Why was the original Phenom so bad, again?

[BlueBlazer]

You re proposing superpi in response of this :

Do you read before answering?..

Yups, I saw. The biggest point here is SuperPi is widely used and isn't optimized for any CPU, period. Doesn't care if its AMD or Intel, it can even run on an ancient Pentium (I). If you want one optimized for AMD then look no further than ScienceMark, which unfortunately hardly used in reviews (and benchmarks) anymore thus hard to make comparisons with all processors (from old until the current generation).

SP FP are not used at all in Fluent ,LS Dyna and any other
FPU intensive soft as precision would be just ridiculous for
any scientific computing that use at least DP if not X87 when
more precision in rounding is needed so your exemple is just pointless,
as was the one above....

But perhaps when using Windows s hand calculator ?..

Incorrect, Single Precision FP is also used in quite a number of HPC applications and not necessarily always using Double Precision type. Example here: CERN HLT Analysis of Intel Knights Ferry on High Energy Physics (do note that Intel's current MIC are primarily single precision, only the next generation will have double precision).

Same pointlessness since all this doesnt hide the fact that
Barcelona has FAR better FPU than Intel clovertown and even
Harpertown as proved by really professionnal softs wich are unlikely
to be rigged since they are ultra optimised for both uarch.

Faster x87 but slower SSE2. Nowadays modern applications no longer use raw x87 but the faster SSE and SSE2 (and AVX) instructions. Only older (and ancient) applications still use raw x87 instructions. Unfortunately, again your argument is unfounded because many applications (e.g. Excel as discussed above) will use different compilers (not necessarily ICC). Even if K8/K10 is using SSE2 instructions, it would be slower than Core2. Likewise using K8/K10 x87 against Core2 SSE2 would be incorrect (SSE2 will be still faster than x87).

Other fact is then you said that INTEGER matters more in server
applications , and then said that a Harpertown was 56% faster
than a Barcelona without specifying that the INtel chips was a 3G
and the comparison a 2G Barcelona.

56% more performance with only 50% increase in clock speed (2G to 3G). That doesn't sum up. The extra 6% more performance can be the result of better IPC.

Moreover, this bench is an Intel made bench, so we can
assume that this was compiled efficently to make full use
of the Core Uarch, yet, considering the frequency difference,
Harpertown is no better in Integer than Barcelona and far
less performing in FP heavy apps, period.

Your bias is showing. SuperPi doesn't care about architecture. :thumbsdown:

 

[Abwx]

If you want one optimized for AMD then look no further than ScienceMark, which unfortunately hardly used in reviews (and benchmarks) anymore.

It s not by chance since Bapco s Sysmark suit far better..

Incorrect, Single Precision FP is also used in quite a number of HPC applications and not necessarily always using Double Precision type. Example here: CERN HLT Analysis of Intel Knights Ferry on High Energy Physics (do note that Intel's current MIC are primarily single precision, only the next generation will have double precision).

You once said cherry picked applications about Fluent/LS dyna...??.:thumbsup:

Faster x87 but slower SSE2. Nowadays modern applications no longer use raw x87 but the faster SSE and SSE2 (and AVX) instructions. Only older (and ancient) applications still use raw x87 instructions. Unfortunately, again your argument is unfounded because many applications (e.g. Excel as discussed above) will use different compilers (not necessarily ICC). Even if K8/K10 is using SSE2 instructions, it would be slower than Core2. Likewise using K8/K10 x87 against Core2 SSE2 would be incorrect (SSE2 will be still faster than x87).

56% more performance with only 50% increase in clock speed (2G to 3G). That doesn't sum up. The extra 6% more performance can be the result of better IPC.

All this SSE blah bla wont change the fact that Barcelona has as much
as 50% better FP perf , so you re insisting on being pointless.

As for IPC , it is specified at the same frequency , since higher
frequency can yield lower or higher IPC , yet anotehr failure to
make apple /apple comparisons.but even then , 6% is miserable
compared with its massive loss in FP.

