Switching to the dark side....

[MrTransistorm]

I remember right before Conroe officially launched, when Apple switched from PowerPC chips to x86 architecture, hearing that they had inked a deal with Intel and wondering WTF they were doing. At that moment AMD had far better overall performance and much better performance/watt. Of course, by the time the new Apple machines rolled off the line, they contained the shiny new C2D/Q chips and it all made much better sense.

Your memory is a little hazy. The first Intel Macs in early 2006 had Core Duo (Yonah) CPUs:

MacBook
MacBook Pro
iMac

EDIT: Sorry, I may have misread your post a little. At first it seemed like you were saying that the first Intel Macs had C2D. If that's not what you meant, then I apologize. :whiste:

 

[XiandreX]

My last AMD cpu was a Phenom II 940 black. It was decent but after shifting to Intel I am hard pressed to see a reason why anyone would use AMD in the performance segment.

I am talking gaming and the likes. Yes the Cpu's are more expensive but run cooler,
single threaded performance is vastly superior and consume less electricity.
Seems like AMD will have to pull an incredible rabbit out of the hat to catch up.

 

[tential]

My last AMD cpu was a Phenom II 940 black. It was decent but after shifting to Intel I am hard pressed to see a reason why anyone would use AMD in the performance segment.

I am talking gaming and the likes. Yes the Cpu's are more expensive but run cooler,
single threaded performance is vastly superior and consume less electricity.
Seems like AMD will have to pull an incredible rabbit out of the hat to catch up.

I just buy the best CPU for the task. Right now that's Intel for the performance segment.

This fanboyism I'll never understand. I have 0 brand loyalty on a PC. I won't lose performance just because "I bought AMD before so I'll buy again." I'll read the reviews and make the relevant decision from there. When AMD is in the lead in the performance segment I'll buy them. Right now there isn't an AMD chip where the loss of performance for me is worth the price drop. Even when comparing Kabini to the J1900/J1800 I'll take the J1900 EVERY time due to the better benches in webbrowsing/office productivity which is what I'd use those processors for.

 

[VirtualLarry]

Even when comparing Kabini to the J1900/J1800 I'll take the J1900 EVERY time due to the better benches in webbrowsing/office productivity which is what I'd use those processors for.

I thought that Kabini still had higher single-threaded IPC than Bay Trail?

 

[DrMrLordX]

Arh please, could have is a bad excuse. JEDEC set the demands too high for RAMBUS to accept it. And JEDEC members didnt have much interest in leapfrog memory performance. The DRAM cartel only wanted status quo to avoid having to change production lines. Low risk, high profit was all they wanted.

Hey, whatever floats your boat. Personally I could care less what the JEDEC members did back then . . . I was too busy watching review sites pan RDRAM as being nearly useless for Pentium 3 systems.

More BS. RAMBUS wanted to sell RDRAM to everyone. AMD still got a RDRAM license today. SIS for example licensed it in 2001 as well.

SiS had several interesting licenses back in the day. AMD having an RDRAM license is news to me, though. Got a link?

What you defend is a criminal and convicted price cartel that not only hurt innovation but also technological progress.

Who's defending the JEDEC members? I'm not. I doubt it hurt innovation or technological progress by all that much, though, but it's all ancient history now.

Just because you saw yourself angry on Intel. And today we pay the price for it. We sit with huge bottlenecks and subpair memory technology. And DDR4 is a showcase on that the pee in the pants got cold now as well. GDDR is at its end too, and desperately hope for HBM/HMC etc to come soon. Since they only managed to double the bandwidth the last 6 years.

Actually, JEDEC memory specs have been a bit funny over the last 4-5 years. Yeah, we still have a lot of systems stuck on DDR3-1333 or DDR3-1600, but you can actually go out and get DDR3-3000 if you want to pay for it. DDR3 has legs, it's just that the upper end of performance isn't being pushed that hard as a standard.

If AMD or Intel wanted to push DDR3-2133 as their standard for memory for all desktop PCs, the board manufacturers could handle it, the CPUs could handle it, and the DRAM manufacturers could probably supply it. It would take voltages above 1.5v in most cases, but it would work. I think the reason why AMD (for example) hasn't done that is the cost associated with using RAM of that kind.

