X-bit labs

Started by: Anemone | Date 10/14/04

Comments: 22 | Last Comment:  08/30/06

[1-20 | 21-21]

1. I think what is most disturbing about this is the almost complete lack of letting us know what lies beyond the 2mb cache chips. Frankly we see the extreme chips now, and we have a great idea how the 2mb Prescotts will perform. In fact we have a pretty good idea what would happen if they moved to 4mb of cache, but frankly we have no idea what kind of cpu we'd drop in after that.

LGA 775 is "promised" to be compatible with a future chip beyond those Prescott 2mb's and I find it hard to believe that promise when they honestly don't know, or won't tell exactly what that chip entails.

Why is this valuable? Why do we care to know now? Because, for me, gaming still favors a quality single core. What has been valuable up till now is that the HT of the P4 gives us good multitasking as well as "decent" gaming. If the future of P4 is to branch out, and this is what this news sounds like, then they are going all out multitasking and they are seemingly giving up on the gaming area entirely.

They will focus on mass volume markets. They have outlined NO functional platforms for the various tasks.

So do we salute AMD for giving us a "what you can get now and what you can plug in later"? Yes we do. Because Intel is failing on yet another front they used to do well at, letting us be assured that those super expensive parts we are buying today, motherboards and the rest, are useful for a chip generation beyond what we have. AND they would usually indicate what those chips were likely to do for the various market segments.

Now we have a thorough black hole. It's like their ability to compete with AMD is collapsing in one huge "don't worry we'll figure a way out" press release. Heck they aren't even saying this openly, which is even worse.

I'll make a bold prediction. They have NO plan beyond adding cache to Prescott. They have no idea how to get a Dothan to scale past 2.5ghz, let alone how to add 64bit ability to it, and they are seriously considering copying everything AMD has done right down to the on chip memory controller, meaning they would throw out every single compatibility they currently have out the window.

When corporations get worried about drastic things, they get real quiet. And you know what, Intel is VERY, very quiet...

2. This also impacts AMD plans :-(. No need to race with itself. They are not planning anything faster than 4200 (on Q3'05) - probably 2.6GHz with 1MB cache = FX55 which will be released soon. At that time fastest Intel will be 3.8GHz 2MB cache 6XX for wstations and P4EE 3.76Ghz 2MB cache 1066FSB.
Will this be the "slowest" year on CPU market ? Probably :-\.

3. I guess what I find interesting is that this isn't the "first" cancellation Intel's made this year.

If ya'll recall, earlier this year Intel junked the "Tejas" and some other cores after "Prescott" because of heat issues and announced that the Dothan/Banias PM architecture would form the base of the next generation of chips. What bit most of all about that was BTX, Inte's "next generation case." Without the massive heat monsters coming down the pipleine, there's no reason to sub out the current ATX case designs for anything new. Now that the 4ghz is canned...

Then I have to look over at AMD. AMD just finally completed a switch to 90nm, and 90nm parts are already shipping. While thermal benifits have not yet been realized (as the parts are 2.4ghz and under) compared to shipping 130nm parts, AMD has some room to grow (as evidenced by repeatable 45% overclocks on Athlon64 3000+ 939 90nm chips).

AMD could take the performance crown back at this time, simply by releasing a 2.7ghz 90nm A64 part.

As if the problem couldn't get worse for Intel, Itanium is dropping line a stone, and not one single analyst really believes that Intels P4 Dual-Core demo was "truely" a dual-core demo.

Where-as AMD is already demo-ing and showing working dual core silicon.

Like KHysiek says, this might be the "slowest" year of the CPU market.

But, if you ask me, it's the signs of Intel's dominance starting to tip over and collaspe. AMD is already the default choice for gamers, and is also making inroads in HTPC systems. Add in a maturing X86-64 market, a lack of P4-EMT's, and the fact that P4-EMT's won't work on current P4 platforms

I really would not want to be an exec in Intel right now.

4. Well BTX is starting to show up in lots of places apparently, but that is just going to piss off the market even more.

Agreed, think being executive management @ Intel right now would be a less than pleasant place to go into every morning.

Locked down chipsets, at a time when Nvidia is putting out not only an Intel quality chipset (Nforce 4), but also one with features Intel isn't even prepared to adopt yet in desktop form factors (SLI).

