Sign in with
Sign up | Sign in
Your question

.13 AXP core??

Last response: in CPUs
Share
March 5, 2002 6:09:50 AM

um... so when are we going to see the .13 AXP core?

<font color=red><b>land of wonder is the next door down</b></font color=red> :wink:

More about : axp core

March 5, 2002 6:41:00 AM

Are you named after the dragon lord from dragon warrior?

If so, probably within the next month or so.


If not,go away!

JK.


Within a month or so is my guess, amd said march last I heard.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 5, 2002 8:06:24 AM

thanks matisaro, and lol i don't know what you are talking about! sorry.

<font color=red><b>land of wonder is the next door down</b></font color=red> :wink:
Related resources
March 5, 2002 10:44:18 AM

the tbred is .13?

<font color=red>DO NOT LIGHT YOURSELF ON FIRE</font color=red>
March 5, 2002 10:48:03 AM

Yes griz, the tbred coming in about a month, is .13micron, it will overclock JUST like the northwood!

I can easily imagine a tbred 1700+ hitting 2.5ghz with good cooling, which performance wise, is just as scaleable as the Northwood.


Now you see why I suggest you get a good mobo and ram, and use your tbird untill the axp arrives, or even the barton, because the good benifits of the northwood, will be even better on the tbred due to the ability to unlock.

(2.6ghz on 166fsb would probably out perform even a northwood 3.5ghz on ddr or higher, it would be amazing.)

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 5, 2002 10:55:26 AM

Quote:


I can easily imagine a tbred 1700+ hitting 2.5ghz with good cooling, which performance wise, is just as scaleable as the Northwood.

That's highly unlikely. I doubt a 1.47GHz processor will go up to 2.5GHz.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
March 5, 2002 10:57:55 AM

LoL


NW 1.6a> 3ghz
Tbred 1.47=2.5

The tbred % is lower, they should gain the same % clockability. Think about it amd_man.

If anything the 2.5 is conservative.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 5, 2002 11:00:05 AM

Why do you think its highly unlikely, a 1.6ghz p4 can hit 3ghz with good cooling, or more. The same applies for the 1.47ghz thunderbird, remember, its % of increase which counts, not pure clockspeed.

If the 1.6a can get 100% clockspeed gain, then so could the 1.47ghz tbird. Its all about the process shrink.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 5, 2002 11:09:47 AM

Matisaro, you said it yourself that many 1.6A P4s max out at 2.2-2.5GHz even with good cooling.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
March 5, 2002 11:14:39 AM

Only because of the ram/mobo, remember.

I will sum it up now.


1.6a p4's in practice top out at 2.2-2.6ghz due to MOTHERBOARD/RAM limitations.

The chip themselves have a much higher maximum.(probably in the 3ghz region).

The topping of the p4 is due to the fsb nature of the overclock, the tbred will NOT have that limitation.

Caveats: There ARE some p4 1.6as which top out at 2.2(the chip) these are probably the low end of the bell curve, there will be chips like that for the tbred as well.(expect a 400mhz overclockish).


Since the raw clockspeeds of the p4 1.6a and the axp 1700+ are so close, the general rule of thumb is that the average overclock of the axp will be slightly HIGHER pure mhz wise than the average 1.6a overclock(due to the ability to unlock the processor and run the cpu at its max regardless of fsb issues).


We clear now my friend?

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 5, 2002 11:17:08 AM

PS: There is NO physical reason I can think of to think that the tbred will not overclock the same as the nw due to process, however I reserve the right to be shocked and amazed incase something happens which renders this untrue!

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 5, 2002 11:33:31 AM

Yes, clear, thank you. If that holds to be true at the launch of the T-Bred, then I'm getting one instead of the Northwood!

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
March 5, 2002 12:29:32 PM

So indeed the multiplier unlock on Tbreds is gonna make it the top overclocker from then!
Man just imagine...less limitations by the bus being too big, instead having bigger multis....
Not to mention 1.47GHZ being even cooler than 1.6A by default, with less cache too!