Its the chicken and the egg. And we had the same with all previous new serial standards too. We get it before we need it. Yet you keep defend inferiour technology due to your personal issues with Intel.

But, that's the problem, AMD chips arguably never needed it, and Intel pretty much designed a CPU around it (which wound up on dual-channel DDR anyway, because it wound up being better than existing RDRAM).

Again you are completely wrong. Not to mention Slot A and Slot 1 are similar. What you refuse to accept is that companies and their CPU uarchs are simply too incompatible at this point. And too many sacrifices would have to be made to support the same socket. Intel for example used GTL+ while AMD used EV6 protocol.

And here I am kicking myself for not including a link. Was Steve Ditlea wrong, all those years ago?

I remember right before Conroe officially launched, when Apple switched from PowerPC chips to x86 architecture, hearing that they had inked a deal with Intel and wondering WTF they were doing. At that moment AMD had far better overall performance and much better performance/watt. Of course, by the time the new Apple machines rolled off the line, they contained the shiny new C2D/Q chips and it all made much better sense.

It was kind of a shock, though you could sort of see it coming with all the Dothan overclocking and reports on Yonah. Yonah was a pretty good chip, all things considered. Not a Conroe, mind you, but it had its merits. In any case, you could tell that Intel had some serious non-Netburst stuff going on that looked pretty good.

Of course, once the Conroe ES samples were in the wild, getting benched, yeah it wasn't hard to understand that move by Apple at all.

This fanboyism I'll never understand. I have 0 brand loyalty on a PC.

Brand recognition and brand loyalty are big deals in marketing. It is not uncommon for people to continue buying a product from a particular vendor due to one or two positive experiences. It is also not uncommon for people to avoid a specific brand due to some negative experience. Rational thought factors into buying decisions less often than you might think.

I can sort of understand people buying AMD products to try and "keep competition alive", but to me, the only reason to adopt an AMD platform recently has been to tinker with something different or weird.

 

[DigDog]

So were do we put VIA in this whole Dark/Light side (admittely very Star Warsy) thing?

no, they are the ewoks.
cute and useless.

 

[ShintaiDK]

SiS had several interesting licenses back in the day. AMD having an RDRAM license is news to me, though. Got a link?

http://www.internetnews.com/infra/article.php/3574856/AMD+Licenses+Rambus+Patent.htm
http://www.electronicsnews.com.au/news/amd-rambus-renew-patent-licence-agreement

Not to mention Cirrus Logic, Nintendo, Sony and so on. The N64, PS2 and PS3 used rambus products.

And here I am kicking myself for not including a link. Was Steve Ditlea wrong, all those years ago?

Wrong? You bet he was. AMD used DIB as well:
http://www.informit.com/articles/article.aspx?p=130978&seqNum=11

You really need a whole new view on your past, if what you think is right is so wrong.

 

[tential]

I thought that Kabini still had higher single-threaded IPC than Bay Trail?

http://www.anandtech.com/show/8067/amd-am1-kabini-part-2-athlon-53505150-and-sempron-38502650-tested

Try reading reviews on this site...

@DrMrLordX
I mean I understand that part for non hardware related things or even products that aren't HEAVILY benchmarked. But I mean come on, we know EVERYTHING there is to know about these CPUs. Is it really that hard to make a decision? I love Mercedes because I've had a relatively positive experience with the car. I love Ralph Lauren because the clothes fit my body and Brooks Brothers does not.

I can't benchmark how Mercedes makes me feel or how Ralph Lauren makes me feel. I can benchmark how much FPS I get in Crysis or how fast a processor loads webpages. It's ridiculous to me to continue buying a processor you KNOW gives you less performance. I mean there are AMD processors at the budget end with OCing that I guess are competitive but IMO it's a waste of time to pick those up. 1 full day of work for the majority of people will allow them to move from a budget AMD processor to a 4670k+ intel processor. That's a HUGE performance jump and longevity jump but that's just my opinion on that part.

But props to marketers to doing their job. There's a reason I did poorly when it came to marketing in school. It just isn't something that can sway me heavily.

