AMD announced its second-quarter revenue figures ahead of schedule on Thursday, and investors are not going to be pleasantly surprised. AMD’s revenue for the second quarter is expected to be USD 1.21bn, a 52 percent increase compared with the same period last year. However, analysts had been expecting AMD to record USD 1.3bn in revenue, according to estimates polled by Thomson First Call.
The whole financial carnival is such a shell game. AMB revenue jumps by 52 percent over last year, but baecause they didn’t meet the expectations of the “great unamed analysts”, there is to be gloom. What a pile of “Packard-Bells”. {If you played with PC’s about 10 years ago, you’ll get the reference]
On the other hand, AMD’s stock is down ‘only’ 1.59%, less than Sun’s stock (2% down). A bit dissapointing in itself, I had hoped to invest in AMD
As expected, there is a great deal of misunderstanding about how stocks are priced. Consider this: analysts predicted that AMD would hit $1.3bil in revenue. Investors acted accordingly, and the AMD stock was priced at a certain level, say $A. AMD only hit $1.2bil in revenue, _so the stock was overvalued_: $A was the price investors would pay based on the $1.3bil figure!
So now that AMD has failed to reach $1.3bil, potential investors have realized it was overvalued and are no longer willing to pay $A for it. Similarly, stock owners realized that AMD was not worth the market price $A, so they sold off as much stock as they could at $A before the price adjusted. Thus, the market price quickly adjusted to some value less that $A.
It’s not that AMD is being punished for not hitting some magic number, and it’s not a shell game. Price changes, up or down, are natural adjustments based on investors’ expectations.
Bah….Analysts…52% increase compared with Q2 2005 and they freak out because of a 9% drop compared to Q1…pfff….
Fab 36=300mm Fab is coming
Fab 30=200mm will be refurbished as a 300mm Fab
-> 4 times more fab capacity than before
this will help AMD Grow (no more fab capacity constraints)
Dell is coming
This will ensure that there will be enough customers for the chips built (if Dell is serious).
Opteron is a good CPU and Conroe production will be limited to 20% of X86 intel chips till december
-> Opteron sales won’t slump that much though P4 sales will.
sound AMD strategy
How conroe is priced against competing athlon will matter a lot…
And for the coming (little) price war, halving the cache size with a bump of speed to reduce manufacturing costs (without losing too much on the performance side) sounds like the sensible thing to do
K8L is coming and later HT3, Directconnect 2, etc…
-> AMD keeps inovating / bringing more features & IPC to the market.
Partners keep working with AMD.
-> AMD really builds there.
Having a socket compatible with sun’s sparc,
licensing it’s technology, allowing third parties to create silicon for specialized tasks and connect it to the CPU…
This should really frighten intel : There is a shift in the industry. Now, AMD is recognised as a serious player, big firms are closely working with it (microsoft, chipset makers, sun, ibm, hp, dell ?) and the cpu business is changing
For all these reasons, I think that AMD should have a bright future ahead, despite it’s slight disadvantage over INTEL’s manufacturing prowess (theyr are at 65nm and AMD won’t be before a few months yet).
On the other hand, INTEL won’t go away…but I can’t see how it could GROW in the future (concerning the CPU/chipset business). In fact, I think it can lessen the bleeding (market share loss) over the next year but, in the long term, It’ll still lose more market shares.
As an AMD Fanboy and stock owner (lol…I lose money today..but…oh…well….stocks !), what I would really like too, is an AMD effort to build a good compiler for AMD CPUS. (to compete with Intel’s)
Edited 2006-07-07 14:45
As an AMD Fanboy and stock owner (lol…I lose money today..but…oh…well….stocks !), what I would really like too, is an AMD effort to build a good compiler for AMD CPUS. (to compete with Intel’s)
Better yet it would be nice to see AMD contribute to GCC to enhance the comiler’s capabilities to compile AMD64 code.
