On June 27, 2005, AMD filed an antitrust lawsuit against Intel in the United States
District Court in Wilmington, Delaware. The complaint details how Intel has unlawfully
maintained its monopoly power in the market for x86 microprocessors by, among other things,:
• forcing major customers such as Dell, Sony, Toshiba, Gateway, and Hitachi into
Intel-exclusive deals in return for outright cash payments, discriminatory pricing or
marketing subsidies conditioned on the exclusion of AMD;
• forcing other major customers such as NEC, Acer, and Fujitsu into partial
exclusivity agreements by conditioning rebates, allowances and market development
funds (MDF) on customers’ agreement to severely limit or forego entirely purchases from
AMD;
• establishing a system of discriminatory, retroactive, first-dollar rebates triggered
by purchases at such high levels as to have the intended effect of denying customers the
freedom to purchase any significant volume of processors from AMD;
• threatening retaliation aga inst customers for introducing AMD computer
platforms, particularly in strategic market segments such as commercial desktop;
• establishing and enforcing quotas among key retailers such as Best Buy and
Circuit City, effectively requiring them to stock overwhelmingly or exclusively, Intel
computers, artificially limiting consumer choice;
• forcing PC makers and tech partners to boycott AMD product launches or
promotions;
• and abusing its market power by forcing on the industry technical standards and
products which have as their main purpose the handicapping of AMD in the marketplace.
Intel’s economic coercion is pervasive and extends to customers at all levels of the x86
ecosystem – from large computer or original equipment manufacturers (“OEMs”) like Hewlett-
Packard, to small system-builders, to wholesale distributors, to retailers such as Circuit City. All
face the same choice: accept conditions that exclude AMD or suffer discriminatory pricing and
competitively crippling treatment. The Japanese Government recognized these competitive
harms on March 8, 2005, when its Federal Trade Commission (JFTC) recommended that Intel be
sanctioned for its exclusionary misconduct directed at AMD. Intel chose not to contest the
charges.
Through its exclusionary conduct, Intel has avoided competition on the merits and
deprived AMD of the opportunity to stake its prices and quality against Intel’s for potential
microprocessor sales. The absence of competition in this important industrycomes at a high cost:
artificial constraints on innovation, higher prices and the loss of the consumer’s right to choose
the products that best suit his or her needs. As such, Intel’s conduct violates the antimonopolization
provisions of Section 2 of the federal Sherman Antitrust Act, as well as
California’s state law prohibitions against secret rebates and tortious interference with
prospective economic advantage. Accordingly, AMD seeks: 1) an injunction to stop Intel’s
anticompetitive conduct; 2) treble damages as provided for under the Sherman Act; and 3)
punitive damages as provided for under California law.
In 2003, AMD began to pull away from Intel technologically and now AMD’s
microprocessors are widely hailed as superior to Intel’s. AMD’s breakthrough came when it
introduced Opteron; the industry’s first x86 backward compatible 64-bit chip. The computing
industry hailed AMD’s introduction of backwards compatible 64-bit computing as an
engineering triumph. In April 2005, AMD was named “Processor Company of 2005” at an Intelsponsored
industry awards show. Bested in a technology duel over which it long claimed
leadership, Intel increased exploitation of its market power to pressure customers to refrain from
migrating to AMD’s superior, lower-cost microprocessors.
SUMMARY OF INTEL MISCONDUCT
The following describes only a sampling of Intel misconduct.
a. Exclusive and Near-Exclusive Deals with OEMs
Dell. Dell has never purchased an AMD microprocessor despite acknowledging Intel’s
shortcomings and its own customers’ clamor for AMD solutions. According to industry reports,
Intel has bought Dell’s exclusivity with outright payments and favorable discriminatory pricing
and service. Dell executives have conceded that they must financially account for Intel
retribution if they decide to launch even one AMD product.
