Is the DARKSIDE that much better?

The AMD Athlon XP was far less competitive with the Intel Northwood Pentium 4's than the Athlon 64 was. All three versions of the Athlon 64 (S754 S940 (FX-series) & S939) were better performers than the Pentium 4 was in almost every application.

That is how I remember things.

Intel had already developed IA-64 but that wasn't available in the consumer space and it was awhile before EM64T was integrated into the Pentium 4 line and enabled.

I remember around this time (2003 - 2004) Intel was publicly stating that they were not going to produce a 64 bit x86 however after AMD started chipping away at their server market their position on this changed. Good thing for AMD users at that time was that Microsoft forced (would not implement a second x86_64) Intel to use the same instruction set that AMD had for x86_64.
 
EDIT:
The two points I"m trying to make here are that AMD can be underlooked and underappreciated, and that Intel is viewed as an evil empire because some people don't like all the power concentrated in once place.

You missed my point.

The REASON AMD doesn't have money, has financial issues, is BECAUSE they started making inferior chips. They started bringing their price up to Intel's and couldn't keep up with the performance and they lost market share.

I get what your saying, but at the same time I see AMD as the dark side, because they are in the dark right now, trying to find the light but not having much luck.
 
The REASON AMD doesn't have money, has financial issues, is BECAUSE they started making inferior chips.

I believe the largest reason is they got overconfident. They had a two or so really good years where they could do no wrong, they had the fastest chips, their stock price and the market share were both at all time highs. They assumed this would continue and made a very bad decision in buying ATI when they did. This ATI purchase has been a serious burden on their financial situation. Consider the fact that AMD paid 5.4 billion for ATI and 2 years later the combined company is worth 1.2 billion. On top of that the Phenom had two major issues that delayed it over 1 year. One issue is that it took too much power so it could not be clocked at 3 GHz. The other issue was the TLB fiasco. To make matters worse Intel hit a homerun with core2. And finally Intel is also causing AMD serious pain by transitioning to smaller process nodes at a faster rate. AMD with their significantly smaller budget can not afford to retool its FAB(s) every year.
 
I do not know where people are getting their history from. The AMD Athlon XP was far less competitive with the Intel Northwood Pentium 4's than the Athlon 64 was. All three versions of the Athlon 64 (S754 S940 (FX-series) & S939) were better performers than the Pentium 4 was in almost every application. The Athlon 64 X2 compared as well against the Pentium D as the Athlon 64 did against the Pentium 4 considering that both chips were the same as their predecessors just with dual cores in the same package. I know the Athlon 64 X2 was a "native" approach but that didn't mean squat in the real world. The Athlon 64 and Athlon X2 provided the most decisive victory regarding performance AMD has ever seen against their competition. Plus the Athlon 64 used less power and generated less heat than the Pentium 4 Prescott CPUs did. It was also first with x86-64 instruction sets. Intel had already developed IA-64 but that wasn't available in the consumer space and it was awhile before EM64T was integrated into the Pentium 4 line and enabled.

To be fair the last time I built an AMD rig for personal use was over 4 years ago, for my sister (Athlon 64 3000), and that was quick, still is cause she's using it to this day, I bought my AMD Athlon XP based off of benchmarks at the time (it wasn't doing that badly against the P4 of that time, from what I remember, although I may have bought it rooting for the underdog at that point), and my Athlon 700 was faster than it's Intel comparison as I spent over a month digging up reviews and results all over the net before I finally bought it, a lot of sites said it was as good as buying the Intel 733MHz and cheaper too (that's still running in my old room back home as an office system for my dad).
 
That is how I remember things.



I remember around this time (2003 - 2004) Intel was publicly stating that they were not going to produce a 64 bit x86 however after AMD started chipping away at their server market their position on this changed. Good thing for AMD users at that time was that Microsoft forced (would not implement a second x86_64) Intel to use the same instruction set that AMD had for x86_64.

Exactly.

