Only a 5% overclock on the T-Bred?

FUGGER · Jun 10, 2002

To shut Matisaro and Kelledin up and prove they did not take the time to read the article that they are desperately relying on.

On the conclusions page of amdmb.com it states this:

"A few summing up points on the new 2200+ processor: the color of the Athlon XP 2200+ processor packaging will be in either a green or brown color. Soon all AMD processors will be moved over to a green package, but currently the fabs are producing both colors. The color in no way affects performance, so it’s merely an aesthetic difference. "

The color indicates the FAB!

You are limited to what your mind can perceive.

FatBurger · Jun 10, 2002

Raystonn says:
You folks were expecting this?

I was, everyone knows that. Mat was the main one that expected huge overclocks.

BTW, you shouldn't have tried returning to HardOCP's forums, dumbass :lol:
Good to have you back here, though

Raystonn says:
Whatever happened to the claims that AMD's shrink to 0.13 micron would enable overclocks similar to those seen with the Northwood Pentium 4?

They apparently died today...

So bye bye, Miss American Pie
Left my Chevy to the levy-

Oh, sorry :redface:

noko says:
Maybe the Barton can even it up somewhat

Doubt it. It's only doubling the L2. That'll make it harder to scale, if anything.

noko says:
Well I have 3-6 months before I will do that, hopefully things will change for the better

Hammer will be around in that timeframe, but most likely in limited quantities and expensive.

AmdMELTDOWN says:
get a load of the clamp for the hsf, you'll need a f'n crow bar to get that sucker on LOL!

Those have been around for quite a while, just not very common. And they don't require any more presure than the single lug HSFs.

SammyBoy says:
On the topic of overclocking, I have yet to see any sites drop the multiplier down so that a 166MHz FSB would still be at the stock speed of the 2200+. Those are benches I'd like to see. Might allow them to win more than 2 out of 32 benches, eh?

HardOCP did just that, and the results were pretty nice. Nothing that couldn't be done with a lower-priced Palomino, though.

Matisaro says:
Did anyone else notice the tbred they used in the test had RED PACKAGING, amd switched the green several months ago, this clearly indicates that tom is using an OLD chip, take this test with a grain of salt.

You mean AMD was sending Tbred review samples "several months ago"? Not likely, sorry.

Matisaro says:
only green chips should be accepted as examples of the cores performance.

If the green ones are the only ones that ship, then you could have a valid argument. I thought that the color of the packaging made no difference, however? That's what I've always heard. Or are you saying it's just because they would be newer?

Matisaro says:
Sites all over are showing 2ghz+ overclocks with air cooling

I think you mean to 2GHz+

And no, HardOCP mentions 100MHz on air, Anandtech had similar results. Haven't checked Ace's yet.

lhgpoobaa says:
still think amd should have forced thermal protection from the start of the XP.

I agree, it would've helped them out immensely.

BTW, I wasn't online when you sent me a message on MSN. I was in bed asleep :tongue:

Eden says:
AMD will probably react to this article soon, and resend the real hand-picked chips like Intel does.

So only the cream of the crop will overclock well?

Eden says:
Then upon looking at the results, I can only deduce, that behind logic, something is wrong. Come on people how can you not realize! It's obvious there is something wrong in the chip.

Sending out chips made "several months ago" is something way too stupid for AMD to let happen, I doubt that is the case.

Eden says:
So personally, I won't beleive this chip's OCing on this website, nor Anand's until AMD or someone professional in these chips confirms it.

Like...? Every website on the intarweb is having the same results with a couple of exceptions, but it's the exceptions that are actually the norm? I doubt it.

Matisaro says:
A: the hardware sites got chips through normal channels not from amd. The majority of chips RIGHT NOW are red.

So the chips that all of us will be able to buy won't be overclockable in the least?

Matisaro says:
B: The review sites nda's have finally been lifted and they have had these samples for months but could not review them till now

Possible, but I doubt it.

Matisaro says:
Ray has been lurking waiting for a moment when intel is ahead to gloat

Don't think so, he returned to HardOCP's forums at the same time, and he hasn't mentioned Tbred there.

Matisaro says:
as the results obtained are on the low end, but within the results I predicted.

The best result is 17%, and only one site got anywhere near that. The rest are more like 5-10%. Not exactly "just like the Northwood".

Matisaro says:
1: heatspreaders increase temprature, burger has some interesting links on that, they are to protect cores from damage, not to help with heat, why they are called heatspreaders is beyond me.

What? Someone actually remembers that? :tongue:
I had to post it enough times, but anyway...

They live up to their name, they do spread the heat. They just don't help dissipate it, since they don't conduct the heat well.

JCLW says:
The thing that shocked me most was the lack of an IHS on a 0.13 chip.

Not gonna help, as Mat said above.