 

[jhu]

http://www.anandtech.com/show/8067/amd-am1-kabini-part-2-athlon-53505150-and-sempron-38502650-tested

Try reading reviews on this site...

From the review:

Kabini does have higher single thread IPC than Silvermont/Baytrail/Avaton.

C2750 is 2.4 GHz. Athlon 5350 is 2.0 GHz.
So that's a score of 790/GHz for Atom C2750 and 1020/GHz for Athlon 5350.

 

[tential]

From the review:

Kabini does have higher single thread IPC than Silvermont/Baytrail/Avaton.

C2750 is 2.4 GHz. Athlon 5350 is 2.0 GHz.
So that's a score of 790/GHz for Atom C2750 and 1020/GHz for Athlon 5350.

But the point wasn't about singlethreaded performance in a benchmark like this or in Cinebench. The point was J1900/1800 perform VERY well in the Office/Web benchmarks which is what I'd use these processors for and at a lower power consumption than Kabini.
They're passively cooled as well.

It's a better value to me to go with the intel solution and I'll stick by that statement. If someone wants slighlty more performance in a Cinebench benchmark then hey, more power to them. I use my processors in real tasks though and the J1900/Kabini are basically neck and neck in browsing benchmarks and J1900 beats it in basic office type production.

In reality, like what reviewers have said, they couldn't see a difference in using EITHER processor in every day tasks, so the lower power consumption and passive cooling solution from the J1900 is far more favorable to me.

But hey, you can debate useless benchmarks that emulate tasks that these processors will NEVER be used for. Cinebench and the benchmark I posted as well mean nothing to me. All that matters is can I web browse, and do basic office work on this type of processor. For BOTH the answer is yes. The J1900 just does it in formfactors I want with the noise levels I want (which is no fan).

 

[chrisjames61]

It has always amazed me that people can associate morality with their choice of CPU.

Does this sophisticated form of morality also apply to your watch, your camera, your TV, your car, your clothing items?

And if not, why not?

Pretty sure it was tongue in cheek. You read way to much into it.

 

[MongGrel]

I just buy the best CPU for the task. Right now that's Intel for the performance segment.

This fanboyism I'll never understand. I have 0 brand loyalty on a PC. I won't lose performance just because "I bought AMD before so I'll buy again." I'll read the reviews and make the relevant decision from there. When AMD is in the lead in the performance segment I'll buy them. Right now there isn't an AMD chip where the loss of performance for me is worth the price drop. Even when comparing Kabini to the J1900/J1800 I'll take the J1900 EVERY time due to the better benches in webbrowsing/office productivity which is what I'd use those processors for.

I agree.

But I'm still OCing a X5650 ATM

 

[SlowSpyder]

It has always amazed me that people can associate morality with their choice of CPU.

Does this sophisticated form of morality also apply to your watch, your camera, your TV, your car, your clothing items?

And if not, why not?

My boss is a Bear's fan. She went to Lambeau to watch a Packer's game last season because her brother got the tickets for nothing. I'm a Packers fan and told her she's coming over to the dark side. I didn't really mean there was any real morality in any way when I said it... just kind of a saying.

 

[DrMrLordX]

http://www.internetnews.com/infra/article.php/3574856/AMD+Licenses+Rambus+Patent.htm
http://www.electronicsnews.com.au/news/amd-rambus-renew-patent-licence-agreement

Uh.

You are aware that AMD signed this licensing agreement in 2006, right? Which was wayyyyyyyy after the Pentium III era when Intel was using RDRAM as a market leverage tool to kill off JEDEC-dependent competitors? There is no way in the Nine Hells that RAMBUS, Inc. would have signed a licensing agreement with AMD back in the Pentium III days. By '06, the PC market was dead to them. A licensing agreement with AMD is free money for nothing.

AMD obviously signed this agreement in 2006 to avoid getting their pants sued off by RAMBUS, Inc. that was wiping the floor with most of the JEDEC "cartels" through patent infringement lawsuits. Obviously AMD didn't want to be sued for developing memory controllers that might have infringed on some RAMBUS memory controller patent.

Wrong? You bet he was. AMD used DIB as well:
http://www.informit.com/articles/article.aspx?p=130978&seqNum=11

You really need a whole new view on your past, if what you think is right is so wrong.

