running higher volts...

[ZeroWing29]

Go for Epox 8RGA+ maybe? Or Abit NF7-S?

Very nice scores anyway JC, that is absolutely incredible for a $70 chip, despite the fact that it IS on a $600 cooler, it just shows that the chip really has a lot of juice in it...and it's half the price of the cheapest Northwood...

[JCviggen]

I have 2 EPoX 8RGA boards lying around... but they have the same cut-out problem even with R678 removed.

Dont have the NF7 1.2 :( but I've been promised one so... lets hope

[Kunaak]

odd observation.

why is the dhryston and whetstone score so low, even with 2.99 ghz and a 460 FSB on the CPU arithmatic?

I can get 9300, and 3800 running only 166x15 at my speed at the moment with my XP 2100, which clearly is more.
thats abit weird.
would this be apart of that thing, where people were arguing that a XP 1700 just cannot put up the same numbers even when overclocked farther then say a Xp 2100?

[JCviggen]

I think that sort of depends on the version... also I think I have a lot of crap disabled in sandra and never bothered to re-enable it...
Also, keep in mind Sandra is about the ****tiest benchmark ever... Super Pi and Pifast turn in the exact same numbers as any other XP I've tried, and they are "real" benches

http://users.pandora.be/JCviggen/superpiXPMW.JPG


http://users.pandora.be/JCviggen/pifastXPMW.JPG

[JCviggen]

Kunaak,

I just ran Sandra 2003 now (instead of 2002) at my cruising speed of 2.8 GHz, and I got 10488/5159. sound better?

[Holst]

Very nice clock as ever JC :D

I agree that sandra is the worlds wost benchamark..

Kick ass superPI, 36 is very nice indeed :)

[Walrusbonzo]

Originally posted by BrantUnger
The statement about 2700+s 'basically' being overclocked 1700s isn't quite accurate.

In each processor revision they add more transistors, and the Instructions per Cycle is upgraded, this is why an authentic 3000+ runs better than a 1700+ overclocked to a 3000+


Performance = IPC * mhz

I believe the 3000+ has somewhere around 20000 IPC while the 1700 would have more like 15000. Dont quote me on those numbers they are off. By a quite a bit, (2000 give or take maybe). In either case, the 3000 is better.


Keep that in mind next time you buy a CPU!


EDIT: Grammar, spelling.

That's nonsense m8!

[Walrusbonzo]

Originally posted by dropadrop


If they have the same fsb and multiplier then they will give the same excact results. (with a slight error margin in one way or the other)

We just tested it with two identical systems based on 8rda+ and ati 9100. One had a 1700+, and the other had a 2100+. Both had the same kind of powersupply, ram, and our corporate win2k. At the same speed, the results are identical.

The diferance starts to show when you push it as high as you can, the 2100+ will overclock alot higher then the 1700+. Still, as long as the multi and fsb are the same, the results are identical.

OK, to try and end this problem........

ALL Tbred Bs have the same number of transistors, I.E. a 1700+ TBRED B will have same number of transistors as a 2800+ TBRED B.

If you have a 1700+ and clock to same speed as a 2800+ with all else the same, then they will benchmark the same too!

Finally, NOT ALL 2100+ will clock better than 1700+, this is a myth that if you have a higher CPU it will have a higher final overclocked speed. NONSENSE!!!!! It's about luck half the time.

I've now owned 3 1700+ Tbred Bs(JIUGB, JIUHB and a DLT3C JIUHB) and ALL do better than a 2100+ AIUHB I had. My M8s 2400+ doesn't even clock as well as my 1700+s!

I hope you are now all clear on these few points?

[dreammmatt]

i heard that this guy couldnt get his week 10 to do much, do u guys think that it was just him? that others will have better experience? i remember seing some people with earlier weeks, and some did well, and some didnt...well, i suppose i just answered my own question...

[altec]

everyone with the 0310 date codes are gettin high oc's and if you consider 800mhz not much i think you may want to re-evaluate :p

[dreammmatt]

well...i dont consider 800 high for a 1700+, seing as their scale is different than that of older chips. 400MHz is low for them, 1.5GHz is high for a 1700+, 800MHz is mid-range, not bad, but not the best either, lol.

LOL...my AGKGA nightmare only got 200MHz...LOL LOL LOL LOL LOL

[macci]

Excellent OC there JC :)
Wish I can find a CPU like that soonish. The POS JIUGB I've got is maxin out at 2660MHz <- and that is at 2.16V -6C under load :P

btw its most likely the CPU and not the mobo(s) thats causing the shut off at high Vcore.

[JCviggen]

Originally posted by macci

btw its most likely the CPU and not the mobo(s) thats causing the shut off at high Vcore.

