About to pull the trigger on an Opty 170...

[Matt2]

Ok, so I've been looking around for an Opty 170 and the only place I found it at a decent price is tankguys for $385.99, just under $400 shipped. Dave Graham is the only place i know that sells it for cheaper, but the wait is far too long for me.

The stepping I would get from Tank Guys is CCBWE 0546 XPMW. Good stepping? I hope to hit at least 2.7ghz on air with decent voltages, no higher than the 1.54v I am running now though.

My neighbor across the street hits 2750 1.55v on his CCBWE 0544 XPMW, can I expect the same results?

So what do u guys think? $400 is quite a bit of cash compared to the $150 I paid for this Venice in June.

I'm driving myself crazy here! Help me decide!

 

[lopri]

Yes 2.7GHz is very possible with 0546XPMW. It's one of the most consistent steppings I've observed.

 

[Diogenes2]

I have that stepping.. Got it from ZZF.. 2.75 at 1.5, 2.65 at 1.4 ..

 

[nealh]

do not waste your money..I went from Venice 3200@2750 to opty@2.7 and I can not say it was worth it..unless you need it for multitasking or the fun of playing with a new cpu...you may not see these amazing gains

my 0546 xpmw hits 2.7 @1.5v and hits a wall.....there are some real sweet chips out there but mine was not one of them...you will get 2.7 but many of the 0546 xpmw I have seen 1.5v(many need 1.55v for 2.75)

but there sre sweet chips like Diogenes and Lopri

beware of the rports are all these great OCers....these IMHO are not the norm..most guys cross post in other forums so it looks like alot hit 2.75-2.8 but they do not

you will also find many 170s hit only 2.6

 

[davegraham]

*cough*

the xpmws can get around 2.8 (that i've seen). they're not bad...

and, sorry 'bout the wait on the 170s. i can't get 'em to save the life of me right now!

dave

 

[nealh]

dave..here I have to disagree...if you look at the forums..its the same people hiting 2.8...more often these are hitting 2.7, some 2750 requiring 1.55v

also some hittinh 2.8 are doing so on the DFI mobos that overvolt and are giving over 1.5v in the bios setting which will equal closer to 1.6v or they are using more exotic cooling than air

so just beware..if you want 2.7ghz..this is likely but expect to give it 1.5v and if you need less great....

 

[Absolute0]

0546 XPMW is a great stepping. Dave is right, check out Xtremesystems, 2.8+ Ghz is qutie common.

 

[davegraham]

Originally posted by: nealh
dave..here I have to disagree...if you look at the forums..its the same people hiting 2.8...more often these are hitting 2.7, some 2750 requiring 1.55v

also some hittinh 2.8 are doing so on the DFI mobos that overvolt and are giving over 1.5v in the bios setting which will equal closer to 1.6v or they are using more exotic cooling than air

so just beware..if you want 2.7ghz..this is likely but expect to give it 1.5v and if you need less great....

OVERALL, it might require a bit more voltage than other steppings (0550s, for example) to get there, but....there seems to be a high incidence of people with 0546xpmws hitting 2.7-2.9 barriers with relative ease. You've also got to look at the variables within the overclocking (which is why i'm a little uncomfortable with "guaranteed" overclocking) aspect that may influence the ability of a particular part to hit a golden speed.

dave

 

[nealh]

Originally posted by: Absolute0
0546 XPMW is a great stepping. Dave is right, check out Xtremesystems, 2.8+ Ghz is qutie common.

most of those guys do not use air..alot water cooling and phase change..alot of mods as well

they get great results but I would not use more than 1.55v with overvolting for a given mobo

alot of the 2.8+ are using like 1.57v + without taking into account the overvolting that the DFI NF4 produce under load..so they are giving 1.6v+...not sure how long the opty will survive

moreover, not all prime 95 stablity, many use Superpi 32mg ..or other ?? methods

I am sorry, I strongly disagree that 2.8+ is common on this stepping at safe vcore(most would agree 1.55v 24/7 is max)

I am not saying it is a bad stepping but do not be stupid like I was ..I have opty 165@2600 and hoped to get 2750+ on 170 given what everyone was saying

so I bit and got a 0546 xpmw and I am lucky to get 2.7v....

I am no novice to overclocking(my venice hit 2750 with valuram running 250mhz)...I use quality components and take my time with overclocking


Also I have not been wowed by the speed improvement over my venice..surely not worth the extra $$$...I know I am more futureproof

All I am trying to convey is keep your expectations in check..OCing YMMV..you may get a better chip than me....

