The US Federal Trade Commission ruined Intel's Christmas by hitting the firm with wide ranging charges of anticompetitive business practices. The Feds also charge that Intel rigged its compilers to put AMD at a disadvantage and that it is trying to smother its rivals in the graphics chip business. Intel has rejected the charges …
Until Intel is forced to pay a $500 BILLION fine, nothing will change
Anti-trust lawsuits are just the cost of reaping profits from violation of law for the WinTel cartel. As long as they derive huge profits from crime they will continue their crime. When they generate tens of billions of dollars in profits per year from crime and get fined a few hundred million, it's all good at the house of crime.
Intel rigging their compilers not to use certain code paths even on CPUs that support them as long as they are non-intel is something which the industry has noted very long ago. FTC is about 5 years late here.
How is rigging their compiler illegal? It's their compiler surely they can make whatever decision they want about supporting the CPU features they decide are important. If they fail to support AMD's hardware AMD should write a compiler optimized for AMD hardware. All CPU companies write compilers that target their hardware. The only difference between Intel and others is that most CPU companies do not have to worry about supporting competitors offering binary compatible processors.
If you were managing a team of compiler writers at Intel how much time would you spend on developing features that Intel CPUs do not support?
What next sue Microsoft because their programs don't work on Linux?
Intel wrote a compiler optimized for their hardware. Is it Intels job to call up AMD and ask them for a whole list of their instruction calls so they can add it into their program.
Hell while were at it why doesn't Chevy sue Toyota for not making their parts compatible.
Intels done alot of shady things but the fact they spent THEIR money making THEIR own compiler optimised for THEIR hardware is probably the stupidest thing they have been accused of yet.
You miss the point...
The issue here is not what you think. There are features supported in both processors. I.E. Intel did the legwork to include those abilities in order to compiler for its processors.
They then "tweaked" their compiler in such a way as those features do NOT get used if the chip is not made by Intel.
It's not a case of not building in feature support for AMD, but rather one of denying features already in the software from working on AMD processors, thus ensuring the same code running on processors supporting the same features runs slower on AMD, and faster on Intel.
Interesting definition of 'Optimization' you're using...
As I understand it Intel's 'optimization' essentially consisted of code along the lines of
"If GenuineIntel, then run normally. Else run really slowly"
AMD and VIA were apparently finding that they could get performance increases of 10-20% on their processors just by changing their CPUIDs to say 'Intel'.
That's the sort of thing you can perhaps get away with if you are a bit-part player, but not when you have 90% of the market.
No, You miss the point..
This is not an ethics class. We are talking about a major US regulatory agency suing Intel for "damages" and using the compiler shenanigans as proof of systematic disregard of legal fair trade practices.
The question is to what extent if any Intel is required to support other x86 compatible CPUs with their compiler. I maintain that their was nothing illegal about Intel's action. They were under no contractual obligation to support AMD CPUs with the compiler. They are a private enterprise and their first duty is to maximize shareholder value and this sometimes precludes playing nice.
I agree what they did might be considered "underhanded" but they are competing in business. Why should AMD get a free ride and not have to pay the cost of developing their own compiler? Surely it was a bad decision on AMD's part to rely on Intel's technological largess and not develop their own compiler that fully optimizes for their CPU features in a way that only their design engineers could possibly accomplish.
Just because Intel was at one point supporting AMD chips better and then changed their compiler to support them to a lesser extent does not constitute material harm to AMD. This logic is like if I bought you a free lunch on Monday and Tuesday and then did not pay for your lunch on Wednesday would you say I had harmed you financially?
Again with missing the point.
This isn't about supporting the CPUs of their competitors, but purposefully crippling their software if it isn't run on their own hardware. This is most certainly underhanded, and if you are a monopoly, 100% illegal.
Anti-trust law is all about the use of your monopoly position in one area to edge out competition in other areas. In this case, Intel has several monopolies. At the time, (and even to a large extent today,) Intel’s complier is one. They use their near-absolute monopoly on CPUs to make their compiler the ubiquitous preferred choice, and they use their near-absolute monopoly on the compiler to cripple software working on their competition’s processor.
The issue gets even more stick when you descend into the intellectual property rights regarding who is allowed to write compilers for certain features. The holy grail of compilers would be one tool that would compile your application to work on all devices. Even restricting this to the x86 market, as a programmer, you want to push one button, and receive a binary that is capable of detecting which features a CPU has available, and using all of them.