Japan. In 1999, AMD began to make notable inroads into Intel’s sales to major Japanese OEMs,
which export PCs internationally including into the U.S. In 2002, Intel paid Sony, Toshiba and
Hitachi multimillion dollar sums, disguised as discounts and promotional support, in exchange
for worldwide exclusivity. Intel also paid multimillion dollar sums to NEC and Fujitsu to cap
AMD’s share of their business.
b. Product-Line, Channel or Geographic Restrictions
Intel has also bought more limited exclusivity from OEMs as a means of excluding AMD
from the most profitable lines or from channels of distribution best tailored to take advantage of
AMD’s price/performance advantage over Intel. For example, Intel has sabotaged AMD’s
attempts to create a successful commercial desktop product at both HP and IBM.
c. Exclusionary Rebates, Predatory Pricing
Intel has also imposed on OEMs a system of first-dollar rebates that create exclusivity or
near-exclusivity and artificially foreclose AMD from competing meaningfully. While in many
industries, a seller might offer “volume discounts,” Intel’s rebate schemes are quite different and
substantially more odious to competition. Intel’s “penetration” or “loyalty” rebates are not based
on efficiencies or cost savings, but instead are designed to avoid head-to-head price competition
with AMD and leverage Intel’s market position. Intel intentionally sets a customer rebate at a
level of purchases it knows to constitute a dominant percentage of a customer’s needs. Intel’s
retroactive discounts then operate to price additional microprocessors at or below cost so that
AMD cannot compete for this business.
d. Threats of Retaliation
Intel has also resorted to old- fashioned threats, intimidation and “k nee-capping” to deter
OEMs from dealing with AMD. For instance, in late 2000, Compaq’s CEO, Michael Capellas,
disclosed that because of the volume of business he had given to AMD, Intel withheld delivery
of server chips that Compaq desperately needed. Reporting that “he had a gun to his head,”
Capellas informed an AMD executive that he had to stop buying AMD processors. NEC’s
European subsidiary, NEC-CI, which operates NEC’s European and non-Japanese Asian
divisions, reported that Intel executives said they would “destroy” NEC-CI for engaging with
AMD in the commercial desktop segment. Intel told NEC-CI’s retailers that the company’s
AMD dealings could impair NEC-CI’s ability to supply products to its customers. When NECCI
resisted the pressure, Intel imposed a discriminatory price increase.
e. Interference with AMD Product Launches
A successful and impressive product launch is essential to generating confidence among
computer professionals, who will be the potential audience for a new microprocessor, and is key
to gaining market acceptance. Aware of the importance of product launches, Intel has done its
utmost to undermine AMD’s. For instance, in 2003, Intel’s CEO Craig Barrett went so far as to
travel to Taiwan to personally threaten Acer’s Chairman, President and CEO with “severe
consequences” for publicly supporting AMD’s product rollout of Athlon64. The Barrett visit
coincided with an unexplained delay by Intel in providing $15-20 million in market development
funds owed to Acer. As a result, Acer withdrew from the launch in the U.S. and Taiwan, pulled
its promotional materials, banned AMD’s use of a video Acer had prepared, and delayed the
announcement of its Athlon64-powered computers.
f. Exclusionary Practices Directed At Retailers
In Germany, AMD has been entirely shut out from Media Markt, which operates retail
stores throughout Europe and accounts for 35% of Germany’s retail sales. Intel provides Media
Markt between $15-20 million of MDF annually, and since 1997 Media Markt has carried Intel
computers exclusively. Similarly, in the U.S., Intel provides full MDF payments to retailers,
such as Best Buy and Circuit City, only if they agree to limit to 20%, not just the shelf space
devoted to AMD-based products, but also the share of revenues they generate from selling AMD
platforms. If AMD’s share exceeds 20%, the offending retailer’s marketing support from Intel is
cut by 33% across all products.
EFFECTS OF INTEL’S MISCONDUCT
Despite its technological leadership, AMD’s market share remains artificially stunted by
Intel’s exclusionary actions. Since 1999, AMD’s worldwide volume share has hovered at 15%,
while Intel has captured at least 80% of x86 microprocessor unit sales in seven of the last eight
years. By capping AMD’s market share, Intel has prevented AMD from expanding to reach the
size necessary to become a predominant supplier to major customers. As a result, those in the
microprocessor industry continue to be beholden to Intel, which requires them to pay monopoly
prices, to be exposed to Intel’s coercive tactics, and to submit to artificial limits on purchases
from AMD. Consumers ultimately foot the bill for Intel’s conduct, in the form of inflated PCprices and the loss of choice in computer products. Finally, society as a whole is worse off for
the lack of innovation that only a truly competitive market can drive.
Thursday, July 31, 2008
Step 3: Navigating Between DDR and DDR-2
Here's a little secret, on the whole DDR-2 RAM has been a bit of bust. Touted as the memory of the future, able to leap small buildings in a single bound for everything from videocards to motherboards.
It promised a lot and delivered little in the real world. Yet since the entire computer industry is shifting towards DDR-2 RAM, we're all resigned to the fact that it's here to stay until FB-DIMM and DDR-3 RAM break out in 2007/2008.