Your bias is showing. SuperPi doesn't care about architecture. :thumbsdown:

Desesperate move , since the real point was to see how intel
cpu compare to amd when the tables are turned optimisation
wise , and all you find is a bench that favour none of the CPU,
and then you talk of bias...??..

 

[BlueBlazer]

It s not by chance since Bapco s Sysmark suit far better..

Do you know what is actually used in Sysmark? Excel is one of them unfortunately. Then there are those pesky Adobe programs.

You one said cherry picked applications about Fluent/LS dyna...??.

You cherry picked that one first to generalize Phenom's performance when overall it was the other way around. :thumbsdown:

All this SSE blah bla wont change the fact that Barcelona has as much
as 50% better FP perf , so you re insisting on being pointless.

I guess you are going in circles believing that raw x87 instructions are still being used today. If you ever checked the FP benchmarks in programs like SiSoft, you will notice SSE and SSE2 tend to inflate the scores compared to x87 (because SSE/SSE2 is much faster than traditional FP instructions). Like I've said, nowadays modern applications uses at least SSE or SSE2.

As for IPC , it is specified at the same frequency , since higher
frequency can yield lower or higher IPC , yet anotehr failure to
make apple /apple comparisons.but even then , 6% is miserable
compared with its massive loss in FP.

These are server applications and of course memory bandwidth and I/O sometimes comes into play (e.g. processing transactions). Despite the lack of that bandwidth, somehow able to get an extra 6%. Go ahead and scale the Opteron 2GHz to 3GHz (assuming perfect extrapolation), you will still get short of that 6% advantage that Xeon has. Unfortunately Barcelona is short of clock speed.

Desesperate move , since the real point was to see how intel
cpu compare to amd when the tables are turned optimisation
wise , and all you find is a bench that favour none of the CPU,
and then you talk of bias...??..

Your bias showing again in the very first line of your reply (the Bapco part first). Sure, you can cherry pick one benchmarks such as this ScienceMark to make your point, but overall Core2 is much faster than original Phenom and that's a fact.

 

[formulav8]

Your bias is showing. SuperPi doesn't care about architecture. :thumbsdown:

After reading the last few boring posts your biased showed more than his. :\

Anyways, one big reason Ph1 was such a letdown is because AMD hyped it up like a Core2 killer. Low clockspeeds was very bad. Clock for clock Phenom1 brought about 10%-15% increase in performance compared to the current Athlon 64's and not even close to the 25% ipc increase it would have needed.

Also Ph2 was much improved by the cores no longer having to fight for cache access. Memory controller Improved overall. Touch faster L3 cache (Only 2 cycles though). Slightly improved BP, buffers, ect.. The biggest boost came from much higher clockspeeds.

 

[Abwx]

Do you know what is actually used in Sysmark? Excel is one of them unfortunately. Then there are those pesky Adobe programs.

I was ironic about it, since it s well known that Bapco is almost an Intel
org, yet, you did take it as a serious proposition..

I guess you are going in circles believing that raw x87 instructions are still being used today. If you ever checked the FP benchmarks in programs like SiSoft, you will notice SSE and SSE2 tend to inflate the scores compared to x87 (because SSE/SSE2 is much faster than traditional FP instructions). Like I've said, nowadays modern applications uses at least SSE or SSE2.

Sisoft sandra, that s quite another usefull bench...

These are server applications and of course memory bandwidth and I/O sometimes comes into play (e.g. processing transactions). Despite the lack of that bandwidth, somehow able to get an extra 6%. Go ahead and scale the Opteron 2GHz to 3GHz (assuming perfect extrapolation), you will still get short of that 6% advantage that Xeon has. Unfortunately Barcelona is short of clock speed.