You state that Intel has some innate “right” to set up their compiler to cripple the function of binaries it compiles when run on non-Intel CPUs. (Your argument is that AMD could just as easily write their own compiler.) The problem here is that AMD would not be legally allowed to write a compiler that took advantage of all the various Intel CPU features, so AMD’s compiler would not be able to produce a binary that would run full speed on CPUs from both companies. (The reverse, BTW, is not true, Intel has IP access to all features in AMD’s CPUs.)
Thus as a programmer you are faced with two options: compile a single binary using Intel’s compiler, which you are fully aware will run the same code slower on an equally-capable AMD processor, or produce two binaries, one for AMD, and one for Intel. Using different binaries per architecture adds huge layers of complexity, even more so now that everything is virtualised, and VMs can, and are, moved across architectures.
So the programmer uses Intel’s compiler, knowing that Intel has a larger chunk of the market, and their program then runs slower on an equally-capable AMD product because of the anti-competitive nature of Intel’s business practices.
If the market is under heavy competition, this is (sadly) a legal practice. This is, however completely illegal under anti-trust law, as it is abuse of a monopoly to a) extend your monopoly into other areas, and b) maintain that monopoly by restricting competition.
You may not like it, but this is the law in several countries around the world, including the US. Intel got spanked for this, and similar practices in several countries and it is now up to the lawyers to prove everything I just said above actually happened. The entire industry knows that it has, but without evidence Intel can’t be spanked for it. If the proof still exists, and the documents weren’t destroyed, then Intel is in some very deep poo.
@frankg778 Seriously, you missed the point...
So, frankg778, by your (flawed) logic, if i ran a software house, like, lets say microsoft, i would have to compile my software twice, and sell 2 versions of it, one for intel and one for everyone else, despite the fact that both versions are x86? what a nightmare that would make.
Then there's me the network administrator, who uses servers with both AMD and intel processors, each having their strengths, so by your logic, i would have to have 2 different versions of each of the apps i use, installed on different versions of the same OS, one for intel and one for AMD, solely because Intel decided to cripple their compilers.
Got any more retarded statements you'd like to throw out there?
Re: Seriously, you missed the point
Look I do not want to upset you but you are sort of living in a dream world. Intel is not a compiler company they are a CPU company. If you want to trust in Intel and then feign astonishment when they pursue their naked self interests then good for you!
But the assertion that Intel has a monopoly on x86 compiler development is just wrong. I understand that developers are lazy and dp not want to have to deal with multiple binaries. So go ahead use the dominant companies tools and believe their lies. Do you think AMD would have done anything different if the situation was reversed?
What companies like MS could do if they cared to optimize for all vendors CPUs is compile the installer with vanilla x86 instructions detect AMD or CPU in the installer. Then copy the correct binaries onto the users hard-drive. This calls for a lot of discipline in library design but it can be done. The problem is most software projects are habitually late on their deadline so this extra effort is seen as impossible in reality.
Just remember as you complain about the Wintel monopoly that when you use Intel's compiler exclusively you are helping to ensure that the monopoly lives on. What if your users actually care about performance and you have a competitor that knows how to ship multiple binaries?
Here's a list of C++ compiler's for anyone who cares:
Re: Seriously, you missed the point
Don't worry no one expects a network admin to be able to figure this stuff out, that's why all windows software comes with an install program. During your work as a network admin have you ever encountered Linux? During Linux installs they typically test the target system and optimize for the installed hardware. Although you may think it impossible, unpaid OSS developers make it happen all the time.
I am a developer from an embedded background, and the only thing that strikes me as "retarded" about this conversation is that you believe one company should supply a full featured compiler for another company's silicon that they make no money on. The fact that Intel could do this for free by virtue of AMD's decision to implement identical capabilities and implementations is irrelevant to the economic argument. AMD was attempting to get a free ride from Intel and they got burned. Somehow AMD survived through the pro-business Bush administration and now the Obama administration may save their bacon, it still does not make AMD look smart.
BTW here's a list of C++ compilers, notice the other companies besides Intel.
Now, please correct me if I am wrong, (with facts and documentation,) however no corporation or individual is allowed to build a compiler for Intel's processors that takes full advantages of all instruction sets. Some are open enough that anyone can write a compiler, (the generic x86 stuff,) some Intel will sell you a licence for, and some Intel absolutely refuses to grant licences for, as their method of insurance that you will choose Intel's compiler for all your binary generation needs.
