5.0 ghz processor

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Vee

Senior member
Jun 18, 2004
689
0
0
There's no longer any pressure for GHz any more. Intel have dropped GHz-marketing and the P4 lineage like hot potatos, because they have pretty much got them into a deep cul de sac.

Even considering just single core, GHz aren't needed for performance, because you can have execution units in parallel, and feed them out of order, AMD style.

GHz will continue to rise. But more slowly, as new manufacturing processes becomes available.
Heat rises much faster with clockrate, than speed.

My guess is, we might see 5GHz by the end of 2006, early 2007. But it really doesn't matter what clock. Those CPUs are going to be so much more powerful than "5GHz" would indicate.

My thinking is that prefetch, cache handling and memory controller is the big challenge area for increased performance.
 

Dug

Diamond Member
Jun 6, 2000
3,469
6
81
Originally posted by: Zebo
Never? Scaling and netburst is dead. Too much heat and fake representation of chip capabilites.

Pentium 3..I mean Dothan desktop is the Intels future.

A 2.0 Dothan, with HT and 800-1000Mhz fsb added would kill a 4.0 Ghz presshot.

I wouldn't say it would kill it.
Dothan benchmarked
 

Avalon

Diamond Member
Jul 16, 2001
7,571
178
106
Originally posted by: Dug
Originally posted by: Zebo
Never? Scaling and netburst is dead. Too much heat and fake representation of chip capabilites.

Pentium 3..I mean Dothan desktop is the Intels future.

A 2.0 Dothan, with HT and 800-1000Mhz fsb added would kill a 4.0 Ghz presshot.

I wouldn't say it would kill it.
Dothan benchmarked

That's a 2.0 Dothan without HT running on a 400mhz bus. It was meant that a similar bus speed Dothan with HT would kill a prescott with the same thing.
 

Falloutboy

Diamond Member
Jan 2, 2003
5,916
0
76
Originally posted by: Avalon
Originally posted by: Dug
Originally posted by: Zebo
Never? Scaling and netburst is dead. Too much heat and fake representation of chip capabilites.

Pentium 3..I mean Dothan desktop is the Intels future.

A 2.0 Dothan, with HT and 800-1000Mhz fsb added would kill a 4.0 Ghz presshot.

I wouldn't say it would kill it.
Dothan benchmarked

That's a 2.0 Dothan without HT running on a 400mhz bus. It was meant that a similar bus speed Dothan with HT would kill a prescott with the same thing.

HTwill not work on a dothan. the closest thing we will get on dothan is dual core
 

DrMrLordX

Lifer
Apr 27, 2000
22,696
12,650
136
Originally posted by: clarkey01
Intel said they would make duel core prescott's. not p-m's, yes i agree p-m's seems a better choice for duel core with less heat and power consumption and the higher IPC, but maybe duel core's with prescott's hyperthreading seems more golden in Intels eyes.

When did Intel say their dual-core cpus would be Prescott based? That would be utter madness. The last Intel roadmap I saw with dual-core CPUs prominantly featured showd clock speeds ranging from 2-2.4 ghz for their dual-core CPUs. Why would they bother running dual-core Prescott cpus at such speeds? No P4 core would do well at that speed as compared to Dothan(or a successor to Dothan).
 

Zebo

Elite Member
Jul 29, 2001
39,398
19
81
Originally posted by: DrMrLordX
Originally posted by: clarkey01
Intel said they would make duel core prescott's. not p-m's, yes i agree p-m's seems a better choice for duel core with less heat and power consumption and the higher IPC, but maybe duel core's with prescott's hyperthreading seems more golden in Intels eyes.

When did Intel say their dual-core cpus would be Prescott based? That would be utter madness. The last Intel roadmap I saw with dual-core CPUs prominantly featured showd clock speeds ranging from 2-2.4 ghz for their dual-core CPUs. Why would they bother running dual-core Prescott cpus at such speeds? No P4 core would do well at that speed as compared to Dothan(or a successor to Dothan).

