“Advanced Micro Devices has hired JPMorgan Chase & Co to explore options, which could include a potential sale, as the chipmaker struggles to find a role in an industry increasingly focused on mobile and away from traditional PCs, according to three sources familiar with the situation.” Woah. Bad news for competition in the x86 space.
http://stream.marketwatch.com/story/markets/SS-4-4/SS-4-16464/
According to this (if I’m not mistaken), they are not selling.
Good to know. Although Intel has future competition from the entire ARM ecosystem, right now their only true competitor is AMD. If AMD were bought out (especially by Intel) we’d have a behemoth that would make Microsoft and Apple look like babies.
i don’t think they could even find a buyer
nobody in their right mind would go headon with intel, and intel can’t buy them thanks to some anti-competitive laws
I was actually thinking someone would want to buy them for the graphics hardware, and transition the CPU side to focus on mobile. That way they aren’t directly competing with Intel (yet).
Stranger things have happened, but you’re right, it probably wouldn’t be Intel. At least, not without some interesting courtroom antics.
Edited 2012-11-13 23:24 UTC
Nvidia buying AMD for the GFX part, ceasing AMD x86 line? (since the license is non-transferable*, and since Nv is focusing on their Project Denver ARM – where AMD CPU talent should come handy)
Now that would be some behemoth… and maybe even without many formal or legislative roadblocks?
*this x86-licensing thing is weird, though – apparently, Nvidia did sell x86 processors at some point: http://en.wikipedia.org/wiki/List_of_x86_manufacturers#x86-processo…
The license can be market-specific, for example embedded only? ~i386-only?
Edited 2012-11-14 00:11 UTC
–The license can be market-specific, for example embedded only? ~i386-only?– No
i386 is 1985 for the instruction set. Add 20 years. 2005 patents died on instruction set.
http://bbs.66club.cn/uc_home/space.php?uid=672797&do=blog&id=592924 The Same design as Nvidia x86 chip is still in production. Nvidia sold that design off to a different maker.
Embedded x86 mostly uses patent expired.
Right. Hmm, so in a decade AMD64 & SSE3 should be clear, and probably perfectly enough to run virtually all legacy stuff… (perhaps that’s what the Chinese aim at with hardware-assisted virtualisation of x86 in Loongson CPUs?)
Hi,
Microsoft could buy them, and then start using their CPUs and GPUs in things like Xbox, and then start optimising Windows for their CPUs while optimising their CPUs for Windows. They could ignore any new features Intel creates, and Intel would have to support Microsoft’s new features just to maintain market share (while competing against ARM at the same time).
I’m scared.
– Brendan
That would be awful.
Apple have tried many times relying on CPUs from a single manufacturer. It has arguably never worked well on a large scale except with Intel. So why would Microsoft bother?
Their tight control on UEFI sounds much more worrying to me.
Edited 2012-11-14 09:50 UTC
Apple’s CPU suppliers had other businesses besides CPUs, and they weren’t as focused as Intel is.
MS and Intel wouldn’t have to compete directly. MS could use Intel consultants and fabs. Plus, the two have a long history, so I would expect it to be more of a mutual thing rather than antagonistic.
And isn’t AMD semi-supportive of OSS alternatives to UEFI? Now, that would be a nice thing for MS to kill off… ;p
MS don’t need to kill OSS alternatives to UEFI: they control the UEFI spec, and they can force pretty much any mobo manufacturer to implement it through the Windows monopoly, that’s more than enough.
Edited 2012-11-16 06:46 UTC
What about Samsung? They have more money and power than Intel has. That sure would be interesting.
And they are innovative, competitive and they have fabs. Samsung would give Intel some beating.
Well, lets hope someone will continue doing just that so Intel won’t have a practical monopoly.
Don’t Intel already have a practical monopoly on x86 processors, ever since they have released the Core 2 line and AMD have found no good answer?
No.
I beg to differ, my netbook is quite handy with a dual core AMD Brazos processor.
The only ones with a proper OpenGL support for netbooks.
As someone that does graphics programming as hobby, I’ll never own an Intel GPU, it is already enough to endure them on my work laptop.
Yup, I have heard similar praise of AMD’s IGPs, but as far as I know they have also been late on the CPU front for a while now, both in terms of performance and power consumption, which is why there are almost no notebooks with AMD processors around nowadays.