You have provided link to server benchmarks, and unfortunately for you,
you didnt pay attention to the fact that Harpertown was clocked 50%
higher, now you re using this 6% figure as a mean to stay afloat, but
alas, uselessly...

Your bias showing again in the very first line of your reply (the Bapco part first). Sure, you can cherry pick one benchmarks such as this ScienceMark to make your point, but overall Core2 is much faster than original Phenom and that's a fact.

Typical Intel fanboy trolling...

AMD Barcelona beats Xeon in SPECcpu2006, FP up to 50% , Integer up to 20%

Keep on the good job of spreading falsehood on Intel s behalf...

 

[BlueBlazer]

Honestly, after reading the last few boring posts your biased showed more than his. :\

The problem is that he continuously implied that the benchmarks (such as the previously discussed POV Ray and SuperPi) were Intel biased ("made for Intel", "optimized for Intel", etc). Then he kept harping on how fast Barcelona was on a select few (cherry picked) applications while ignoring the rest. My point is that nowadays SSE and SSE2 are used, and that Core2 was faster then K8/K10 in that area (which can contribute to the many of benchmark scores). Of course, relating that point with him is like going in circles....

 

[BlueBlazer]

Sisoft sandra, that s quite another usefull bench...

You can't argue that, can't you. You can clearly see how SSE and SSE2 is way faster than pure x87.

You have provided link to server benchmarks, and unfortunately for you,
you didnt pay attention to the fact that Harpertown was clocked 50%
higher, now you re using this 6% figure as a mean to stay afloat, but
alas, uselessly...

Like I mentioned earlier, extrapolate the Opteron to 3GHz (50% higher clock speed to match Xeon's 3GHz clock speed) and hypothethically the total performance should be 50% higher (assuming perfect extrapolation). But then the Xeon has an extra 6% advantage (56% minus 50%). Explain that.....

Keep on the good job of spreading falsehood on Intel s behalf...

First link says "Yahoo - Document Has Expired".

I was ironic about it, since it s well known that Bapco is almost an Intel
org, yet, you did take it as a serious proposition..

Typical Intel fanboy trolling...

Your bias showing up more.... :thumbsdown:

 

[Idontcare]

The problem is that he continuously implied that the benchmarks (such as the previously discussed POV Ray and SuperPi) were Intel biased ("made for Intel", "optimized for Intel", etc). Then he kept harping on how fast Barcelona was on a select few (cherry picked) applications while ignoring the rest. My point is that nowadays SSE and SSE2 are used, and that Core2 was faster then K8/K10 in that area (which can contribute to the many of benchmark scores). Of course, relating that point with him is like going in circles....

This is the problem/challenge of anyone who wants to take their arguments down the path "well anything that shows my preferred CPU maker in a bad light is because that thing was optimized for the other guy"...be it CPU's, SSD's, GPU's, etc.

If you start from the outset with the initial condition that the method of generating/acquiring data itself is invalid then there is nothing more to discuss, it is a simple QED.

You are wasting your time attempting to rationalize anything at this point because the other side of the debate has already retreated to the point of taking up the position that anything can be invalidated by way of the "but but but it's an Intel optimized bench" argument.

Seriously just don't waste your time. We all see the lunacy in the posts and the lack of robustness in the logic, we just aren't as interested as you are in attempting to reason with it. Just let it go. There is no arguing or debating to be done when one party cannot agree to reasonable terms of engagement at the outset.

This other party feels anything that shows AMD in poor light must be an Intel optimized compiler situation. That's a QED. They may as well make it their sig so that everyone knows to not bother engaging them in debate because they are going to pull out the lamest/weakest excuse on the internet to "support" their arguments. Who wants to waste their time on that?

 

[Abwx]

This is the problem/challenge of anyone who wants to take their arguments down the path "well anything that shows my preferred CPU maker in a bad light is because that thing was optimized for the other guy"...be it CPU's, SSD's, GPU's, etc.