Hehe this relates to our other thread too...embrase the madness

http://arstechnica.com/cpu/004/prescott-future/prescott-1.html
 

Hulk

Diamond Member
Oct 9, 1999
5,118
3,660
136
Originally posted by: DAPUNISHER
I couldn't care less if a 5ghz chip happens in the next couple years, I would rather have a 2 core CPU with strong IPC and a fully supporting OS thanks.


and fully supporting apps.
 

clarkey01

Diamond Member
Feb 4, 2004
3,419
1
0
"When did Intel say their dual-core cpus would be Prescott based? That would be utter madness. The last Intel roadmap I saw with dual-core CPUs prominantly featured showd clock speeds ranging from 2-2.4 ghz for their dual-core CPUs. Why would they bother running dual-core Prescott cpus at such speeds? No P4 core would do well at that speed as compared to Dothan(or a successor to Dothan)"


trust me i know my stuff DrMrLordX
 

clarkey01

Diamond Member
Feb 4, 2004
3,419
1
0
i agree DrMrLordX P-m duel core would be a better idea but they seem to think hyperthreading with duel cores seems a better way to go, cept not with power consumption. a duel core with hyperthreading would appear as 4 virtual cores with presccot, with p-m's the pipeline is too short to implement hyperthreading.
 

Dug

Diamond Member
Jun 6, 2000
3,469
6
81
Originally posted by: Avalon
Originally posted by: Dug
Originally posted by: Zebo
Never? Scaling and netburst is dead. Too much heat and fake representation of chip capabilites.

Pentium 3..I mean Dothan desktop is the Intels future.

A 2.0 Dothan, with HT and 800-1000Mhz fsb added would kill a 4.0 Ghz presshot.

I wouldn't say it would kill it.
Dothan benchmarked

That's a 2.0 Dothan without HT running on a 400mhz bus. It was meant that a similar bus speed Dothan with HT would kill a prescott with the same thing.


Excuse me, but where did you get that increasing the bus speed would increase performance so much. I know HT doesn't so I don't even need to comment on that.

Last time I checked bus speeds gave at most ~10% performance gains. This has been true for a long time. Unless you have something to show different.
 

DrMrLordX

Lifer
Apr 27, 2000
22,696
12,650
136
Ugh, dual-core Prescott = pain and agony. I can only imagine how they'll work out a cooling solution for that . . . monstrosity. It's a shame there aren't more Voodoo 5 6000 video cards around. One of those would be perfect for a duallie rig running dual-core Prescotts. Urk.

And Clarkey, looks like you're right, at least for now. I'd really rather see Intel release dual-core Pentium Ms. Let's hope someone at Intel is going to design one as a backup plan in case the dual-core Prescotts fail.

Thanks for the article link, Zebo.
 

Vee

Senior member
Jun 18, 2004
689
0
0
Originally posted by: Zebo
Originally posted by: DrMrLordX
Originally posted by: clarkey01
Intel said they would make duel core prescott's. not p-m's, yes i agree p-m's seems a better choice for duel core with less heat and power consumption and the higher IPC, but maybe duel core's with prescott's hyperthreading seems more golden in Intels eyes.

When did Intel say their dual-core cpus would be Prescott based? That would be utter madness. The last Intel roadmap I saw with dual-core CPUs prominantly featured showd clock speeds ranging from 2-2.4 ghz for their dual-core CPUs. Why would they bother running dual-core Prescott cpus at such speeds? No P4 core would do well at that speed as compared to Dothan(or a successor to Dothan).

Hehe this relates to our other thread too...embrase the madness

http://arstechnica.com/cpu/004/prescott-future/prescott-1.html
That's kinda bad news. :disgust: What will happen to the Conroe now? :frown: Looks like I'm onboard that AMD ship for a very long cruise. :sun: - Oh well, might as well relax, cheer up and enjoy it. Bye, bye Intel. :wine:

(I don't care about GHz or heat or whatever. I never liked the P4 because it immediately disappointed me with its 'real life' performance, compared to PIII (not to mention Athlons). I don't want dual cores of that crap. I want a CPU from Intel that does things, other than benchmarks and media encoding, well.)

