ClawHammer: Everything Goes As Scheduled - Page 2 - CPUs

Last message on previous page:
you don't use it if you don't need it ...

if you know you don't know, the way could be more easy.

Reply to labdog

Register or log in to remove.

as far as I know - this is not the case.
however I posted the quistion on AnandTech HIGHLY TECHNICAL forum for further confermtion... those guys know what thier talking about (not that I say you dont) as some of them actually work at this buisnes ...

This post is best viewed with common sense enabled

Reply to IIB

2.33 for 3.4+, i think Hope make Life

if you know you don't know, the way could be more easy.

Reply to labdog

ALU for basic integer instructions not for float, high calculations & others...

so:

Business Apps don't care about high end newest ALU technologies.

but maybe you intergrate eg 3D Studio Max as an Business App ?

just a little note
my post is string related

if you know you don't know, the way could be more easy.

Reply to labdog

Have You lost Your mind, or are You just trying to master things You know nothing about.

Just curious.

Reply to Era

oh yes, i have lost my mind.

sorry if you don't understand nothing about.

if you know you don't know, the way could be more easy.

Reply to labdog

Jerk..

Reply to Era

the ALU is the most basic of a processor execution unit.
it deals with basice logic and Integer arithmetecs.
ANY progrem who uses alot of basic Integer caulcaltions should run faster on a processor with Faster and More ALU and AGU units...

though a processor could have bootlenecks preventing this - like a larger pipline (without a very good BPU) or small cache...

This post is best viewed with common sense enabled

Reply to IIB

great.

thanks for the lesson

if you know you don't know, the way could be more easy.

Reply to labdog

well... heres an answer I got over the anandtech Highly technical forum:

Quote :

does processors gain the same scalbilty (in %) on the same process shrink? >>

No. There are two main sources of delay in an MPU: gate (transistor) delay and interconnect (wire) delay. While in an ideal process shrink gate delay reduces by the ratio of the Leffs (channel length), wire delay stays the same (ignoring technology improvements, such as copper interconnects). In addition, since MPUs stay roughly the same size over time due increased cache size and microarchitectural additions, absolute wire delay gets worse over time compared to gate delay.

The critical path delays of a given microarchitecture can be more or less interconnect delay driven than another microarchitecture. For an empirical example, just look at the Athlon and P3...they both started at or around 750MHz at .18um. While the P3 hit a wall at 1GHz, the Athlon is at 1.67GHz...though admittedly the current Athlons use copper interconnects, so use the 1.3GHz (aluminum) Duron as a comparison. Its not uncommon for a given microarchitecture to be designed to have a "sweet spot," with the knowledge that its demand will be the highest along one or two particular process shrinks.

if its not uncommon for a processor to have a "sweet spot" then it is posible that P4 will scale better with die shrinks...

you can find the thread I opend
<A HREF="http://forums.anandtech.com/messageview.cfm?catid=50&threadid=716359" target="_new">here</A>

This post is best viewed with common sense enabled

Reply to IIB

dude i couldnt agree with you more !!
i feel exactly the same way but there is no point in raving on about it as it will get you nowhere.
the majority of diehard AMD and Intel fans has such a closed mind its ludicrous. I dont see why ppl cant be open to all products and acknowledge the gd and bad points of each. AMD fans always go on about how amazing the XP is cus it competes with p4nw cuse of the p4 pipeline. At the end of the day there both pretty freakin similar in performance !

Trying is the first step to failure

Reply to Grifter

Thanx!

maybe your right and im fighting a lossing battle...

This post is best viewed with common sense enabled

Reply to IIB

Quote :

1) Say what? Slot (Classic) Athlon started at...can't remember. Tbird started at...750? So that's not quite double.

Classic Athlon started at 500MHz, and Thunderbird was at 700MHz (I used to have 2 of them and they were unlocked), so it is double.

:smile: Good or Bad have no meaning at all, depends on what your point of view is.

Reply to khha4113

i love flipping burgers, don't you ?

if you know you don't know, the way could be more easy.

Reply to labdog

Man, discussions are what make this forum go on. I know it is tiring, but don't you also learn new stuff?
Heck I just learned quite about die shrinks and ALUs here! I love debates, it makes the mind work to find different alternatives...so it is not a battle, it's a mind's work.

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...

Reply to eden

One thing to point out regarding the respective shifts down to .13 is that P4 had some significant enhancements made to the core, including increased cache. So, the scalability/performance may be quite different.

mark-

When all else fails, throw your computer out the window!!!

