AMD @ 5ghz - when? (unlcocked)

S

sacon

n00b
Joined
Jun 25, 2004
Messages
13
I promised to buy my self a brand new PC when AMD had hit 5ghz (unclocked)

I have had a 2ghz AMD for two years.

I thought I would have stashed some cash for it by now, but I also thought that moores law would have seen AMD getting closer to 5ghz.

How long have I got left to save???
 
Aiming for a raw mhz number is fairly short sighted IMO.

That being said, keep holding your breath.
 
quite a while as we're only on 2.6ghz, and focus is moving from clock speed and pure performance with one core to TLP, with performance from multible cores at a slower speed. we may not even hit 5ghz for 10 years :p

also, moore's law has nothing to do with speed. just transistor density ;)
 
a long, long, long time.

dual core is the way amd and intel are both going. I don't expect to see anything like a 5ghz cpu for quite some time.
 
Short sited, no. Just a goal I set. I have been building and rebuilding PC for 10 years costing who knows what. I thought that a 5ghz goal (2 years back) would give me a 2/3 year break from constantly spending money.

10 years to hit 5ghz, realy !!
 
(cf)Eclipse said:
also, moore's law has nothing to do with speed. just transistor density ;)
Click to expand...

not *exactly* right, as the law is referring to the *number* of components.

From Moores original paper:

"The complexity for minimum component costs has increased
at a rate of roughly a factor of two per year (see
graph on next page)."
 
eh, close enough :p

and we just did have a breakthrough, unless you've been living under a rock when X2 and pentiumD were launched quite recently :D
 
Not for a while. Artificial diamonds is something theyr'e thinkin about currently.

Quantum computing is another.

I wonder how much smaller they can get with silicone wafers ..
 
Looking at the quad core rumors at the inq and stuff.... quite possibly never. I am a bit disappointed in that it seems single core progress is a complete and total dead - we dont even want to try it anymore - duck. Yep, single cores significantically better than today's aren't COMING OUT! We will probably get an FX59 only!

I mean to avoid nitpicking, might be some advances through just SHEER TIME. A significantly better single core than today probably wont come until time and technology has moved so far that a budget single core based off the latest uber-four-core would be faster than today... but then you'll want the uber-four-core. You're better off hoping the programmers get things in-line so that you wont care. A core that can get to 180% of the single-threaded performance of today's FX-55, IS NOT coming in anything that wouldnt be considered an ETERNITY in comp terms. Look, they don't even want to try! No one would have thought last year that 2005 could come and go without so much as 2 or 3 speed bumps in single cores. I wouldnt hold my breath on a 3.0GHz... never mind 5.
 
mohammedtaha said:
Not for a while. Artificial diamonds is something theyr'e thinkin about currently.

Quantum computing is another.

I wonder how much smaller they can get with silicone wafers ..
Click to expand...

yes, how small, seems that 5ghz would be on or too far for this technology

a new fabriaction process must be being invented in some dark lab somewhere
 
Well the Cell processors aren't making it any easier ..

God damm silicone...

Would be funny if Silicon breasts had computer chips in them .. make use of all that silicone ..
 
well, amd and intel are both working on 65nm
i dunno if we'll see much below 30nm, as it's.. just too small
 
Silicon is used in computer chips. Silicone (with an e) is used in fake boobs.
 
Conker said:
Amd didn't hit 5ghz yet but toms hardware guide showed them overclocking a northwood p4 to 5.25ghz with liquid nitrogen. Theres a video with them doing it. :)

http://www20.tomshardware.com/cpu/20031230/index.html

check out the video
http://www20.tomshardware.com/cpu/20031230/5ghz-03.html

I am a AMD fan but back then p4 had those extreme overclocks. :)
Click to expand...

it only seems like a few years ago that someone had overclocked an intel to 1ghz before a 1ghz had ever been sold

AMD have focused on chip design as compared to intel that have been focused on ghz, which is why Intel win the ghz war and AMD win the clock for clock war
 
mohammedtaha said:
well ... 2.6 + 2.6 = ?

:p .. If Intel doesn't hit 5Ghz .. AMD won't .. That's my law ..
Click to expand...

Actually, I believe that AMD will at some point be a more mainstream name than intel is. No more little blue guys flying around.
:rolleyes:
 
rhashem said:
Silicon is used in computer chips. Silicone (with an e) is used in fake boobs.
Click to expand...


Actually I'm pretty sure they do not use it in fake boobs anymore; because of the danger of them leaking and poisoning the individual. I think all they use now is saline.
 
mohammedtaha said:
I wonder how much smaller they can get with silicone wafers ..
Click to expand...

presently I think the smallest logic node is expected to be 22nm, but that presumes some fairly signifigant changes in the way transistors are made, including high-k metal gates and halo-doping. And that's not to say things like the delay in metal wires (Copper or AL) won't make such a process prohibitively difficult.
 
Maybe we are looking at the question the wrong way. I would say that we are currently at 4.8Ghz 2x2.4Ghz for the X2 4800+

I know that doesn't mean twice the performance, but neither would a 5Ghz AMD (over a 2.5 Ghz part).


So, when the X2 5000+ comes out, the OPs goal's should be met, by my way of looking at it.


 
rhashem said:
Silicon is used in computer chips. Silicone (with an e) is used in fake boobs.
Click to expand...

And both provide hours of entertainment for teenagers.
 
You can't look at clock frequency as frequency per core times number of cores. That's incorrect on many levels. At the fabrication process level, it's currently impossible for an unmodified AMD K8 processor to hit 5 GHz. At the chip logic level, that's incorrect because both cores are synchoronized on a common clock generator running at the nominal "single-chip" frequency. From a performance standpoint, that's incorrect because performance is entirely dependent upon the software's ability to use both cores.
 
jordan12 said:
Actually, I believe that AMD will at some point be a more mainstream name than intel is. No more little blue guys flying around.
:rolleyes:
Click to expand...