Quite right – a contribution to compiler development would be prefeerable (OpenWatcom, GCC, javac and so on).
Intel C++ Compiler might be a great compiler suite, but it is commercial and does not integrate seamless everywhere with no hazzle.
Having a socket compatible with sun’s sparc,
I’ve only heard theories. Anything solid on this?
No, nothing solid on that yet. I only heard theories too.
On the speculation side, there could be a ring of truth to it : sun has been working closely with AMD
(btw are they going to release some woodcrest servers ? )
and it could be interesting for both firms
(and the idea is so out of places, come on sharing a socket with sun !, that it should be true….lol)
On the other hand, it could be a complete dud.
Like that anti-hyperthreading thing, that I can’t understand : how could it happen ?
if there was an easy way to make 2 cores act like a single core 1.5 faster, they would have done it a long time ago !
As for sound plans, there are a few things I like about AMD nowadays :
1) I get the impression that they spend a lot of time trying to figure what would the smartest thing to do next.
I mean, they don’t have a lot of resources compared to Intel, so they really have to and they seem to manage.
2) they shut up and deliver which is one of the thing that makes them more credible (I guess that it would be nice to have more info on their future roadmap though).
As an AMD Fanboy and stock owner (lol…I lose money today..but…oh…well….stocks !), what I would really like too, is an AMD effort to build a good compiler for AMD CPUS. (to compete with Intel’s)
You do realise that SUN Microsystems make a x86/64 compiler specifically optimised for Opeteron/AMD64, and is completely free to download and use – available on both Solaris and Linux; too bad there is no Windows version.
You got to admit that these F.A. who don’t seem to produce any useful products themselves but complain about a company with 52 percent growth are a bunch of dips. Hech, I don’t believe the computer industry itself is anywhere close to that either.
Umm, hello. They already did that, how do you think AMD64 support got into GCC in the first place? AMD partnered with SUSE to make it happen.
Now if you want some of that fancy stuff intel does like auto vectorization etc, then I can see your point.
The AMD64 Port and extensions for GCC are not nearly as heavily optimised as the ICC Compiler on Intel chips. There can be greater than 20% improvement in performance just by using ICC over GCC, that much performance on a large scale web farm could mean saving tens of thousands of dollars or more, Its unlikely that many people go down that road but the options are there. AMD really need to start to invest in the optimisations which will bring AMD upto speed. Especially as they are being used more and more on super computers et al.
Most of that improvement comes from ICC being a simpler compiler (it does C/C++ to x86/x86-64) compared to GCC, which handles far more languages and architectures and pays a performance penalty for it.
Without getting into a flame war over licenses, I’ll just point out that Intel’s icc and Sun’s cc are good optimizing compilers; by comparison, gcc is mediocre. You get what you pay for.
All that said … autovectorization only helps in benchmarking. Any real world code either has so little vectorized code that there is no difference, or is already hand-tuned assembly, or has to be sufficiently cross-processor-generation that processor-specific optimizations like autovectorization are turned off. The market for benchmark-inflating but not useful optimizations isn’t worth AMD’s time.
(A supercomputer is going to use BLAS or some other optimized math library which already has any relevant vectorization; AMD has BLAS for their processors.)
With early Conroe test leaks, staged by Intel, many of potential AMD buyers simply decided to wait a bit more for Conroe, officialy Core 2 Duo, myself included
Performance and power consumption gains over X2 and FX series of AMD chips are too significant to be ignored. In some cases ~50% for cheaper Intels.
Since all of tests have now been confirmed there is not one reason to buy AMD over Intel unless You are fanboy, but fanboyism is long gone in real world.
AMD made several big mistakes, the worst being underestimating Intel. They will not have K8L till 2008 by latest reports and transfer to 65nm will not help much, if only to lessen the performance gap. They will try with cutting prices down, but in the end their manufacturing process is more expensive then Intels since they are still on 90nm, so it is hard to beleive that they will be able to keep up with price race.