I believe the largest reason is they got overconfident. They had a two or so really good years where they could do no wrong, they had the fastest chips, their stock price and the market share were both at all time highs. They assumed this would continue and made a very bad decision in buying ATI when they did. This ATI purchase has been a serious burden on their financial situation. Consider the fact that AMD paid 5.4 billion for ATI and 2 years later the combined company is worth 1.2 billion. On top of that the Phenom had two major issues that delayed it over 1 year. One issue is that it took too much power so it could not be clocked at 3 GHz. The other issue was the TLB fiasco. To make matters worse Intel hit a homerun with core2. And finally Intel is also causing AMD serious pain by transitioning to smaller process nodes at a faster rate. AMD with their significantly smaller budget can not afford to retool its FAB(s) every year.

Not quite. i don't know if it was really a matter of being over confident. In fact I think if anything, the Athlon 64 caught Intel off gaurd and that is the reason for AMD's success during that time. It isn't that the Athlon 64 was that good but rather because the Pentium 4 was really just that bad. This is obviously in regard to performance alone and not sales wise.

You are correct though. When it comes to manufacturing Intel pretty much sets the bar most of the time in the world of semi-conductors. Very few companies can do what they do and do it as often or as fast. On the subject of the Phenom though, the TLB bug did NOT delay the Phenom processor. Not at all. That bug was discovered after the Phenom had been released. This no doubt cost AMD quite a few server sales which obviously hurt them. You can also blame the acquiisition of ATI as part of the reason for their financial woes. ATI wasn't doing so hot product wise when AMD picked them up. This trend continued until the Radeon 4870 and 4850 came out which wasn't all that long ago.

To be fair the last time I built an AMD rig for personal use was over 4 years ago, for my sister (Athlon 64 3000), and that was quick, still is cause she's using it to this day, I bought my AMD Athlon XP based off of benchmarks at the time (it wasn't doing that badly against the P4 of that time, from what I remember, although I may have bought it rooting for the underdog at that point), and my Athlon 700 was faster than it's Intel comparison as I spent over a month digging up reviews and results all over the net before I finally bought it, a lot of sites said it was as good as buying the Intel 733MHz and cheaper too (that's still running in my old room back home as an office system for my dad).

The Athlon XP really competed against the Pentium 4 socket "Willamette" parts which sucked. Clock for clock the Pentium III had them beat. The only thing it could do well was scale clock speed wise. The Athlon XP was pretty much taking a backseat to the Pentium 4 socket 487 "Northwood" CPUs once they released. It was when the Athlon 64 came out that AMD really supplanted their performance crown. Though video encoding was still the Pentium 4's forte it was just about the only thing they beat the Athlon 64 at.

On the subject of the original Athlon, yes, it was faster than the Pentium III was. However you couldn't have paid me to run one of those in my rig. (Well you could have, but it would have been expensive. :) ) because the motherboards were SHIT. They had under spec'ed voltage components and many of them were catching fire. We had stacks of dead boards from NEC's, Packard Bell's, HP's, Compaq's etc, due to these problems. All the boards were substandard. Of course the next generation of boards were better but the chipsets and chipset drivers sucked horse anus. It wasn't until NVIDIA stepped in that this began to change. The original Athlon also ran hotter than the Pentium III did as I recall and it used more power. Again I think there were a few benches the Pentium III still cleaned house in, but I don't remember for sure how many and what those were. I'm too lazy to research it and frankly I don't care at this point.
 
The strength of the Athlon/Athlon XP was not that they were faster than Pentium III or 4 per se, but that they gave you pretty much the same performance for less money. A lot less in some cases.

These days AMD still tries to go the 'less money' road, but since the performance is no longer there, AMD just drops off the radar for a lot of buyers... Most people can afford a mid-range Core2 system, like the Q6600 or such. So unlike in the days of Pentium 4 and rambus and such, people aren't really looking for a cheap alternative to Intel. Intel provides the cheap alternatives itself. The heat issues are now on AMD's side aswell, rather than on Intel's side.
 
On the subject of the Phenom though, the TLB bug did NOT delay the Phenom processor. Not at all. That bug was discovered after the Phenom had been released.

I think it delayed AMD from ramping up the speeds. AMD surely wanted to ramp the speeds up a bit after the intro and this bug prevented them from selling anything faster than 2.4 GHz for several months.
 
ATI wasn't doing so hot product wise when AMD picked them up. This trend continued until the Radeon 4870 and 4850 came out which wasn't all that long ago.