Lower performing chips with vague or no listed replacements.

Top end chips which are slower and locked down vs the AMD solution being open, cheaper, and better performing.

The very real potential that they might have to copy the memory controller on chip idea and thus trash the expensive 915/925 line.

Biggest innovation in a year is to have added 10 instructions to SSE and added cache to their desktop chips...

A roadmap that has been torn up and has zero information past 3 months as to what Intel will be producing past 3 months from now (desktop)

The practical fact that they have just about given up both the gaming and performance crowns to AMD...

hmmm

5. Intel is in that position they're in now because they believed speed was the answer. Management took over Engineering...And Management seriously f**ked up.

This is what happens when you let Management take the lead...Instead of Engineering! That's a big lesson to learn Intel!

Its unsurprising that the P4 4Ghz is canned. All signs for the past year has showed us they've hit a big ass brick wall with the P4 architecture.

Time to dump it and use Pentium-M. They've already showed us the technical diagram (Are the called Lithograph picts?), of a dual-core Pentium-M ("Yonah")...If there's Intel's last hope, that's it. Its the only core they have that can compete clock for clock (close enough) to the A64/Opteron and do it with less heat under 32bit. (In pure 64bit, Itanium II whips ass...Don't believe me? Ask about NASA's Itanium II based cluster for weather simulation).

I doubt AMD will bring anything faster than 2.8Ghz in the near future. And there's less likelyhood now with the canned P4 4Ghz.


And Anemone, that's the most uninformed prediction I've heard to date.

Just compare the R&D capacity of Intel and AMD.

AMD has to do alot with alot less (they've done well with the K8, and with IBM's help in 90nm and SOI). Intel can crankout several different designs simultaneously. (as demonstrated by Pentium-M, P4 and Itanium).

They could be working on several more designs that your "bold prediction" seriously fails to address.

They will never copy anything AMD unless their customers put them in that position. And that's only one case...EMT64/AMD64, because Dell, one of Intel's biggest customer needed an Intel equivalent.

And even then Intel has yet to release any mainstream CPU with EMT64. (Because according to them "there isn't a need for it yet". Yeah right...Its more like "We can f**k up AMD64 scheme by delaying its acceptance as long as possible".)

The more they delay EMT64/AMD64 progression, the more "waste of time" it feels. (It doesn't help when MS is ever delaying the AMD64/EMT64 compatible Windows!)

And BTW, Anemone, how come AMD hasn't increased speeds with the newer 90nm CPUs? What's with this varying of memory channels and cache size bullshit with several speed ratings? What happened?

Here's my prediction.

AMD will never go faster than 2.8Ghz. Simply because the competition isn't going further than 3.8Ghz. (No one can forget AMD's "me too" attitude...Their pricing premium of their K8 line as demonstrated this).

Both AMD and Intel's dual-core will be slower than the regular single core desktops in terms of speed. (I wonder how AMD is gonna use their rating system then).

Multithreaded apps that fully utilise multiple cores or processors are a bitch to write, so software is gonna be a few years behind. I predict we won't see much benefit of dual-core until 2 to 3 yrs later, if we're lucky.

***They will never copy anything AMD unless their customers put them in that position. And that's only one case...EMT64/AMD64, because Dell, one of Intel's biggest customer needed an Intel equivalent.
***

Try Dell is Intel's biggest customer, period. Dell has also been hammered for a long time from gamers and the corporate business it makes sales to for AMD parts. There is strong evidence to suggest that if Dell didn't get such a large price-break from Intel, AMD units would already be shipping in bulk.


***And BTW, Anemone, how come AMD hasn't increased speeds with the newer 90nm CPUs? What's with this varying of memory channels and cache size bullshit with several speed ratings? What happened?
***

Woah, and you want to accuse someone of being uniformed?
Let me ask you this 22 : When experimenting with a new process, do you go with "newer and faster" products or "older and slower" products?

Let's look at some examples. When Intel dropped the core size of P4 and attempted to make the new core sizes the fastest most advanced products, Intel wound up with very low yields and had great trouble when meeting stock requests.

When Nvidia tried to move GeforceFX to a new process from high end to bottom end, they were unable to provide sufficient parts at any level, giving ATi almost a full complete year of lead time in the DirectX 9 market.

