Why can't AMD make 3.8GHz processors?

[Fraggable]

I was wondering, If Intel can make (single-core) 3.8GHz Prescott processors using a 90nm process, why do AMD processors top out at 2.8GHz (FX-57) if they're 90nm too? Everyone knows that Prescotts run way hotter than AMD 64's have ever thought about running so heat shouldn't be a problem. I also know that an FX57 will outrun a 3.8GHz Prescott in most applications but still, why can't they make them any faster?

Maybe it's a stupid question and you'd have to know more about their architecture to understand it, but it doesn't make sense to me.

 

[jtworldwide3]

you'd have to know more about their architecture to understand it

Think of it this way, an AMD cpu is like an 8 lane highway at 55 MPH and an Intel CUP is like a 4 lane at 70. They get the same amount of cars thought but at different speeds.

 

[biostud]

oh noes, the most asked question on CPU's ever once again shows its ugly head.

some simple searches on these forums and on the interweb would give you answers.

 

[LTC8K6]

Intel got the higher clocks by making their pipeline longer, but this also decreased performance. At first, the higher clocks bested the Athlons, but as AMD got their speeds a little higher, Intel's pipeline penalty began to show more and more. Amd's FPU also got better.

 

[MrX8503]

proccessors have been stuck around 2-3ghz for a reason. Its better to make a processor more efficient rather than make it 3-4ghz.

 

[StrangerGuy]

Think about which is more impressive at the same 0.09u process:

12-stage pipeline @ 2.8GHz, runs warm, or
30+ stage pipeline @ 3.8GHz, runs oven-hot

The former is, right?

 

[Fraggable]

Ok, I understand what the reason is now. Intel chose to raise their clocks without changing the pipeling structure, while AMD shortened the pipeline and made their CPUs smarter.

I should have known that, but for some reason it didn't occur to me.

Thanks for the answers!

 

[biostud]

Originally posted by: Fraggable
Ok, I understand what the reason is now. Intel chose to raise their clocks without changing the pipeling structure, while AMD shortened the pipeline and made their CPUs smarter.

I should have known that, but for some reason it didn't occur to me.

Thanks for the answers!

actually it's more intel that lengthened their pipeline going from p3 to p4 in order to be able to reach higher clocks while AMD kept their short pipeline

 

[Shimmishim]

you know... it sounds dumb of intel to do that... to increase the pipeline to get higher speeds...

but in the consumer market (normal buyers who don't visit forums like anandtech) don't know any better...

they have no concept of IPC or pipelines...

and we've always been taught.... "faster is better"

so if you were a consumer with no knowledge of processors and the "faster is better" mentality.... which would you buy... a 2.0 ghz AMD processor or a 3.0 intel processor assuming they were the same price?

edited for "same price" comment

 

[Fraggable]

Originally posted by: Shimmishim
you know... it sounds dumb of intel to do that... to increase the pipeline to get higher speeds...

but in the consumer market (normal buyers who don't visit forums like anandtech) don't know any better...

they have no concept of IPC or pipelines...

and we've always been taught.... "faster is better"

so if you were a consumer with no knowledge of processors and the "faster is better" mentality.... which would you buy... a 2.0 ghz AMD processor or a 3.0 intel processor assuming they were the same price?

edited for "same price" comment

Which is why Intel is changing their processor model numbers to 3XX, 5XX, 7XX, 9XX instead of just listing clock speeds. They're trying to prime the market for when they're forced to lower clocks and change the pipeline structure.

 

[pulsedrive]

Originally posted by: Fraggable

Originally posted by: Shimmishim

Which is why Intel is changing their processor model numbers to 3XX, 5XX, 7XX, 9XX instead of just listing clock speeds. They're trying to prime the market for when they're forced to lower clocks and change the pipeline structure.

They have already changed things, that is where the Pentium M comes into play.