When it comes to general usage, what does “late on the CPU front for a while now” even mean? Because CPUs have been more than powerful enough, in such usage, for a while now.
There are only really two home PC areas which can be still starved for power – gaming and video editing. In both, the GPU kinda matters more; so in both, AMD is the one with an edge, thanks to more capable integrated GPUs.
For the low-performance market, I believe AMD chips also perform worse on the power efficiency front, though I haven’t checked in-depth benchmarking tests in a while. Again, might explain why almost no middle-end to high-end notebook uses their stuff.
It seems to me that you’re forgetting one thing: anyone who really needs lots of GPU power won’t be satisfied with the low-end stuff that gets integrated in CPUs, whether it’s done by Intel or AMD.
In the dedicated GPU area, AMD are competing with NVidia, not Intel, and it’s a different battle altogether.
The AMD APU line that makes up their mid to bottom end lineup are actually very competitive with performance per watt due to having rolled the GPU and northbridge chips into the CPU, so the rated power draw for those chips has to take that into account.
As for those that need more oomph, take a look at the new stuff stating to filter out now, like the MSI GX60 laptop. They are taking AMD’s fastest mobile APU and pairing it up with AMD’s fastest mobile dedicated GPU, it’s getting you plenty of CPU power for anything you might be CPU bound in as well as giving you a ton of GPU performance when you need it as well as still being able to have good battery life as t can turn off the HD7970m and just use the APU. On the flip side when you need that performance the APU’s GPU can be used for GPGPU physics in the same way that Nvidia uses PhysX in games or as just another GPGPU in your GPGPU capable software.
More AMD laptops are just now filtering to the market as well, Everything kind of died off with AMD launching new chips either due to chip shortages, companies holding off on orders till the new chip dropped and whatnot. But now that the new stuff is starting to hit the shelves it should be picking up again from the low end tablet and netbooks with the Z-60 to the mid range laptops with the A10-4600m to the upper mid range desktops with the FX-3850.
Well, let’s see how it pans out then. I’d be happy to see AMD back into the game, having the Intel name written everywhere on new computers is kind of depressing.
It all depends where you look for stuff.
I want to do graphics programming on the move, as such I took the time to research where I could buy netbooks with proper hardware.
In Germany there are plenty of shops that will sell you one.
I’m fairly certain they ship internationally http://www.xoticpc.com/
The Asus N56DP-DH11 or MSI GX60 both have the fastest A10 APUs in them, the GX60 though has an HD7970m as well as a few bits that may not play well with non windows OSs, but it can be fitted with 3 HDDs in only a 15″ laptop.
What Kivada says (Brazos does fine in netbooks), plus this isn’t about those “who really needs lots of GPU power” – in plenty of usage scenarios it’s more about “good enough” while keeping costs down; AMD offering generally nice bang-per-buck.
For example, once that video editing app offers realtime performance, you don’t really need greater performance that much. It’s also nice to be able to comfortably play more games on your inexpensive AMD laptop, than on an inexpensive Intel one.
Yeahm somethiung like the Asus N56DP-DH11 is more then enough for the vast majority of people CPU wise and the HD7660G GPU http://www.notebookcheck.net/AMD-Radeon-HD-7660G.69830.0.html is much faster then the Intel HD Graphics 4000 http://www.notebookcheck.net/Intel-HD-Graphics-4000.69168.0.html without the added heat, power consumption or noise of an addon card nor the driver issues of the Intel/Nvidia solution of trying to manage the GPUs from 2 different vendors.
I see large hosting companies buying systems with AMD processors for doing virtualisation (where they actually prefer more cores instead of really high performance cores).
The real question is whether anyone can buy them (practically)… From information disclosed after their 2001 lawsuit settlement with Intel, their x86 license is non-transferable. They in fact had their license agreement amended (after their 2009 lawsuit) just to make Global Foundries legitimate, since originally they were restricted to manufacturing x86 chips themselves (i.e. they could not use a 3rd party).
http://download.intel.com/pressroom/legal/AMD_settlement_agreement….
