Conroe debunked

[ToastMaster]

Then bring them on? I gave links and showed where AMD was Faster and showed links where they were not. I don't have many posts about AM2 vs 9xx, find a link? My biggest gripe is that X2 cost too phucking much, I didn't say they were slower and even said the newer Presler only closed the gap!

I just got through saying "While AMD was kicking Intel ass with X2, their prices reflected that", "they're poud of their X2 and the prices show that". I'm very pro Conroe and I'll make no bones about that=P Pro Intel, I don't think so. I bought a 3500+ because it was $208 the time to by Opterons had passed then. Around most of you guys, almost anyone is pro Intel so I see your point.

Um, a 3800+ spanks the living hell out of a lot of different Intel processors at the moment, for $300, easily outperforming Intel's current $300 offerings in gaming (at stock speeds), as well as many more-expensive processors. You're telling me that that is too expensive for you, and yet you'll jump at the opportunity to buy a $318 Conroe? Seems a bit hypocritical to me...

Oh, I also find it amusing that while you say: "I'll get the X2 when Conroe comes out, crushes it, AMD then drastically cuts the prices, then I'll upgrade it. I", you then follow that up a few posts later to someone's response of never needing to buy AMD again due to performance charts with a "QTF! ". So which is it, you won't ever buy AMD again, or you will once prices drop? Seems a bit hypocritical to me...

Oh, and it also smells of a little dislike of AMD there. As well as short-signtedness. But alas, I digress.

While you'll be finally buying an X2 when Conroe comes out, I'll actually be using a Conroe, while building my mom a new system around the 3800+ X2. It's great being platform-non biased in anyway.

 

[Emission]

Least its always fun overclocking an intel. AMD and its cheap turn-key overclocks, sheesh. The Opterons were the first open show of that. Intels may be hard to clock but at least its more fun lol. People got a lot out of the 2.4 Pressys. jacked em up to 3.8 and beyond . Wonder how it will change with Conroe...And lets be honest, why spend ungodly amounts of money on something that you can get the same performance for, for less than half price?

 

[ToastMaster]

Regarding Anandtech's Yonah preview:

The Core Duo ran 4-4-4 DDR2-533 while the X2s ran 2-2-2 DDR. If you look at the comparison table for AM2 DDR2 vs 939 DDR, 3-3-3 DDR2-667 offers about the same performance as 2-2-2 DDR400.
The memory used in the Yonah benchmarks is probably more equivalent to 3-3-3 DDR2-400 on this same table. The difference is from 4 to 30%, let's say 5 or 10% - Yonah over X2 with equivalent RAM and the assumption that Yonah behaves similarly to X2 w/rt latency. Of course, we didn't have this comparison table when the Yonah preview came out. But, it wasn't impossible to infer something to that effect. And again, this pertains to certain games. We knew Conroe was going to have twice the cache of Yonah and feature 4-issue wide execution and we did have Intel's claims of performace + performance/watt from . So, some of us did come to slightly different conclusions than Anand Lal Shimpi.
If you read reviews for reviewers' opinions, fine. From time to time, I find them amusing, too. But, if you debate using their opinions, you run the risk of having misunderstood what they said and also accepting any fallacious conclusions they may have drawn from the data and available information. If you did fall into that trap, then you would not have been alone. But, some did not fall into it with you.

If you had actually read what I had said, I never actually said that I concured with what some review sites had said, concerning Yonah being a fair predictor of what the performance of Conroe would be. I'll repeat what I said, in an attempt to finally clear this up (and hopefully so I won't have to answer this once more):

What I meant, is that when Yonah samples were first coming out for preview/review, some sites (incorrectly) tried to use Yonah as a type of prediction of what Conroe performance might be like. I was simply trying to point this viewpoint out, hence my links. It was nothing more than that. Then, because of this viewpoint from some prominent sites, coupled with how Yonah was still outperformed by the 3800+ and all other X2s, made some people fearful that Conroe wouldn't be able to truly beat AMD's 2006 offerings. As we know, if anyone had properly looked at the development of Conroe, they'd have known better. All I was trying to do, was show that that viewpoint, that Intel would have nothing truly competitive to offer in 2006, is similar to how many people here view the idea that AMD won't have anything competitive to offer in 2007. Noone had seen Conroe yet, and yet people still bought it. Noone here has actually seen what AMD has planned for 2007, and thus to be making all of the ignorant statements that we are now seeing concerning AMD not having anything for a few years, etc. etc., is just ignorant at best.