Of course DDR-2 memory isn't all bad, it offers a greater level of bandwidth between memory and processor, and that's a good thing. It's just that single-core Intel systems (the current largest segment consuming DDR-2 RAM) aren't very inspiring, and the wonderfully low timings associated with DDR memory have been cast aside for a pointless frequency game. There's a difference between PC2-6400 with high lanencies, and PC2-6400 with low latencies when it comes the benchmarks, and so far the latter has been sadly overlooked for far too long.
By the end of May, AMD's Socket AM2 Athlon64 processor will be running along on DDR-2. The socket AM2 Athlon64 isn't expected to demand a ton of bandwidth from the get-go, but rather benefit more from DDR-2 memory with tighter CAS latency timings. Unfortunately at the moment these types of parts are missing from the DDR2 memory equation so it's hard to offer commentary on where this will all be headed.
It's very likely that initial Socket AM2 Athlon64's will perform no better, or no worse than equivalently paced Socket 939 counterparts. Between now and then, perhaps AMD will have tweaked the memory controller to utilize more memory bandwidth, or DDR-2 memory latencies will have dropped somewhat. Like you, I'm still waiting to see.
The saga isn't yet written, and pre-release glimpses of Socket AM2 performance by way of Engineering Sample CPUs are only telling half the story. It will be interesting to see what happens, but certainly the prevalence of DDR-2 RAM is unrelenting.
It promised a lot and delivered little in the real world. Yet since the entire computer industry is shifting towards DDR-2 RAM, we're all resigned to the fact that it's here to stay until FB-DIMM and DDR-3 RAM break out in 2007/2008.
Of course DDR-2 memory isn't all bad, it offers a greater level of bandwidth between memory and processor, and that's a good thing. It's just that single-core Intel systems (the current largest segment consuming DDR-2 RAM) aren't very inspiring, and the wonderfully low timings associated with DDR memory have been cast aside for a pointless frequency game. There's a difference between PC2-6400 with high lanencies, and PC2-6400 with low latencies when it comes the benchmarks, and so far the latter has been sadly overlooked for far too long.
By the end of May, AMD's Socket AM2 Athlon64 processor will be running along on DDR-2. The socket AM2 Athlon64 isn't expected to demand a ton of bandwidth from the get-go, but rather benefit more from DDR-2 memory with tighter CAS latency timings. Unfortunately at the moment these types of parts are missing from the DDR2 memory equation so it's hard to offer commentary on where this will all be headed.
It's very likely that initial Socket AM2 Athlon64's will perform no better, or no worse than equivalently paced Socket 939 counterparts. Between now and then, perhaps AMD will have tweaked the memory controller to utilize more memory bandwidth, or DDR-2 memory latencies will have dropped somewhat. Like you, I'm still waiting to see.
The saga isn't yet written, and pre-release glimpses of Socket AM2 performance by way of Engineering Sample CPUs are only telling half the story. It will be interesting to see what happens, but certainly the prevalence of DDR-2 RAM is unrelenting.
Step 2: Heat and Reliability
Cooling has always been the one major area where Intel processors were always considered to be far superior to AMD's offerings... remember the days of AthlonXP's going up in smoke? While the Socket 775 Pentium 4 heatsink architecture offers more room to grow, allows for larger heatsinks to be installed, and a bit more scalable in the long run, AMD's not totally out of step either. AMD has dramatically improved the shape, size and quality of heatsinks that it uses to keep Athlon64 processors running cool and quietly. With the de-emphasis of OEM processors, the company has better control over the retail heatsinks that come bundled with its Athlon64 processors, and hence the end user experience. So far, this generation of 'K8' heatsinks have been quiet running, and well designed so temperatures remain at acceptable levels.
To make things easier for the end user, heatsinks can be installed in any direction without damaging the processor. Back in the days of the socket A Athlon and AthlonXP CPU, if the heatsink was installed in the wrong direction you'd end up with a dead chip in under 4 seconds. In the unlikely event that the heatsink fan fails nowadays, that little tiny Athlon64 processor below will not cook itself to death. All current AMD processors employ thermal throttling which lowers the speed of the processor automatically should the CPU temperature rise too high.
On the whole, AMD and Intel are pretty even in thermal loads this year. From the consumers point of view it makes no difference if one processor or the other is used as both will operate reliably and quietly.