 

[Shimmishim]

Originally posted by: Fraggable

Originally posted by: Shimmishim
you know... it sounds dumb of intel to do that... to increase the pipeline to get higher speeds...

but in the consumer market (normal buyers who don't visit forums like anandtech) don't know any better...

they have no concept of IPC or pipelines...

and we've always been taught.... "faster is better"

so if you were a consumer with no knowledge of processors and the "faster is better" mentality.... which would you buy... a 2.0 ghz AMD processor or a 3.0 intel processor assuming they were the same price?

edited for "same price" comment

Which is why Intel is changing their processor model numbers to 3XX, 5XX, 7XX, 9XX instead of just listing clock speeds. They're trying to prime the market for when they're forced to lower clocks and change the pipeline structure.

yes... that is very true!

as much as their processors (with netburst and super long pipelines) suck... they are really clever about their marketing scheme....

 

[Shimmishim]

Originally posted by: pulsedrive

Originally posted by: Fraggable

Originally posted by: Shimmishim

Which is why Intel is changing their processor model numbers to 3XX, 5XX, 7XX, 9XX instead of just listing clock speeds. They're trying to prime the market for when they're forced to lower clocks and change the pipeline structure.

They have already changed things, that is where the Pentium M comes into play.

that is true as well...

the pentium M has been hugely successful in the mobile world...

but i think the mindset for desktop consumers and laptop consumers is different.

with laptops, you want long battery life, cool running, yet still fast.... at least in my opinion so speed isn't such a big factor with these guys....

but what do i know... i'm a desktop consumer!

 

[aigomorla]

Originally posted by: StrangerGuy
Think about which is more impressive at the same 0.09u process:

12-stage pipeline @ 2.8GHz, runs warm, or
30+ stage pipeline @ 3.8GHz, runs oven-hot

The former is, right?

u forgot to mention, how the top will use 1.37V while the second one will most probably need double if not tripple the voltage hence making all the electricity goto waste

 

[Acanthus]

Originally posted by: aigomorla

Originally posted by: StrangerGuy
Think about which is more impressive at the same 0.09u process:

12-stage pipeline @ 2.8GHz, runs warm, or
30+ stage pipeline @ 3.8GHz, runs oven-hot

The former is, right?

u forgot to mention, how the top will use 1.37V while the second one will most probably need double if not tripple the voltage hence making all the electricity goto waste

So pentium 4s run at 4v now? :laugh:

 

[Acanthus]

There are a lot of reasons that AMD has the lead they do right now...

1. On die memory controller, dramatically reduces latency, as the memory controller is on the actual CPU die, instead of the the motherboard north bridge.

2. HT bus architecture, this distributed serial bus architecture is more efficient than intels, and lanes can be added for multiprocessor motherboards, giving a TREMENDOUS advantage in 4 - 8 and higher processor systems.

3. SOI, AMD and IBM developed Silicon On Insulator quite a while ago, this reduction in leakage and power draw has been a small advantage for AMD for quite a while. Intel is supposed to move to SOI with a new 65m, or 40nm process. They have already selected a substrate and are experiementing with it.

4. Shorter Pipeline, their architecture was based on a shorter pipeline that relied on efficiency over clockspeed. Intel went the opposite and increased the pipeline to lean on clockspeed for performance. Intel expected netburst to scale to at least 6ghz by now, but they hit a clockspeed wall that wasnt expected by anyone in the industry. This has been intels major fumble thats going to take time to recover.

Edit: Nehalem, the 40nm variant of the netburst architecture, was supposed to scale to 10ghz by 2007. Had that scenario materialized, Intel would have a scarily commanding lead in the industry.

 

[jazzboy]

Originally posted by: Acanthus
Edit: Tejas, the 65nm variant of the netburst architecture, was supposed to scale to 10ghz by 2007. Had that scenario materialized, Intel would have a scarily commanding lead in the industry.

I thought Tejas was only supposed to go to 7ghz and Nehalem (think thats what it's called) was supposed to be the 10ghz one.

Anyway, I guess thats all in the past now. I wonder if 10ghz cpus will ever be made in our lifetimes.