--
For the first time, Hookers are hooked on Phonics!!
Anonymous
a b à CPUs
March 5, 2002 12:41:38 PM

Following your logic, a Tualatin should be able to overclock to over 2,5 Ghz with ease.. no ? Its .13 and available at what ? 1.4 Ghz ? guess what ? Not very likely. Im sorry mat, but your logic doesnt mean a thing. We all know P4 was designed to reach absurdly high clock speeds, it is beginning to reach them. We also know these 1.6A nortwoods are in fact 2+ Ghz northwoods rebranded to suit market needs. Maybe when AMD would start fazing out "old" AXPs, we would see some .13 Tbred that are very overclockable (much like early Durons) and rebadged with slower speeds, but I very much doubt an aircooled 100% overclock. Thats something we have NEVER seen from AMD. The most impressive overclocker from AMD I ever layed hands on was my good old Duron 600 with achieved just over 50%. I'll be VERY happy if my next Tbred does 50% overclock. Every other cpu from AMD I have had so far managed from 10% to about 40%.. aircooled. Thats good enough for me though.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
March 5, 2002 12:47:52 PM

You're right though, and in no way I think the Tbred could do this aircooled. However remember, the Tually you mention is at FULL speed 1.4GHZ, not its stock 1.13 that came out first. Always choose the least fast one which runs the coolest. P3s have their own scalability like Athlons, since they topped at 1 GHZ from 500MHZ or so. Tuallys should be able to reach less than 2GHZ, but they are limited by their own old technology. Tbred will go aircooled, OC by 400MHZ possibly. Watercooling will lead the way up to 2.3GHZ and more fer sure.

--
For the first time, Hookers are hooked on Phonics!!
March 5, 2002 1:54:03 PM

Who said anything about 100% overclock.
March 5, 2002 2:36:18 PM

Quote:
The topping of the p4 is due to the fsb nature of the overclock, the tbred will NOT have that limitation.

Huh? That makes no sesnse.

The Tbred will NOT see 100% overclocks. The P4 was meant to ramp in speed, and the 1.6A is actually a 2.0, don't forget. The Tbred should overclock better than the XP, but it won't be anywhere near 100%

Did we see these huge overclocking gains from the .25->.18 shrink? No, because the P4 wasn't around then, there were no CPUs designed like it is (to ramp in clockspeed). That's why the Tbred won't overclock like the P4.


Tell you what, Mat. If you can get a 100% overclock on your Tbred (perfectly stable) with air cooling, I'll pay for it. If not...hmm...you buy me a Lian-Li case?
Feel like putting your money where your mouth is?

<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
Don't step in the sarcasm!<P ID="edit"><FONT SIZE=-1><EM>Edited by FatBurger on 03/05/02 08:39 AM.</EM></FONT></P>
March 5, 2002 2:57:01 PM

Quote:


Tell you what, Mat. If you can get a 100% overclock on your Tbred (perfectly stable) with air cooling, I'll pay for it. If not...hmm...you buy me a Lian-Li case?
Feel like putting your money where your mouth is?

hehe, personally I don't like the way the Lian-Li cases look. I prefer the beige or black look of Antec SX1030 and SX1040 beauties, though.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
Anonymous
a b à CPUs
March 5, 2002 3:04:47 PM

Matisaro did. I quote:

>If the 1.6a can get 100% clockspeed gain, then so could >the 1.47ghz tbird. Its all about the process shrink.

Comming to think of it, he might have meant 100% compared to the previous generation when overclocked.. in that case, not even the P4 applies, since it went up to 2 Ghz on .18 stock frequency, and now 50% higher overclocked on .13; if you apply that logic to the athlon, which now tops out at 1.6 Ghz (right ?), you might expect 2.4 Ghz from Tbred. I'd be impressed, but you never know.

If you follow this logic completetely, it would really mean that if AMD would bring out a, say 1.33 Ghz Tbred, it would overclock to almost 2 Ghz.. which might be feasable.


= The views stated herein are my personal views, and not necessarily the views of my wife. =
March 5, 2002 5:29:07 PM

Ahhh...new shiny cases....is there anything not sexy about them?!... :wink:

--
For the first time, Hookers are hooked on Phonics!!
March 5, 2002 5:34:29 PM

thouroughbreds should be out on the 20th.

<font color=red><i>I refugee from Guatanamo Bay,
dance around the border like I'm Cassius Clay
</i></font color=red>
March 5, 2002 5:41:34 PM

fact of the matter is we won't know how well it'll overclock until we actually start testing them with quality parts and various cooling methods.
I personally would be happy getting a 1733 MHz Tbred running 150 FSB (close to 2000 MHz) or maybe 166 FSB if I were lucky. You do have to take into account that the voltage is going to drop greatly, and you can ramp it right back up for more stability when you overclock. Should be interesting to see what these buggers can do, but only time will tell.