If you start from the outset with the initial condition that the method of generating/acquiring data itself is invalid then there is nothing more to discuss, it is a simple QED.

You are wasting your time attempting to rationalize anything at this point because the other side of the debate has already retreated to the point of taking up the position that anything can be invalidated by way of the "but but but it's an Intel optimized bench" argument.

Seriously just don't waste your time. We all see the lunacy in the posts and the lack of robustness in the logic, we just aren't as interested as you are in attempting to reason with it. Just let it go. There is no arguing or debating to be done when one party cannot agree to reasonable terms of engagement at the outset.

This other party feels anything that shows AMD in poor light must be an Intel optimized compiler situation. That's a QED. They may as well make it their sig so that everyone knows to not bother engaging them in debate because they are going to pull out the lamest/weakest excuse on the internet to "support" their arguments. Who wants to waste their time on that?

Things are documented...
Just imagine that it s not by chance that Intel is leader on cpu.
When products are not really superiors , use any means to spin
the real numbers..

"Many software developers think that the compiler is compatible with AMD processors, and in fact it is, but unbeknownst to the programmer it puts in a biased CPU dispatcher that chooses an inferior code path whenever it is running on a non-Intel processor," Fog writes, "If programmers knew this fact they would probably use another compiler. Who wants to sell a piece of software that doesn't work well on AMD processors?"

http://www.osnews.com/story/22683/I...quot_Cripple_AMD_quot_Function_from_Compiler_

 

[Idontcare]

Yes yes, and men never landed on the moon, xtors came from alien technology, there are carburetors that can give you 300 mpg but the oil cartel bought the patents and shelved the designs, and world governments are a front for the actual hegemony of unified power brokers faced by the Rothschilds.

I don't care why Intel's products are seemingly superior to AMD's for the apps I use...be it satanic rituals involving the blood of baby seals or compiler engineers that are paid $500k a year to unoptimize code for AMD processors...at a pragmatic level I am interested in performance of product A using application B regardless the magickery behind the creation of product A or application B.

If AMD can't outmaneuver Intel and beat them at their own game then I have pity for AMD's employees and shareholders but I am still going to buy the product that delivers best price/performance/power numbers for my app.

AMD's processors might be capable of performing 50 times better than Intels, make me a sandwich, clean my house, and walk the dog...but if it doesn't actually do those things, if for whatever reason it cannot deliver that performance, then the capability is pointless and useless to me at a practical level and I don't buy products based on their "potential", I buy them based on what they actually deliver.

 

[AnandThenMan]

If AMD can't outmaneuver Intel and beat them at their own game then I have pity for AMD's employees and shareholders but I am still going to buy the product that delivers best price/performance/power numbers for my app.

We can only hope that AMD will continue to maneuver themselves enough to survive in the marketplace, otherwise things could get really ugly for us consumers.

 

[Abwx]

I am interested in performance of product A using application B regardless the magickery behind the creation of product A or application B.

That is what all was about, but then :

I buy them based on what they actually deliver.

And you know what they can deliver thanks to benchmarks,
yet , some notorious benches as Sysmark were completely
modded as soon as it did show Amd s product being quite
competitive.

So , how can you know what the product deliver if benches
are rigged this way ?..

http://www.vanshardware.com/reviews/2002/08/020822_AthlonXP2600/SYSmark%202002%20Analysis%20Presentation%20FINAL.pdf

http://www.vanshardware.com/reviews/2002/08/020822_AthlonXP2600/020822_AthlonXP2600.htm

 

[Idontcare]

We can only hope that AMD will continue to maneuver themselves enough to survive in the marketplace, otherwise things could get really ugly for us consumers.

I use to strongly believe in this school of though as well, but with the rise of ARM these past couple of years I now question this.

Were AMD to falter I doubt that Via would sit still and not move to re-enter the mainstream performance tier.

Likewise if AMD were to falter I doubt that ARM would not make a move to roll out a seriously high-power version of its architecture. Or IBM would roll out a desktop version of its Power7. Or Oracle would enter with a Sparc version for the consumer space.