I agree. I believe the point of buying ATI was not just for graphics cards. It was for the killer CPU+GPU hybrid that they were supposed to deliver late next year. My reasoning here is they could have waited and got a better price (even last year).
 
but rather because the Pentium 4 was really just that bad
I think Intel underestimated its competition. And then there was the problem with IA-64, they had invested over 10 billion dollars into IA-64 that and needed it to be far ahead of x86 cpus in server space so that they could convince people to adopt a new binary incompatible server only platform. P4 was initially planned to be ramped up to 5GHz+ but I believe the plans were to do that at a much slower pace. AMD's competition caused intel to rethink its plans.
 
These days AMD still tries to go the 'less money' road, but since the performance is no longer there

Perhaps that will change in the next year. However when Intel moves to the 32nm process and 6 to 8 core cpus there will be a lot more pressure on AMD and it will probably be back to where it was before the Phenom 2 was released.
 
Don't buy a q6600. Their slow and are going to be EOL soon. If you have to have a quad, the 9xxx or Core i7
 
I think it delayed AMD from ramping up the speeds. AMD surely wanted to ramp the speeds up a bit after the intro and this bug prevented them from selling anything faster than 2.4 GHz for several months.

The TLB bug had nothing to do with the slow speeds Phenom was released at. That was due entirely to the amount of heat the chips put out; the initial Phenoms had a TDP of 130W while being clocked at 2.3GHz or lower, while all of Intel's quad-core CPUs had a TDP of 95W, were clocked considerably higher, and performed much better as well.

AMD had a ton of problems with their 65nm process, which is also one of the reasons why the 65nm Athlon X2s tended to actually be worse overclockers than their 90nm counterparts, as opposed to Intel's CPUs which were able to obtain consistently higher overclocks with each process shrink.
P4 was initially planned to be ramped up to 5GHz+ but I believe the plans were to do that at a much slower pace. AMD's competition caused intel to rethink its plans.

Netburst was designed to scale up to 10GHz. Intel originally introduced their first dual-core CPUs because they realized 10GHz wasn't feasible, and they needed some way to scale the architecture to faster designs. Then they started developing the Core architecture.
 
The TLB bug had nothing to do with the slow speeds Phenom was released at. That was due entirely to the amount of heat the chips put out; the initial Phenoms had a TDP of 130W while being clocked at 2.3GHz or lower,

That is not exactly what I remember. I do remember this was the claim for low ferquency at one point but looking at the following link (and the follow up links) I am not sure what the real reason was being that AMD was releasing conflicting information:

http://techreport.com/discussions.x/13724

Remember back then TDP for AMD was maximum possible power and not the same as what Intel called TDP. I believe not releasing chips because they have a 130W maximum power was a mistake for AMD. I mean some Hyperthreading P4s got that high with out overclocking.
 
Intel originally introduced their first dual-core CPUs because they realized 10GHz wasn't feasible

The big problem is it consumed significantly more power compared to AMD X2 at a similar level of performance.
 
Remember back then TDP for AMD was maximum possible power and not the same as what Intel called TDP. I believe not releasing chips because they have a 130W maximum power was a mistake for AMD. I mean some Hyperthreading P4s got that high with out overclocking.
AMD's CPUs run much closer to their TDPs than Intel chips do. Most Intel CPUs actually consume significantly less power than they're rated for most of the time. AMD chips on the other hand have much less conservative ratings and often run close to their TDPs under load.
The big problem is it consumed significantly more power compared to AMD X2 at a similar level of performance.

No, the problem was that the P4s and PDs didn't have a similar level of performance compared to the A64s and X2s. The difference in power consumption wasn't actually that large, but Intel CPUs at that time just didn't have the performance to go along with their clock speeds.
 
AMD's CPUs run much closer to their TDPs than Intel chips do. Most Intel CPUs actually consume significantly less power than they're rated for most of the time. AMD chips on the other hand have much less conservative ratings and often run close to their TDPs under load.

I have not seen this to be the case with dozen or so X2 cpus I have at work. If I believe the UPS on my desktop at work it consumes around 65W total system under full load and that is with a 65W 2.9GHz 5600+ + 4GB of DDR2 + 1 750GB hard drive and a nvidia 256 MB 7200GS and it does with cpu temp of less than 45C and mid to high 20s at idle.