"Why can't I be the man? I mean, I DO have harmony balls..." -epoth
March 5, 2002 5:44:05 PM

Quote:
thouroughbreds should be out on the 20th.

Thats a rumor started by the inqwell and has been shot down by Hardocp.
I guess the guys at Hardocp are confident in saying that IF Tbred was coming out the 20th they would already have at least one of them to start benchmarking with. And their site said they had not received any for testing yet.

"Why can't I be the man? I mean, I DO have harmony balls..." -epoth
March 5, 2002 6:00:18 PM

The Athlon 2100+ will be released between the 13th and 20th of this month depending on some market conditions.

The Athlon 3400+ will be released during the first week of April. It's initial debut price will make the 2100+ very attractive, but it's performance is meant to stun the community. They will release newer processors very slowly but very steadily after that up until the Athlon 4100+ at least.

<font color=red>God</font color=red> <font color=blue>Bless</font color=blue> <font color=red>America!

DELETE THE "ITS ALL GONE QUIET" THREAD!</font color=red>
March 5, 2002 6:01:56 PM

DH what the hell are you talking about?
PR3400 in April? 4100 later on?? Wut? You must be confusing it to Hammer next year!

--
For the first time, Hookers are hooked on Phonics!!
March 5, 2002 6:38:50 PM

Nope, just adding more rumors to the pile. Let me go on...

The Northwood will be released at 5 Ghz in September and the at 10 Ghz in February of next year depending on the weather.

<font color=red>God</font color=red> <font color=blue>Bless</font color=blue> <font color=red>America!

DELETE THE "ITS ALL GONE QUIET" THREAD!</font color=red>
March 5, 2002 7:07:43 PM

Hey FatBurger, if I overclock 100% will you buy it for me? Because the way things are going right now it looks like im going to get a thoroughbred. im not sure though..... if they suck then i'll get a northwood.

<font color=red>DO NOT LIGHT YOURSELF ON FIRE</font color=red>
March 5, 2002 7:34:00 PM

Quote:
The Northwood will be released at 5 Ghz in September and the at 10 Ghz in February of next year depending on the weather.

And then in summer of 2003 you can get a pentium 6 implanted in your brain.

"Why can't I be the man? I mean, I DO have harmony balls..." -epoth
March 5, 2002 7:35:44 PM

Yes, but it's not as easy to prove stability. Mat lives in the same city as me.

<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
Don't step in the sarcasm!
March 5, 2002 7:38:12 PM

No sense of humor? Where's the love!?

<font color=red>If you were to have sex with your clone would that be considered incest or masturbation?</font color=red>
March 5, 2002 8:56:08 PM

no I'm completely serious.
I thought you were too :wink:

"Why can't I be the man? I mean, I DO have harmony balls..." -epoth
March 5, 2002 9:03:30 PM

sure, um.... oh yeah AMD will come out with the mlouwerb soon too. oh at 10Thz : ) hihihi

<font color=red><b>land of wonder is the next door down</b></font color=red> :wink:
March 5, 2002 11:43:11 PM

I'll make a complete video, if I do it and it runs stable.


<font color=red><pre>i so good i jealous of me</pre><p></font color=red>
March 6, 2002 12:11:39 AM

*pulls up a chair and some popcorn*

When AMD's new thing drops on the market, make sure we get some results posted to this thread. I just wish I could see Matisaro's face when he realizes he is wrong about it scaling the same percentage in clockspeed as the 1.6A. We do have people getting a 100% overclock on the 1.6A with something like a Vapochill. So in order to verify his claim, we will need to see a 100% overclock for AMD. Just how long after introduction are we going to wait before declaring it a no-go anyway?

Let us make it clear now that we are talking about clockspeed and not PR rating. This is about overclocking the processor, not the marketing department.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
March 6, 2002 12:24:46 AM

can i have some of your popcorn? please.

<b> There is no greater sin against oneself than the failure to believe in one's own worth. </b>
March 6, 2002 12:32:35 AM

I don't think anyone cares about clock speed alone anymore though. We'll stick to the benches and most importantly the real life applications, costs, and so on.

I hope Matisaro is wrong. I like it when performance is close. If AMD gains the crown though it will be a plus since the Northwood could use some price drops and Intel needs someone to keep them on their toes.

I just don't a 30% difference in performance. That's too much.