There's a lack of competitors now, but that's because there are currently much more profitable grounds for those alternative architectures to compete in. In the vacuum that would come from AMD's faltering I have every confidence that such a vacuum would gain attention and notice by any number of existing, well monied, microprocessor manufacturers.

Granted there will be a time delay involved, a good four years at best, before the alternatives are brought to market. But 4 yrs is not such a long time in this industry. Consider that we, as consumers, have been suffering from AMD's faltering with phenom now for 4 yrs. Its survivable.

The key for AMD, or Via, or Oracle, or Nvidia (ARM), is GloFo and TSMC. If the foundries can't develop competitive process technology on a timeline that parallels that of Intel's then none of these alternative architectures stand a chance at competing with Intel.

 

[dac7nco]

Abwx,

you seem to be on this forum exclusively to defend year 2007 performance in the year 2011, and I don't understand why. The Phenom-II would've been a stellar chip in 2006, a great one in 2007, but still a midrange one in 2008. AMD failed by making a C2Q competitor as Nehalem arrived, and it wouldn't be here if not for the ATI buy. We'll see bulldozer pretty soon, and I have high hopes for it, but let's not kid ourselves, kid... The greats at AMD jumped ship a long time ago.

This argument over compilers and optimizations doesn't matter one bit. Anyone buying CPUs for a specific application will be buying serious chips, not desktop ones. Intel, unfortunately, kills in both spaces.

Daimon

 

[tulx]

I'm running a Phenom 9850 BE to this day (on 3,1 GHz) and still playing Crysis 2 on max settings. Of course, I get 25 fps instead of 250, but not that it would matter.

 

[dac7nco]

I'm running a Phenom 9850 BE to this day (on 3,1 GHz) and still playing Crysis 2 on max settings. Of course, I get 25 fps instead of 250, but not that it would matter.

More power to you! My HTPC (Q8300/5850) can manage that, but just barely.

 

[jhu]

Likewise if AMD were to falter I doubt that ARM would not make a move to roll out a seriously high-power version of its architecture. Or IBM would roll out a desktop version of its Power7. Or Oracle would enter with a Sparc version for the consumer space.

Unlikely. IBM tried that with POWER4, and then Apple switched to Intel processors. Consumer Sparc? Also unlikely given their price/performance disadvantage compared with x86.

 

[AnandThenMan]

I use to strongly believe in this school of though as well, but with the rise of ARM these past couple of years I now question this.

I've thought about this as well, but the timeline for ARM to become prevalent in the x86 world may be years, if ever. I hope it happens, I've said many times it is absolutely absurd that x86 and the market it serves is largely being restricted to only two vendors, one being only a fringe player.

Were AMD to falter I doubt that Via would sit still and not move to re-enter the mainstream performance tier.

Likewise if AMD were to falter I doubt that ARM would not make a move to roll out a seriously high-power version of its architecture. Or IBM would roll out a desktop version of its Power7. Or Oracle would enter with a Sparc version for the consumer space.

IBM I am pretty sure will never enter the consumer processor space again (consoles being the anomaly), no opinion on the others listed, except VIA has no chance.

There's a lack of competitors now, but that's because there are currently much more profitable grounds for those alternative architectures to compete in.

Looking at the profits Intel is raking in and sheer volume they ship, the x86/desktop/workstation space is VERY profitable, although the price of entry is pretty scary. And I've said it over and over again, no one wants to compete head to head with Intel. If it was a viable thing to do, you can bet there would be companies lining up. It doesn't make sense that the only thing keeping other competitors out is AMD.

In the vacuum that would come from AMD's faltering I have every confidence that such a vacuum would gain attention and notice by any number of existing, well monied, microprocessor manufacturers.

Like who? Also who wants to be another AMD, spending so much on R&D, only to end up with the scraps of the market?