PR497 says:
isnt the new TBred supposed to overclock like a champ?....

Yeup, and now that Mat has been proven wrong and won't admit it, we get to have a huge debate that doesn't get anywhere.

I'm just glad Mat isn't as stubborn as I can be, all debates would be like this.

Hi mom!

gal128 · Jun 10, 2002

Would everyone be this upset if the T'bred could meet or beat the 2.53P4 w/o being overclocked? I was expecting that. Same performance at half the cost would have been great. Can you match T'bred's performance with an OC'd 1.6 or 1.8 P4?(By performance work done not clock speed) If that is possible and I were AMD, I would be embarassed. It would be great if there was a huge drop in athlon prices when the Clawhammer comes out.

HULK SMASH!!!

BigRollTide · Jun 10, 2002

In reply to:
--------------------------------------------------------------------------------

BTW, I wasn't online when you sent me a message on MSN. I was in bed asleep

--------------------------------------------------------------------------------

You sleep?

FatBurger · Jun 10, 2002

You sleep?

I'm trying to quit :redface:

EDIT:
And Hot Hardware agrees with me:

The Athlon XP 2200+ we tested was packaged in brown organic material (scanned below). AMD will eventually produce all of their CPUs using green packaging. Don't worry if you get a brown CPU though, the color has absolutely no bearing on performance.

<A HREF="http://www.hothardware.com/hh_files/CCAM/amd_axp_2200.shtml" target="_new">Link</A>

Hi mom!Edited by FatBurger on 06/10/02 11:02 AM.

FUGGER · Jun 10, 2002

Lets be fair, I think 2 weeks was the deal for Matisaro to obtain a Tbred and break 2.4Ghz or 50% with a locked multiplier and stock HSF.

For someone who claims to have worked in a fab he sure let that chip color thing slap him in the back of the head.

You are limited to what your mind can perceive.

jclw · Jun 10, 2002

A real IHS _should_ provide a larger surface area for moving heat from the core to the heatsink. If the PIII FC-PGA2 package actually does it is another question.

In my opinion the 0.13 micron AMD chips need an IHS that works.

- JW

FatBurger · Jun 10, 2002

True, 2 weeks for a 50% overclock with any model of Tbred.

Hi mom!

Copenhagen · Jun 10, 2002

Mat, Mat, Mat ... sigh.

I warned you long time ago about the lack of clockspeed headroom that could be expected from Thoroughbred.

The Palomino was a result of an excellent fabrication process (0.13 micron class gates) together with a new core layout. Both resulted in lower temperatures and more clockspeed headroom than the TBird. Almost every possible clockspeed potential was then squeezed out of the Palomino (XP2100+). This way AMD ended up "using" quite a bit of the gain which would normally be expected from a die shrink. I told you before, and I think it's fair to say that your expectations have been proven wrong, to say the least.

HardOCP was able to get 100 MHZ overclock using air-cooling (PAL 8045) and about 200MHz using water cooling.

It looks like it's difficult for the new shrinked core to get rid of the heat, due to the small size. Fact is, that the old Palomino can be clocked even higher than the new Thoroughbred, with proper cooling. What a big disappointment, though expected. It's certainly not a time to buy Thoroughbred with all those "bad" chips around, unless price goes sufficiently down. I expect that we are not going to see any substantial improvements until Barton, both due to the added L2 cache and the maturing of the manufacturing process.

Here is some of the final words from HardOCP:

So while the overclocker will be able to utilize a 333MHz DDR bus or better with his TBred CPU, it will take some very good air or water cooling to do so and you still won't get any big MHz increases on the CPU clock.

I hate to say it, but our Thoroughbred samples really reminded me of our last Palomino CPUs. 

/Copenhagen - Clockspeed will make the difference... in the end 😎

vagabond · Jun 10, 2002

From what I've seen of the new CPU, I believe I'll keep to my tried and true Athlon T'bird core 900Mhz.

Still whoopin' that a** in Counter-Strike!!! 😱)

Peace y'all!!!

Vagabond

You can lead a horse to water but you can't make him drink it!

eden · Jun 10, 2002

If the color indicated the FAB, then that means one fab has the bad timing issue, the other produces the green ones with the right fixed timing...

That is by your logic.

--
Meow

FUGGER · Jun 10, 2002

That is your theory, god forbid AMD produces them in one color.

"A great deal of anticipation about the Thoroughbreds has gone into the question of whether the die shrink will allow further overclocking of the next generation of XP processors. Unfortunately, at present, it does not appear as if there will be a huge gain over the current Palomino CPUs, the rationales were laid out earlier in this review. The XP2200 running at a phisical speed of 1800 MHz would overclock to 1938 MHz albeit with some stability problems. For those only interested in CPUID scores, we could up it beyond 1950 MHz, however, stable operation was not possible beyond 1900 MHz."