From what I can tell the 1.6A scales nice due to the yields, but the average user isn't going to want to buy a 1.6A and overclock it so we're probably going to start seeing a drop in the scalability of the 1.6A as the yields move to higher clockspeeds that are more expensive. The same thing happened with the T-bird 1 Ghz as well as other processors. Intel isn't going to sell the 1.6A forever and the 1.8A and 2.0A don't scale as nice.

<font color=red>If you were to have sex with your clone would that be considered incest or masturbation?</font color=red>
March 6, 2002 12:43:53 AM

BBayens your entire post is just wrong, so lets start with the obvious incorrectness.

Quote:
Following your logic, a Tualatin should be able to overclock to over 2,5 Ghz with ease.. no ? Its .13 and available at what ? 1.4 Ghz ? guess what ? Not very likely.

A: the p3(.18) topped out at 1.1ghz, the tualitin has hit 1.7ghz(and some 1.8's have been reported) But ill stick with 1.7ghz, this is roughly a 60% overclock on the maximum.

B: The axp tops out at 1.8ghz(ythe 2100 is due for release, and most axps hit 1.8ghz easily on overclock). Multiply this by 160%(the 60% gain from the tually) and you get. 2.88ghz, You with me, still following SIMPLE logic here, and while you may debate the % gain from the p3, even if the gain is only 50% conservative estimate) the new max for the tbred will still be at 2.5+.

C: if the new max for the tbred is 2.88, then if they sell a slower processor as a tbred(like the 1.6a northwood) then you could EASILY, get a 80=90% overclock out of it.

Quote:
We also know these 1.6A nortwoods are in fact 2+ Ghz northwoods rebranded to suit market needs.

They are JUST THAT! They are backfilling the line to maintain product flow, the SAME would hold true for slower Tbred cpus, thus this point is invalid.

Quote:
We all know P4 was designed to reach absurdly high clock speeds, it is beginning to reach them.

What clock speed the chip was designed for has NOTHING(or very little) to do from the % of clockspeed gain due to a process shift, if the willamette topped out at 2ghz, and the northwood can now go to 3ghz(made up numbers) that is a 50% increase in max clockspeed, thus ANY chip going from the .18>.13 micron process would gain 50% clockspeed.(the fact the axp is designed to run slower is already accounted for in the simple fact that its top clockspeed is lower, so its 50% gain, in real clockspeed terms, is smaller than the northwoods.

Quote:
Maybe when AMD would start fazing out "old" AXPs, we would see some .13 Tbred that are very overclockable (much like early Durons) and rebadged with slower speeds, but I very much doubt an aircooled 100% overclock. Thats something we have NEVER seen from AMD. The most impressive overclocker from AMD I ever layed hands on was my good old Duron 600 with achieved just over 50%. I'll be VERY happy if my next Tbred does 50% overclock. Every other cpu from AMD I have had so far managed from 10% to about 40%.. aircooled. Thats good enough for me though.

A: the tbred is cheaper to make than the axp.
B: dresden is amds only fab, and they have promised a total .13micron conversion before the end of the year,
Add those two factors together and you are GUARENTEED to see low rated tbreds released.(1700+ would be my guess hence my example). These chips will be full speed tbreds marked at a lower speed to backfill, JUST like 1.6a northwoods.




"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 12:48:17 AM

Quote:
The Tbred will NOT see 100% overclocks. The P4 was meant to ramp in speed, and the 1.6A is actually a 2.0, don't forget. The Tbred should overclock better than the XP, but it won't be anywhere near 100%


see my post above as to why this is an incorrect statement, and the 1700+ tbred will also be actually a 2400+ or whatever is the top speed for the tbred on release.

The 100% thing is highly unlikely, but I never claimed a 100% overclock.(although, if they relased a 1500+ tbred it could do 100% IMO.

Quote:
Huh? That makes no sesnse.



Ill spell it out for you.

The average overclock of the 1.6a northwood is at betwen 2.2-2.4ghz NOT because of the cpu, but because of the ultra high fsb and its effects on the ram mostly, follow me? The tbred will not have this limitation due to the fact it can be unlocked.


Quote:
Tell you what, Mat. If you can get a 100% overclock on your Tbred (perfectly stable) with air cooling, I'll pay for it. If not...hmm...you buy me a Lian-Li case?
Feel like putting your money where your mouth is?



If they release a 1500+ tbred, I will take that bet.

PS: I never said either 100% overclock, or aircooling only btw, but I will still take that bet if they release the cpu.(assuming I have money to buy it when released due to unemployment lol).