There is alot of history between the two companies when it comes to x86 licensing, going back all the way to the eighties… Suffice to say that, as things stand now, a buyer of AMD would not be able to manufacture x86 compatible chips without negotiating their own license with Intel, at least not anything beyond the 486 instruction set.
That said, imo, the only buyer than makes any sense at all is someone who already has a current x86 license. Which unfortunately, afaik, is no one. Really, the only current, full, non-expiring x86 licensee of Intel is AMD, everyone else is limited to manufacture only licenses or has a limited/partial license.
VIA, for example, negotiated a 10 year license in 2003, so their license expires next year. They may be able to re-negotiate it, but I don’t see how they could scrape together the money to afford AMD anyway. Outside of VIA? There isn’t anyone I know of left… IBM may still have a license, but I doubt it covers some of the newer stuff in the ISA. Nvidia? Nope.
AMD’s only bankable assets are skilled worker and a GPU business – their CPU line will likely die with them. Ironically, the only way that AMD could sell their x86 IP to someone else is if Intel brokered and blessed the deal…
Unless someone does a reverse triangular merger, whereby the purchaser creates a shell subsidiary that purchases AMD and then merges the subsidiary into AMD. In this construction AMD becomes a wholly owned subsidiary, but it survives as a business entity and all non-transferable assets endure.
If the Intel – AMD licensing agreements don’t have specific verbiage about this scenario, someone could buy AMD and keep the x86 licenses intact.
It’s the same trick Attachmate used to acquire Novell and keep Novell intact.
Read the agreement…
I am certainly not fluent in legalize, but I interpret item 2 there to specifically address a triangular merger scenario.
Additionally, if I understand the agreement correctly (big if) – a buyer would be able to continue manufacturing existing AMD designs as the dissolution of the agreement would not retroactively affect existing products… But without a new agreement with Intel they buyer could not design new microprocessors because they would no longer have license to do so.
Also, interestingly… the “existing design” clause explicitly limits the buyer to “Personal Computer” microprocessors, which is defined in the agreement to be servers, workstations, desktops, laptops, tablets, netbooks, and notebooks but not smartphones, cell phones, pocket pcs, or consumer electronic devices. It does not mention game consoles, but I would think that would fall under a consumer electronic device.
That means (again, imo) that the buyer could not sell any AMD microprocessors (even existing ones) for use in game consoles… I.e. they would lose the ability to sell Microsoft processors for the xbox. For the 360 that is assuming there is IP in the xenos GPU that is covered by license form Intel. I certainly don’t know the answer to that, but I wouldn’t be surprised if there is. And rumor was that AMD was making a fusion derivative for the xbox 720 – that would definitely be off the table.
Edited 2012-11-14 18:23 UTC
Samsung would like to take Intel on. They are already hinting at that direction with the new Chromebook sporting an Exynos 5 processor.
Samsung actually has the resources and fabs to compete head to head with Intel, so they could be a very dangerous competitor.
ARM != x86(-64). Don’t compare apples to oranges.
Besides, the x86 war is pretty much over, and Intel is the clear winner, so it doesn’t matter if AMD survives or not. The next war will be fought over who has the superior instruction set, the ARM conglomerates (Samsung, NVIDIA, Apple, Qualcomm, Broadcom etc. etc.) or Intel with its x86-64.
That is one of the reasons I’ve been urging people to put their money where their mouth is and buy AMD when they need a new PC or laptop.
The other of course is that frankly the “bang for the buck” on AMD chips is just nuts and probably less than 5% of those actually buying PCs have tasks that would benefit from the IPC advantage Intel has. you can get AMD quad kits on Tiger starting at $150, and you can buy Thuban X6 chips for $105. heck I even love gaming and my X6 just blows through the latest games while giving me plenty of cores for other tasks, truly a great chip.
So next time you are buying a PC or laptop take a serious look at AMD, I’m old enough to remember what it was like when Intel stood alone and it was NOT fun, it was stagnation and high prices. I put my money where my mouth is and my entire family is on AMD,we’re talking 5 desktops, 1 laptop and 1 netbook, and I can say all of them do every task we can come up with with plenty of cycles left over and plenty of cash saved over going Intel.
Yes, people do not realize this. I can easily collect a PC (especially now, considering black Friday is near) using AMD components for less than $200 (including case, PSU, motherboard, CPU, RAM, GPU, excluding HDD and monitor). But the cheapest Intel option would be almost twice that.