That doesn't mean they used Intel's implementation of DIB (the article you linked doesn't say one way or the other, nor does it say exacly which "sixth generation" AMD processors used DIB. AMD didn't move L2 cache onto the CPU until the K6-III, which was their first chip to have a "backside" bus).

Ditlea even directly quoted an Intel guy (Larsen) as saying that they weren't licensing it to anyone. Whether or not it was covered by the x86 cross-licensing agreement, I'm not sure. It certainly looks like AMD took a long time to implement DIB, possibly due to the fact that they were unable to simply ape Intel's design.

 

[DrMrLordX]

http://www.anandtech.com/show/8067/amd-am1-kabini-part-2-athlon-53505150-and-sempron-38502650-tested

Try reading reviews on this site...

@DrMrLordX
I mean I understand that part for non hardware related things or even products that aren't HEAVILY benchmarked. But I mean come on, we know EVERYTHING there is to know about these CPUs. Is it really that hard to make a decision? I love Mercedes because I've had a relatively positive experience with the car. I love Ralph Lauren because the clothes fit my body and Brooks Brothers does not.

I can't benchmark how Mercedes makes me feel or how Ralph Lauren makes me feel. I can benchmark how much FPS I get in Crysis or how fast a processor loads webpages. It's ridiculous to me to continue buying a processor you KNOW gives you less performance. I mean there are AMD processors at the budget end with OCing that I guess are competitive but IMO it's a waste of time to pick those up. 1 full day of work for the majority of people will allow them to move from a budget AMD processor to a 4670k+ intel processor. That's a HUGE performance jump and longevity jump but that's just my opinion on that part.

But props to marketers to doing their job. There's a reason I did poorly when it came to marketing in school. It just isn't something that can sway me heavily.

As in politics, different factions cling to different versions of reality. Here are some of the justifications for defending AMD hardware (or casting aspersions upon Intel) that I have seen over the past several months:

My usage pattern requires an octocore, and AMD sells me one for relatively cheap.
Intel has a compiler advantage.
I can't tell the difference in performance between 4-core Haswell + HT and an 8320/8350 in <insertnameofappcategoryhere> when actually using the machines.
Intel's anti-competitive behavior during the K7 and K8 days is the reason AMD took so long in updating K8 and took so long in fixing K10, which is why they're behind now.

Whether or not I agree with any of the above statements is irrelevant. What is relevant is that hardly any benchmark can address those statements. Or, in the case of the guy who really is pushing as many cores as he can get (and, for whatever reason, cares nothing about heat and power costs), Vishera @ ~5ghz can actually do pretty well until you start comparing it to LGA2011 processors. The benchmarks actually might show that that guy got the best bang/buck from AMD, provided he doesn't run it long enough for the 200-250W power draw to burn a hole in his wallet (or anything else for that matter).

 

[ShintaiDK]

Uh.

You are aware that AMD signed this licensing agreement in 2006, right? Which was wayyyyyyyy after the Pentium III era when Intel was using RDRAM as a market leverage tool to kill off JEDEC-dependent competitors? There is no way in the Nine Hells that RAMBUS, Inc. would have signed a licensing agreement with AMD back in the Pentium III days. By '06, the PC market was dead to them. A licensing agreement with AMD is free money for nothing.

AMD obviously signed this agreement in 2006 to avoid getting their pants sued off by RAMBUS, Inc. that was wiping the floor with most of the JEDEC "cartels" through patent infringement lawsuits. Obviously AMD didn't want to be sued for developing memory controllers that might have infringed on some RAMBUS memory controller patent.

RAMBUS wanted to sell technology. They had nothing as such to do with Intel. Intel just saw the bottleneck that memory is for the PC and tried using incentive to change that. You make it sound like RAMBUS is some kind of daughter company of Intel that Intel fully controls and only wanted to hurt AMD. Intel tried to subsidize the memory manufactors to get cheaper RDRAM on the market. However the DRAM cartel had other anti consumer friendly interests.

RAMBUS technology was already used by multiple manufactors and licensed by even more.

That doesn't mean they used Intel's implementation of DIB (the article you linked doesn't say one way or the other, nor does it say exacly which "sixth generation" AMD processors used DIB. AMD didn't move L2 cache onto the CPU until the K6-III, which was their first chip to have a "backside" bus).