This is official: "Vee" took another step deeper into the AMD camp today. I'm going to order components for an A64 system.
 

Zebo

Elite Member
Jul 29, 2001
39,398
19
81
Cheers:) Since the first Tbird, I never liked P4s high prices compared to thier fantastically cheap AMD equivalent counterparts. Now we have a heat issue and thus a correspondingly noise issue which totally puts me in the AMD camp even at the same price as the A64's are. Not to mention Intels new socket practically designed to break and overclcoking block issues as yet unresolved.:)
 

oldman420

Platinum Member
May 22, 2004
2,179
0
0
if it does happen it will be in the next 2-3 years but you will need a hsf the size of a greyhound bus to cool it so they will have to make a bigger case for it he he
give me 2 fully integrated 2.5 ghz cpus and a os and apps that support it and i will be happy
 
Jul 12, 2004
154
0
0
It's been interesting to see the wall that most chip fab companies are hitting with the 90 nm process. It puts a lot more speculative interest for me on how AMD will make this transition. With Apple moaning about the inability of the G5 to ramp to where they hoped it would go (3 GHz), the 2.5 GHz G5 being liquid cooled and Prescott's under-whelming performance and over-whelming power consumption, it doesn't seem likely that AMD can beat this problem?
But maybe I'm being overly pessimistic, as Dothan has moved from 130 to 90 nm without a significant power output increase. I'm not a tech head but the Athlon 64 seems to be closer to the Pentium-M in being a high IPC part, so maybe the die shrink won't be so negative for it either?

The recent Anandtech Dothan article surprised me in just how well the Dothan performs against Prescott & Athlon 64, except in media encoding & gaming. For general usage though, it has to be the most balanced x86 design out there, period. An excellent IPC, low power, versatile dynamic clock/voltage switching. I see a problem though. If you're not a gamer or hard core media encoder, you don't really need any faster than today's lowest spec CPUs. So who are the next generation CPUs being marketed at. Sorry, I mean designed for :)

Well if they're being designed to match people's needs, then they should be aimed at people who need more power, so we're back with gamers and encoders. But the P-M isn't so good at these tasks! Hmm. Either way it seems as if dual-cores are the way forward. With media encoding, dual cores shouldn't pose a problem, as it's relatively easy to get close to maximum theoretical benefits using parallelism. But what about for gaming performance? Isn't the ball suddenly going to be passed to games developers to optimize their games for dual cores? I imagine that's going to be a hell of a lot of extra work for no extra revenue. But aren't next generation games consoles moving towards multiple cores anyway? I think I'm correct on this, which means that all this extra development work is going to be done anyway, at least by software houses that are targeting multiple platforms.

It points to the games platform designers being more prescient than the likes of Intel, who seem to have been caught short recently. In the future, making comparisons between different platforms might very well get a lot more complicated. Certain platforms might well shine in certain areas but be very weak in others, much more so than today.
So a dual core Prescott running at a high clock rate, marketed at gamers, might be released, but built into a fridge! At least now I can start to see the logic behind selling fridges with a built in web browser. You'll be able to view your virtual electric meter whizzing around via your utility companies website, whilst fragging at insane frame rates simultaneously. If the thought of your next electric bill gets too much, just open the fridge door and pull out a cold one :)
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Tejas, the 5.0ghz netburst variant that was to be on an .065 process and with 2MB of cache, was canned.

5.0ghz is a long way off.
 

DrMrLordX

Lifer
Apr 27, 2000
22,696
12,650
136
The answer to all our heat problems is: diamond transistors. Then it will no longer be the transistors threatened by heat. It'll be the user instead. Ha ha!