From <A HREF="http://www.lostcircuits.com/cpu/xp_2200/" target="_new">lost circuits</A>

You are limited to what your mind can perceive.

eden · Jun 10, 2002

I agree the future Athlon K7s are not meant to ramp much further, but still it does not explain in any way, that the thing only OCed less than 100MHZ.

We can however assume that since there was 0.13m technology in the Palomino, that it was a 0.13m product. This way, it would explain the transition not being as much beneficial. It would also mean that the XP2200 is one of the last of the saga, thus would explain a bit of the limited OCing, but not 100% logically.
Plus with that, comes also the fact they didn't try out the XP1700+ 0.13m! It is 1.5V! Maybe THAT would OC!

--
Meow

Matisaro · Jun 10, 2002

Yeup, and now that Mat has been proven wrong and won't admit it, we get to have a huge debate that doesn't get anywhere.

I'm just glad Mat isn't as stubborn as I can be, all debates would be like this.

I wont have a problem admiting I was wrong, I made a prediction which went with all available evidence, and if its wrong its wrong.

The issue im having is the fact the .13 cores overclock worse than the .18, this is HIGHLY STRANGE, strange to the point of me suspecting somethings wrong, coupled with the fact the green packaging is the only packaging they use now(despire what amdmb says btw fugger, there is only 1 fab).

I am perfectly willing to throw in the towel as soon as I get more information.

There have been reports of 2.1ghz overclocks and higher, amd demoed a 2.2ghz aircooled tbred and 2500+ tbreds are on the roadmap, all this from a core that according to thg,anand etc has 100mhz TOPS left to go, highly doubtful.

The Cash Left In My Pocket,The BEST Benchmark

Matisaro · Jun 10, 2002

If the color indicated the FAB, then that means one fab has the bad timing issue, the other produces the green ones with the right fixed timing...

THERE IS ONLY ONE FAB, so the theory greens are from one, and reds are from another is retarded.

AMD switched to ALL GREEN PACKAGING months ago, the fact these tbreds are red ARE PROOF they are older chips, period.

The Cash Left In My Pocket,The BEST Benchmark

dhlucke · Jun 10, 2002

Matt, back off of this one. You can't win it. Granted the new core seems worse than the prior one, but you're better off backing off the claim until you actually have something in your hand that says otherwise.

What's the deal with releasing a processor that really isn't a big improvement but requires a new HSF?

I have a computer and it does weird stuff. please help.

FatBurger · Jun 11, 2002

Matisaro says:
I am perfectly willing to throw in the towel as soon as I get more information.

Last time you said you'd take the results of the major review sites. Well, you're not.

Matisaro says:
There have been reports of 2.1ghz overclocks and higher

I saw one with a Vapochill, that hardly counts.

Matisaro says:
AMD switched to ALL GREEN PACKAGING months ago, the fact these tbreds are red ARE PROOF they are older chips, period.

No, it's not. It's proof of the fact that they're red instead of green. That's all. Period.

The fact that everyone except for Eden is accepting the results and telling you you're wrong should be enough.

Hi mom!

Matisaro · Jun 11, 2002

Last time you said you'd take the results of the major review sites. Well, you're not.

I said no such thing, and for the record I am not debating my prediction is right, I am debating we should wait for more evidence before saying it is wrong, I have highlighted reasons the major review sites overclocks are to be taken skeptically, I am waiting for more results before saying the tbred sucks.

No, it's not. It's proof of the fact that they're red instead of green. That's all. Period.

Do you deny that amd has been packaging all of its axps with green packaging, and have been doing so for the past 2 months?

I submit the question, why would amd be selling red packaged tbreds when they announced a transistion to green packaging months ago, and all axps fabbed after said date 2 months ago have come in green packaging!

The Cash Left In My Pocket,The BEST Benchmark

Matisaro · Jun 11, 2002

The fact that everyone except for Eden is accepting the results and telling you you're wrong should be enough.

And im not saying they are wrong, I am saying I am waiting for more confirmation and explination of the OBVIOUS strange things I have pointed out.

I am in NO WAY denying these overclocks, I am waiting for more reports from GREEN tbreds to see what is going on!

Thats all, nothing more.

Any attempts to paint me in a corner of debating against these overclocks are incorrect, I am doing no such thing.

There is no issue, if these oc's are truely representative of the tbred, I fully admit my prediction was wrong, I have 0 problem with that.

I do however see several things in these articles which make me desire to wait for more confirmation of the scaleability of the tbred.

I ask you, how do you think amd will release a pr2500 tbred which is on the roadmaps if the core wont even do 2ghz vapochilled?

You have to admit, something is fishy here, I dont know what, or what it all means, but something dosent jive, and all I want to do is find out more.