When ATi moved to a new process, they moved the mid-range Radeon 9600's first. Not only was ATi able to meet stock requests and inventory, prices stayed down, and they made a profit.

When AMD moved to a new process, they moved the new process products into midrange or lower-high range products. Again, AMD made profits and was able to meet inventory and stock requests better than competitors.

Do I need to also look at Sun's Sparc, Intel's Intanium, IBM's Power, IBM's PowerPC, Via's C-series, and Transmeta chips as well for this?

There is firm evidence in place that you do NOT run your high end products on newer processes until you have the kinks worked out. ATi was able to ship Radeon 9800 Pro and XT's built on the .15u process while Nvidia was choking to get out slower and decidedly inferior cards on .13u. Same thing with AMD.

Why have new speeds not gotten here yet on .09u? Simply because loosing a few yields of FX-55 2.6ghz 90nm parts could cripple AMD. Loosing a few yeilds of Socket 939 3000+ 90nm parts isn't as crippling as the same part is available in another process.


****
AMD will never go faster than 2.8Ghz. Simply because the competition isn't going further than 3.8Ghz. (No one can forget AMD's "me too" attitude...Their pricing premium of their K8 line as demonstrated this).
****

I highly doubt this. AMD isn't stupid. If they can release 2.9ghz product on 90nm that meets the thermal requirements, they will do so.


***Both AMD and Intel's dual-core will be slower than the regular single core desktops in terms of speed. (I wonder how AMD is gonna use their rating system then).
****

Same way they do now. Unless you missed it, there already is a seperation in in Sempron vs. AXp vs. A64. Sempron processors with model rating numbers are known to be less powerful than AXp or A64 chips with the same model number.


****Multithreaded apps that fully utilise multiple cores or processors are a bitch to write, so software is gonna be a few years behind. I predict we won't see much benefit of dual-core until 2 to 3 yrs later, if we're lucky.***

Um... just out of wondering here... why exactly are the software's going to be behind? From a runtime standpoint, a dual core processor will closely look like current dual system's or P4 HyperThreaded systems. We already see a benifit in using dual processors, quad processors, and 8-way processors in today's servers. We also see a (small) benifit in HyperThreading in some applications today (though granted, some apps are preformance penalized)

Why exactly would we not see a performance benifit in dual core?


***They could be working on several more designs that your "bold prediction" seriously fails to address.***

They could, but they aren't. The process teams working on P4 and beyond are slowely being moved over to the Dothan/Banias based teams, adding things like X86-64 and hyperthreading to the PM design. This is something that's been known and documented. The Intanium team is kinda of in limbo.


***Time to dump it and use Pentium-M. They've already showed us the technical diagram (Are the called Lithograph picts?), of a dual-core Pentium-M ("Yonah")...If there's Intel's last hope, that's it. Its the only core they have that can compete clock for clock (close enough) to the A64/Opteron and do it with less heat under 32bit. (In pure 64bit, Itanium II whips ass...Don't believe me? Ask about NASA's Itanium II based cluster for weather simulation).***

Technical diagrams are not shipping products.

Also, Itanium only whips in certain situations that require extremely high Floating Point throughput. Your example here 22 is extremely misplaced and mis-used. Nasa's weather simulation is a dedicated computing system. It is not a general purpose workstation, desktop, or otherwise personal computer. The OS and softwares running on I2 in this case are more optomized than you will see in a "typical" computing enviroment. Let's pull that I2 out and place it in a mixed server enviroment where it has to deal with Web-requests, mail requests, database access, and other assorted typical server items. Lets pull that I2 and put into a gaming computer. Sorry 22, the verdict is already in. In pure 64bit, X86-64 whips the I2 in every case in today's computing applications.

You are forgetting an imporatent fact here. The Intanium2 is not meant to compete with P4, A64, AXp, or anything that "we" would use. It was meant to compete against Sparc, Alpha, and MIPS. In that respect, I2 has a future. It's just not in generalized computing.


****Just compare the R&D capacity of Intel and AMD***

You also appear to be forgetting that AMD does more than desktop chips and server settings. AMD also happens to be big in Flash Memory and other designs.