To make things easier for the end user, heatsinks can be installed in any direction without damaging the processor. Back in the days of the socket A Athlon and AthlonXP CPU, if the heatsink was installed in the wrong direction you'd end up with a dead chip in under 4 seconds. In the unlikely event that the heatsink fan fails nowadays, that little tiny Athlon64 processor below will not cook itself to death. All current AMD processors employ thermal throttling which lowers the speed of the processor automatically should the CPU temperature rise too high.
On the whole, AMD and Intel are pretty even in thermal loads this year. From the consumers point of view it makes no difference if one processor or the other is used as both will operate reliably and quietly.
Step 1: It's the Performance, Stupid!
There's little doubt that AMD's K8 Athlon64 processor is currently the fastest architecture available. The Athlon64 architecture is superior to Intel's Netburst (the architecture that drives the Pentium 4) in every which way, and Intel's band-aid fixes have not been enough to keep up with the perpetual underdog from Austin. It's true that Intel does have a real winner with its Pentium M and Pentium III pedigreed Core Duo, but these are primarily mobile CPUs, and consequently beyond the scope of what I'll be speaking on.
What was it that happened to so dramatically shift the position between Intel and AMD's processors?Why is Intel faltering on the desktop front and AMD winning the hearts and minds of geeks world wide? It certainly isn't for lack of advertising, but that's another story.
The real reason for all of this upheaval and change is Intel's Netburst architecture. It was supposed to last for 10 years when it was introduced in 2000, however that lifespan was cut short in 2003 when Intel struggled so publicly with the Prescott core. The initial product was full of kinks, its performance was lousy, it suffered from voltage leakage, and it was pretty obvious that many of its faults were due to the way Intel "improved" its processors from one speed generation to the next. The days of the good old die shrink and ramp up are certainly dead now.
After some initial questions to the necessity of a 64-bit processor in a 32-bit world, AMD's Athlon64 processor was well on its way to becoming the sweetheart of computer geeks. It's efficient core architecture allows the Athlon64 to handle more work per clock cycle than the Pentium 4/D (which was also the case with AMD's previous generation), so more gets done with less so to speak.
Intel's wildcard has always been its special CPU SSE series instructions, but that advantage has also dwindled away. While AMD's parts often do not support the latest Intel instructions at the time of introduction, the company does tend to integrate them in time to coincide with the release of software that uses these new features. In fact, if you look at the enhanced instruction sets in the latest AMD Athlon64 processors, you'll notice that it supports more instructions than an equivalent Intel Pentium 4 processor!
Perhaps Intel's one saving grace is that the Pentium 4/D can still overclock quite well, with a little inventive cooling it will achieve frequencies that AMD users can only reach with extreme cooling. Realistically though as nice as the round numbers are, these are empty goals. An Athlon64 may be clocked a whole gigahertz slower than a Pentium 4, but it still performs much better in benchmarks; the correlation between frequency and performance is pretty much dead.
On the horizon, Intel's upcoming 'Conroe' core is starting to look like it might give AMD a run for its money, but it's not available yet so comparing it with current technology is not appropriate.
What was it that happened to so dramatically shift the position between Intel and AMD's processors?Why is Intel faltering on the desktop front and AMD winning the hearts and minds of geeks world wide? It certainly isn't for lack of advertising, but that's another story.
The real reason for all of this upheaval and change is Intel's Netburst architecture. It was supposed to last for 10 years when it was introduced in 2000, however that lifespan was cut short in 2003 when Intel struggled so publicly with the Prescott core. The initial product was full of kinks, its performance was lousy, it suffered from voltage leakage, and it was pretty obvious that many of its faults were due to the way Intel "improved" its processors from one speed generation to the next. The days of the good old die shrink and ramp up are certainly dead now.
After some initial questions to the necessity of a 64-bit processor in a 32-bit world, AMD's Athlon64 processor was well on its way to becoming the sweetheart of computer geeks. It's efficient core architecture allows the Athlon64 to handle more work per clock cycle than the Pentium 4/D (which was also the case with AMD's previous generation), so more gets done with less so to speak.
Intel's wildcard has always been its special CPU SSE series instructions, but that advantage has also dwindled away. While AMD's parts often do not support the latest Intel instructions at the time of introduction, the company does tend to integrate them in time to coincide with the release of software that uses these new features. In fact, if you look at the enhanced instruction sets in the latest AMD Athlon64 processors, you'll notice that it supports more instructions than an equivalent Intel Pentium 4 processor!