If you want the absolute top performance, Core i7 has no competition, but if you want to most bang for your buck AMD is the best choice.
Could this be the long awaited beginning of x86’s downfall? I’m ambivalent, but it would make a whole lot of people happy, for whatever reason.
I’d be a lot happier if DEC was still cranking out Alphas.
What advantages would you say the Alpha has over something like ARM or MIPS, or SPARC even? I know ARM doesn’t have 64 bit yet, but the others do, so assuming ARM64 is just like ARM…
Arm64 is a nice odd ball. It requires less silicon than its Arm 32 bit relations. Uses less power per core than its 32 bit relations and is faster. Yet it also has backwards compatibility support.
Arm pulled of some nice creative design voodoo with the Arm64 bit design.
Arm64 bit also allows you to probe and find out what soc chip you are in.
If that’s the case, I wonder how viable it would be to sell nothing but ARM64 cores but have some branded as 32 bit until the user pays for the upgrade.
I wouldn’t be surprised if the ARM64 is not cheaper to produce than ARM 32-bit.
Part of the improvements probably came from using going with a smaller process, for example from 32nm to 22nm.
I just wonder if a company focused on nothing but the ARM64, would their fabrication or production process be heavily simplified by being able to focus their lines on just one design rather than multiple designs, especially if the ARM64 is going to be indistinguishable from ARM32 when running in 32 bit.
I don’t know the specific answer to that question. What I do know is that alphas were a generation or two ahead of x86 back in the 90’s in terms of performance and they were effectively killed by politics as soon as compaq and HP merged.
Sparcs were better too, but Sun/Oracle have always ben kind of a pain to deal with. But yeah, if Sparcs were doing better than they are now for workstation/desktop class computers I’d feel better.
MIPS? Don’t know much about the arch other than its not really around, other than the Chinese processor and in the embedded side everyone seems to be choosing arm over mips.
ARM? I do kind of feel better now that arm is making its way up from the bottom. It will be interesting to see if they can really match x86 on the higher end rather than just being more energy efficient.
I remember reading Sparc was hamstrung by it’s sliding register window. It gets wide really well because of the window, but that window keeps it from scaling to high clockspeeds.
Early to mid-90s …but IIRC P6 largely bridged the gap. BTW, K7 was designed also by people who came from Alpha team, it uses the EV6 bus of Alphas (there were some plans for Alphas on Slot A, IIRC – too bad they didn’t come to fruition, it could be interesting)
MIPS is used in plenty of routers and such.
Generally, WRT x86 versus ARM – curious that the former started as something meant really for the embedded market, the latter as a desktop processor for Acorn Archimedes …but they both came to dominate the “other” market.
Samsung did ship a series of motherboards using the irongate/irongate-2 chipset (AMD 751/761 Northbridge), the UP1000, UP1100, and UP1500. They were, for all intents and purposes, pretty much identical to standard Athlon motherboards – they were PCI/AGP boards using standard PC southbridge chips. They just had different sockets (originally Slot B and later Socket B – which had more pins than the standard AMD packaging design to hold Samgung fabbed Allpha EV67/EV68 processors).
Edited 2012-11-16 21:40 UTC
Ah yes, Slot B; though Socket B is new to me. And did I remember correctly about some plans for Slot A (and/or Socket A?) Alphas? The world of computing could be somewhat different if we had those…
Oh well, too bad the Itanium was so successful… (in persuading most of the competition to exit the market)
With everything converging on ARM it’s no big deal if x86 dies and would be better.
tablets and smartphones are all shiny and trendy, but are hardly “everything”..
a larger slice of real world IT (from servers down to workstations, desktops and notebooks for work – not to mention pc gaming -and industrial PC’s) all rely on X86 and 10+ years of application compatibility
i for one would have liked the “standard” platform to be one based on a cleaner ISA rather than an ugly one like X86, but that wont happen any time soon
(not even with AArch64… before 64 bit ARM becomes the norm in servers it’ll take a while plus some iterations of multighz + massive multicore / threaded high IPC out of order ARM cpu’s at least as fast as the X86 processors that dominate now…)
Edited 2012-11-14 16:50 UTC
AMD and Sun should have merged and gone into business selling ARM based hardware for the enterprise. Its only a matter of time before ARM starts making in roads in the business world.
