Buy your Opty's while you can, AMD code change:

I just got off the phone with a source (It's a retailer who I know very well). He just talked to an AMD distributor and it seems the part codes on the 939 pin Opterons are going to change.

Supposedly, this always happens when there are Core changes (san diego, venice etc).

This may mean that AMD doesn't like us using all their 939 Opty's for overclocking, and will be changing them from their FX-57 roots and unlocked multi's.... But who really knows for sure :confused:

It wouldn't surprise me if the golden age of sweet opterons will change soon with these part numbers....

as always, you must get in early to get the good stuff.................

If only changing #s then all its ok.

why would AMD bother? They sell all line excellent

144/146/148/150/165

they all pretty much out of stock all the places.

I seriously doubt were a killing distribution to OEMs. Believe me, we make up about a 1% of the total CPUs sold world wide.

Quote:

---

Originally Posted by phadeout

This may mean that AMD doesn't like us using all their 939 Opty's for overclocking, and will be changing them from their FX-57 roots and unlocked multi's.... But who really knows for sure :confused:

---

But they have locked multipliers :confused:

Opty's are unlocked "downwards" only.

Who knows, maybe with the part number changes they will make the 939 opty's more like X2's...

Just thought I'd let you guys know about the part number changes is all, for better or worse...

Maybe it is just the Memory Controller they are changing?

Quote:

---

Originally Posted by TEDY

why would AMD bother? They sell all line excellent

144/146/148/150/165

they all pretty much out of stock all the places.

---

their A64 line is hurting

Wow I feel lucky, my 165 came in the mail today, good thing I didnt wait!

Quote:

---

Originally Posted by dinos22

their A64 line is hurting

---

How so?
We've bought plenty of A64's and FX's before Opty's came out :p

Chances are the average user doesn't consider Opterons before regular A64's yet.

Yea, I would think that the extremist community would do nothing but help the sale of a product by producing a worldwide hype that will boost sales. Especially with the opteron.

Quote:

---

Originally Posted by zenzog

How so?
We've bought plenty of A64's and FX's before Opty's came out :p

Chances are the average user doesn't consider Opterons before regular A64's yet.

---

if you look at wanted to buy section at OCAU..........almost everyone is being told about Opterons and every sale that was supposed to go A64 ends up going Opteron instead because of overclocking potential

:woot: Makes me happy I have 4 - 144 optys just sitting here.

their a64 line? don't know about other countries but here in my country a64 systems sell is great.

i think amd released these amazing opterons for the hype we would create.
the hype over the normal desktop core took them from 15% to 50% marketshare in USA retail in about a year and a half.
now they're using us to hype the opterons in the hope the corporates will hear all the noise thus causing a rise in AMD's marketshare there.

i know they'd rather sell opterons to corporates than venices to us
they make huge profits in that sector and their market penetration needs plenty of work.

and lets remember that very often a core revision can be very benificial to overclockers. otherwise we might still be running 486's :D

Large companies don't base their buying decisions on the opinions of guys who use liquid nitrogen to overclock. They base them on total cost of ownership, performance per watt, supplier reliability, etc. (And if AMD is to be believed, a fear that Intel will screw them over if they dare to go green.) These guys do their homework, and AMD knows it. There's no bait-and-switch going on here.

-aMp-

um if they did their homework they'd all be running opterons already
they behave more like sheep
they're not going to recommend a cpu their boss has never heard of for fear of their jobs

I suspect the new single core OPN will be...

OSA1??DKA5CF

Why would a code change be negative? If I remember correctly, that happend when Winchester became Venice, and that stuff.
Seems good to me!
I don't think they would change the core of the opterons, just to stop us from overclocking em like mad.

Quote:

---

Originally Posted by TEDY

their a64 line? don't know about other countries but here in my country a64 systems sell is great.

---

same here. opterons are pretty much out of reach (both pricing- & availability-wise) for most ppl in our region.

There are no plans for another Rev. before 65nm are there? Such as D0 to E3/E4???

Quote:

---

Originally Posted by dinos22

if you look at wanted to buy section at OCAU..........almost everyone is being told about Opterons and every sale that was supposed to go A64 ends up going Opteron instead because of overclocking potential

---

Everyone is being told here about Opterons, at [H]ardforum, at Anandtech... probably at every enthusiast forum. Still, it shouldn't make too much of a difference as most resellers are OOS and AMD controls the distribution. If they didn't want the word to get out and cause competition with their other line, they could've (and still can) do something about it.

