Sign in with
Sign up | Sign in
Your question

Mat, I dont wanna say I told you so....

Tags:
Last response: in CPUs
Share
June 10, 2002 8:53:50 AM

But I did tell you so.

I wins I win!



This post is best viewed with common sense enabled

More about : mat dont wanna told

June 10, 2002 10:52:50 AM

Red packaging, Green packaging tbreds are overclocking to the lower end of my prediction scale, I if anything, slightly over rated the overclockability.


I am thinking on why, and I think its due to the copper transistion, I did not give enough credit to the addition of copper on the tuallatin line.

Either way many tbred samples are clocking@2.3ghz and that number will grow, that is about 30% topspeed increase, which was on the low end of my origional prediction.
(25-50% topspeed increase).


So, while you did disagree with me, anyone who wants to dig up my origional claim, can see I am not incorrect at all.

Disappointed?, possibly, incorrect?, No.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 10, 2002 11:05:14 AM

Also, my prediction for big overclocking was for downbinned cores, and I havent seen any reports of a 1700+ overclocking, but if it hits 2.2ghz thats a 50% overclock.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
Related resources
June 10, 2002 12:53:14 PM

Mat,

I think you are being overly defensive. It is apparent that the Athlon line is now dated, even Barton has no chance against the high GHZ capabilities of the P4.

But consider this: With the upcoming Hammer line of processors, the older Athlon line (Tbred, Barton) will become the economy processor, replacing the Duron. The .13 architecture will allow a greater yield, dropping the price significantly. The present Athlon line will eventually be selling near the price range of the current Duron line once Hammer is established. That is a lot of processor for that price, regardless of the level of the competition.

AMD needs the Hammer series to compete against Intel for the processor crown. If it does well, then the Athlon line will continue as the economy chip. If the Hammer does not deliver, then the Athlon line will disappear as the Duron soon will.

Until it is released, you can only guess at Hammers performance. If Hammer cannot compete performance wise, we will see AMD settle in as the ecomomy chip alternative to Intel (both price and performance wise). The days of paying less for more performance will then be at an end. But if Hammer performs, then the race for the high end processor line will continue for a bit longer, at least until the next Intel processor release.

But for now, face it. While the Athlon is the champ clock for clock, it cannot keep pace with the P4 while having a 700mhz clock deficit.

<font color=blue>This is a Forum, not a playground. Treat it with Respect.</font color=blue>
June 10, 2002 1:52:03 PM

Nobody expected it will, AMD is here waiting for Hammer, they are just adding chips to keep some kind of competition, but nothing extreme. It is now a period of relaxing...

--
Meow
June 10, 2002 1:54:23 PM

Hey Mat I also forgot to tell you this, but I was amazed they made 1700+ chips!
If those aren't downbinned, technically they might have a lower heat than the downbinned ones, no?
Well at 1.5V, NW's usual Vcore, and with the green packaged Tbreds, they just might get at least 45% OC.
Let's just hope, cuz the current tests are flawed IMO, I see too many holes here with no justifiable reason that they were meant this way.

--
Meow
June 10, 2002 3:10:26 PM

1: im not defensive, I was mentioned by name, and I presented facts which at least should cause us to explore whats up with the red green issue.


secondly

Quote:
Until it is released, you can only guess at Hammers performance. If Hammer cannot compete performance wise, we will see AMD settle in as the ecomomy chip alternative to Intel (both price and performance wise). The days of paying less for more performance will then be at an end. But if Hammer performs, then the race for the high end processor line will continue for a bit longer, at least until the next Intel processor release.

But for now, face it. While the Athlon is the champ clock for clock, it cannot keep pace with the P4 while having a 700mhz clock deficit.



I have always said amd is betting on the hammer, I know the athlon core is dated, as for the hammers performance, an 800mhz hammer beat a 1.6ghz p4 on quake 3, intels traditional strong point, wanna extrapolate that?