So this means that your alternatives are to produce multiple binaries, (flat out not an acceptable solution,) or to compile your binaries only for Intel chips, using Intel's compiler. Otherwise your binaries will take a performance hit when running on the majority of x86s CPUs in the world.
That's why it's anti-competitive.
Oh, and about Intel not being a compiler company, because it's a CPU company? It's both. It's also a flash company, a GPU company and many, oh so very many more things than just those.
Also, in the interests of full disclosure, I have to ask…do you work for Intel, one of it’s major partners or their affiliates?
Curious about the FTC case
How did Intel alter their compiler to put AMD and other CPU firms at a disadvantage? This is a complex issue because AMD and Cyrix were trying to make x586 chips that could execute classic i486 instruction sets faster than a 486. Clearly Intel will optimize for their chip set. Why in Intel under any obligation to support other people chips with it's compilers? Was this part of a previous agreement?
I think Intel may have stifled competition but they also delivered a lot of value to end users. The true cost of inhibited competition is probably incalculable because we have to imagine how PC architecture would have evolved with a less dominant Intel. The discount strategy of requiring OEMs to use nearly 100% Intel seems like a stronger point in FTC's case. This may be illegal typically you get bulk discount for purchasing a certain lot size not a certain percentage of total inventory.
However I keep going back to who were the victims of these monopoly practices? It will be interesting to see which if any OEMs are cooperating with the government.
The issue concerns post 486 extensions to the set
For example, it is possible to use MMX, SSE, SSE2, etc as MOV for multiword instead of REP MOV. Intel compiler uses that only on Intel. It is totally disabled on Athlons despite the CPUID being able to provide full information on what is supported and whatnot. Similarly, there is a number of other optimisations which the compiler allows only on a genuine intel CPU. In total for some integer and single precision FP code the speed difference can approach 50%.
The fact is Intel has never been obliged to support non-Intel CPUs here. So the only way the FTC can have a case if it has gotten its hands on internal docs which specify why it was done (the fact that everyone knows it is a different story). Considering that these docs should not exist as per Intel document destruction policy makes this even more interesting (someone has a big grudge here).
In any case all linux distros have these optimisations as loadables/alternatives for libc. Similarly, key routines in the linux kernel use them as well. Not sure about microsoft, but I would expect them to have implemented something similar as well. Any advantage icc was providing was purely temporary. It was a welcome relief to Intel during the period when their development was crippled as a result of moving it to a country division which was more interested in fiddling expenses than delivering. It helped them keep the high end of the market and the embedded part even when Opteron was beating P4 into a pulp on all benchmarks. It is not that much relevant any more. After moving their CPU design to Israel Intel has pretty much caught up with AMD. From being a hot topic on most mailing lists about 5 years back icc is now something that is rarely mentioned.
Thanks anonymous coward
Thanks for the detailed explanation.
But even if they somehow find a "smoking gun" memo that says "cripple MMX extensions on AMD chips" is that really against the law? This seems more like a PR stunt then a salient business/legal issue.
Here is a good write up of the problem http://www.swallowtail.org/naughty-intel.shtml (I think that is the right URL, work proxy blocks it here). I know it is http://www.swallowtail.org
reaping what they sow
Glad to hear that Intel's settlement with AMD didn't end the matter.
@frankg778: victims include everyone who couldn't get AMD gear through their normal supplier during those years when AMDs products were faster and cheaper, everyone who hasn't been able to get ION based netbooks, both because of intel's threats to break discount agreements and because of intel's practice of selling atom + chipset cheaper than atom alone (go figure). Everyone who is potentially denied high resolution, high performance netbooks because intel has used its market position to define the netbook segment as not capable of high-def.
This is a pretty clear case of a repeat offender using their market clout to hold back innovation on all fronts. I just hope that the FTC's action will help clean things up. Big brother, 'cause he's got it right for once.
The victims are AMD and consumers
You paid a lot more for Intel crap because the distros and channel was blackmailed by Intel. And Intel laughed all the way to the bank to the tune of Billions per year for at least 15 years or more. AMD lost sales and market share and you paid more. That's who the victims are.
Anti Monopoly Law Killed Bambi
'According to Intel, "These new rules would harm consumers by reducing innovation and raising prices."'
Somebody please think of the consumers!