While you are correct in that Intel does have a larger overall employment, Intel is also involved in area's that AMD is not involved in.

je.saist
www.gamenikki.com

6. Once again, Intel following the AMD lead.

7. I eventually went back and yanked that response out and put it up in html format.

http://saist.fateback.com/Intel_Industry_Viewpoint_Response.html

8. Well this topic sure put Intel in the news, yes?

Its going to be quite a difficult world sorting out whether a given user benefits more from dual core or more from a faster single core. But when the industry shows such an all out push for multiple cores as it is now, maybe that will help fuel development of making software perform better under SMT conditions across the board. It'll be a while though till we see that bear fruit.

While a nice technical achievement, I don't consider Itanium a reasonable portion of a discussion on desktop cpu development. Just my $.02.

The who is bigger and who is smaller argument has a number of interesting business facets that could be considered however.

A smaller manufacturer has lower overhead costs (less factories that cost money irregardless of useage)

A smaller manufacturer has the ability to change plans and react faster to changes in the business environment.

A smaller manufacturer is likely to be more in touch with its customers, simply because it must be that way.

You've already listed, as have others, the numerous detractions on being a smaller manufacturer, so we'll leave them as is.


I'd like to see what kind of a roadmap Intel produces for the upcoming full year, not just the Prescott 2m's and just what kind of lifecycle they want to give the 915/925 chipsets (if there even is a lifecycle).

I want to know if the memory is available, and the chipsets are available, why the P4's seem to have no yield @ 1066 bus.

On a positive note, since it seems that enthusiasts can reach and exceed 4ghz on air cooling alone with Prescott now, the likely reason for cancelling the 4ghz model is not ability, but that whatever is coming brings enough power with it, that there would be simply no need for a 4ghz model. 2mb L2 cache is a good idea, if it makes manufacturing and cost sense.

I'm pleased my first post was responded point by point, and glad we got some discussion going on this news :)

ty

9. This is funny, I can't believe people argue over this BS, get over it folks.

And look how much wasted effort goes into it! Pages and pages of stuff no one cares about until the products mentioned actually arrives. This is a good laugh!

Why waste time on this folks? Intel vs AMD, who cares! As long as they're both here and in competition, slugging it out, I don't really give a sh*t.

I use either...Heck, I even use Transmeta and VIA! You know why I don't care? Because at the end of the day, the competition among the processor companies furthers innovation! Whether its dual-core, SMT, low power, hardware accelerated encryption, etc...Its all good.

Why not make something better of yourselves than sit around and type about what's gonna happen with endless debates? If you're an idiot before, you're still an idiot after...Arguing over the web is as useful as hearding 10000 ferrets. It gets you nowhere.

Learn some new skills folks. How about doing some programming? Maybe start on a new OS? Linux? BSD? Other Unix variants? You really wanna rely on MS for your computing needs in the future? Expand yourselves folks!

10. Anemone, don't ever under-estimate Intel...They may screw up some things, but they a plenty resourceful...

Intel dual desktop core chips to be called x20, x30, x40
http://www.theinquirer.net/?article=19147

Intel's Yonah: a dual core thin and light Pentium M
http://www.theinquirer.net/?article=19149
(This shows Pentium-M variant will be EMT64 and "execute disable" bit featured)

Intel Dempsey ready for Blackford, Greencreek punch-up
http://www.theinquirer.net/?article=19150

11. dude that is off the hook

12. Hello, very nice site!
http://ch-fioricet.myk.pl/
Thanks!

13. Hello, very nice site!
http://cialis-on.any.pl/
Thanks!

14. It is an excellent site, we can learn many things perfectly in line with vaishnava sampradaya.
http://online-poker-game.c-o.cc
Thanks!

15. it is very good web site for all.
http://med-on.up-a.com
Thanks!

16. A wonderful site.
http://fioricet-online.sinfree.net
Thanks!

17. Hello, very nice site! Please also visit my homepages:
http://phentermine4you.be/cheap-order-phentermine.shtml
Thanks!

18. Hello, Excellent site
http://yasmin-on.any.pl/
Thanks!

19. Hello, Very good site with nice info's
http://viagra4you.be/cheap_viagra/buy_viagra_cheap.html
Thanks!

20. Hello, Very good site with nice info's
http://hair-care.wepaw.org http://healthy-hair-care-tip-for-today-black-woman.wepaw.org http://hair-care-tip.wepaw.org
Thanks!

[1-20 | 21-21]