1.6ghz hammer > 3.2ghz p4, hows that for extrapolation. .13+soi+2 additional stages shows the hammer clocking above the top tbred clock, which even by PESSIMISTIC reports seems to be ~2ghz, therefore you see hammer which will compete with the p4 till 4ghz at least.


MY points are thus.

A: the red core green core issue needs to be explained
B: there are reports of 2.2ghz+ overclocks aircooled for the tbred, and amd demoed a 2.2ghz aircooled processor
C: the tbred line ONLY has to compete with the p4 till the end of this year.
D: My predictions on the clockability of the tbred line were generous, overly so, due to the fact I miscalculated the role of copper in the tuallatin core upgrade, however the actual results were within my predicted range, and I have no reason to be defensive cause I was right.
E: eveyrone should wait for the resolution of the green core issue before passing judgement on the tbred as obviously something is up.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 10, 2002 3:23:55 PM

While it's interesting to note that Hammer running at 800mhz outperforms a P4 at 1600 mhz on Quake benches. HOWEVER, Intel the comparison was between the Hammer and a P4 Willamette, a much lower IPC P4 than the current Northwood. So, while a 1.6ghz Hammer may match a theoretical 3.2ghz P4 Willamette, I doubt the same would hold true between a Hammer and Northwood.

Mark-

<font color=blue>When all else fails, throw your computer out the window!!!</font color=blue>
June 10, 2002 6:06:53 PM

Quote:
<i>Matisaro says:</i>
Either way many tbred samples are clocking@2.3ghz


Haven't seen any yet, just the rumor of Gigabyte having one, which needs to be confirmed or refuted.

Quote:
<i>Eden says:</i>
Nobody expected it will


I thought you were around when Mat was telling everyone and their dog to pass up on 1.6As and wait for Thoroughbred, because it'll overclock "just like the Northwood".

<font color=blue>Hi mom!</font color=blue>
June 10, 2002 6:25:03 PM

Big high five IIB, I would also like to add.

Matisaro, I told you so too!

Post some links to those 2.3Ghz Tbreds please, we have already gone over core color as being fab related and nothing more.

You are limited to what your mind can perceive.
June 10, 2002 6:55:20 PM

Again in your logic, one fab uses red, one uses green. The green works better, thus that fab has better timing, while the other has the bad ones.

--
Meow
June 10, 2002 7:19:11 PM

Quote:
I thought you were around when Mat was telling everyone and their dog to pass up on 1.6As and wait for Thoroughbred, because it'll overclock "just like the Northwood".


Low binned ones just might. Wait and see.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 10, 2002 8:39:48 PM

They didn't even bother to try out the XP1700+ at 1.5V. For all you could know, this XP2200 might not be a good OC stepping at all while the 1700+ is.

--
Meow
June 10, 2002 10:30:52 PM

Quote:
They didn't even bother to try out the XP1700+ at 1.5V. For all you could know, this XP2200 might not be a good OC stepping at all while the 1700+ is.


Heres a theory for you, if you were amd, and you had 10000 non scaleable chips you had to sell to a community you knew to be overclocker friendly, what would you sell those chips as?

Conventional wisdom would tell me to bin them at the lowest possible and mobiles, simply because that would give them the most headroom etc.

But, knowing the community, why not sell them as the topspeed chip, because most people who buy to overclock NEVER get the top binned chip, therefore the people who buy the 2200+ usually run them@2200+.

I am perfectly willing to accept the tbred dosent clock well, I just want more evidence before hand.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 10, 2002 11:12:19 PM

Like all prosessors it takes a little while until the process and yields are good, it'll overclock better later, and as xbitlabs tested. Running the athlon at 166mhz fsb gives it a 10% boost in speed with DDR333, and if you take that into account I find the xp to be very competitive with P4 still!