Can anyone blame me or say I am grasping at straws for wanting more data?

I have admitted fully that my prediction was higher than reality, even gave the reason, and if these results are borne out by other tests then by all means, I was wrong.

Whats the big deal about keeping an open mind?

The Cash Left In My Pocket,The BEST Benchmark

Kemche · Jun 11, 2002

Since I don't know much about packaging. My question to you is that how much difference does the package makes. If you think Red(Brown) package is the reason that T-bred can't be overclocked well can you prove that. Because I think when AMD Switch Palomino core from Red to Green it didn't made any difference in Overclockability of Polomino. So why would it make a difference on T-bred core?

KG

"Artificial intelligence is no match for natural stupidity." - Sarah Chambers

Raystonn · Jun 11, 2002

This is not really a packaging issue. AMD rearranged the components on their core. It is now rectangular. Why did they do this? So they could fit on some more cache with the Barton without having to redesign the core again, making it square again as a Barton. What does this mean? The T-Bred is not necessarily optimized for best performance as it currently exists. It is optimized for upgradability to the Barton.

AMD does not have the resources to redesign their cores for an optimal layout every time they release a new version.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =

Matisaro · Jun 11, 2002

Since I don't know much about packaging. My question to you is that how much difference does the package makes. If you think Red(Brown) package is the reason that T-bred can't be overclocked well can you prove that. Because I think when AMD Switch Palomino core from Red to Green it didn't made any difference in Overclockability of Polomino. So why would it make a difference on T-bred core?

You misunderstand, the packaging color has no effect itself.

Let me lay it out step by step for clarity.

~2 months ago, amd switched the packaging color on all of its axp processors from red to green, every processor I have seen which was fabbed after said date came with a green core color.

This required new machines new plastic, whatever.

The fact is that for the last 2 months no chips made have been on a red organic package.

OK, now.

It was heavily rumored that the tbred was delayed because the first stepping contained timing issues which limited the cores scalability, it was further rumored that many chips were completed with this issue, but a core relayout was completed which repaired the issue, it takes 3 months for a chip to get from fab start to fab out generally(give or take depending on the fab and process).

OK, with me so far?

Now, we know months ago amd produced bad chips which did not clock well, we also know that they fixxed this issue several months ago, add that to the fact all new chips come in green packaging gives you this.

The red packaged tbreds are at least 2 months old, and this makes it likely they were from the first flawwed stepping, there have been reports I have seen in other threads, and bear in mind it is day 1 of nda removal, many other reports will come out im sure, which is what I am waiting for. Which say that some tbreds are clocking very well, amd itself has demoed 2.2ghz tbreds aircooled, I find it hard to believe that a core which anand cant even get to 1.9ghz with a vapochill is going to be able to be released at pr2500+ as amd currently has on the roadmap.

I think MAYBE something is up with the red cored tbreds, and my theory is sound as to the time of their creation, and all I am waiting for is either

A: reports saaying amd is making red and green chips at the same time(from a reputable source) which would be fine
B: green tbreds are overclocking as poorly as the red ones
C: more overclock results from red chips to further show they are limited, etc.

I am waiting for more data.

The Cash Left In My Pocket,The BEST Benchmark

Matisaro · Jun 11, 2002

AMD does not have the resources to redesign their cores for an optimal layout every time they release a new version.

I suppose you have evidence to back up this fud huh raystonn?

The Cash Left In My Pocket,The BEST Benchmark

Phelk · Jun 11, 2002

Since you're getting a beating in here I'll stick up for you Mati.

Seems to me that something just ain't quite right. It is true that the Athlon is past it sweetspot and the P4 is just starting to hit it's stride. But a 180nm to 130nm change should yield better results than what we have seen so far.

I think we should wait for the two weeks and see what the 'greenies' can deliver.

 The Opteration was a success... I'm now a full-wit 😱

Matisaro · Jun 11, 2002

Thats all I am saying, something isnt right.

All of the review sites topped out at 1.9ghz give or take, this is way lower than it should be, and all the review sites all have red cores, red cores I may add which should not exist since amd went to green packaging months ago.

Either the cores are old, or amd is doing dual packaging color(which makes no sense) or amd switched back to red for the tbred(which makes no sense).

Something just isnt right and I want more overclocking results before I pronounce the tbred dead, thats all.

The Cash Left In My Pocket,The BEST Benchmark

Only a 5% overclock on the T-Bred?

Distinguished

Illustrious

Distinguished

Distinguished

Illustrious

Distinguished

Distinguished

Illustrious

Distinguished

Distinguished

Champion

Distinguished

Champion

Splendid

Splendid

Polypheme

Illustrious

Splendid

Splendid

Distinguished

Distinguished

Splendid

Splendid

Distinguished

Splendid

Share this page