Ditlea even directly quoted an Intel guy (Larsen) as saying that they weren't licensing it to anyone. Whether or not it was covered by the x86 cross-licensing agreement, I'm not sure. It certainly looks like AMD took a long time to implement DIB, possibly due to the fact that they were unable to simply ape Intel's design.

We already know the socket wasnt patented. And the DIB isnt patented either. Whats your next excuse?

I dont understand why you are so fixated on Intel being evil here. AMD had ZERO interest in using the GTL+ bus for their design. They wanted the EV6. The dream of socket compability is moronic.

 

[ShintaiDK]

As in politics, different factions cling to different versions of reality. Here are some of the justifications for defending AMD hardware (or casting aspersions upon Intel) that I have seen over the past several months:

My usage pattern requires an octocore, and AMD sells me one for relatively cheap.
Intel has a compiler advantage.
I can't tell the difference in performance between 4-core Haswell + HT and an 8320/8350 in <insertnameofappcategoryhere> when actually using the machines.
Intel's anti-competitive behavior during the K7 and K8 days is the reason AMD took so long in updating K8 and took so long in fixing K10, which is why they're behind now.

Whether or not I agree with any of the above statements is irrelevant. What is relevant is that hardly any benchmark can address those statements. Or, in the case of the guy who really is pushing as many cores as he can get (and, for whatever reason, cares nothing about heat and power costs), Vishera @ ~5ghz can actually do pretty well until you start comparing it to LGA2011 processors. The benchmarks actually might show that that guy got the best bang/buck from AMD, provided he doesn't run it long enough for the 200-250W power draw to burn a hole in his wallet (or anything else for that matter).

AMD is entirely guildty of their own situation. AMD when in front on purpose delayed 65nm and fab expansions. They also delayed the next uarch after K8. Not to mention their extreme high prices. All because they couldnt imagine Intel catching up while they only wanted to sell the top end CPUs. Specially server segment. Hector at the time already saw the server segment as AMD forever. Thats your precious AMD that you keep defending like saints.

Not to mention the lie after lie with Phenom, Bulldozer and so on.

 

[XiandreX]

I just buy the best CPU for the task. Right now that's Intel for the performance segment.

This fanboyism I'll never understand. I have 0 brand loyalty on a PC. I won't lose performance just because "I bought AMD before so I'll buy again." I'll read the reviews and make the relevant decision from there. When AMD is in the lead in the performance segment I'll buy them. Right now there isn't an AMD chip where the loss of performance for me is worth the price drop. Even when comparing Kabini to the J1900/J1800 I'll take the J1900 EVERY time due to the better benches in webbrowsing/office productivity which is what I'd use those processors for.

I can see how my post sounded like Fanboyism but it couldn't be further from the truth. I buy what is the best at that moment. I love AMD as a company.
I was so much an advocate that back in the day that I ordered one of the first Athlons.

My first overclock was the AMD K6II-233, using a peltier cooler. I have owned mostly AMD cpu's but as Intel pulled away for the higher end cpu's it became difficult to choose AMD.

If AMD had a CPU that matched intels single thread performance I would jump back to AMD in a heartbeat. The reality is they don't.
Shrug, I buy what is the best performance/money. The i5-3570k was that unit.

To each his own :thumbsup:

 

[KingFatty]

AMD is entirely guildty of their own situation. AMD when in front on purpose delayed 65nm and fab expansions. They also delayed the next uarch after K8. Not to mention their extreme high prices. All because they couldnt imagine Intel catching up while they only wanted to sell the top end CPUs. Specially server segment. Hector at the time already saw the server segment as AMD forever. Thats your precious AMD that you keep defending like saints.

Not to mention the lie after lie with Phenom, Bulldozer and so on.

Causing harm to yourself, shooting yourself in the foot etc., is one level of being evil. But there can be multiple "levels" of how evil/dark you can be. Crossing the line into illegal activity, I would think, could be on a different level than self-harm/ambitious marketing. http://consumerist.com/2009/12/17/ftc-sues-intel-for-decade-of-illegal-sales-tactics/