Where there's smoke there's Intel asH
Below is a time line of news articles that actually lead to any sane persons understanding of todays FTC's decision against Intel. (web links are included) Why Intel decided to go to trial is the question.
SUNNYVALE, CALIF.-AUG. 31, 2004-Today, AMD (NYSE: AMD) announces it is demonstrating the industry's first x86 dual-core processor. During demonstrations held at the company's Austin facilities, AMD is showing an HP ProLiant DL585 server powered by four dual-core AMD Opteron(tm) processors manufactured on 90nm silicon-on-insulator process technology.
With a simple upgrade path to more efficient computing, based on AMD's existing system infrastructure and industry-standard architecture, enterprise customers can expect more efficient processing power without the penalties of increased power consumption and heat dissipation. The dual-core AMD Opteron processor for servers and workstations is expected to offer the best performance per watt in the market when AMD plans to make it available in mid-2005.
"This industry milestone changes the dynamics of the computing business," said Dirk Meyer, executive vice president, AMD Computation Products Group. "Once again, AMD is delivering a simplified approach to more efficient processing power, with products that will deliver multi-core 64-bit computing to our strong customer base." ........................
Intel dual-core chips set to ship
By Michael Kanellos, CNET News.com
Monday, April 18, 2005 09:28 AM
A few PC makers will start selling PCs containing dual-core chips from Intel later today, three days ahead of the debut of the dual cores.
Dell, Alienware and a few others are preparing to take orders on Monday for PCs containing these chips, an Intel spokesman said. The companies will also be able to ship these PCs to customers. Although volumes of the Extreme Edition Pentium 4, code-named Smithfield, will initially be low, Intel will ship millions of dual-core chips by the end of the year, the spokesman said.
By the end of 2006, Intel expects that 70 percent of its server chips and 85 percent of its desktop and notebook chips will be dual core, the company has said.
Intel's pre-emptive strike will likely give it the right to claim being the first to ship dual-core x86 chips. Advanced Micro Devices is slated to release dual-core Opteron chips for servers and workstations on April 21. Hewlett-Packard, however, is currently taking preorders on dual-core Opteron servers.
Intel First to Ship Dual Core
UPDATED: The chipmaker begins releasing its Pentium Extreme Edition 800-series ahead of AMD's Opteron
.........Dell announced it would be one of the first PC makers to ship Intel's new dual core Pentium Extreme. The Round Rock, Texas-based hardware vendor said it would begin shipping its Dimension brand of PCs with the new chips relatively soon with prices starting at around $3,000.
HP and Gateway are expected to follow suit. ....
Workstation Processors Duel: AMD Opteron against Intel Xeon
[12/21/2005 08:44 PM | CPU]
by Ilya Gavrichenkov
Today we are going to compare the performance of dual-processor workstations built on single-core and dual-core processors from AMD Opteron and Intel Xeon processor families.
........................The CPU market has changed greatly throughout this year of 2005 and Intel’s position is not as superior in every market sector as it had used to be. Yes, the microprocessor giant enjoys the global advantage over AMD as concerns the total quantity of units shipped, but AMD has pressed the rival in the retail sector since individual users react more quickly to the changing market situation. It is a fact that AMD is currently the technological leader and has been such for quite a while already. AMD was the first to propose and implement 64-bit extensions to the classic x86 architecture and in 2005 became the first x86 CPU manufacturer to start shipping dual-core processors.
We’ve always kept our eye on the competition in the desktop CPU market and our numerous test sessions have proved that dual-core CPUs from AMD are more appealing than Intel’s due to a number of reasons. And today we are going to touch upon another subject of great controversy between Intel and AMD – processors for high-performance workstations................
May 18, 2006 2:00 PM PDT
Dell opts for AMD's Opteron
By Tom Krazit , Michael Kanellos and Ina Fried
Staff Writers, CNET News
Last modified: May 18, 2006 4:24 PM PDT
Dell has agreed to use Advanced Micro Devices' Opteron chip in multiprocessor servers by the end of the year, ending a long-standing policy of sticking exclusively with Intel.
The PC maker made the move public in its first-quarter earnings press release on Thursday. Speculation has mounted for years as to whether Dell would adopt the company's chips, despite Dell's exclusive relationship with rival Intel to this point. AMD has enjoyed a performance lead in server benchmarks over Intel's Xeon processors. ...........