Good Luck with your decision

 

[Matt2]

Well... I just ordered the 170 from Tank Guys. I almost cancelled it twice now and its only been 15 minutes.

I definately understand where you are coming from nealh. $400 is a lot of money to spend for possibly marginal speed increases, possibly none if you get a dud chip.

We shall see though. Anyone else have any opinions? I doubt Tank Guys is gonna ship it today, so I have all day to decide. BTW - it's an OEM processor, no warranty, always scares me.

 

[davegraham]

Matt2,

those 170s are great chips....I'm pretty sure TankGuys offers a DOA warranty so...check with them. However, AMD does not provide US with a warranty, so it's disengenuous for us to say otherwise.

btw, those 170s are hand-picked by someone.... *wink*

dave

 

[nealh]

Originally posted by: davegraham
Matt2,

those 170s are great chips....I'm pretty sure TankGuys offers a DOA warranty so...check with them. However, AMD does not provide US with a warranty, so it's disengenuous for us to say otherwise.

btw, those 170s are hand-picked by someone.... *wink*

dave

hand picked??? I am pretty dense what are you suggesting

 

[Cooler]

Get it !

 

[Absolute0]

Hand picked means good overclockers.

As for the Opty overclocking, nothing wrong with 1.6 volts. I know the Ultra-D overvolts, but i ran my X2 @ 1.6 volts 24/7 for months, no harm done, on my watercooling max load was 38C . I also ran my 3000+ Venice at 1.66v 24/7 for months, no harm done. Too much voltage for too long is dangerous... but i have no qualms with using 1.6-1.65 volts 24/7 so long as my load temps are sub 40C.

My X2 3800+ was happily doing 2.8 Ghz with 1.50 volts and 2.85 Ghz with 1.55 volts, and many of these new Opterons kill my 3800+. 0530, 0546 ,and 0550 are the known best weeks. Some 0547s also i think.

 

[nealh]

Originally posted by: Absolute0
Hand picked means good overclockers.

As for the Opty overclocking, nothing wrong with 1.6 volts. I know the Ultra-D overvolts, but i ran my X2 @ 1.6 volts 24/7 for months, no harm done, on my watercooling max load was 38C . I also ran my 3000+ Venice at 1.66v 24/7 for months, no harm done. Too much voltage for too long is dangerous... but i have no qualms with using 1.6-1.65 volts 24/7 so long as my load temps are sub 40C.

My X2 3800+ was happily doing 2.8 Ghz with 1.50 volts and 2.85 Ghz with 1.55 volts, and many of these new Opterons kill my 3800+. 0530, 0546 ,and 0550 are the known best weeks. Some 0547s also i think.

ok..but intel engineer said too high(who knows the amount) vcore alone is bad and increased temps with it is worse

you are on WC setup..so 1.6 maybe ok...I am on SI 120 so I will hit 48/49C at full load at 1.5v

Hand picked..is tankguys pretesting these cpus????? if so will they tell us the result if so I would buy that for sure

 

[TankGuys]

Originally posted by: nealh

Originally posted by: Absolute0
Hand picked means good overclockers.

As for the Opty overclocking, nothing wrong with 1.6 volts. I know the Ultra-D overvolts, but i ran my X2 @ 1.6 volts 24/7 for months, no harm done, on my watercooling max load was 38C . I also ran my 3000+ Venice at 1.66v 24/7 for months, no harm done. Too much voltage for too long is dangerous... but i have no qualms with using 1.6-1.65 volts 24/7 so long as my load temps are sub 40C.

My X2 3800+ was happily doing 2.8 Ghz with 1.50 volts and 2.85 Ghz with 1.55 volts, and many of these new Opterons kill my 3800+. 0530, 0546 ,and 0550 are the known best weeks. Some 0547s also i think.

ok..but intel engineer said too high(who knows the amount) vcore alone is bad and increased with it is worse

you are on WC setup..so 1.6 maybe ok...I am on SI 120 so I will hit 48/49C at full load at 1.5v

Hand picked..is tankguys pretesting these cpus????? if so will they tell us the result if so I would buy that for sure

Pre-testing, no. It's been suggested, but I don't think it would be a good idea.... for lots of reasons. What it DOES mean is that we try to nab the best steppings we can. It's tough since we can't really request specific ones, but we can try and pull strings at our various suppliers to try and get some guidance

Pre-testing is just not a cost-effective option. Too many risks, and the potential return on the time investment would be even more abysmal than selling the parts already is!