Early conroe/woodcrest adopters are big CG studios that in the begining helped K8 to penetrate high-end market and move to high performance servers, now it’s the other way around with Pixar leading the transition to Intel.
These will be 2 very hard years for AMD, I hope they make it through with something like K8 was.
The areas that AMD has learnt better than Intel ever will is in the area of spare parts, Having the same socket and motherboard etc for entire generations of processors is much better than what you get from Intel where many of the interim processors XEON 2.4 is different from a 2.6 etc in the actual socket, We switched many of our servers over (and the process continues) to AMD primarily because we can hold much fewer spares and still use the same boards. I know that is about to change with the Opteron rev F but its still been so good for the last few years that there is no reason for us to move away. Its especially important as we do not buy gold service etc from our hardware providers because we have to have as close to 100% uptime and those pesky 4hour replacement times are not good enough. Much cheaper to buy a pile of processors and ram etc and run with it.
Performance and power consumption gains over X2 and FX series of AMD chips are too significant to be ignored. In some cases ~50% for cheaper Intels.
…Since all of tests have now been confirmed there is not one reason to buy AMD over Intel unless You are fanboy, but fanboyism is long gone in real world.
You miss the fact, that by the end of this year AMD will largely transfer it’s production to 65nm, where it’s claimed cpu’s will have up to 40% increase in frequency performance alone. It will be available within half year. So much of Intel performance crown (which BTW still belongs to AMD). Half year and it’s gone. And a switch from one technological process to another takes (even for Intel) about 1.5-2 years.
Since Conroe isn’t in retail yet (and we will have to wait for reasonable quantities on market), the whole point of excluding AMD from current plans is plain stupid.
AMD made several big mistakes, the worst being underestimating Intel.
Who made mistakes ? Let’s see … 1GHz threshold, x86-AMD64, Hypertransport, ccHT, Integrated Memory Controller. Indeed all shameful mistakes. Intel response you ask ? 4MB caches. It’s David vs Goliat.
They will not have K8L till 2008 by latest reports and transfer to 65nm will not help much
Not help much ? And who is fanboy here ?
These will be 2 very hard years for AMD
First – only half year, second – not so hard, since main AMD problem was lack of brand recognition and lack of manufacturing prowess. It now has DELL and new fab becomes operational.
Did I mention AM3 will be compatible with AM2 ? Will Intel do the same for me ?
sorry man but transfer to 65nm will allow AMD to move towards ghz celling at around 3.6 ghz, which is also estimated celling for conroe and woodcrest. That is nowhere near 40% performance increase. Now take conroe vs AM2 and You will see that conroe is clock by clock faster than AMD so when AMD switches to 65nm conroe will still hold the lead.
2008 is too late for AMD since Intel will have completely new platform by then.
Intel has no plans in changing lga775 in near future, that has been on market for 2 years now.
Past triumphs wont mean nothing if AMD You cant keep up with biggest rival, as it is the state of the fact now.
I think announcement made by nVidia and ATi about their chipset plans made very clear which platform will dominate next period till arrival of new architecture from intel and amd.
I wish that AMD wins this one but it simply isnt possible. Woodcrest smashed FX in every single server test. If I was company chief and could choose between building cluster with woodcrest and similary priced FX I wouldn’t think twice where to put my money.
Pixar is serious company, I dont think they would jump to intel if they havent had all their options investigated.
Last but not least, You are forgeting SSE4, from what I understood this will be a huge boost for conroe once software starts to appear with optimizations for it.
Just being realistic.
Edited 2006-07-08 01:59
Come on…
Amd leads then it’s Intel then it’s AMD and so on…
That’s called competition. In the past, it happened too and the situation was really worse at the time for AMD (debts, capacity consttrained, no industry recognition, etc…).
Sure, Intel will win (a bit) by becoming the performance leader again but that won’t mean the death of AMD.