Perhaps Intel's one saving grace is that the Pentium 4/D can still overclock quite well, with a little inventive cooling it will achieve frequencies that AMD users can only reach with extreme cooling. Realistically though as nice as the round numbers are, these are empty goals. An Athlon64 may be clocked a whole gigahertz slower than a Pentium 4, but it still performs much better in benchmarks; the correlation between frequency and performance is pretty much dead.
On the horizon, Intel's upcoming 'Conroe' core is starting to look like it might give AMD a run for its money, but it's not available yet so comparing it with current technology is not appropriate.
AMD VS INTEL
Once again it's time to get into the boxing ring for another battle of AMD vs. Intel! And despite the strong opinions this editorial is likely to stir up, I always appreciate hearing back from you. I'm sure plenty of people will disagree with my take on the processor industry, but hey that's why I'm the tech analyst!
With that in mind, it's time for some up to the minute commentary on the state of the computer world. The AMD vs. Intel battlefront has changed dramatically over the last three years, and like you, I have gone the path that most enthusiasts have. I want the best performance for my hard earned cash, so I choose the fastest available hardware without really considering who manufacturers the parts... after all, does it really matter who makes the fastest CPU?
Age has mellowed my thirst for speed, and my upgrade schedule has slowed to a yearly pace, but that doesn't mean I'm settling for any less. To fit into this leisurely schedule I've had to focus a bit more on evolving technologies, and do my best to avoid the lemons (hello RDRAM!?) and technological evolutionary branches which aren't going anywhere... say for example ATI's Crossfire.
Here's how it all plays out in five easy steps.
With that in mind, it's time for some up to the minute commentary on the state of the computer world. The AMD vs. Intel battlefront has changed dramatically over the last three years, and like you, I have gone the path that most enthusiasts have. I want the best performance for my hard earned cash, so I choose the fastest available hardware without really considering who manufacturers the parts... after all, does it really matter who makes the fastest CPU?
Age has mellowed my thirst for speed, and my upgrade schedule has slowed to a yearly pace, but that doesn't mean I'm settling for any less. To fit into this leisurely schedule I've had to focus a bit more on evolving technologies, and do my best to avoid the lemons (hello RDRAM!?) and technological evolutionary branches which aren't going anywhere... say for example ATI's Crossfire.
Here's how it all plays out in five easy steps.
Time to set the record straight...
"Oh no, Colin has turned 180 degrees and is now a total Intel fanatic!!!" I'm sure that's what a lot of you will be saying after reading this. No, I'm not biased towards Intel, I'm simply trying to point out some facts. I don't love either company, Intel or AMD, but I do like them. They're good for each other. Just think about a world where only Intel or AMD exists. With a virtual monopoly on x86 processors, who's going to stop either company from charging you through the nose for processors? I actually like AMD more then Intel, AMD seems to listen to the public on what they want and in general they usually do try and deliver. Intel on the other hand seems a little more arrogant, pushing things down people's throat (RDRAM + P3...) instead of listening.
When people e-mail me asking about systems, I almost always tell them to go with an AMD based system. Why? AMD is, and probably always will be a better value then Intel. Value doesn't mean lousy performance either! AMD CPU's are damn fast and can give Pentium 4's a good run for its money but in the end P4's are faster!
In terms of CPU "care" Intel has this down. Their retail heatsinks are designed very well, not only to cool the CPU but the larger foot print of the socket 478 heatsinks helps shield against EMI. AMD has a lot of catching up to do. They're moving in the right direction in terms of heatsinks, the current ones bundled with the Thoroughbred XP2200+'s are a heck of a lot better in design and cooling capability then the ones that originally came with the AthlonXP (Palomino core). AMD CPU's still have severe heat issues to conquer; even the Thoroughbred (0.13 micron) runs extremely high. At those temperatures it takes only a few seconds before the little processor dies of over heating.
As an independent hardware tester and reviewer I'm quite offended when someone attacks my integrity. The benchmarks that are being used on the web are not biased towards Intel just because they use SSE2. if I wasn't allowed to test with SSE2 enabled benchmarks, it would be like testing a car's top speed without using the 5th or 6th gears.
I'm not biased towards Intel but since with all the crap floating around I just wanted to try and set the record straight. This is more of a personal rant then anything else, but hey, I'm also a consumer. If you want to flame me and my opinions you're more than welcome to e-mail me here.
Oh and just for the record, I went with a P4 1.8A which runs at 2.4 GHz on an Abit TH7II-RAID with 512MB PC1066 RDRAM and it's running faster (and quieter) then my AthlonXP 2000+ on an Abit KR7A-RAID with 512 PC2100 CL2 DDR.... for the moment that is.