 

[Acanthus]

Originally posted by: jazzboy

Originally posted by: Acanthus
Edit: Tejas, the 65nm variant of the netburst architecture, was supposed to scale to 10ghz by 2007. Had that scenario materialized, Intel would have a scarily commanding lead in the industry.

I thought Tejas was only supposed to go to 7ghz and Nehalem (think thats what it's called) was supposed to be the 10ghz one.

Anyway, I guess thats all in the past now. I wonder if 10ghz cpus will ever be made in our lifetimes.

Youre correct, ill edit, it was so long ago the roadmap actually said that

 

[aigomorla]

Originally posted by: jazzboy

Originally posted by: Acanthus
Edit: Tejas, the 65nm variant of the netburst architecture, was supposed to scale to 10ghz by 2007. Had that scenario materialized, Intel would have a scarily commanding lead in the industry.

I thought Tejas was only supposed to go to 7ghz and Nehalem (think thats what it's called) was supposed to be the 10ghz one.

Anyway, I guess thats all in the past now. I wonder if 10ghz cpus will ever be made in our lifetimes.

they'll probably bring about another system in determining clock speed. And if ur wondering, it hasnt been more then 28 yrs since the first 8086 and this was clocked at a mind whooping 5, 6,, 8, and 10 MHz versions! Even 10 yrs ago, the fastest chips were 233 MMX P1's and now we see dual cores and systems with over 4ghz in clock freq.

So if u really think about it, 10ghz, or its equavalient is very possible in our lifetimes.

 

[markkleb]

I remember my VIC-20. It was 1mhz, not 1k mhz or 3000 mhz but 1mhz. Amazing how stuff has changed.

 

[Continuity28]

Originally posted by: aigomorla

Originally posted by: jazzboy

Originally posted by: Acanthus
Edit: Tejas, the 65nm variant of the netburst architecture, was supposed to scale to 10ghz by 2007. Had that scenario materialized, Intel would have a scarily commanding lead in the industry.

I thought Tejas was only supposed to go to 7ghz and Nehalem (think thats what it's called) was supposed to be the 10ghz one.

Anyway, I guess thats all in the past now. I wonder if 10ghz cpus will ever be made in our lifetimes.

they'll probably bring about another system in determining clock speed. And if ur wondering, it hasnt been more then 28 yrs since the first 8086 and this was clocked at a mind whooping 5, 6,, 8, and 10 MHz versions! Even 10 yrs ago, the fastest chips were 233 MMX P1's and now we see dual cores and systems with over 4ghz in clock freq.

So if u really think about it, 10ghz, or its equavalient is very possible in our lifetimes.

We don't want 10ghz. We want efficiency and shorter pipelines until the fabrication process allows 10ghz naturally - that certainly won't happen any time soon. Getting to 10ghz is not hard, if you like having a multiple gigawatt processor with a heatsink the size of a room. Aim for efficiency and scale speed as you go along, it's better in every way.

 

[dmens]

We want efficiency and shorter pipelines until the fabrication process allows 10ghz naturally

LOL, I don't know who "we" are, but intel, amd, ibm and everyone certainly don't want shorter pipelines and "efficiency", whatever the hell that means. Are you implying that long pipelines are inefficient? OK, maybe if the machine nuked and/or the frontend mispredicted every 50 cycles.... oh wait, that doesn't happen.

Throughput (and performance) is achieved by width, frequency (longer pipelines... gasp!) and smart speculation, not short pipelines and stalling for every flow uncertainty in the pursuit for "efficient" execution.

 

[richardrds]

Fraggable,

Just want to make sure you are aware of the " x 1.6 " rule.

You multiply a A64 by 1.6 to find out the approximate P4 equivelant.

So a 2.4Ghz A64 (like a Venice core 3800+) is about the same performance as a P4 at 3.8Ghz (2.4Ghz x 1.6 = 3.8Ghz). That is why AMD calls thier 2.4Ghz A64 Venice core a 3800+, the 3800+ means a little better then 3800Mhz (i.e. 3.8Ghz).