The market is a bit more complex than that…
sorry man but transfer to 65nm will allow AMD to move towards ghz celling at around 3.6 ghz, which is also estimated celling for conroe and woodcrest. That is nowhere near 40% performance increase
(3.6/2.6)-1 = 38.46% – actually very close to 40% (when you consider Opteron instead of “entusiast’s” FX)
Woodcrest smashed FX in every single server test.
Except I think SMP, which on server market is pretty important, don’t you think. And if you consider server you would like to compare Woodcrest with Opteron, not with FX. FX is for workstation/gamers. How does Woodcrest compare with sets of 2+ cpu’s ?
Pixar is serious company, I dont think they would jump to intel if they havent had all their options investigated.
Hmmm…. if they havn’t had all those discounts from Intel…
Last but not least, You are forgeting SSE4, from what I understood this will be a huge boost for conroe once software starts to appear with optimizations for it.
Yeah – and those SSE4 will be so importing on server side. End on desktop/workstation we will have to wait and see what impact it will really have, if any.
Past triumphs wont mean nothing if AMD You cant keep up with biggest rival, as it is the state of the fact now.
Oh boy. Intel are trying to catch up with AMD – today.
I wish that AMD wins this one…
No you don’t. Just admit it.
Woodcrest smashed FX in every single server test. If I was company chief and could choose between building cluster with woodcrest and similary priced FX I wouldn’t think twice where to put my money.
The FX Intel test that I saw looked pretty dodgy to me. They had even overclocked the Intel processor.
Besides, it’s Opteron on servers you want to concentrate on. I don’t see Intel being too confident there. Once Opteron moves to 65 nm the fundamentally better architecture will show up and I’d bet on AMD moving well ahead again.
“With early Conroe test leaks, staged by Intel, many of potential AMD buyers simply decided to wait a bit more for Conroe, officialy Core 2 Duo, myself included”
Actually, it wasn’t only the Core 2 Duo chipset causing people to wait. AMD is also releasing a newer Athlon 64 architecture with support for DDR2 RAM so many people decided to wait and see.
“These will be 2 very hard years for AMD, I hope they make it through with something like K8 was.”
Not necessarily. AMD still has a superior architecture for SMP. If the market move to computers with 16 cpus/cores quickly, Intel’s performance will lag behind dramatically. If the market stagnates at 4 cores, it is not like Athlon is a poor performer either. AMD’s product line still is very competitive so sales will end up depending heavily on the costs of new systems.
Fair enough. Now here’s some benchmarks:
http://www.gelato.org/pdf/apr2006/gelato_ICE06apr_parallel_novillo_…
Check out the gcc4.2 vs icc 9.0 benchmark about halfway down. Yes this is largely parallel stuff, but if you’re talking scientific then this is your benchmark. I think this shows the kind of parity GCC is getting to. Still not all the way there, but it’s getting better.
Sure, there’s room for improvement — but nobody should be jumping on AMD’s case over 52% growth. Pretty darn good results, if you ask me. Show me other tech companies that are performing that well.
As intel has to continually increase the front side bus to improve the performance of their processors the use of faster memory etc is required. With the introduction of FBDIMM Memory the power gains that the Woodcrest et al make against the AMD based processors will simply be null and void, If the figures I have read are correct the 6watt per dimm makes the Intel cost more in the long run (when your talking about servers with 8->32GB which is the high end market AMD is so appealing in) AMD’s marketing on power of a whole server is still going to be a better bundle/power saving.. Times that by thousands of servers in a datacentre and that is BIG Money, Any CFO that is told he could save between 10k and 100k on power alone will just about always want to travel down that road..
Edited 2006-07-08 11:32
The funny thing about the intel compiler, is that is also generate code that run fast on amd64/opteron processores, so it’s not really a reason to use intel chip. You can use Amd chip, and still use the optimizations that Intel build with their compiler :}