Wednesday, November 28th 2007

AMD Drops Out of Top 10 Chipmakers

More bad news about AMD. According to the research firm iSuppli Intel will increase its market share by the end of 2007 to 12.5 percent, keeping its place as the world's top chipmaker, while rival AMD will drop out of the top 10. Samsung Electronics will remain the world's second-biggest chipmaker with 7.4 percent of the market while Toshiba will rise to third place, pushing Texas Instruments to number four, iSuppli said. Infineon will rise from 15th to 10th place thanks to its wireless business.
Intel successfully defended much of the market share that it won from AMD in the first quarter in the PC microprocessor segment due to the success of its lines of dual- and quad-core chips,
said iSuppli's head of market intelligence, Dale Ford. Read the full report at eWeek.Source: eWeek
Add your own comment

24 Comments on AMD Drops Out of Top 10 Chipmakers

#1
InfDamarvel
This is why I regretted them merging with Ati.
Posted on Reply
#2
HAL7000
by: InfDamarvel
This is why I regretted them merging with Ati.
Agreed............:toast:
Posted on Reply
#3
intel igent
by: InfDamarvel
This is why I regretted them merging with Ati.
no patience :shadedshu
Posted on Reply
#4
btarunr
Editor & Senior Moderator
Rumor: AMD is working on a new x86 micro-architecture where it's a 4 core RISC processor + 500 stream processors (derivative of the AMD Stream computing) + a native multi-level cache and HT 3.0. The chip runs at 2.5 GHz and is comparable to the CELL micro-architecture. No, this isn't Sandtiger. Another nut-kick for Intel is taking shape deep in the bowels of Dresden.

....but hey this is just a rumor.
Posted on Reply
#5
HAL7000
by: btarunr
Rumor: AMD is working on a new x86 micro-architecture where it's a 4 core RISC processor + 500 stream processors (derivative of the AMD Stream computing) + a native multi-level cache and HT 3.0. The chip runs at 2.5 GHz and is comparable to the CELL micro-architecture. No, this isn't Sandtiger. Another nut-kick for Intel is taking shape deep in the bowels of Dresden.

....but hey this is just a rumor.
Rumors, paper, press releases.....this is the nut kick for AMD....:nutkick:
This is why AMD sucks in todays market......unfortunately
Posted on Reply
#6
kwchang007
by: btarunr
Rumor: AMD is working on a new x86 micro-architecture where it's a 4 core RISC processor + 500 stream processors (derivative of the AMD Stream computing) + a native multi-level cache and HT 3.0. The chip runs at 2.5 GHz and is comparable to the CELL micro-architecture. No, this isn't Sandtiger. Another nut-kick for Intel is taking shape deep in the bowels of Dresden.

....but hey this is just a rumor.
Hey if that's true....it's good and bad. Good for the reasons that you get an a**load of processing power and if you ran graphics on that thing it'd be crazy as it has that many stream processors (granted if you had fast memory too). Bad in that things are just being developed for 4 cores, much less 504+. Second you need uber fast memory need to take full advantage of this. Third, it'll emit a ton of heat (think water cooling=necessary for overclocking) unless the fab size is dropped by alot. Fourth, that's alot of silicon they need=expensive=lower clocks on average due to worse speed binning. Fifth, alot of power consumption, think in terms of a 2900xt (even at lower fab size, it still has more stream processors) + Barcelona's 125 watt TDP for 2.4 ghz. So talking about 300+ watts of power (worse than presscot :twitch:) and heat. Not bad if that's cpu+gfx...but horrible if that's cpu only.

Back on topic things really aren't going well for AMD. But it's got to be remembered it's playing with big people and they really aren't that big of a company.
Posted on Reply
#7
mandelore
I think of it this way, Nvidia focuses solely on GPU / chipset technology and can pour all its money into that, Intel is the same, so AMD/ATI is fighting a war on 2 fronts, having to divide its resources to combat 2 giants throwing ALL theirs into the market they play in.

AMD is doing great to keep above water, and I do fear for ATI, but I know they will fight back, but having half the resources (if they were split evenly between gpu development and cpu development) is not a good position to be in
Posted on Reply
#8
erocker
I am losing money with AMD.
Posted on Reply
#9
cool_recep
Don't worry be happy AMD! They must have something to release..