AMD gives you most bang for your bucks!!
June 10, 2002 11:24:04 PM

Oh please - why do you keep on going?
its Clear you we're wrong - it happens.

early bin 1.6NW USED to give ya a 50% OCing range - now its even higher - somthing like 2.6-3Ghz Air-cooled on regular basis - NO athllon XP gives ya 50% OCing, not on regular basis, not AT ALL.

and the 2200+ is a OCing joke. if we compare it with the 2.4NW and say it regularly OCes to 3Ghz air-cooled. then it has 25% regular Overclocking space.
the 2200+ cant achive half of that on regular basis air cooled - a 25% overclocking for it would take it to 2.25Ghz - [-peep-] regular basis - I've NEVER seen the Athlon XP does that speed, Air cooled, Proven stable, or else for that matter.




This post is best viewed with common sense enabled
June 11, 2002 1:11:51 AM

Quote:
Oh please - why do you keep on going?
its Clear you we're wrong - it happens.


Whos keeping on going, I am waiting for more results, if you read all the various threads on the subject I have admitted I was wrong countless time.


Quote:
and the 2200+ is a OCing joke. if we compare it with the 2.4NW and say it regularly OCes to 3Ghz air-cooled. then it has 25% regular Overclocking space.
the 2200+ cant achive half of that on regular basis air cooled - a 25% overclocking for it would take it to 2.25Ghz - [-peep-] regular basis - I've NEVER seen the Athlon XP does that speed, Air cooled, Proven stable, or else for that matter.


And I say to you the damn cores been out 1 day, and I have put forth possible issues with the red core testing samples, whats your problem with waiting a week or so for more data?



:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 11, 2002 2:17:31 AM

10000 chips is nothing......they sell over 3 million a year. They would rather throw em out then send them to reviewers......Who wants bad reviews? Thats just stupid.

I sold my sig for $50.
June 11, 2002 5:55:39 AM

Quote:
10000 chips is nothing......they sell over 3 million a year.

Do we have to keep going over the numbers!?!? AMD is on pace to sell 30+ (thirty) million CPUs in 2002. Not 3 million

10+ TIMES that.. I provided linkage in the last thread when this came up.

Mmmm... <font color=red>Red Hot</font color=red>
June 11, 2002 10:28:43 AM

10000 is a made up guess, it is probably signifigantly higher than that.

As for sending reds to the reviewers check out my 4 possible reasons for that. In this thread or the other one.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 11, 2002 4:55:40 PM

IIB it is clear the K7 is at its end days. We are NOT discussing why that. We are discussin technical stuff, we are intrigued by the fact a 0.13m chip clocked less than a 0.18m one. This isn't about defending Tbred at all!

BTW not many would usually test the TOP END chips. I think enthusiasts are those who buy the low clocked chips of a core, which cost less, and OC them high like Saturn, since those make it worth the money! They should test the XP1700+ at 1.5V, until I start changing my mind on this technical weirdness.

--
Meow
June 12, 2002 12:08:40 AM

NO athllon XP gives ya 50% OCing, not on regular basis, not AT ALL.
and the 2200+ is a OCing joke

well, no [-peep-] Sherlock, that's like saying a 2200+ is going to overclock as well as a 2200NW, if you are LUCKY, you can get the 2200NW to 3000, thats only like 36 percent, not the same as a 1600NW to 2500/2600 or so

now...take an AXP 1700+ Tbred, and overclock that to around 2500, then you get a reasonable oc

Just stop flaming Mat already

"When there's a will, there's a way."
June 12, 2002 2:29:42 AM

Quote:
Just stop flaming Mat already


Heh, some weeks just arent yours ya know?

I made a prediction based on the evidence available at the time, according to the evidence now my prediction was overzealous, however there is some question as to the veracity of the evidence we have now, so I am waiting.