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 12:51:12 AM

Quote:
Comming to think of it, he might have meant 100% compared to the previous generation when overclocked.. in that case, not even the P4 applies, since it went up to 2 Ghz on .18 stock frequency, and now 50% higher overclocked on .13; if you apply that logic to the athlon, which now tops out at 1.6 Ghz (right ?), you might expect 2.4 Ghz from Tbred. I'd be impressed, but you never know.


I was speaking on rated clockspeed overclocks, and the 1.47 would need to reach 2.93 which is possible, but would be on the extreme edge of the tbreds range.


What I am talking about, is simple semiconductor mechanics, the design of the core, is taken into account BEFORE the process shift is determined, what I mean is that, the p4 was designed for faster clockspeeds, and when they shift to a smaller process, they will gain MORE real clocks simply because 50% of 2000 is MORE than 50% of 1.66ghz!



"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 12:53:49 AM

Quote:
fact of the matter is we won't know how well it'll overclock until we actually start testing them with quality parts and various cooling methods.
I personally would be happy getting a 1733 MHz Tbred running 150 FSB (close to 2000 MHz) or maybe 166 FSB if I were lucky. You do have to take into account that the voltage is going to drop greatly, and you can ramp it right back up for more stability when you overclock. Should be interesting to see what these buggers can do, but only time will tell



This is entirely true, however semiconductors behave in a predictable way, and reasonable deductions can be made from our simple knowledge.

The tually gained 50-60% max clock due to the process shrink.

Thus the Axp should gain the same % max clock.


The northwood gained slightly more max clock, but it had other process enhancements over the willamette(namely copper interconnects IIRC.)

It is all pure numbers~!

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 12:56:56 AM

Quote:
When AMD's new thing drops on the market, make sure we get some results posted to this thread. I just wish I could see Matisaro's face when he realizes he is wrong about it scaling the same percentage in clockspeed as the 1.6A. We do have people getting a 100% overclock on the 1.6A with something like a Vapochill. So in order to verify his claim, we will need to see a 100% overclock for AMD. Just how long after introduction are we going to wait before declaring it a no-go anyway?



Yes, but not a 100% overclock of the fastest tbred, the key to a 100% overclock, is if amd chooses a 1600 or 1500+ to backfill the market with. All the statements I have made, are correct and follow semiconductor mechanics.

PS: There are more enhancements to the northwood core than just a process shrink, this muddys the pools of debate somewhat.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 12:58:13 AM

Quote:
Matisaro's face when he realizes he is wrong about it scaling the same percentage in clockspeed as the 1.6A.

I said they will scale the same % clockspeed in total maximum of the core, not any specific speed rating. If amd were to release a 1ghz tbred then huge ocs would result, thats kind of the same thing intel did with the 1.6a.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 1:00:05 AM

Quote:
What clock speed the chip was designed for has NOTHING(or very little) to do from the % of clockspeed gain due to a process shift

This is incorrect. The Pentium 4 core is designed to scale up to over 10GHz. The process shrink mostly just eliminates heat from high clockspeeds. The components are all capable of running at about 10GHz if it were not for heat.

The Athlon core was designed to run up to about 2GHz. Its components will run into timing issues beyond that which have nothing to do with the heat that is eliminated due to a process shrink. Remember, a core has clockspeed limitations that are not addressed by a process shrink.

As far as your percentages go, this is just bad math. If a processor from some company is capable of running at 10 times the speed at which it is introduced, it will be able to scale up much further due to a process shrink than one which is capable of running at only 1.5 times the speed at which it is introduced. The introduction speed has a great deal to do with your 'percentages' here. You are completely ignoring this denominator in your equation.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
March 6, 2002 1:02:13 AM

Quote:
I hope Matisaro is wrong. I like it when performance is close. If AMD gains the crown though it will be a plus since the Northwood could use some price drops and Intel needs someone to keep them on their toes.


The performance will still be close, not super revolutionary, remeber the NW is going to 133fsb in like may, However the rated speeds of the chips will remain in step, and overclocking will be done only by us enthusiasts.

The NW can scale just as fast as the tbred, if amd were to put alot of pressure on intel, they may have to spend too much resourses combating the NW, thus draining from their hammer launch.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 1:14:50 AM

The axp core was relayed out, when you do a process shrink of course you adjust your timings to compensate for shorter trace paths and any timing issues, this is simple logic. They dont just take masks and shrink them, they retest timing paths and change anything which causes problems, this is why .13 requires engineering wafers, to test the new core changes, not the .13 process itself.