Quote:

---

Originally Posted by Haltech

I seriously doubt were a killing distribution to OEMs. Believe me, we make up about a 1% of the total CPUs sold world wide.

---

Uh.. my message is too short so I'm going to add three characters...

hmm... hope they don't mess them up..... I love my Opetons :)

Quote:

---

Originally Posted by largon

I suspect the new single core OPN will be...

OSA1??DKA5CF

---

My guess would be the same ;) (If, if in fact, the rumor the OP heard was true.)
IMO, the s939 Opterons were a means for AMD to rid themselves of the remaining E4 (San Diego) cores before moving production to dual core dies. Once all the E4 single cores are depleted, they either would discontinue the s939 Opteron line or transition the OPN to the E6 OSA1??DKA5CF as largon stated above.
Oh btw, sometimes there are changes to the OPN that actually have little to do with the type of core:
http://img39.imageshack.us/img39/6572/x24gd.jpg

It seems every retailer for example in germany are OOS. Maybe AMD ran out of E4 silicon already. ETA's are like 2 weeks...
:para:

Quote:

---

Originally Posted by largon

It seems every retailer for example in germany are OOS. Maybe AMD ran out of E4 silicon already. ETA's are like 2 weeks...
:para:

---

Sounds about the right ETA.......for the arrival of the E6 Diegos ;)

so how are they likely to clock?

flytek,
I believe s7e9h3n is the only one yet who's had pleasure punishing E6 SDs. I myself might be able to tell something about it - in about 2-3 weeks. I already booked some model 146 samples...

Edit: Opterons i received were E4 after all. Oh well... ;)

Is anyone else thinking, "NOOOOOOOOOOOO!"

I am, my birthday is in 5 months, and I can only build a new computer then. I can only hope that this code change will lead to improvement, so I can still overclock the ownteron.

Quote:

---

Originally Posted by manomanx2

Is anyone else thinking, "NOOOOOOOOOOOO!"

I am, my birthday is in 5 months, and I can only build a new computer then. I can only hope that this code change will lead to improvement, so I can still overclock the ownteron.

---

THe switch to rev6 isn't necessarily a "bad" thing, just consider it a "different" type of overclock :D

Also, I've many times reffered to the new E6 1mb l2 cache cpu's as "E6 Diegos". Please, let me just say that I use the "Diego" name as a habit and the new 1mb cache cpus aren't in fact based on the E4 "San Diego" cores. So technically, the term "E6 Diego" is impossible :p:

stephen, politicians use words like "different". Any chance of cuttting out the politics and telling us what different means.

Paranoia reigns. AMD Conspiracy to take the well-overclocking chips away! :)

It seems Venice will also get a replacement. New OPN for the 512kB L2 core:
ADA3?00DKA4CG

I think it's a bit strange that they would end the big Opteron thing going on. They have made a big success from day one, AMD saw what was happening. All OC'ers completely loved them. So what did AMD do? They removed the 3000+, and changed the prices so the 144 and the 146 became cheaper (oct 31) than the 3200+ even though Opterons are more expensive to produce (1 MB cache).
In short, they made the opterons even more good looking after it became the overclockers dream, why would they suddently change again?

Mats,
They probably get better yields from E6 silicon than E4. Better yield -> more working cores -> more $$$ (-> price drop?)

AND

better silicon -> better OC

thank god.

Quote:

---

Originally Posted by manomanx2

thank god.

---

Don't get too excited... These are pure speculations.

Quote:

---

Originally Posted by largon

Mats,
They probably get better yields from E6 silicon than E4. Better yield -> more working cores -> more $$$ (-> price drop?)

AND

better silicon -> better OC

---

More likely, it's a shift in production process. It's easier and more cost effective to produce one basic type of cpu and "lock/unlock" certain parts of the core to produce the different models ;)
The E4 Opterons will become cheaper since AMD will eventually stop producing that core, therefore they get thrown in the "clearance bin" :p:

Most of you don't realize that overclockers only make up 1-3% of overall sales. AMD doesn't care if their CPUs OC well or not, they will do whatever gets them the most profits.

If they did figure out a new process that improves OC and is cheaper or the same as the current process, then good, all new CPUs will OC better overall. But if the demand for these low end Opterons has exceeded supply (which it has), it's possible that the code change will mean lower OCs. They could of tweaked the process to produce higher yields, but lower OCs overall, for example.