Enterprise would prefer all servers to be x86 which is a big reason why Sun went tits up.
There are enough problems as it is with software fragmentation, tossing ARM into the mix only compounds it.
As I have said before server hardware is a solved problem. A $200 quad core cpu is a solved problem. It’s peanuts in cost compared to software and administration. Big corps don’t even look at the price of the cpu, they hand a company like Dell a blank check when it’s time to upgrade.
Edited 2012-11-14 17:16 UTC
Actually, where I see interest in the enterprise for ARM is in the “Just enough hardware” segment. Sure a quad core is cheap. But a room full of them draws a lot of power and requires a lot of cooling. There are plenty of cases where a server is needed but anything x86 you would buy is likely overkill. I don’t see ARM replacing x86, but supplementing it.
Edited 2012-11-14 19:01 UTC
The problem is that dealing with a few ARM related bugs or issues will offset any power savings. There are enough issues with running LAMP stacks on alternative distros, toss in the ARM factor and the problem becomes exponentially worse.
I could see a company like Google looking into ARM since they run server farms and manage their own OS but don’t expect the typical corp to have any interest. ARM power savings are more meaningful to mobile devices, for servers you’re talking about turning off a few lightbulbs. It’s peanuts for the typical corp.
Maybe it’s true that where there’s smoke there’s fire, but there are no actual news yet and, as I see it, these rumours should be treated as such rather than a matter of fact.
Think about it: Reuters starts with “Advanced Micro Devices has hired JPMorgan Chase”, which seems to be a matter of fact, only to conclude with “according to three sources familiar with the situation”, which is NOT a matter of fact. And no, citing “three sources” instead of one or even the generic “sources”, doesn’t give it any more credibility.
Also, in the “article” there’s a statement from AMD (which I sort of missed the first time):
“AMD’s board and management believe that the strategy the company is currently pursuing to drive long-term growth by leveraging AMD’s highly-differentiated technology assets is the right approach to enhance shareholder value. AMD is not actively pursuing a sale of the company or significant assets at this time.”
Oh, and if we ever needed proof that Wall Streets analysts are fecking eejits rather than just crooks, mentioning Intel as a possible suitor should make it clear once and for all.
RT.
Lets see how this plays out for Linux support, if they are really bought.
I would not be surprised if ARM buys AMD, especially now that AMD has announced it will build server chips based on ARM cores and AMDs interconnect technology. It may be precisely this technology that could make AMD attractive to ARM.
Though the x86 license is non-transferable, I believe this means AMD can’t sell the license to someone else, but if someone buys AMD as a whole, I would think the license would be part of it. I would expect a model similar to the MIPS take-over: A group of companies led by ARM would aquire the patents while the x86 business is split off from this, retaining the AMD identity to keep the x86 license but with new ownership. Not unlike the recent MIPS takeover.
how come could AMD exist and be successful in pre-Athlon times when it only produced chips that performed worse than Intel’s but were price competitive and now after it has developed (relatively) much better stuff in CPU, innovated interconnects, has leading GPU tech it struggles.
In addition while PC market is on the verge of slide down it’s still much larger than it was then.
Edited 2012-11-14 11:26 UTC
both Phenom’s and Bulldozer’s have higher TDP but lower IPC and real world performance (especially the second, with its statically partioned multithread architecture – static partioning is generally a bad idea btw) than Core’s, this makes current FX’s chips that few people would want to buy…
AMD really should have fixed their architecture’s weaknesses asap, and churned out new models like there’s no tomorrow (since their competitor hasnt rested on its laurels either) – instead it seems they’ve even taken some sort of development pause, like a sabbatical…
plus, they had a much better management back then…
http://semiaccurate.com/2012/10/15/amd-is-imploding-because-managem…
It could be a perception problem. AMD makes good silicon, but in comment threads on the net you find a lot of vitriol against AMD. If you read some threads, you’d believe Intel made something near sentient with the Core architecture and AMD went back to 486 with Bulldozer.
AMD makes good performers with very reasonable pricing, but since they don’t take the performance crown (Intel does with a much higher price tag), you have a lot of people proclaiming that AMD sucks and they are getting an Intel part.