The key for AMD, or Via, or Oracle, or Nvidia (ARM), is GloFo and TSMC. If the foundries can't develop competitive process technology on a timeline that parallels that of Intel's then none of these alternative architectures stand a chance at competing with Intel.

It's been proven that no one can out spend Intel in fabrication, so unless Bill Gates himself decides to get into the fab business, Intel will continue to lead the industry. AMD/GloFo has done remarkably well considering their relatively miniscule budget, in fact at a given node, GloFo is quite close to what Intel offers, especially when their process matures. Problem is they are always 12+ months behind Intel in moving to a smaller process.

 

[Abwx]

Abwx,

you seem to be on this forum exclusively to defend year 2007 performance in the year 2011, and I don't understand why. The Phenom-II would've been a stellar chip in 2006, a great one in 2007, but still a midrange one in 2008. AMD failed by making a C2Q competitor as Nehalem arrived, and it wouldn't be here if not for the ATI buy. We'll see bulldozer pretty soon, and I have high hopes for it, but let's not kid ourselves, kid... The greats at AMD jumped ship a long time ago.

This argument over compilers and optimizations doesn't matter one bit. Anyone buying CPUs for a specific application will be buying serious chips, not desktop ones. Intel, unfortunately, kills in both spaces.

Daimon

Fairly right , but this is an environement were competition
is extremely ferocious despite there s only two firms that are
competing, although this unbalanced duopoly is a by product
of a progressive elimination of many firms from this market,
and i was pointing that Intel used falsehood as a mean to
downplay the competition.

After all, when the opteron 64 was king, Intel remained
first in server marketshare with P4 based xeons that
were vastly inferior.

Hope that Bulldozer will somewhat have a few traits
of its glorious predecessor...

 

[Idontcare]

That is what all was about, but then :




And you know what they can deliver thanks to benchmarks,
yet , some notorious benches as Sysmark were completely
modded as soon as it did show Amd s product being quite
competitive.

So , how can you know what the product deliver if benches
are rigged this way ?..

Its simple, I do what any other well-intended level-headed individual would do, I bench them with my own apps and decide based on the very performance data I care about:

Its not your job to defend the world's less-motivated consumers against the marketing machine that is Intel and Bapco. That is AMD's job.

Under-motivated consumers will get exactly what they deserve, name an industry where this is not true.

 

[dac7nco]

this unbalanced duopoly is a by product
of a progressive elimination of many firms from this market,
and i was pointing that Intel used falsehood as a mean to
downplay the competition.

After all, when the opteron 64 was king, Intel remained
first in server marketshare with P4 based xeons that
were vastly inferior.

Time and Inertia, Abwx, time and inertia.

In the A64 era AMD was the only choice if you wanted performance, and Hypertransport literally changed my world. In my time I've used x86 chips from TI, AMD, NexGen... and going onward. I use Sparc and Power chips at work, and my favorite machine of all time ran on a Motorola 68040.

I lost a LOT of money when Irving Gould ran Commodore into the ground in the early '90's, and a lot of other people did too. Commodore was to slow and too late... It takes a PLAN to deal with the big picture, and AMD doesn't have that. If Intel was interested in wrecking AMD's CPU division, it could & would. Intel is likely considered "critical infrastructure" by the U.S., and would get a slap on the wrist by selling 2500Ks at $100 until AMD was done for. AMD may not get the same backing since they sold their semicon division to a middle-eastern country. The tarriffs the EU would give Intel would be a joke: what are they going to do... by CPUs from Olivetti?

ALL corporations eliminate their competitors by shady dealings. Any community college economics course will tell you that American capitolism is survival of the wickedest.

Daimon

 

[Idontcare]

I've thought about this as well, but the timeline for ARM to become prevalent in the x86 world may be years, if ever. I hope it happens, I've said many times it is absolutely absurd that x86 and the market it serves is largely being restricted to only two vendors, one being only a fringe player.