Remember, Opterons are meant for SERVERS. The volume of sales for the server market is WAY lower than the desktop market. This is why Opterons are so rare compared to normal A64s, hence the price hikes.

Look back, and it's the same story when AMD introduced the Winchester, Venice, and even X2. All of the early samples were generally better OCers than the latter ones. Most 3700+ Diegos and X2 4400+ these days have a hard time reaching 2.6Ghz stable.

This is where all started :) .....http://www.vr-zone.com/?i=746&s=3
Imo, every cpu's week or Cxxx codes are always different in term of OCability...

That was a very interesting link, thanks for sharing :)

One thing worth noticing is that the multiplier is pre-programmed by the speed grading machine through the pins, not by the die or the package.

WTF? This means there's no laser cut or any other "hard" programming. I guess the guy who thought he could unlock A64 with the JTAG interface was right.

So, opteron 154 = fx57 :)

Quote:

---

Originally Posted by HKPolice

Most of you don't realize that overclockers only make up 1-3% of overall sales. AMD doesn't care if their CPUs OC well or not, they will do whatever gets them the most profits.

---

The OC'er market is not even an afterthought for AMD. However in the server market we have a chance to have a bigger impact than the general desktop market.

Quote:

---

Originally Posted by HKPolice

Remember, Opterons are meant for SERVERS. The volume of sales for the server market is WAY lower than the desktop market.

---

How funny would it be that the 939 Opteron turns out to be a stunning sales success (based on non-server sales) and AMD goes on to claim they're winning a bigger share of the server market which in turn gives them more sales leverage with corporate sales. A brilliant plan. ;)

Quote:

---

Originally Posted by Dumo

This is where all started :) .....http://www.vr-zone.com/?i=746&s=3
Imo, every cpu's week or Cxxx codes are always different in term of OCability...

---

Not entirely... Notice that ALL A64 cpus have "assembled in malaysia" written on them, yet the article you linked was from an AMD plant in Singapore. It seems pretty stupid to assemble the CPUs in malaysia then ship it to Sinapore to be speed graded, packaged and then shipped to the rest of the world. I know that the actual silicon is made in Dresden Germany, but couldn't it be assembled in Singapore as well as Malaysia?

Hence why there are different CXXX codes maybe? Each one representing a certain stepping from a certain assembly plant.

As you all know, certain countries make better OCing CPUs overall. IE Made in China P4s are the worse, while Made in Malaysia is usually the best.

Quote:

---

Originally Posted by HKPolice

Not entirely... Notice that ALL A64 cpus have "assembled in malaysia" written on them, yet the article you linked was from an AMD plant in Singapore. It seems pretty stupid to assemble the CPUs in malaysia then ship it to Sinapore to be speed graded, packaged and then shipped to the rest of the world. I know that the actual silicon is made in Dresden Germany, but couldn't it be assembled in Singapore as well as Malaysia?

---

Read it all :) ...

http://img391.imageshack.us/img391/8...shot0106ua.jpg

Quote:

---

Originally Posted by Dumo

Read it all :) ...

http://img391.imageshack.us/img391/8...shot0106ua.jpg

---

There are probably varying calibrations in the test equipment. A plant in Malaysia may test a core to be 4000+ while the same chip could be tested as only 3700+ in another plant.

That would explain the varying OCs from different C codes

It could be also possible they drop the 144 and the 146. I hope that doesn't happen :( But it could also the fact they run out of E4 which seems to be more logical.

I ordered mine over an week ago in Germany, but they are out of stock...
Nearly impossible to get one :(

Quote:

---

Originally Posted by gundamit

The OC'er market is not even an afterthought for AMD.

---

It quite obviously is, or the FX chips wouldn't even exist, let alone be unlocked :)

Quote:

---

Originally Posted by Kameleon

It quite obviously is, or the FX chips wouldn't even exist, let alone be unlocked :)

---

The point he is trying to make is that yes, AMD does cater to the oc market but it's revenue is a very, very small percent of its total sales.

Quote:

---

Originally Posted by Kameleon

It quite obviously is, or the FX chips wouldn't even exist, let alone be unlocked :)

---

Not really, FX chips are always the top of the line - fastest stock speeds. Companies always charge an arm and a leg when you want their best. Hence why these chips are always at least 2x more expensive than their normal A64 counterparts. When you're paying THAT much for a CPU already, you're basically also paying for that unlocked multiplier. Hell, it's probably EASIER to make an FX since they can skip the multiplier programming stage.