Doubtful.... I used to have this with my 2100+ .... when I removed R678 I suddenly could get 2.18V without shutdowns, coming from only 2.0V ... so the CPU was the same, but the voltages which would be tolderated went up considerably...
The only thing which seems to effect it is how much power the CPU is drawing... so at high speeds, the max. voltage you can reach is lower...
I cant rule out that the CPU has something to do with it... but I can only prove my point when I get my hands on an NF7 or smth

[Kunaak]

what is the "R678"?

is that the hard drive shut down chip?

[JCviggen]

Originally posted by Kunaak
what is the "R678"?

is that the hard drive shut down chip?


No, its a resistor in the over-current protection circuitry :)

[Jen]

JC thats one hell of a OC




Jen

[zabomb4163]

there are many p4's that can't hit your clockspeed JC

*wishes to self that i could post a link to this in the intel section without getting threatened/flammed*

[dreammmatt]

one word...lol
sad but true, that intel fan-boys go ballistic whenever they hear that chipzilla has been shot out of the air. Ya know what is really funny though? that a 3.0GHz Celeron is like poop compared to a one-point-anything athlon ^_^ - the only reason that intel must sell thoes things is that they "have a big number" (3,000MHz) and a lot of people think that = good.

Athlon people know better ^_^, man...i wanna see what 3DMark that 3GHz 1700+ got...did he say? or did i miss it?

(Disclaimer: at no time do i mean to insult or make fun of intel fan-boys; I was mearly having fun pointing out the obvious; please dont flame me >_<)

[zindane18]

I have a 2100+ and Im only hitting 2.3 ghz. a 600 mhz overclock about....Dont really know what holding me back....I think the chip just might be maxed out....Im running around 1.825 vcore at 38c load.... Any ideas?? specs below.

[jdmcnudgent]

JC is the oc king. nice job.;)

[wildbilly2k]

where did you buy this chip?

[Cowboy X]

I must chip in ( pun not initially intended ) to congratulate jc on another stupendous overclock . Good to see you participatingin our forums again .

[pwned2k]

hey wut do you guys think about the 310UPMW at excaliber? it looks to be the newest week, but not an xp. i might go for it, if anyones tried it, let a brotha know.

[JCviggen]

Its boring to be at the same speed the whole time so I tried a little more this morning ;) shame I cant run past 2.00V on this board really...

http://users.pandora.be/JCviggen/XPMWmax.JPG

I also improved my Pifast time slightly, but my board wont run past 228 DC :(

http://users.pandora.be/JCviggen/pifast228.JPG

[Walrusbonzo]

You are an overclocking God!

Keep up the good work ;)

[macci]

Damn JC!
I just beat ur old record two days ago (51.08 (http://www.solidhardware.com/macci/P43060/hex4132.gif) w/ P4) and now you improve once again. aarrgghh!
Excellent :D

[JCviggen]

Originally posted by macci
Damn JC!
I just beat ur old record two days ago (51.08 (http://www.solidhardware.com/macci/P43060/hex4132.gif) w/ P4) and now you improve once again. aarrgghh!
Excellent :D


LOL ! :)

Well its more impressive with a P4... although I have a feeling my rig was cheaper :D 62 euro CPU, 115 euro board and 2x54 euro mem ;)
I need more FSB...

[Warlord2]

damn thats awesome, looks like that chip can even do a little more. I see you unlocked 13, have you tryed 13.5 or is 3 gigs all its got at that voltage?

[dreammmatt]

ok...now...how sane does it seem to say "is 3 gigs all its got?" LOL. those chips are awesome! they defy natural chip marketing!

[JCviggen]

On an 8RDA, multipliers 13.5 and 14x (as well as 12.5 btw) hinder overclocking a bit. Multipliers 13 and 15 are what you need. At 13.5 or 14 it will not even POST at 2960... it'll show "FF" and immediately reboot at 100 FSB. But at mults 13 and 15 I can post at more than 3050 :D
So you see, its a matter of getting the multipliers right...

And yes... i'm afraid 3 gigs is all she's got captn'

[PMM]

Funny you should say that I experience the same on my Asus
with some Multipliers yielding better performace than others.


Thanks for mentioning OC JCviggen was wondering where you
all were hiding :)

[umekami10]

Excellent Ocing JC :)

It is also impossible to run Epox over 230 fsb dual channel cas2 with any other bios except C16(no matter what Vddr ).

I have tried C16 bios and highest I could run was about 241 4-2-2 Cas 2 but with low multiplier 8, Vddr around 3.18V. If I changes multiplier to 10 or 11 or 12 etc I should reduce the fsb according to multiplier. For instance highest Fsb with multiplier 12.5 and 13 is 230 with 12 is 233 . With 11 is 235. with 10 is 238…etc

Why it is so I believe it has to do partly with Nforce 2 chipsett and partly with power sources on board. More voltage lower fsb. And More CPU Mhz more memory bandwidth and less FSB.

We se the same pattern a single channel . But some differences is there. Memory can run in high fsb 243 cas 2 in all bios ….

[yogi99]

Is xtreme done?? Seem to see some familiar ppl here.