 

[Absolute0]

They can't pretest because then they would be selling used chips. But when someone says "handpicked" that means they choose the steppings that are known to be better overclockers.

 

[nealh]

Originally posted by: Absolute0
They can't pretest because then they would be selling used chips. But when someone says "handpicked" that means they choose the steppings that are known to be better overclockers.

ok...thanks Guys

Pretesting would be too time consuming, I agree...I for one would not consider it used but a better guarantee that I got what I wanted but would not suggest anyone start this business...

 

[Chesebert]

Originally posted by: nealh

Originally posted by: Absolute0
Hand picked means good overclockers.

As for the Opty overclocking, nothing wrong with 1.6 volts. I know the Ultra-D overvolts, but i ran my X2 @ 1.6 volts 24/7 for months, no harm done, on my watercooling max load was 38C . I also ran my 3000+ Venice at 1.66v 24/7 for months, no harm done. Too much voltage for too long is dangerous... but i have no qualms with using 1.6-1.65 volts 24/7 so long as my load temps are sub 40C.

My X2 3800+ was happily doing 2.8 Ghz with 1.50 volts and 2.85 Ghz with 1.55 volts, and many of these new Opterons kill my 3800+. 0530, 0546 ,and 0550 are the known best weeks. Some 0547s also i think.

ok..but intel engineer said too high(who knows the amount) vcore alone is bad and increased with it is worse

you are on WC setup..so 1.6 maybe ok...I am on SI 120 so I will hit 48/49C at full load at 1.5v

Hand picked..is tankguys pretesting these cpus????? if so will they tell us the result if so I would buy that for sure

I use to work on the AMD Athlon processor (circuit design) and I can attest to high vcore kill processor independent of the temp. break down gate oxide for one is a problem at 90nm it was a problem at 130nm when I worked as an engineer. AMD outsourced SOI from IBM and I know we tested chip running extremely high volts through them to simulate accelearted death.

 

[nealh]

Originally posted by: Chesebert

Originally posted by: nealh

Originally posted by: Absolute0
Hand picked means good overclockers.

As for the Opty overclocking, nothing wrong with 1.6 volts. I know the Ultra-D overvolts, but i ran my X2 @ 1.6 volts 24/7 for months, no harm done, on my watercooling max load was 38C . I also ran my 3000+ Venice at 1.66v 24/7 for months, no harm done. Too much voltage for too long is dangerous... but i have no qualms with using 1.6-1.65 volts 24/7 so long as my load temps are sub 40C.

My X2 3800+ was happily doing 2.8 Ghz with 1.50 volts and 2.85 Ghz with 1.55 volts, and many of these new Opterons kill my 3800+. 0530, 0546 ,and 0550 are the known best weeks. Some 0547s also i think.

ok..but intel engineer said too high(who knows the amount) vcore alone is bad and increased with it is worse

you are on WC setup..so 1.6 maybe ok...I am on SI 120 so I will hit 48/49C at full load at 1.5v

Hand picked..is tankguys pretesting these cpus????? if so will they tell us the result if so I would buy that for sure

I use to work on the AMD Athlon processor (circuit design) and I can attest to high vcore kill processor independent of the temp. break down gate oxide for one is a problem at 90nm it was a problem at 130nm when I worked as an engineer. AMD outsourced SOI from IBM and I know we tested chip running extremely high volts through them to simulate accelearted death.

having worked at AMD do you feel 15% increase in vcore is too dangerous..most seem to say 10% is ok and 15% will shorten some but probably still safe for long term use

I understand any increase in vcore can be lead to decrease life...but would 15% lead to less than 3-5 yr life span on these Opty DC

 

[Chesebert]

15% won't do much.....well not in the short term, like as in few days, at least I remember talking to test engineers that puts few volts in the Ahthlon SOI parts.....but that's back when the vcore was at 1.6 to start with IIRC. Been out of the engineering field for a few years so no clue what 90nm SOI can handle but SOI gets really testy when you put more volts through it the leakage current is ridiculously high. But the reason they lower vcore with each core shrink is to keep constant electric filed which protects gate oxide as they are in the few A now days (few dozen atoms...LOL)

 

[Chesebert]

Originally posted by: nealh

Originally posted by: Chesebert

Originally posted by: nealh

Originally posted by: Absolute0
Hand picked means good overclockers.