IBM I am pretty sure will never enter the consumer processor space again (consoles being the anomaly), no opinion on the others listed, except VIA has no chance.

Looking at the profits Intel is raking in and sheer volume they ship, the x86/desktop/workstation space is VERY profitable, although the price of entry is pretty scary. And I've said it over and over again, no one wants to compete head to head with Intel. If it was a viable thing to do, you can bet there would be companies lining up. It doesn't make sense that the only thing keeping other competitors out is AMD.

Like who? Also who wants to be another AMD, spending so much on R&D, only to end up with the scraps of the market?

It's been proven that no one can out spend Intel in fabrication, so unless Bill Gates himself decides to get into the fab business, Intel will continue to lead the industry. AMD/GloFo has done remarkably well considering their relatively miniscule budget, in fact at a given node, GloFo is quite close to what Intel offers, especially when their process matures. Problem is they are always 12+ months behind Intel in moving to a smaller process.

If this were true, that the future is destined to be as we see it to be from our vantage point, then Intel itself would never have come about because they were attempting the impossible in their time by competing against the very likes of Fairchild Semiconductor, IBM, Cray, DEC, etc.

The Intel of today is not assured to be the Intel of tomorrow. If this were 1960 and you were attempting to prognosticate who would be the technology leaders of 1970 or 1980 or 1990 you would have had an equally dismal outlook for any company by the name of Microsoft of Nvidia or Intel.

I'm not worried about who (or how) an alternative to Intel will come to be, I need only look to the past to acknowledge the likelihood that one (or more) competitors shall come to be in the future.

You might laugh at this, but there once was a time when Britain ruled the sea and the sun never set on her imperial lands. Those people died, the people that replaced them weren't necessarily cut from the same cloth. Andy Grove and Paul Otellini won't live forever. Intel will hire their own Hector Ruiz for CEO someday, as DEC did and as Fairchild Semi did, and history will repeat itself.

 

[Abwx]

Its simple, I do what any other well-intended level-headed individual would do, I bench them with my own apps and decide based on the very performance data I care about:

That kind of selection tool is out of reach for common people.
The most "critical" are those who would ask a friend that is
a geek, and this latter will respond according to the benches
hanging here and there while most will simply ask some Bestbuy s
recently hired techno guy..

Its not your job to defend the world's less-motivated consumers against the marketing machine that is Intel and Bapco. That is AMD's job.

Under-motivated consumers will get exactly what they deserve, name an industry where this is not true.

Right, but in a geek site , we can perhaps point the marketing
claims innaccuracies, it wont hurt the quality of the debate, quite
the contrary.

 

[Abwx]

DP

 

[Abwx]

Time and Inertia, Abwx, time and inertia.

In the A64 era AMD was the only choice if you wanted performance, and Hypertransport literally changed my world. In my time I've used x86 chips from TI, AMD, NexGen... and going onward. I use Sparc and Power chips at work, and my favorite machine of all time ran on a Motorola 68040.

I lost a LOT of money when Irving Gould ran Commodore into the ground in the early '90's, and a lot of other people did too. Commodore was to slow and too late... It takes a PLAN to deal with the big picture, and AMD doesn't have that. If Intel was interested in wrecking AMD's CPU division, it could & would. Intel is likely considered "critical infrastructure" by the U.S., and would get a slap on the wrist by selling 2500Ks at $100 until AMD was done for. AMD may not get the same backing since they sold their semicon division to a middle-eastern country. The tarriffs the EU would give Intel would be a joke: what are they going to do... by CPUs from Olivetti?

ALL corporations eliminate their competitors by shady dealings. Any community college economics course will tell you that American capitolism is survival of the wickedest.

Daimon

At the other end, competition allowed us to have really versatile
machines with relatively open standards that allow for affordable prices.

Back in 1990 , an Intel based 486 DX2 66 PC was some 5000$...