Same with the P4EE series, top of the line stuff that costs wayyyyy more, and AFAIK, they're not even unlocked.

So it's more like AMD saying "Meh, you enthusiasts are paying double the normal price already, might as well throw you a bone and leave it unlocked, saves us an extra step anyways"

Quote:

---

Originally Posted by gundamit

The OC'er market is not even an afterthought for AMD. However in the server market we have a chance to have a bigger impact than the general desktop market.
How funny would it be that the 939 Opteron turns out to be a stunning sales success (based on non-server sales) and AMD goes on to claim they're winning a bigger share of the server market which in turn gives them more sales leverage with corporate sales. A brilliant plan. ;)

---

That's a good theory but I doubt corporations are dumb enough to fall for it. They will research on chipset sales too. If there's 2-3 times more CPU sales than chipset, then obviously not all of the CPUs are actually going into servers. IMO, almost all of Opterons in corporate servers will use ECC ram anyways, there is no real reason for them to use normal DDR ram and risk memory corruption.

Quote:

---

Originally Posted by manomanx2

thank god.

---

i have been trying to get an amd64 stable @3ghz on air for over 2 years now.
my cab2e looks like it is maxing out @ 2978mhz.

there is no god.


very truly yours,
politenessman

Quote:

---

Originally Posted by HKPolice

IMO, almost all of Opterons in corporate servers will use ECC ram anyways, there is no real reason for them to use normal DDR ram and risk memory corruption.

---

What was AMD's intention for S939 Opterons, anyway? What purpose could they possibly serve that normal A64s don't, apart from being lovely lovely clockers for us lot?

Quote:

---

Originally Posted by Kameleon

What was AMD's intention for S939 Opterons, anyway? What purpose could they possibly serve that normal A64s don't, apart from being lovely lovely clockers for us lot?

---

There is high demand for low-memory (<= 4 GB) servers with no excessive reliability requirements either.

Socket 940 Opteron systems had a very good reputation. But less than perfection-requiring businesses are hesitant to buy because they have to buy expensive motherboards that require registered RAM, which usually means ECC, too, and the board vendors are giving out paranoid lists of qualified RAM.

Socket 939 is percived as much easier to deal with, but businesses would not buy anything labeled "Athlon 64" which they assume "Athlon == gaming".

Even in our company which is certainly smarter than average when it comes to computers did people say "now that AMD is offering Opterons for 939 maybe that's an option". It's very hard to realize the difference between a 1 MB cache Athlon 64 and a socket 939 Opteron - none.

People also ignore the fact that socket 939 mainboard also have published qualified vendors lists for RAM and some board are pretty picky. Such things are overlooked when the assumption socket 939 == easy is so convenient. Same for the assumptions that socket 939 systems have faster RAM.

Quote:

---

Originally Posted by Kameleon

What was AMD's intention for S939 Opterons, anyway? What purpose could they possibly serve that normal A64s don't, apart from being lovely lovely clockers for us lot?

---

Smaller companies who aren't running such mission critical apps. Notice I said "almost all of the opterons in corporate servers" in my previous post.

If you look around in other forums, there are quite a few Optys for sale, that AREN'T good OCers or are "OEM, never tested" :stick:

Speculation, speculation. The only option now is to wait.

Or kidnap an AMD official.

Quote:

---

Originally Posted by HKPolice

There are probably varying calibrations in the test equipment. A plant in Malaysia may test a core to be 4000+ while the same chip could be tested as only 3700+ in another plant.

That would explain the varying OCs from different C codes

---

The assembly and test plant in Malaysia merely does the assembly to the substrate and then tests to see if the cpu is functional - since K8 they have done no speed grading. ALL K8 speed grading but no assembly is done in Singapore.

Quote:

---

Originally Posted by largon

Mats,
They probably get better yields from E6 silicon than E4. Better yield -> more working cores -> more $$$ (-> price drop?)

---

In theory, yes. But that's not what AMD did. The bigger, older 146(115mm², 1 MB L2 cache, E4) is $30 cheaper than the smaller, newer 3200+ (83mm², 0.5 MB L2 cache, E6). It's not because they've not updated the prices, they've done that two times last months. Yeah 146 is newer on the market but you know what I mean, E4 is older than E6.