AMD counters Intel with speedier Opterons
Alert PrintWaiting for Xeon
By Ashlee Vance in Mountain View • Get more from this author
Posted in Servers, 26th September 2005 21:23 GMT
Free whitepaper – Guidelines for specification of data center power density
AMD today won a round of its ongoing chip marketing battle with Intel by pumping out a new fleet of speedier dual-core Opteron chips. Intel did its best to counter the move with a couple of variations to its single-core Xeon line but obviously fell short of its rival.
Customers can start buying 2.4GHz versions of the Opteron 280 and 880 chips. That's a boost from AMD's 2.2GHz mainstream Opteron parts and puts the rest of the line on par with an existing premium 2.4GHz chip used by Sun Microsystems. The Opteron 180 chip for single processor servers and workstations will ship at 2.4GHz as well within 30 days
AMD Files Antitrust Complaint Against Intel In U.S. Federal District Court
– Complaint Details Worldwide Coercion of Computer-Makers, System-Builders, Distributors and Retailers from Dealing with AMD –
– Intel’s Illegal Acts Inflate Computer Prices and Limit Choices for Businesses and Consumers –
SUNNYVALE, Calif. -- June 28, 2005 --AMD (NYSE: AMD) announced today that it filed an antitrust complaint against Intel Corporation (“Intel”) yesterday in U.S. federal district court for the district of Delaware under Section 2 of the Sherman Antitrust Act, Sections 4 and 16 of the Clayton Act, and the California Business and Professions Code. The 48-page complaint explains in detail how Intel has unlawfully maintained its monopoly in the x86 microprocessor market by engaging in worldwide coercion of customers from dealing with AMD. It identifies 38 companies that have been victims of coercion by Intel – including large scale computer-makers, small system-builders, wholesale distributors, and retailers, through seven types of illegality across three continents.
“Everywhere in the world, customers deserve freedom of choice and the benefits of innovation – and these are being stolen away in the microprocessor market,” said Hector Ruiz, AMD chairman of the board, president and chief executive officer. “Whether through higher prices from monopoly profits, fewer choices in the marketplace or barriers to innovation – people from Osaka to Frankfurt to Chicago pay the price in cash every day for Intel’s monopoly abuses.”
AMD Files Antitrust Lawsuit Against Intel
By Rob Wright
10:40 AM EDT Tue. Jun. 28, 2005
From the June 27, 2005 issue of VARBusiness The rivalry between the industry's two biggest chip makers heated to a boiling point Tuesday as AMD (NYSE:AMD) announced it had filed an antitrust complaint against Intel (NSDQ:INTC) in U.S. federal district court.
AMD's bold move rekindles a long-dormant legal feud between the two companies and shifts the competition away from the technology front, where AMD had been gaining mind share with its 64-bit computing microprocessors. The lawsuit, filed in U.S. district court in Delaware, accuses Intel of unlawfully maintaining its market-leading position in the x86 microprocessor field by coercing customers from dealing with AMD, violating the Sherman Antitrust Act as well as California law. The complaint identifies 38 companies that AMD says have been victims of Intel's illegal business practices, including large computer makers, distributors, systems builders and retailers.
AMD filed an antitrust suit against Intel in the early 1990s with similar allegations, but the suit was settled out of court, which led to a legal cease fire that lasted until now.
@ the intel fanbois
Their compliler basically goes....
Are you an intel chip ? if yes have industry standard optimisation. If no be nerfed even if you run industry standard optimisations.
I recall someone hacking an AMD chip to show an intel CPUID and getting better performance
Re: the intel fanbois
I don't think I am an Intel fanboy, but my point was AMD designs great chips and no one is disputing that. Why shouldn't a great tech company like AMD design a great compiler that shows their chips to their best advantage?
It seems kind of looney for AMD to think wow Intel is pretty good at writing compilers I am sure they will watch our backs. It is the technology BUSINESS, companies have to be smart about technology and business. Companies run by tech wizards that have no business savvy will get crushed eventually.
- Twitter: La la la, we have not heard of any NUDE JLaw, Upton SELFIES
- China: You, Microsoft. Office-Windows 'compatibility'. You have 20 days to explain
- Apple to devs: NO slurping users' HEALTH for sale to Dark Powers
- Is that a 64-bit ARM Warrior in your pocket? No, it's MIPS64
- Apple 'fesses up: Rejected from the App Store, dev? THIS is why