Check out the most recent word on the street about processors right here, the latest memory technology here, motherboards at the end of this link, and heatsinks and cooling solutions at FrostyTech.
Find out about this and many other reviews by joining the Weekly PCstats.com Newsletter today! Catch all of PCSTATS latest reviews right here.
When people e-mail me asking about systems, I almost always tell them to go with an AMD based system. Why? AMD is, and probably always will be a better value then Intel. Value doesn't mean lousy performance either! AMD CPU's are damn fast and can give Pentium 4's a good run for its money but in the end P4's are faster!
In terms of CPU "care" Intel has this down. Their retail heatsinks are designed very well, not only to cool the CPU but the larger foot print of the socket 478 heatsinks helps shield against EMI. AMD has a lot of catching up to do. They're moving in the right direction in terms of heatsinks, the current ones bundled with the Thoroughbred XP2200+'s are a heck of a lot better in design and cooling capability then the ones that originally came with the AthlonXP (Palomino core). AMD CPU's still have severe heat issues to conquer; even the Thoroughbred (0.13 micron) runs extremely high. At those temperatures it takes only a few seconds before the little processor dies of over heating.
As an independent hardware tester and reviewer I'm quite offended when someone attacks my integrity. The benchmarks that are being used on the web are not biased towards Intel just because they use SSE2. if I wasn't allowed to test with SSE2 enabled benchmarks, it would be like testing a car's top speed without using the 5th or 6th gears.
I'm not biased towards Intel but since with all the crap floating around I just wanted to try and set the record straight. This is more of a personal rant then anything else, but hey, I'm also a consumer. If you want to flame me and my opinions you're more than welcome to e-mail me here.
Oh and just for the record, I went with a P4 1.8A which runs at 2.4 GHz on an Abit TH7II-RAID with 512MB PC1066 RDRAM and it's running faster (and quieter) then my AthlonXP 2000+ on an Abit KR7A-RAID with 512 PC2100 CL2 DDR.... for the moment that is.
Check out the most recent word on the street about processors right here, the latest memory technology here, motherboards at the end of this link, and heatsinks and cooling solutions at FrostyTech.
Find out about this and many other reviews by joining the Weekly PCstats.com Newsletter today! Catch all of PCSTATS latest reviews right here.
Hardware websites and benchmarks are cheating AMD...
I don't frequent forums much myself anymore but whenever I see people claiming that the benchmarks being used are skewed towards Intel, it really makes my blood boil. That has got to be the most uninformed thing I've ever heard. Yes, there's a huge conspiracy against AMD and all the hardware sites are secretly being paid off by Intel to run "Intel based benchmarks", that's the only reason why Intel beats AMD... ludicrous!
Most software like the MadOnion benchmarks (PCMark, 3DMark) have SSE2 optimization but is that cheating? They also have 3D Now! code written in them too! How about the notorious SysMark2001 with the Windows Media Encoder 7.0 not using AthlonXP's SSE? AMD themselves stated that SysMark 2001 is a valid benchmark, AMD did release a patch to allow Media Encoder 7.0 to use SSE but they liked the original SysMark 2001 because even without using SSE, the Athlon scored well and it showed off the mighty FPU of the Athlon.
Quake III Arena does love the bandwidth available to the Pentium 4, it always has, but that is something which is isolated to that one game. Games like Return to Castle Wolfenstein or Jedi Knight II which use the Q3 engine don't show the same signs of loving the P4.
Like it or not SSE2 is here to stay, heck Intel even licensed SSE2 to AMD to run in their "Hammer" line of processors!
Most software like the MadOnion benchmarks (PCMark, 3DMark) have SSE2 optimization but is that cheating? They also have 3D Now! code written in them too! How about the notorious SysMark2001 with the Windows Media Encoder 7.0 not using AthlonXP's SSE? AMD themselves stated that SysMark 2001 is a valid benchmark, AMD did release a patch to allow Media Encoder 7.0 to use SSE but they liked the original SysMark 2001 because even without using SSE, the Athlon scored well and it showed off the mighty FPU of the Athlon.
Quake III Arena does love the bandwidth available to the Pentium 4, it always has, but that is something which is isolated to that one game. Games like Return to Castle Wolfenstein or Jedi Knight II which use the Q3 engine don't show the same signs of loving the P4.
Like it or not SSE2 is here to stay, heck Intel even licensed SSE2 to AMD to run in their "Hammer" line of processors!
Subscribe to:
Comments (Atom)