The p3 core was introduced at what, 400mhz at .25 micron, the clockspeed gains of said core have ALL followed the same general pattern.

I will use the athlon as my example because I am more familiar with it.

Athlon was introduced at 500mhz on the .25 micron process. That cpu topped out at 750mhz,(50% overclock) then it was reduced to .18 micron. A substancial core redisgn took place at this time, putting the cache on chip. The tbird topped out at 1.5ghz(a 100% core gain over the slot a) This was due to A: the process shrink, B copper interconnects and C: ondie cache.) The axp core was a redesign, which gained 20% on the same process due to shifting of the core.



<A HREF="http://www.arstechnica.com/cpu/2q99/clock-intel-2.html" target="_new">http://www.arstechnica.com/cpu/2q99/clock-intel-2.html&...;/A>

Process not design is the limit of cpu speed.

<A HREF="http://www.howstuffworks.com/microprocessor1.htm" target="_new">http://www.howstuffworks.com/microprocessor1.htm&lt;/A>

The table is missing the .18 p3 and the .13 p3, but as you can see, the transistion from .35>.25 showed the same rate, and if we extrapolate the p3.18 speed from this chart, it follows the pattern.



"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 1:16:25 AM

Quote:
This is incorrect. The Pentium 4 core is designed to scale up to over 10GHz. The process shrink mostly just eliminates heat from high clockspeeds. The components are all capable of running at about 10GHz if it were not for heat.


This is incorrect, in the fact that a process shrink dosent just reduce heat, it changed timings and when you design a shrink you account for this, you dont just shrink the current masks, dosent work that way.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
March 6, 2002 1:41:38 AM

Let me make myself clear, so I don't come off on the wrong foot.

Quote:
<i>Fat Burger says:</i>
The Thoroughbred will overclock nicely and be AMD's best processor to date.


That said, I'm willing to bet that you will not get a 55% overclock on your Tbred (you can use any you want), within a week of it landing in your hands, using the retail heatsink, and no physical mods on the motherboard (voltage mods, etc.). You may use any RAM you want, the side of the case must be on, and your computer must be perfectly stable through everything I throw at it. You can also test my computer for stability if you wish.
Well?

<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
Don't step in the sarcasm!
March 6, 2002 2:03:05 AM

Any comment on the rumor that T-bred will have 1.65 volt vs 1.5 for the northwood.

Maybe is was true after all that PAL allready use some 0.13 micron

cheap, cheap. Think cheap, and you'll always be cheap.AMD version of semi conducteur industrie
March 6, 2002 2:25:15 AM

Sshhht, I sense another debate which will fill my mind with stuff before sleeping, I'll be dreaming about process shrinks eating me!
Pass me some popcorn Ray, not for your bet, but for the debate that's coming!

--
For the first time, Hookers are hooked on Phonics!!
March 6, 2002 2:30:25 AM

Quote:
The axp core was relayed out, when you do a process shrink of course you adjust your timings to compensate for shorter trace paths and any timing issues, this is simple logic. They dont just take masks and shrink them

I did not mean that the only thing done was a shrink. However, the basic core design remains the same. The components get rearranged a bit, but the execution pipelines and major defining parts of the core are not modified. A few bells and whistles are added, heat is removed, and a few things are moved around to try to get a bit more performance from the core at that particular die size.


Quote:
they retest timing paths and change anything which causes problems

You cannot fix everything with a shrink. There are multiple types of timing problems. For example, clock jitter is a serious problem as frequencies increase. Making the components smaller does not by itself reduce jitter. Problems such as these must be addressed in the initial design of the core. Major problems and limitations cannot be addressed by a simple die shrink. This would require a completely new core design, with an execution pipeline designed for high clockspeeds.


Quote:
Process not design is the limit of cpu speed.

In theory this might be true. In practice this is incorrect. Process helps speed up a design. The design itself has firm limits based on its specifications. Some components have real limits to them that shrinking will not change. When the limit of these components is reached, you have reached the limit of your design. You will then need to redesign the core to allow these components more time while accelerating the rest of the components. This is what a redesign of the execution pipeline does. This is what occurred during the transition from the Pentium III to the Pentium 4.

-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
    • 1 / 6
    • 2
    • 3
    • 4
    • 5
    • More pages
    • Next
    • Newest
!