I wonder though if they really go for the part that squarely beats AMD in performance and pay the premium for it, or they go for a cheaper part that could have been equally filled performance wise by an Intel or and AMD and then bask in the idea that they got a chip from the company that makes the fastest chips (just not the particular chip they actually bought).
1) Because when AMD sold the K6 and K6-2 a PC that would work at all was relatively more expensive than it is now and saving a £100+ was a big deal. Saving £20-30 at the low-end isn’t now.
2) The margins are so tight that the small companies making mainstream PCs have left. The big guys tend to favor Intel for everything. Only when the Athlon 64 was leagues ahead did AMD become a must have.
3) AMD keeps investing and hoping they will have an architecture to at least draw level with this years Intel release. By failing to do so their CPUs that would be premium products and earn them profits/pay bills have to be sold at knockdown prices to keep their hand in the game and survive another year.
The same in true in video cards market, if you don’t have the full range or halo products no one notices you at all. Nvidia and AMD are very close and competitive. Everybody else S3, Matrox, etc have died a death.
It’s only because AMD is willing to make no money to prolong death that they are still in the game at all.
It is a shame as I would love to give them a crack.
Edited 2012-11-14 16:24 UTC
As far as AMD goes, I still think their biggest mistake was not building their own chip sets. Most chip sets were full of bugs. I remember I bought one of the first thunderbird motherboards with a Via chip set. It had known (after several months) problems with geforce 2, sound blaster live, was unstable as hell.
As time went on, all of my AMD systems died at some point and I threw them away. I have only had 1 or 2 Intel systems ever actually die on me. Usually I get rid of them due to uselessness. None of this is recent, but consumers have a long memory. Now it basically comes down to Intel always having the top performing hardware.
AMD had 750, 760 chipsets in early Athlon era (some data about them in http://en.wikipedia.org/wiki/Comparison_of_AMD_chipsets ). But IIRC they were quite expensive, people were mostly buying VIA-chipped motherboards.
Also, while the northbridges were relatively performant, AMD southbridges had somewhat limited functionality …resulting in a number of “hybrid” motherboards: AMD northbridge, VIA southbridge (made easy by the connection still used by then: PCI). Which reintroduced quirks typical of VIA back then…
i’ve never seen the 760 in the wild, but the 750 was realy popular
could be because it was one of only 2 chipsets for slot-a
I don’t think I’ve ever seen either of them live, period (probably because of being in a more price-sensitive place – mostly only the arrival of Duron made the K7 architecture really popular, and those were typically paired to, ugh, VIA KT133(A); at least VIA did improve over time …I keep here an AthlonXP 1700+ with VIA KT600, it had no issues)
Technically those two aren’t dead (yet?) – Matrox making some niche products ( http://www.matrox.com is there to check), S3 as a GFX component of Via chipsets.
Patents is keeping Microsoft from building a x86 virtual runtime environment for Windows RT.
Maybe it won’t be enough to buy AMD, maybe the legal challenges are to great and old cross licensing agreements will be broken.
anyway just saying
Microsoft is powerful enough to force Intel to the negotiating table. Microsoft buys AMD and tells Intel it wants the x86 license transferred.
Intel says no.
Microsoft announces that the next version of Windows will only run on Power.
Intel cr*ps itself…
The next day Microsoft announces it has changed its mind. Somewhere in the small print something about an x86 license is mentioned.
If a company bought AMD and lost the rights to the x86 instructions what kind of damage would be done by going 64-bit only? Obviously, most 3rd party Windows consumer applications and the current linux compatibility layer in FreeBSD would be affected, but what about the OSes and the instructions in the CPUs?
Does Windows 64-bit Vista/7/8 have 32-bit code still lingering that would keep them from running on AMD chips? Would we lose instructions like sse or do they have 64-bit versions? Most importantly, would any chips be grandfathered in so while new chips would lose 32-bit instructions, the new company could still produce the current line of CPUs while they make the transition?
I would see two problems with that:
1/AMD64 processors start in x86 compatibility mode as an architectural decision, so removing x86 support would effectively imply creating a new incompatible processor architecture (similarly to what was done in IA-64… and it bombed)
2/Even with fully 64-bit operating systems, most user software is still 32-bit. An x86 compatibility mode would thus be necessary in order to execute it.
