Sign in with
Sign up | Sign in
Your question

Intel/AMD price war brings down market

Last response: in CPUs
Share
November 24, 2006 12:41:53 PM

As I noted several months ago, witht e level of prices set for Core 2, all ASPs in the semiconductor have shrunk though the industry has grown 19% in the last year.

Do you think that the Intel price war is really affecting the whole industry? I believe that it is and as long as they have most of their inventory under $300, ASPs will continue to sag as pressure is put on all players in the industry to keep prices down.


Linkage!
November 24, 2006 12:50:08 PM

Yo. Happy Thanksgiving vacation to you to.

Back to the topic, how much of this can really be attributed to the price war? last time I checked sales were up a significant amount. And, do we really want this price war to be over?
I only hope it switches to the RAM industry.
November 24, 2006 2:08:56 PM

If you think about it for a moment, you'll realize that it's not really a price war from the perspective of Intel because they still have processors priced at all ranges of the scale...low-med-high. However, from the AMD side, they are forced to lower their prices across the board because they cannot compete with Intel's VALUE/Price proposition.

When you say "affecting the whole industry", what segment are you referring to? If you are referring to other semi-conductor co's, then yes, they have new competition on the block. If you are referring to resellers, then no. Why would an inexpensive CPU affect the profit margin of a reseller? If anything, it gives them a little more wiggle room to play with the final price of a pc since joe consumer has been relatively conditioned over the last few years regarding what a pc costs.
Related resources
November 24, 2006 2:59:02 PM

I agree, introducing a new single core at this point in the game may not be a great revenue generator. Seems like I read somewhere that Intel decided not to produce a single core Core2 in 2007 for that very reason. No link though...can't remember where I read it, sorry.

I guess the one arguement to this could be that if one of the CPU's on a dual core die is dead, it makes sense to sell it as single core rather than ditch the whole chip. This I can see as plausible.
November 24, 2006 8:54:49 PM

I think we will know for sure when the FX reaches a sane price.
November 24, 2006 9:20:31 PM

Quote:
Not sure that a dual core with one dead chip would really work that well, unless if it had its own driver to eliminate the possibility of using the secind core, otherwise everything would be completely unstable


This is I believe already relatively commonplace - it works fine.
November 24, 2006 9:26:39 PM

Quote:
If you say so, but I personally owuldn't want to buy a processor thats half dead, would you?


In many cases, you wouldn't even know until it arrived. And no, I wouldn't have a problem because the two halves are unlikely to be 'related'.
November 24, 2006 9:32:55 PM

Emerging Markets, India etc.

Single core CPU's whilst not useful for tech heads are great for poor countries and @ 65nm you get a lot of CPU's for your money. Imagine how many single core cpu's with say 256k cache AMD could get from each wafer.
November 24, 2006 9:47:36 PM

I don't know but if you had 1 dead core on a C2D and you set the other's freq higher you end up with a 2.5 ghz clocked at 3.2 or so with lower power consumption and heat than a P4/D series proc. otherwise its just trash. why not sell it in a lo/medium price range in developing countries or domestic markets. Better tech/less waste. Of course I don't know if its possible or if the C2D process has enough thow faulty proccesors to make it practical :roll:
November 25, 2006 5:15:09 AM

Do you remember what the xp3200 cost at release? How about the top of the line P4c or e?
Over the past 3 years, the price of the fastest chip has gone through the roof. It's about time prices settled back down again. After all, Intel spends as much to make the E6700 as they did to make the P4c 3.2 ghz.
For AMD, they need 65 nanos. Then thier cost/chip will be only slightly higher than the xp chips.
November 25, 2006 9:02:23 AM

Intel do have the current die size advantage at the moment but as their cpu's have massive cache they will get less chips per die than AMD @ 65nm with much smaller cache.

If AMD can produce a CPU that competes with C2D but only has 1MB cache then AMD will be in a good position as they can produce more chips per wafer. Bringing down their manufacturing cost and improving margin.
November 25, 2006 9:50:05 AM

Inflation of income over 10 years, specifically in the sub-category of people whom work in InfoTech.