Either way, I am not bent out of shape, and some people would try and use my prediction to attack me, coughfuggercough, I dont care, I am coming off calm and logical and I refuse to drop to the level of flamer in response to peoples comments towards me.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 12, 2002 2:50:22 AM

*grins*

i'm considering cancelling my order on newegg for an AXP 1800+, and just waiting for the 1700+ T-breds...they sound at least somewhat promising, maybe it still will be 1.6a repeated

"When there's a will, there's a way."
June 12, 2002 2:52:06 AM

1700+ should hit 1.9ghz easily, seems that with the current evidence the core itself tops@1.9ghz, the consistancy of the top clocking signifys that.


1700+@2300+ should be expected, which is not a bad overclock in the slightest.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 12, 2002 2:57:27 AM

166Mhz FSB X 11.5 multiplier, 1900Mhz Athlon sweetness...2300+ eh? I'd wager that WILL compete favorably with a 2.53NW, would you agree? 1466Mhz to 1900Mhz isn't a bad overclock, and looking at results from Hardocp, if they can get their 2200+ to do 2025Mhz, then i'm certain a 1700+ can hit at least 1900, maybe even 2000.

"When there's a will, there's a way."
June 12, 2002 2:57:38 AM

Anand's results of the XP1700+ OC was dismal, where it needed 1.85V to reach 1.66GHZ. Once again it proves something is wrong in the 0.13m process, because the 0.18m was clocking aircooled on XP2100 levels or more.


--
Meow
June 12, 2002 2:59:16 AM

No it can't compete the NW 2.53GHZ, it uses a very high FSB and RAM. It just wouldn't. It will however match or outperform a stock 2.4GHZ 400MHZ FSB one.

--
Meow
June 12, 2002 3:00:04 AM

Do you think that AMD hasn't let the T-bred mature enough yet? It would partially explain perhaps why the Clawhammer samples so far only run 800Mhz. .13m and SOI must be causing some production problems with the Hammer, I wouldn't doubt the T-bred has problems with just .13m. 1700+ not worth it? Yet?

"When there's a will, there's a way."
June 12, 2002 3:02:53 AM

I can wait for sure. This has become technical, not just about getting a Tbred or not. Maybe Mat's theories might be true about THG and the rest using sample tbreds which were not clocking well and had issues. Otherwise there are so many things that are odd about the Tbred, which is why I find no sense in this.

--
Meow
June 12, 2002 3:14:15 AM

Do you mean just the thouroughbred or AMD using .13m in general? Something isn't right, I mean, if you read the report from Richard Heye, he "assured" Hardocp that AMD has perfected their .13m process, which I highly doubt.

heres a link http://www.hardocp.com/article.html?art=MzEw

I kind of doubt it's just the engineering samples, I really do believe something is wrong with the T-bred in general. According to THG, the transistors dropped to 37.2 million, since their was a die change. Perhaps some of the core was disturbed with the rearrangement and transition to .13m



"When there's a will, there's a way."
June 12, 2002 3:20:34 AM

I'm not a silicon expert, Mat is, but from what I think I am seeing in the Palomino silicon, there was two areas, top right and bottom right, with seemlessly nothing on it... I dunno if it was an empty transistor area for heat dissipation or something else, this is just a guess.

--
Meow
June 12, 2002 3:30:09 AM

Would that explain the change in shape of the die?

"When there's a will, there's a way."
June 12, 2002 12:23:47 PM

The die is shaped in expectation of another 256k of cache for the barton.


This is not a .13 process issue, but a core issue, I have not discovered why its not clocking well, and am waiting for more data.

I dont think amds .13 process is bad, but the tbred core may just be at its limit, or have a flaw which can or can not be corrected, hard to see the future is.

The darkside clouds everything.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
June 12, 2002 1:51:00 PM

I am personally more sure it's a flaw in the Tbred's process, because otherwise there is no explanation to why it OCs worse than a 1700+ 0.18m would, considering it uses 0.2V less.

BTW those two parts I spoke of, on the silicon, are they empty transistor areas?

--
Meow
!