They are like "Mother Nature"
Posted on Reply
#10
btarunr
Editor & Senior Moderator
by: kwchang007
Hey if that's true....it's good and bad. Good for the reasons that you get an a**load of processing power and if you ran graphics on that thing it'd be crazy as it has that many stream processors (granted if you had fast memory too). Bad in that things are just being developed for 4 cores, much less 504+. Second you need uber fast memory need to take full advantage of this. Third, it'll emit a ton of heat (think water cooling=necessary for overclocking) unless the fab size is dropped by alot. Fourth, that's alot of silicon they need=expensive=lower clocks on average due to worse speed binning. Fifth, alot of power consumption, think in terms of a 2900xt (even at lower fab size, it still has more stream processors) + Barcelona's 125 watt TDP for 2.4 ghz. So talking about 300+ watts of power (worse than presscot :twitch:) and heat. Not bad if that's cpu+gfx...but horrible if that's cpu only.

Back on topic things really aren't going well for AMD. But it's got to be remembered it's playing with big people and they really aren't that big of a company.
It's a general purpose processor with a major architectural drift from current. Four x86-64 cores. Each core utilizing "stream processors" to compute broken-down threads. So it can be a quad-core processor if you want it to be, if you send in more than four threads, no problem, it can change its hat into allocating groups of stream processors to deal with threads. This way, the core-speed stays real low. 2.5 GHz could be it's limit. You've got lots of cache. TDP? Well the HD 3000 series GPUs are 55nm fabbed and run reasonably cool (and energy efficient). but well again.....all this is a blur now.

A (very) rough schematic diagram

Posted on Reply
#11
Greatpumkin628
"Another nut-kick for Intel is taking shape deep in the bowels of Dresden. "

Are you serious? I already feel like poor Charlie Brown trying to kick that football and AMD keeps giving me the Lucie treatment. This long awaited/delayed/hyped "Prescottynom" space heater/CPU and now we'll wait till the clocks come up in Q/1 Q/? 08' or just you wait till' we get it to 45nm then.. oh boy. Then the bulldozer sandtiger yada yada...

They can't seem to pull off Phenom a year late... let alone this abomination of which you speak. I fear they simply are in over thier heads.
Posted on Reply
#12
btarunr
Editor & Senior Moderator
by: Greatpumkin628
"Another nut-kick for Intel is taking shape deep in the bowels of Dresden. "

Are you serious? I already feel like poor Charlie Brown trying to kick that football and AMD keeps giving me the Lucie treatment. This long awaited/delayed/hyped "Prescottynom" space heater/CPU and now we'll wait till the clocks come up in Q/1 Q/? 08' or just you wait till' we get it to 45nm then.. oh boy. Then the bulldozer sandtiger yada yada...

They can't seem to pull off Phenom a year late... let alone this abomination of which you speak. I fear they simply are in over thier heads.
Patient, son. The Clawhammer you're using was the thing that delivered the nut-kick last-time around.
Posted on Reply
#13
kwchang007
by: btarunr
It's a general purpose processor with a major architectural drift from current. Four x86-64 cores. Each core utilizing "stream processors" to compute broken-down threads. So it can be a quad-core processor if you want it to be, if you send in more than four threads, no problem, it can change its hat into allocating groups of stream processors to deal with threads. This way, the core-speed stays real low. 2.5 GHz could be it's limit. You've got lots of cache. TDP? Well the HD 3000 series GPUs are 55nm fabbed and run reasonably cool (and energy efficient). but well again.....all this is a blur now.

A (very) rough schematic diagram
Well I can see why it's good. Cause you can free up your main 4 cores by offloading everything else to the stream processors. But it's also a nightmare because either you need drivers, or really good firmware. Both of those are going to take up processor resources anyways. Then you're talking about lots of cache still=lots of space on die needed. Look at 65 nm, at 8mb cache Intel can get 4 cores in....just barely (I think that it's very little space anyways). For TDP and power consumption, I'd rather see it have a better cool n quiet aka shutting off like all but 100 stream processors. Oh and that's at 55 nm. AMD has been struggling with 65 nm and finally has nailed it. ATI I believe outsources its chips which is why then can bring it down to 55 nm. Even at 55 nm, think two hd 3870s....not an easy thing to cool and power when at max load. (TDP...ok maybe not 300 but like 200s which is still above stock Prescott)
Posted on Reply
#14
Darkrealms
Unfortunately for AMD when they were kicking Intel's ass, Intel still advertised and still sought large contracts and had the money to last. AMD has always had a problem with advertising (perhaps they should let ATI advertise for everything?). They also don't have the finances that Intel has, even before the merge with ATI. AMD has always had the "techie" feeling to them, never the "business professional" feel (another thing they may need to transplant from ATI?).
This is coming from an AMD/Nvidia fan, never an ATI fan. ATI just seemed to have a better business sense.
Posted on Reply
#15
phanbuey
by: Darkrealms