They / We are willing, at times, to pay to have the best.

Still as I don't game as often anymore, just like 70%+ of the population, the 40 core + STREAM processing future looks damn good to me.

The irony is IBM / Apple were trying to do exactly what Intel is doing now, just with a non-x86 architecture (backwards compatibility was not there to hold it back either, but this is what shot them in the foot).

Only the top 10% of gamers / enthusiasts will want to pay for the top end gear. I got an AMD Opteron 270 (2 x dual-core) back in the day, kept it for under 2 years and sold it to finance two new PCs. One of which actually outperforms the AMD Opteron 270 with only half the core count. (No prizes for guessing which chip though - Core 2 Duo E6600 at 3000/1333).

Was it cost effective ?, Depends how you look at it, I'd say "Hell no !", but then again I sold it 3 months before the AMD Opteron 200/2000 series CPU prices got cut in half overnight (October 11th, 2006 if I recall correctly). Took many places a month to notice, Good thing I read MoneyCentral then eh ? :wink: - Also watch / predict industry movements and price-cuts pretty damn accurate - Was near screaming at someone who bought an Opteron 275 (old tech, at a high price, no upgrade path) just half a week before the 50% price slash on AMD Opteron CPUs. (For a game server too :roll: - Performance per Dollar was their biggest concern).

Overclocking the cheap gear is where it is really at. (For gamers / enthusiasts at least). 8) - I hope AMD / Intel never forget about the cheap-skates that always want more... for less. It is the gear of progress.

Least we all end up as dumb consumers, playing on overpriced consoles with fake yield / laser problems. :roll: :lol: 


With 40 cores, a Stream processor / High power video card the future of gaming isn't going to be about having two fast cores, it'll be about having 32+ power / heat efficient cores.

Heck 70%+ of gaming performance (price wise) is the Video Cards GPU anyway.
November 25, 2006 10:18:54 AM

Yeah, they'll be going quad-core per die, with 512 KB L2 cache per core, and a larger 'shared by all cores' L3 cache. This is on Socket-F (1207) for future AMD Opteron processors. The same concept will hit single socket systems likely in Q3 2007, maybe Q4 2007.


8) EDIT: Recall that AMD use exlcusive caches, not inclusive ones, so each cache 'adds' to the total', vs inclusive caches where 2 MB L3 cache would mean only 2 MB of cache is there, and L2 and L1 would actually duplicate instructions/data in each cache.

Still sticking with the 3 pipeline system though on the cores, but tweaking out some look-ahead and 'borrowing' some of the smarter Intel NGMA (Core 2 Duo, Xeon 5100 and newer) features. I find this a bit weird as they've got the IA-64 engineers on their side of the fence now. (Think 6 to 11 pipelines - too much of a leap for software + compilers to scale though of course)

AMD have a few IA-64 engineers working for them now too, and IA-64 is a damn good architecture with damn poor management. (It needs a shrink to 65 nm, less L3 cache, cheaper per unit to make, and can have twice - if not more - the cores with the same transistor count and power consumption).

They can leverage that knowledge into new AMD64 designs ironically, and just do it with better management / over-all design. (IA-64 is too much of a leap, and too early, but much of its design can be 'salvaged' into x64). 8)

If I was AMD here is what I'd do:

- Start development of a C, Obective-C and C++ compiler specifically to optimize for AMD64 + AMD/ATI steaming processors.
- Hardware is good, just remember that without 'optimized' software to run on it hardware is useless.
- This is where, or when, AMD have always historically lost the bulk of their benchmarks, for the ones they do lose in, and they lose potentially large contracts as a result.



- Communicate the MMX, SSE, etc message to developers - Because sadly most are still not using the features for performance. (Intel C++ compiler is a good product - What do AMD have to compete with it ? - Jack shit nothing is the answer and they should get off their hardware focus'd ass and fix it (create software, AMD, never ?).

- Leverage / Salvage 'viable' IA-64 tech R&D into AMD64 / x86/x64 designs

- Improve everything near the registers and L1 cache (+50% larger L1 cache per core perhaps). (Even more so than predicted designs).