Intel has been the the mainstream provideer of X86 CPU's for more than 20 years now. I doubt that this will change anytime soon.
 
sacon said:
I promised to buy my self a brand new PC when AMD had hit 5ghz (unclocked)

I have had a 2ghz AMD for two years.

I thought I would have stashed some cash for it by now, but I also thought that moores law would have seen AMD getting closer to 5ghz.

How long have I got left to save???
Click to expand...

Forever and ever.

Also listed on the backburners of both AMD and Intel's future are Hex chips. 6 cores per wafer. Its not so much of how much speed do you need anymore, its how much info can you pump into a program at a given time. Right now multi-core processors are simply the way both companies are heading because its meets that need. Mhz increasing has pretty much hit a wall. Intel with yields, AMD with its pipe. Just FYI the first 6nmThey can only go so much further and that future is visual now. The time of the single core is pretty much done with. I think the best you can hope for is 3gig from AMD and 4.2gig from Intel under single cores before a complete transision to Dual, Quad and Hex cores takes over. If the prices dont change drastically over single cores, programs start to take advantage of them, then there is no reason for the consumer to care.

I'll take a 4 core AMD at 2 gig any day of the week over a 5gig single core pentium.

AMD will never submit to straight mhz increasing. The fastest processor they have is 2.8 FX-57 which was on display at computex and should be going retail by mid next month. I dont expect them to get anywhere close to 4 gig for another 3years+. More so if they submit to multi core wafers as i suspect they will.

We simply do not need raw power anymore. If they want to eliminate the bottleneck placed by programs on CPU's they need to multicore them and allow the programs to asign certain tasks to each core. This = future.

The first 65nm intel parts (Dempsey) will be released with 2mb L2 cache, 1066FSB 3.80GHz, 3.60GHz, 3.40GHz, 3.20GHz, 3GHz and 2.8GHz.

Will be interesting to see their thermal output and if they try to make a push again. Btw this is the roadmap for early 2006. As you can see, no 4gig + parts.
 
AMD has a dual core 2.2 out, I've seen the page to buy it at.

2.2+2.2=4.4

Not too far to go, dual core will be where you start getting into the 5ghz range if that even counts(?).

~Adam
 
CleanSlate said:
AMD has a dual core 2.2 out, I've seen the page to buy it at.

2.2+2.2=4.4
Click to expand...


That's not how it works at all, perhaps if it was load-balancing... but it's not.
 
As for 5Ghz on AMD, don't count any time soon unless they introduce faster-switching transistors (not to mention cooler-running... which also happens to be the reason AMD chips are cheap), no way in hell could AMD produce chips that run as "cool" as Intel's running at the same frequencies.
 
Moore's law is an incorrect guesstimate that intel has exploited for PR purposes. Whenever it has been proven wrong people switch exactly what the Law predicted. Moore predicted that the number of transistors on a single chip would double every 18 months. This has not happened, so people have quitely shitfted the focus from transistors to the vaugely termed "processing power". It's a myth that grumpy old techies pass onto n00bies. It's not true; let it die. Please.
 
mohammedtaha said:
actually it was 12 months and it was later extended to 18 months to reflect the change ..
Click to expand...
heh, i like how we can change a law like that to make it fit out needs ;)
 
sacon said:
it only seems like a few years ago that someone had overclocked an intel to 1ghz before a 1ghz had ever been sold

AMD have focused on chip design as compared to intel that have been focused on ghz, which is why Intel win the ghz war and AMD win the clock for clock war
Click to expand...

Actually the 1st 1ghz computer commerically avialbe with either an intel or amd chip was a kryotech 1ghz AMD based off of a 700mhz althon I believe.

Intel was the first to announce 1ghz machines, but AMD was the first to actually ship processors that could make the cut.

So AMD in my mind won the race to 1ghz by being the first company to be able to sell me a chip that came from the factory at 1ghz or from the aftermarket with exotic cooling.

The only way intel wins is on paper. This was of course in the day when they had a fairly level playing feild similiar clock for clock perfomance and similiar scaling capabilities per process technology.

ScHpAnKy said:
That's not how it works at all, perhaps if it was load-balancing... but it's not.
Click to expand...

Well and X2 at 2.2ghz is like a 4.4ghz proc on heavily multithreaded apps, which will be games in the next 12mo with all these multicore consoles coming out.

Back when quake3 supported smp, my dually 550mhz celeron ran very comprable times to single 1ghz P3s.
 
obiwansotti said:
Actually the 1st 1ghz computer commerically avialbe with either an intel or amd chip was a kryotech 1ghz AMD based off of a 700mhz althon I believe.

Intel was the first to announce 1ghz machines, but AMD was the first to actually ship processors that could make the cut.

So AMD in my mind won the race to 1ghz by being the first company to be able to sell me a chip that came from the factory at 1ghz or from the aftermarket with exotic cooling.

The only way intel wins is on paper. This was of course in the day when they had a fairly level playing feild similiar clock for clock perfomance and similiar scaling capabilities per process technology.



Well and X2 at 2.2ghz is like a 4.4ghz proc on heavily multithreaded apps, which will be games in the next 12mo with all these multicore consoles coming out.

Back when quake3 supported smp, my dually 550mhz celeron ran very comprable times to single 1ghz P3s.
Click to expand...

it was th Super G and it was on a 500Mhz K6-2 and later a 700 slot-a K7
 
You must log in or register to reply here.
Back
Top