The x 1.6 rule is the rule you use for S939 A64 chips with 512M of L2 cache.

If the cpu has 1G L2 cache modify the rule to " ( x 1.6) +200Mhz "

 

[dexvx]

It's amusing to read posts by enthusiasts who have no idea about CPU archetiture try and be the arm-chair engineer in these kinds of debates.

Originally posted by: Shimmishim
but in the consumer market (normal buyers who don't visit forums like anandtech) don't know any better...

they have no concept of IPC or pipelines...

and we've always been taught.... "faster is better"

so if you were a consumer with no knowledge of processors and the "faster is better" mentality.... which would you buy... a 2.0 ghz AMD processor or a 3.0 intel processor assuming they were the same price?

You obviously don't either. I see a lot of misinformed people who actually believe Sempron64 3000+ (1.8Ghz) can actually top a 1.5Ghz Celeron-M/1MB. Guess what, the Sempron is 10 stage and the Dothan based Celeron-M is 14. Shocker.

Originally posted by: aigomorla

Originally posted by: StrangerGuy
Think about which is more impressive at the same 0.09u process:

12-stage pipeline @ 2.8GHz, runs warm, or
30+ stage pipeline @ 3.8GHz, runs oven-hot

The former is, right?

u forgot to mention, how the top will use 1.37V while the second one will most probably need double if not tripple the voltage hence making all the electricity goto waste

Laughable. Double/Tripe Voltage? Where do you come up with this stuff? You seem to think theres a direct correlation between pipeline length and processor heat. Which there is not. We also seem to be forgetting that even at 3Ghz, the A64's will draw around 100W also. The 12 Stage Itanium2's draw just as much power as the "hottest" single core Prescotts. The IBM Power4's draw around 50% more.

 

[Furen]

Originally posted by: dexvxYou obviously don't either. I see a lot of misinformed people who actually believe Sempron64 3000+ (1.8Ghz) can actually top a 1.5Ghz Celeron-M/1MB. Guess what, the Sempron is 10 stage and the Dothan based Celeron-M is 14. Shocker.

LOL, A Sempron 2800+ can easily crush a Celeron M in most tasks, excepting gaming (where the cache makes too much of a difference). The Sempron has a 12 stage pipeline, by the way, the original K7 is the one that had a 10-stage pipeline.

Originally posted by: dexvx
Laughable. Double/Tripe Voltage? Where do you come up with this stuff? You seem to think theres a direct correlation between pipeline length and processor heat. Which there is not. We also seem to be forgetting that even at 3Ghz, the A64's will draw around 100W also. The 12 Stage Itanium2's draw just as much power as the "hottest" single core Prescotts. The IBM Power4's draw around 50% more.

I think that he said double/triple voltage but he meant double/triple current draw. This is somewhat accurate but it is very indirectly related to the pipeline length. A longer pipeline requires more logic than a comparable shorter pipeline and, in this respect, a longer pipeline leads to higher power draw (Example, the pipeline in prescott requires more logic transistors than the pipeline in northwood but you cant just compare this to the pipeline in the K8 or P6 since the arrangement is very different). Another thing is that a fast-switching transistor should require thinner gate-oxide channels, which makes the dielectric's insulating ability diminish greatly, which leads to higher current leakage. The problem with Prescott was that it was huge (for a 90nm CPU) and it operated at high clock speeds that required fast-switching transistors.

Itanium2 is just massive because of its width, though its clock speed is low (and the first Itanium 2 was a 180nm chip, which makes its transistors close to 4x bigger than Prescotts, which, by itself makes them draw more power). Power4 is also pretty wide and its clock speeds are a bit higher than Itanium2's (not to mention that it also uses a 180nm process). By the way, a K8 at 3GHz may draw about 100W but this is because its wasn't BUILT for those clocks but, as a comparison, a dual-core FX-60 at 2.6GHz draws much less than 100W (something like 70-80W).