Reply to zengeos

BTW concerning the thread you linked us to...
Are they saying that despite the extreme heat and IPC the Athlon had, that at 0.18m it was able to go to 1.67GHZ, when as a P3 only got at least 400MHZ before literally being "called back"?
If so, then how does it explain the P3 wasn't as heat eater and such high performance as the Tbirds?
This does open some speculations about the Northwoods' future speed grades, but we've already seen so many OC articles so it shows it can reach speeds far beyong..

Another thing I wanted to discuss...is the fact that each and every component in a CPU is crucial to its performance... What I mean is, even though the MHZ is no longer the importance and that AMD is right about proving so, it does not mean we should ignore it...
Why? Because since each component is crucial, denying it is denying the CPU potential and the science of building CPUs. When I say that, I mean it. AMD is using IPC now, it's great but they have got to exploit the remaining component, the MHZ. They have the processor power but they need the processor potential! Intel is using MHZ and is not exploiting most of the interal CPU, therefore have a processor potential with barely any processor power.
So give this comment what you want, but if you don't beleive in using the most of every component a CPU offers, then you have no interest in Science, or making chips, and are denying the potential. Right now I can see AMD doing a lot to regain MHZ to get their CPUs further, but I see not many plans about Intel trying to exploit IPC so far... but then again they have their own secrets.

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...

Reply to eden

I tend to agree with you Eden.

Performance is a combination of IPC and MHZ. If you are weak in one or the other then the overall chip performance likely will suffer and thus system performance suffers.

Intel said from day one that P4 was designed to improve performance with speed increases. So, P4 performance might conceivably improve as clockspeed increases.

AMD otoh, is in a situation where Athlon is really reaching performance peak. Even with a .13 shrink AND SOI, I don't see Athlon increasing much beyond 2 to 2.2Ghz in real terms. Taken in perspective, that's a 4x clockspeed increase from the initial .25 Athlon and the Barton. That's certainly nothing to sneeze at.

P4 otoh, will likely clock up to 3-3.5ghz with Northwood, and to 5ghz and beyond with the next iteration in 2003 going into 2004. Taken in perspective, the increase is similar to what Athlon has enjoyed in a similar timeframe...4x.

Hammer wil add some nice twists to all this I suspect. I'll be interested to see how it does at Cebit in March (If it makes it's debut there, which many folks suspect)

Mark-

When all else fails, throw your computer out the window!!!

Reply to zengeos

Heh yeah...can't wait!
Yes the P4 will reach 5GHZ, granted that's a lot but that's also how 10GHZ will feel, it may be alot but if it's the P4 that does it, I wonder if it's that performing... At 5GHZ they musta reached 0.09m for sure, and now I think they must focus on IPC to make the potential of the CPU unlocked like I stated. The P4 has the speeds to reach, now they hafta give it the power. It's like a cargo truck. At first with the P4, the truck is rather empty or carries not a lot even though it reaches fast speeds unlike others. Since it has that potential, now it's time to load it up and prove its efficiency in carrying so much with such speed!
The AMD however is the other way, a slow but very efficient carrier. They both need to expand their capabilities, so don't think for a sec I am pro-AMD here. I may favor them for now, but I think a lot of us do for what they gave us with the AXP.

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...

Reply to eden

But the Tbred itself will make the XP go to at least 2.5GHZ IMO, if that's what 0.18m made the Tbird to Palomino reach. Add SOI, which is something that should be an excellent alternative than wasting huge pipelines and have potential of mispredictions and delays, the XP stands to get to 3GHZ if AMD feels the power to do so. But the Hammer's PR will definitly set a standard, if it's true that it will jump from 3400 to 4400 so suddenly (I am suspecting it might be false). However I am skeptical about how AMD will release the Hammer and when. I mean, if they do get the ClawHammer out end of this year, I don't see Intel reaching 3.4 GHZ that fast, so if it does go AMD's way, might cause a lot of trouble for Intel and the OEMs. Who knows?
I am eager to see how each company will do...this year is bound to be one huge milestone for technology of CPUs and we are there to help it and witness it!
Likewise for the graphics industry!

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...

Reply to eden

Well...I think Athlon will reach a PR 3000, but I have some doubts it will actually hit 3ghz clock speed.

As for Clawhammmer.. the 3400 and 4400 are slated 3 quarters apart .09 version is slated for H2 03) (Claw .13 micron version is slated for Q4 this year and the AND the 4400 is slated for .09 micron. I don't see any problems with AMD gaining that jump in performance with the .09 shrink. I suspect we'll see a 36 or 3800 in Q1 03 and a 4000 or 4200 in Q2 03. This is, of course sensible speculation.