- Share L3 cache, reduce L2 cache (As they are, and as IBM / Apple did in G5, but using 65 nm and aim for power efficency for servers with high core counts).

- Advertise and Market the AMD C, Obj-C, C++ Compiler in every PC magazine and development magazine, and in more management focused magazines

- Research more into 'aggregation' of cache and dual+ channel memory controller performance.

- Encourage 'fat' binaries that work on Win32 / Win.x64 with both an Intel compiled element and a AMD64 compiled element. (Universal binaries like Apple are using for G5 / x86 / x64 software - it just checks CPU flags and jumps to best code path). (Storage is not the concern it used to be for .EXE sizes guys !!!). (They won't be 'fat' as in for any OS Kernel, but 'fat' as in 'many code paths').

- Don't try to legally attack Intel over their compiler, BUILD YOUR OWN. (AMD have no contract with Intel over compiler performance for their non-Pure Intel compatible architecture) - Instead of bitching about it MAKE ONE and then get every developer to use it.
- HELL GIVE IT AWAY FOR FREE, (Worked for Sun Microsystems + JavaDK). Make it like the *nix GCC but for Windows !

- Get on the 'new' Novell / SUSE bandwagon, Help make AMD64 optimized binaries for Novell Netware (the Novell NetWare Kernel is still x86 + PAE-36 based for crying out loud).

- Don't compete with Sun Microsystems, work 'with' them.

- Expect HP to stab me in the back at some point.

- Continue to work with Dell to grow consumer base (quantity of cheap sales, with good margins). Mass produce a PC that actually wins price/performance, RTFs, etc for Government / Large Business contracts. (Because AMD based PCs actually 'lose' most of these contracts for the most simple of reasons - If vendors even offer an AMD based solution at all that is - They need a total back-flip in this area of weakness).

- Don't dictate to the community, communication is a 'full duplex' (two way) street, ask for input, start polls, start 'ads' on websites (IT based + Affordable) with 1+ KB strings asking for input and requests from the community.

- Work more closely with Top500 listed companies (SuperComputers + Businesses)

- and a few other things I'll keep to myself for now.





Yes.... if only.... I had more influence towards AMD or Intel. :p 
... prob never happen though. :roll:

People do realise they can é-mail URL peralinks to AMD, Intel, Sun Microsystems, heck even SGI are back now and request more or specific design directions. (Just don't come on like a retard in the msg and back up your claims with technical data - Heaps of smart people at TomsHardware).


"Remember just 1 or 2 'A4' pages of text can change history"
- Tabris:D arkPeace, Nov 25th 2006
November 25, 2006 10:48:24 AM

Quote:
I only hope it switches to the RAM industry.


Heres to that!!
November 25, 2006 10:53:14 AM

I agree with the compiler issue and have often wondered why they havent done this already considering AMD's claims of anti-optimisation.

Your list is quite long..

With Dell's constant revenue stream you would expect AMD to have more confidence in investing in long term more expensive projects. I think AMD is heading in the right direction they just need to weather the storm till they get 65nm running full steam.

I don't think there has ever been a more exciting time for tech heads. I've always felt that Intel have held back in terms of increasing CPU speed so that they could milk the market as long as possible, thankfully AMD caught up and even 'leapt ahead' which has have forced Intel to get busy.

I've seen people with C2D systems clocked at 4Ghz+ this amazes me. I mean my X2 system running @2.5Ghz is only half the speed of the potential out there. That's almost frightening. In six months the gap will be probably twice the size it's all very interesting.......
November 25, 2006 10:54:11 AM

AMD is too full of it's self. Even now days when they know they are not the top dog at the moment they go around mouthing off like they are. They need to be more humble, but oh well. That's what happens when arrogant people run a company.
At least Intel is trying to chage the way it used to be.
November 25, 2006 10:56:40 AM

Humble? This is business, not a playground fight. Saber rattling and massive egos are par for the course.
November 25, 2006 11:01:39 AM

The fact that you know AMD have been mouthing off, and I presume you are refering to things like the 'quad core for dummies' issue proves that their tactic has proved beneficial. AMD have always been lacking in the marketting department but now more and more people are aware of the choices they have. ANd everytime they mouth off someone new hears about it. All publicity is good publicity.