As for the Opty overclocking, nothing wrong with 1.6 volts. I know the Ultra-D overvolts, but i ran my X2 @ 1.6 volts 24/7 for months, no harm done, on my watercooling max load was 38C . I also ran my 3000+ Venice at 1.66v 24/7 for months, no harm done. Too much voltage for too long is dangerous... but i have no qualms with using 1.6-1.65 volts 24/7 so long as my load temps are sub 40C.

My X2 3800+ was happily doing 2.8 Ghz with 1.50 volts and 2.85 Ghz with 1.55 volts, and many of these new Opterons kill my 3800+. 0530, 0546 ,and 0550 are the known best weeks. Some 0547s also i think.

ok..but intel engineer said too high(who knows the amount) vcore alone is bad and increased with it is worse

you are on WC setup..so 1.6 maybe ok...I am on SI 120 so I will hit 48/49C at full load at 1.5v

Hand picked..is tankguys pretesting these cpus????? if so will they tell us the result if so I would buy that for sure

I use to work on the AMD Athlon processor (circuit design) and I can attest to high vcore kill processor independent of the temp. break down gate oxide for one is a problem at 90nm it was a problem at 130nm when I worked as an engineer. AMD outsourced SOI from IBM and I know we tested chip running extremely high volts through them to simulate accelearted death.

having worked at AMD do you feel 15% increase in vcore is too dangerous..most seem to say 10% is ok and 15% will shorten some but probably still safe for long term use

I understand any increase in vcore can be lead to decrease life...but would 15% lead to less than 3-5 yr life span on these Opty DC

correction: did not work at AMD...worked for AMD a difference...I worked at IBM for AMD...

 

[nealh]

Originally posted by: Chesebert

Originally posted by: nealh

Originally posted by: Chesebert

Originally posted by: nealh

Originally posted by: Absolute0
Hand picked means good overclockers.

As for the Opty overclocking, nothing wrong with 1.6 volts. I know the Ultra-D overvolts, but i ran my X2 @ 1.6 volts 24/7 for months, no harm done, on my watercooling max load was 38C . I also ran my 3000+ Venice at 1.66v 24/7 for months, no harm done. Too much voltage for too long is dangerous... but i have no qualms with using 1.6-1.65 volts 24/7 so long as my load temps are sub 40C.

My X2 3800+ was happily doing 2.8 Ghz with 1.50 volts and 2.85 Ghz with 1.55 volts, and many of these new Opterons kill my 3800+. 0530, 0546 ,and 0550 are the known best weeks. Some 0547s also i think.

ok..but intel engineer said too high(who knows the amount) vcore alone is bad and increased with it is worse

you are on WC setup..so 1.6 maybe ok...I am on SI 120 so I will hit 48/49C at full load at 1.5v

Hand picked..is tankguys pretesting these cpus????? if so will they tell us the result if so I would buy that for sure

I use to work on the AMD Athlon processor (circuit design) and I can attest to high vcore kill processor independent of the temp. break down gate oxide for one is a problem at 90nm it was a problem at 130nm when I worked as an engineer. AMD outsourced SOI from IBM and I know we tested chip running extremely high volts through them to simulate accelearted death.

having worked at AMD do you feel 15% increase in vcore is too dangerous..most seem to say 10% is ok and 15% will shorten some but probably still safe for long term use

I understand any increase in vcore can be lead to decrease life...but would 15% lead to less than 3-5 yr life span on these Opty DC

correction: did not work at AMD...worked for AMD a difference...I worked at IBM for AMD...

oops sorry

so 15% for a few days is ok..but months/yrs is "really bad"

I own a P4 2.4b@2.9 and ran it this way for 3yrs never degraded or had issue..longest I kept a processor and overclocked it..I hate to say I have forgotten the die size and vcore on it

 

[davegraham]

Originally posted by: davegraham
Matt2,

those 170s are great chips....I'm pretty sure TankGuys offers a DOA warranty so...check with them. However, AMD does not provide US with a warranty, so it's disengenuous for us to say otherwise.

btw, those 170s are hand-picked by someone.... *wink*

dave

lol..what i meant was something completely different that what went on through the majority of this thread...

*sorry ben...wasn't trying to imply anything*


i was just being cagey about how i knew about those CPUs....hehehe.....

dave