As for this causing problems for Intel....You may be surprised. I seriously expect to see Intel pump up the P4 speed rating as soon as Hammer starts coming out the door. In fact I wouldn't be surprised to hear that Intel is stockpiling higher speed P4's for sale later this year. While their roadmap may only show around 2.8 or 3ghz by year end, I doubt they will sit still and let AMD fire such a major shot and not have a return barrage. I wouldn't even be surprised if Intel doesn't release a 3.6ghz chip in response to Hammer. They don't have to have large supplies...just the press release of availability soon will put a major crimp in AMD's *glory*. Heck look at how difficult it was/is to get the higher speed grade P4's now.

Mark-

When all else fails, throw your computer out the window!!!

Reply to zengeos

Quote :

the Athlon is at 1.67GHz...though admittedly the current Athlons use copper interconnects, so use the 1.3GHz (aluminum) Duron as a comparison. Its not uncommon for a given microarchitecture to be designed to have a "sweet spot," with the knowledge that its demand will be the highest along one or two particular process shrinks.

This comparison is flawwed, firstly I stated that the archetectures could not change(thus opening a new max performance for the core).

Second, I said that the %gain from the top clockspeed is what matters, the p3 compared to the athlon at .18micon is not even a factor. The p3 topping out at 1ghz, and the athlon making it to 1.67ghz on .18 is a measure of their layout, not their tranistor quality.

Also, the copper interconnects totally negate any comparisons.

What I am saying is that, 2 chips, will gain the same % in clockspeed(based on their origional max) from a process shrink, as it is nothing more than a basic structural change.

However, I will grant that you can optimise designes for certain process sizes to accounr for wire delays etc.

And to the other point, yes, the northwood had things added, so the axp may or may not gain as much % from the die shrink.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

Reply to Matisaro

Quote :

dude i couldnt agree with you more !!
i feel exactly the same way but there is no point in raving on about it as it will get you nowhere.
the majority of diehard AMD and Intel fans has such a closed mind its ludicrous. I dont see why ppl cant be open to all products and acknowledge the gd and bad points of each. AMD fans always go on about how amazing the XP is cus it competes with p4nw cuse of the p4 pipeline. At the end of the day there both pretty freakin similar in performance !

Whos talking about amd and intel, Im talking about .18 and .13 micron fab process.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

Reply to Matisaro

I was confiming my support for this statement
"when you will face that:

High IPC low Clockspeed
&
High ClockSpeed low IPC

are just as good if they blance out
"

get it ?

Trying is the first step to failure

Reply to Grifter

Yes, I suppose, however I am not debating amd/intel here, this is pure semiconductor mechanics.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

Reply to Matisaro

Quote :

it is not a battle, it's a mind's work.

yep.

if you know you don't know, the way could be more easy.

Reply to labdog

Quote :

Another thing I wanted to discuss...is the fact that each and every component in a CPU is crucial to its performance... What I mean is, even though the MHZ is no longer the importance and that AMD is right about proving so, it does not mean we should ignore it...
Why? Because since each component is crucial, denying it is denying the CPU potential and the science of building CPUs. When I say that, I mean it. AMD is using IPC now, it's great but they have got to exploit the remaining component, the MHZ.

this is not true... a prcessor scales linerly as you Increase its Clock speed. it goes like this:

IPC X MHz = Preformance.

if you Increase the clock speed by 20% the preformance will go up by 20%. this is not hard to belive.
lets take P4s long piplne - 20% Increasment in clock speed means that he will miss-predect 20% times more - but it also means it will predect right through out the pipline 20% times more...
this is true for its FPU unit - it will work 20% faster...
the Athlons FADD and FMUL (hes 2 FPU units) will each work 20% faster too... the hole execution architecture will do 20% more instractions - as there is no change in the IPC the instaction it does per cycle.
the liner scaling could only be seen by theortical benchmarks - which runs iner loops of instractions without streesing other system components.

given that knowledge we can say that a theortical 3.33Ghz athlon XP will preforme like a 4GHz nothwood... who will get there faster? if P4 does scale better on 0.13 - then he will be pretty damn close...

also take in to acount - that the FSB might start be a limiting factor as Athlon reaches clockspeeds in which it will be hard for him to fetch instructions fast enougth from the Memory. we already seen that at very high clock-speeds RDRAM+400MHz BUS take first to an even higher clocked DDRRAM which can only use 266MHz of the BUS bandwith...

I guess at those clockspeeds "Intel-Killer 3400+" will come into play with its dual-channel DDR-SDRAM and esstemated ~800MHz bus...

This post is best viewed with common sense enabled

Reply to IIB

Are you saying Hammer is an 800MHZ Bus and Dual-Channeled?
I highly doubt that, as a lot think it will be at 333MHZ. Although I did hear once about some hyper CPU at 800MHZ bus, not from Intel though...

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...

Reply to eden

ClawHammer: Everything Goes As Scheduled - Page 2

Forum CPU & Components : CPUs - ClawHammer: Everything Goes As Scheduled

Please mind