Edited 2012-11-14 15:06 UTC
Plus, isn’t amd64 effectively a superset of x86? Meaning you can’t really implement the former without major parts of the latter?
Yeah, AMD64 also follows a logic that’s very similar to that of x68, but I don’t know if Intel’s license/patents/whatever prevents people from doing what they want with x86 go this far.
Edited 2012-11-15 07:30 UTC
Data is 64 bit, instructions can use 16, 32 or 64 bits.
For example:
mov ah, bh //16 bit
mov ax, bx //32 bit
mov rax, rbx // 64 bit
Intel patents cover instruction set so you can’t have x64 only cpus without infringing Intel’s patents.
So that’s why hardware patents are as bad as software patents because they hinder competition.
God forbid an Intel-only world!
I agree but on the mobile side there is still Samsung Enyo, and nVidia Tegra, and Apple A6. I know mobile != desktop but the success of those other companies in mobile isn’t helping AMD at all.
Mostly gone are the days that AMD was the yin to Intel’s yang. Even Intel is a late entry to mobile with Atom and Apple is talking about eventually using the A6 in MacBook Air. The duopoly is in trouble for sure.
What happened to VIA, still making processors?
VIA/S3 is still around, but to get any of their Nano, C7 or Eden CPUs or Chrome 500 series GPUs you will have to go out of your way to find them.
VIA/S3 has focused more on embedded x86 and ARM SoC designs.
Sure, you could get yourself a VIA Nano Quad core http://www.via.com.tw/en/products/processors/quadcore/index.jsp and an S3 Chrome 540GTX PCI-E GPU http://www.s3graphics.com/en/products/class3.aspx?productId=19 just to say you have the fastest VIA/S3 hardware on the market, but it ain’t saying much and even though they claim Linux support, don’t expect much, it’s best suited to Windows, but doesn’t like that much either last I checked…
ARM is and will be more efficient than x86. but that doesn’t mean it does or will matter.
intel has proven it can do VERY low power x86, and they are just getting started. they have had fully one (“1”) x86 chip low power enough for cell phones. just. getting. started..
amd selling out to samsung or whoever would be fine. amd has been the walking dead since they stopped being competitive with intel at the same time as the ARM ascendancy.
AMD is always one step behind Intel when it comes to efficiency. Intel invests heavily into R&D and a buyer like Apple would have to ramp up R&D to match which isn’t something that could happen overnight even if they were willing to spend the money.
It would be less of an issue for a company looking to build full server stacks with proprietary extensions or simply to put “works best with Oracle cpus” on the box. But AMD has a red balance sheet and I’m not sure how much of their losses could be cut by selling off unneeded divisions.
IPC sucks for AMD. AMD actually managed to release worse quad core processors than their old Phenom II line.
When Intel did really bad, in the P4 days, they decided to change the architecture, not to try to improve Netburst which was doomed.
AMD are trying to make small incremental improvements to the failed Bulldozer architecture. To what end will they succeed, we shall see in a few years. Either they recover fast, either they can fill for bankruptcy.
They recently hired back Jim Keller, the guy who lead the design for their Athlon 64 cpus. Let’s hope Jim Keller helps them make good architectural choices.
Aside for bad IPC, AMD is a whole process node behind Intel. They are on a 32 nm process, while Intel is on 22.
They have to coerce Global Foundries in changing the process much faster or try to negotiate with other fabs, maybe TSMC. Intel will switch to 14 nm in 2013, Global Foundries might stay on 32.
That’s not really how it was. When Intel did really bad with Netburst, it hardly hurt them thanks to the “influence” they had on OEMs (conversely, AMD was rewarded for its tech lead less than they should).
Intel was pumping out Netburst CPUs for over half a decade, including a major redesign, Prescott (basically deserving a Pentium 5 branding), which was “more Netburst than Willamette & Northwood” – and which performed quite poorly considering the amount of changes, how it required ~2x transistors.
Generally, IPC of AMD CPUs doesn’t “suck” …it’s worse, but they are often quite decent (especially when looking at the whole, including GPU; plus http://www.osnews.com/permalink?542521 )