Remember, Intel got rich by controlling the market and stiffling innovation. Now they dont have that control and the result it is unlikely that they will ever see the huge profits they have become accustomed to.

The war benefits us all as long as both sides survive!
November 25, 2006 11:39:27 AM

Which is exactly why they need smarter, more humble, innovative middle management. (read: my ideas above, if I had *any* influence over AMD and their future direction - Even if, hypothetically, I was a mid-low level manager in AMD I would be pressing these issues to head office).

I'll roll dice when I know the potential outcome is more likely to be good than bad. Worse case scenario for AMD nothing changes (relative over time).

With the stream processing idea, and their own compiler, they could make Online Multi-Player games 20x - 200x the scale and dominate that market. (With their 'all new compiler and dev toolkit' :lol:  ).

From there make all 'Business Collaboration' software 3D and support large scale virtual offices, etc. (A holy grail image to project to managers of large companies / government departments that want 'new and shinny' stuff).

They still need some arrogance in upper management, but they need to show a more humble, and open, face at times.

This worked for Apple at times (various tech issues got in way), and worked for Sun Microsystems for a very long time (but again, management issues, etc got in the way).

AMD/ATI don't have that "tech issues get in the way" problem that other large key players have. They are very lucky like this. However this is making them arrogant over time. They should leverage it and present a 'new public face open to two way communication with every potential client', and build the software to enable massive client-services in business related 3D worlds. (with the hardware AND SOFTWARE DEV TOOLS, ie: COMPILERS to pull it off, and help their potential clients pull it off too).

Intel has made an awesome comeback, and the only counter to an awesome come-back it a near absolutely perfect 'suite' of hardware products, dev products, and client-services.

Intel has vPro, sure that is cool, but... once again... what to AMD have to compete with it ? - Constantly responding to innovation looks bad over time, they need to innovate too.


For example, I would like to run an AMD64 optimized version of OpenOffice.org (Sun influence here) under a Win.x64 Kernel (XP x64 or Vista x64 clients), Now I have not the knowledge or time to pull this off myself, nor am I the only one who wants to achieve this. By acting as a central development environment for C, Obj-C & C++ (client services) much like Sun Microsystems do but for Java based apps.... well if you can't envision where I am going by now you shouldn't be reading this thread. :lol: 


In a nutshell:
Good news = Good stock price + Good publicity = More people attracted to product ----> Innovation.
Bad news = Bad stock price + Some publicity to potentially leverage.

It is really that simple.
... also my 'Hypothetical ToDo list' isn't 'that long' for a company the size of AMD with multiple international customers. (Heck that list is just the high priority stuff 8O ).

Still I can see ASP going down, but not too much, March/April it'll turn around, but it'll take 'large corporate customers' 6+ months to really notice.

March / April and October 2007 are going to be very interesting for us all.

If AMD did all the above, and had real balls (I only want a 45/45/10 split in the market, as a consumer and potential investor) they could use ballsy advertising - Much like Sun Microsystems did to compete with Dell. (The 'hot as 'hELL' rack servers).

But it would look something more like this (Mk I and Mk II are original Tabris:D arkPeace creations, the original is available, the irony is I used OpenOffice.org, derived from a Sun product, to make this):


Java Poster - The truth comes out [:p ]

Unlike Sun's 'Dell = Hot as Hell servers' advertising, I think both AMD/ATI and Sun would get a 'in joke' and present that to the community as a trio. Sun Microsystems use/sell more and more AMD Opteron based servers over time, but the software people make for those servers and "ZeroBugs (tm)" cross-platform interoperability is where many problems still are.

- Tabris:D arkPeace
- ZeroBugs (tm) is [not] a registered trademark of DarkPeace Heavy Computing Industries. :wink:
!