It was quite a shocked to see my Intel Q9550 idling over 35C but then again it is supposed to run up till at least 80C. My X2s would smoke long before that...

Its pretty much idle now:
jmd0 ~ # sensors
coretemp-isa-0000
Adapter: ISA adapter
Core 0: +47.0 C (high = +76.0 C, crit = +100.0 C)

coretemp-isa-0001
Adapter: ISA adapter
Core 1: +44.0 C (high = +76.0 C, crit = +100.0 C)

coretemp-isa-0002
Adapter: ISA adapter
Core 2: +45.0 C (high = +76.0 C, crit = +100.0 C)

coretemp-isa-0003
Adapter: ISA adapter
Core 3: +37.0 C (high = +76.0 C, crit = +100.0 C)

jmd0 ~ # cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 23
model name : Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz
stepping : 7
cpu MHz : 2003.000
cache size : 6144 KB
 
Don't buy a q6600. Their slow and are going to be EOL soon. If you have to have a quad, the 9xxx or Core i7

Wow, no wonder people make stupid decisions.

This couldn't be FURTHER from the truth.

The q6600 is far from slow and MORE than nesc. for most people at a very affordable price point. It is a great overclocker as well.

Nothing wrong with going 9xxx or core i7, but to say they are slow is just plain silly.
 
BTW, since I was posting temps. Here is the system that the Q9550 replaced. Both are up 24/7 and in the same room:

Here are the idle temps:
jmd1 mythtv # sensors
k8temp-pci-00c3
Adapter: PCI adapter
temp1: +26 C
temp3: +28 C

jmd1 mythtv # cat /proc/cpuinfo
processor : 0
vendor_id : AuthenticAMD
cpu family : 15
model : 33
model name : Dual Core AMD Opteron(tm) Processor 285
stepping : 2
cpu MHz : 1000.000
cache size : 1024 KB

And as for power I have said the Q9550 system idles at ~100W with 3 hard drives + nforce 7200gs ? + kworld 115 ATSC tuner. And the 2 core 285 with 4GB of pc 3200 idles at around 150W with 4 drives and nvidia 4200 ti card and a hauppage pvr 150.
 
I think it delayed AMD from ramping up the speeds. AMD surely wanted to ramp the speeds up a bit after the intro and this bug prevented them from selling anything faster than 2.4 GHz for several months.

No, the clock speeds had nothing to do with the TLB bug. *IF* AMD could have released faster clocked CPUs, they would have. From the very beginning AMD had problems getting the Phenom clock speeds where they needed to be.

I think Intel underestimated its competition. And then there was the problem with IA-64, they had invested over 10 billion dollars into IA-64 that and needed it to be far ahead of x86 cpus in server space so that they could convince people to adopt a new binary incompatible server only platform. P4 was initially planned to be ramped up to 5GHz+ but I believe the plans were to do that at a much slower pace. AMD's competition caused intel to rethink its plans.

I don't know about 10 billion dollars invested in IA-64 but the rest of it is about right. Except they said that Netburst was going to be good to reach 10GHz. Which never happened. So I think they over estimated their design and of course what AMD was doing made them re-think their plans.

Perhaps that will change in the next year. However when Intel moves to the 32nm process and 6 to 8 core cpus there will be a lot more pressure on AMD and it will probably be back to where it was before the Phenom 2 was released.

I wouldn't count on much out of Phenom II.

AMD's CPUs run much closer to their TDPs than Intel chips do. Most Intel CPUs actually consume significantly less power than they're rated for most of the time. AMD chips on the other hand have much less conservative ratings and often run close to their TDPs under load.


No, the problem was that the P4s and PDs didn't have a similar level of performance compared to the A64s and X2s. The difference in power consumption wasn't actually that large, but Intel CPUs at that time just didn't have the performance to go along with their clock speeds.

Right.

The big problem is it consumed significantly more power compared to AMD X2 at a similar level of performance.

They consumed more, I don't know how significant the increase was. I know many of the Pentium D's actually consumed less power than the Pentium 4 did on a core vs. core comparison.