This is coming from an AMD/Nvidia fan, never an ATI fan. ATI just seemed to have a better business sense.
Completely Agree... advertising, time to market, designing product outside their production capability... AMD's business execs have made some pretty bizzare decisions.
Posted on Reply
#16
corwin155
dont matter

even though ill stick with amd/ati even if they are behind everyone else .
i like amd chips look forward to upgrading to phenom
My rigs for Htpc / light gaming
Msi k9A Platinum AMD 580X CrossFire (ATI RD580) Chipset
amd 64 EE 5200+ mild oc 2.8 ghz AC 64 pro 92mm cpu fan
3 gigs adata ddr2 800 @ 401.9
2x 250 wd sata2 hds
1x ATI HD2600xt gddr4 oc 845 core 1170 mem dx10.0
1x sony cd burner 1x samsung dl-dvd 18x
logisys 575watt psu
WinV home premium 32bit
CPU-Z Database (ID : 249565)
Posted on Reply
#17
F-22
AMD knows better than we do that they HAVE to do something awesome to get back in the game. Only time will tell.

Until then, I'll still be an AMD/ATI fanboy (even tho i have q6600)
Posted on Reply
#18
btarunr
Editor & Senior Moderator
by: kwchang007
Well I can see why it's good. Cause you can free up your main 4 cores by offloading everything else to the stream processors. But it's also a nightmare because either you need drivers, or really good firmware. Both of those are going to take up processor resources anyways. Then you're talking about lots of cache still=lots of space on die needed. Look at 65 nm, at 8mb cache Intel can get 4 cores in....just barely (I think that it's very little space anyways). For TDP and power consumption, I'd rather see it have a better cool n quiet aka shutting off like all but 100 stream processors. Oh and that's at 55 nm. AMD has been struggling with 65 nm and finally has nailed it. ATI I believe outsources its chips which is why then can bring it down to 55 nm. Even at 55 nm, think two hd 3870s....not an easy thing to cool and power when at max load. (TDP...ok maybe not 300 but like 200s which is still above stock Prescott)
Then again, the number of stream-processors will depend on models, higher model, more SPs. Secondly, did you know, AMD has a contract with Fijitsu that is a Japanese fab maker? So a lot of AMD processors today are made at their fab. Like NVIDIA, if AMD has to quantitatively compete with Intel, it has to get fab-less too. NVIDIA is fab-less. AMD has fabs in Dresden, Sunnyvale and Bangalore. Now it's getting to out-source it.

Drivers? No. The resource arbiters do the job of handling the SPs, not a software driver. The OS would recognize the CPU to be a "4-core, 512-threaded" CPU.(508 SPs). Stream processors perform elemental math operations very fast and thus don't require caches.The 4 x86 cores have dedicated L1 C, L1 D and L2 caches of their own. A large Level 3 cache can be speculated.

And this way AMD doesn't really have to worry about Intel coming up with a Core 2 Oct, if that ever happens
Posted on Reply
#20
kwchang007
Huh didn't know AMD outsourced most of their chips....maybe that's why they dropped out of the top 10?
Posted on Reply
#21
WarEagleAU
Bird of Prey
Who cares what they do. Im so tired of everyone pissing and moaning about what AMD is doing. Get off the bandwagon. They dont have to be the fastest or the best. Its just you folks got spoiled on them when Intel had the nutburst architecture. Give them some damn time, shit.
Posted on Reply
#22
Woah Mama!
Hmmm, somehow I think it'll take more then a triple core CPU built on cheap production lines and a shiny new video card to get them out of their $6,000,000,000 debts.

You know the show is nearly over in business when you are up to your neck in debt and have no conceivable way to get out of it. I hand it to AMD to last this long though. Further slipping from the forefront of technology... Yeah... With everything else that is going wrong for AMD, this is nearly icing on the cake IMHO.
Posted on Reply
#23
eidairaman1
i think AMD maybe working on something totally out of the ordinary, since multicore is commonplace.
Posted on Reply
#24
btarunr
Editor & Senior Moderator
by: eidairaman1
i think AMD maybe working on something totally out of the ordinary, since multicore is commonplace.
Read posts #5, #11 and #19
Posted on Reply
Add your own comment