I have not seen this to be the case with dozen or so X2 cpus I have at work. If I believe the UPS on my desktop at work it consumes around 65W total system under full load and that is with a 65W 2.9GHz 5600+ + 4GB of DDR2 + 1 750GB hard drive and a nvidia 256 MB 7200GS and it does with cpu temp of less than 45C and mid to high 20s at idle.

It was quite a shocked to see my Intel Q9550 idling over 35C but then again it is supposed to run up till at least 80C. My X2s would smoke long before that...

Its pretty much idle now:
jmd0 ~ # sensors
coretemp-isa-0000
Adapter: ISA adapter
Core 0: +47.0 C (high = +76.0 C, crit = +100.0 C)

coretemp-isa-0001
Adapter: ISA adapter
Core 1: +44.0 C (high = +76.0 C, crit = +100.0 C)

coretemp-isa-0002
Adapter: ISA adapter
Core 2: +45.0 C (high = +76.0 C, crit = +100.0 C)

coretemp-isa-0003
Adapter: ISA adapter
Core 3: +37.0 C (high = +76.0 C, crit = +100.0 C)

jmd0 ~ # cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 23
model name : Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz
stepping : 7
cpu MHz : 2003.000
cache size : 6144 KB

I wouldn't believe your UPS at work. 65w seems way too low to me for idle power on that system.

Wow, no wonder people make stupid decisions.

This couldn't be FURTHER from the truth.

The q6600 is far from slow and MORE than nesc. for most people at a very affordable price point. It is a great overclocker as well.

Nothing wrong with going 9xxx or core i7, but to say they are slow is just plain silly.

Agreed.
 
I wouldn't believe your UPS at work. 65w seems way too low to me for idle power on that system.

I meant 65W on the 2.9Ghz amd X2 and not the Intel Q9550. The Q9550 system draws ~100W at idle (measured with a kill a watt).
 
I meant 65W on the 2.9Ghz amd X2 and not the Intel Q9550. The Q9550 system draws ~100W at idle (measured with a kill a watt).

Are you talking about the processor alone or the whole system? 65w for the system seems really low. Even for the X2.
 
Are you talking about the processor alone or the whole system? 65w for the system seems really low. Even for the X2.

Whole system under full load. Sometime I will have to bring the Kill a watt in to test the accuracy of this.
 
Except they said that Netburst was going to be good to reach 10GHz. Which never happened.

My opinion (boy I am full of them - you guys probably think I am full of something else...), anyways is that without the competition from AMD, Intel would have a 5GHz Netburst processor running under the 65nm process right now instead of core2 and i7 running under the 45nm process with the 32nm process less than a year away.
 
My opinion (boy I am full of them - you guys probably think I am full of something else...), anyways is that without the competition from AMD, Intel would have a 5GHz Netburst processor running under the 65nm process right now instead of core2 and i7 running under the 45nm process with the 32nm process less than a year away.

I didn't say that Intel didn't need competition or it wasn't a good thing.

With that said, in order for me to buy AMD instead of Intel AMD is going to have to earn it. They are going to have to provide a processor that is either faster than Intel's in the applications I run, or it will have to offer comparable performance for less money. I need more advantages than disadvantages going with AMD. Right now they do not provide that.
 
I need more advantages than disadvantages going with AMD. Right now they do not provide that.

Fully agree. That is one reason why I bought my first Intel cpu at home in 7 years and one for work as well.

To me the future does not look great for AMD unless they either get bought out (at 1.2 Billion they are ripe for the picking if the whole market was not in serious trouble) or unless they manage to ramp Phenom 2 to significantly higher frequencies in 2009. They keep talking about overclocking Phenom 2 chips to 4 GHz on air, increase the allowable motherboard power envelope and sell 3.5Ghz chips even if users have to get a special 175W mobo for $300.
 
Whole system under full load. Sometime I will have to bring the Kill a watt in to test the accuracy of this.
65W from the wall for that system at full load is not possible. Even assuming a single HD, that would only leave about 30W DC draw for the rest of the system once efficiency is factored in. Kill-A-Watts and other similar power-measuring devices (including the one in your UPS) are known to sometimes report very inaccurate readings when dealing with Active PFC PSUs, which is probably what's happening here.
My opinion (boy I am full of them - you guys probably think I am full of something else...), anyways is that without the competition from AMD, Intel would have a 5GHz Netburst processor running under the 65nm process right now instead of core2 and i7 running under the 45nm process with the 32nm process less than a year away.

No. Intel abandoned Netburst when it became apparent that they simply could not get it to scale higher within a reasonable heat and power envelope. They couldn't even hit 4GHz on a production P4. Higher-clocked P4s were actually competitive with the A64s, so if Intel could have produced faster chips, they wouldn't have needed to come up with a new architecture. Even with the 65nm Cedar Mill core, Intel couldn't get higher than 3.8GHz. 5GHz was simply impossible with that design.
They keep talking about overclocking Phenom 2 chips to 4 GHz on air, increase the allowable motherboard power envelope and sell 3.5Ghz chips even if users have to get a special 175W mobo for $300.
It's not just about power, it's about heat. It takes a pretty beefy cooling setup to deal with a 175W CPU. Plus, a 3.5GHz chip would require a lot more voltage than a lower-frequency one.
 
No. Intel abandoned Netburst when it became apparent that they simply could not get it to scale higher within a reasonable heat and power envelope.

My reasoning was at a slower pace they would have fixed the problem with netburst.
 
My reasoning was at a slower pace they would have fixed the problem with netburst.

But they wouldn't have. They already tried to fix the issues by giving the design a longer pipeline and moving to a smaller process (twice). It didn't work. That's why they went with Core 2. Intel stuck with Netburst for a very long time; a lot longer than they should have. Not that Netburst was a bad design - I don't consider it to be a failure the way many other people do. It's just that Intel tried to drag it out for longer than it was good for.
 
Only closed minded AMD fanboys would still build a AMD system at this point in time.

That's not necessarily true. AMD is still competitive in several market segments. They just aren't competitive at the high-end.
 
That's not necessarily true. AMD is still competitive in several market segments.

I consider them very competitive in the < $90 dual core cpu market. I have built a dozen or so such XPSP3 boxes (asus m2n or m3n + 4GB corsiar XMS2 + 750GB Seagate 7200.11+ Antec case + 2.6GHz or greater X2 + some $30 nVidia card) for our developers here at work which translates to sub $500 total machines (no where near the deal I got on eBay but I can not buy there for work). Anyways for what the users do 95% of their time these machines are excellent.
 
If you want to build the absolute most bottom dollar rig, then yes AMD is very competitive. If your goal is performance oriented then AMD's offerings are much less attractive when compared to Intel's offerings. More heat, less performance, less overclocking gains, a worse motherboard platform, (albiet a cheaper one) no thanks.
 
well, I went with a Q6600 and have not been sorry for it... I do still like my FX-60 but am working on finishing up the Q6600 system... and my new Gateway 6860 Laptop... Love that thing... thanks for everything and will chat with ya all later. Happy holiday and have a happy new year.!
 
Yes, it is. I have just built my first INTEL/Nvidia rig since Celeron 300MHz and Geforce mx400 times... And it comes from an AMD/ATI fanboy...

My video cards: HIS ICEQ HD4850<Visiontek HD3870<Asus ATI 1950Pro<ATI x700 Mobile<ATI Radeon 9800Pro<ATI Hercules 8500 (fantastic 2d picture quality)....

My Porcessors: AMD x2 5400+ BE, x2 3800+, San Diego 4000+, AMD Barton Mobile 2500+ @ 3200+, AMD Athlon 2000+ Thoroughbred, AMD Athlon Thunderbird 1000MHZ...

To sum it up, if I decided to switch to the "Darkside", it means things are pretty bad.
 
If you want to build the absolute most bottom dollar rig, then yes AMD is very competitive. If your goal is performance oriented then AMD's offerings are much less attractive when compared to Intel's offerings. More heat, less performance, less overclocking gains, a worse motherboard platform, (albiet a cheaper one) no thanks.

Well, I think you're exactly right. And for the vast majority of people shopping for a computer, price is the #1 factor in buying a PC.

I was noticing this at Wal-Mart and Best Buy when looking at laptops for my brother. Most computers in retail stores don't have Core-2 processors. They have AMDs. That's because most people are just shopping based on price, and an AMD chip will do pretty much everything they need to do.

So, AMD as a company is very competitive in certain segments, because not all markets are purely performance based.

P.
 
Back
Top