Also, I'd like to see benchmarks showing where a stock Yonah is able to beat an equivalently-clocked X2 at stock settings, both equipped with the fastest ram available. I honestly haven't seen any mention of Yonah holding a performance lead, although I could be wrong.

Regarding Conroe vs AM2:

The E6700 was shown to be 20% faster than a 2.8 X2 at Spring IDF. We also have the other Gelsinger benchmarks of Conroe, Merom vs Yonah, and Victor Wang's benchmarks(the Sandra results were all I was looking for) backing these numbers up. This also means that Conroe is 25% faster per clock than X2 as demoed at Spring IDF. But, wait. Recall that Conroe was running DDR2-667 @ 5-5-5-15. That's pretty sad stuff. Applying the same logic as above, we can conclude that Conroe will be 30-35% faster per clock than AM2 X2s when running equivalent memory.

You're correct. You also however forget that Anandtech readjusted the system for 4-4-4 timings, in their Conroe follow-up: http://www.anandtech.com/tradeshows/showdoc.aspx?i=2716&p=5

Ultimately, as they themselves put it, the timings made very little difference, a few fps, and certainly not the 10%+ difference you believe the difference in timing will have. I think it's also important to add that Conroe will no reason to support anything higher than DDR2-667, as dual-channel DDR2-667 offers more bandwidth than Conroe's FSB can utiliize.

Regarding Conroe speeds:

The Woodcrest 3.0 1333fsb model will be priced below desktop XE chips. It will also have a lower TDP of 80w vs 95w. Right now, it's a good bet the top Conroe XE model will be doing 3.33 GHz this year.

Let's not forget, AMD is soon-to-ship X2s that will be at 65w TDP.

Regarding K8L:

When AMD says it's "evolutionary", remember that the evolution talked about in both camps is multicore-scaling, the race to multi-core, multicore is the name of the game, etc. K8L is just a code-name for the quad-core implementation of the basic K8 architecture, just as X2 was a dual-core of the same.
Don't expect 4-issue, Z-Ram, blah blah. These are just wild rumors. Expect AMD to put the same kind of effort and get the same kind of results into quad core as with dual core. Expectig major architectural changes would be expecting too much based on what AMD has been able to do from the first opterons to AM2. Anyway, K8L won't be competing with Conroe. That's what 65nm AM2 will be doing and it will not be doing it well unless it can hit 4 GHz effortlessly....early next year when the market will be saturated with Conroe parts.

Actually, we truly don't know anything really about K8L, except that it is seemingly based off of the K8, as given the K8 in its name. . However, stating that it is simply a probable name for quad-cores is just as much of a speculative statement as anything else. Also, I find it amusing that everyone assumes that AM2 and 65nm are AMD's only response to Conroe for next year. Once again, as I've said before, unless you're a design engineer at AMD, you don't know what AMD's true plans are. AM2 was simply a change of memory, which AMD had to do sooner or later, just as 65nm is another step in the process to reduce power consumption, costs, etc. I don't think Intel moved to 65 nm simply to compete against AMD, and it shouldn't be viewed as AMD's answer to Intel either.


There's alot more to be said. I'm still wondering why anyone would expect Donnie to have been saying positive things for AMD back when no one had anything bad to say against X2's except that they were costing an arm and leg. Why would there be any need to add another voice to the cheerleading choir, the horde of penile fencers, or the fanatical mob, when they lacked any kind of opposition???

 

[coldpower27]

Which also explains *why* Intel has been having all sorts of heat issues (yes, with Conroe, but as far back as the 90nm die-shrink, with Prescott). If you increase both the number of transistors and the transistor density (which is whart Intel did with *every* die-shrink), you are going to have *heat dissipation issues* (imagine the Rose Bowl's capacity fit into the space of the Staples Center). Every transistor (regardless of function) throws off heat, increasing transistor density (which happens due to a die-shrink) while also increasing the sheer number of transistors means that there is that much less area left for the heat to escape from. That means that soon it will actually become cost-effective for water-cooling to replace aircooling (if for no other reason than investment safeguarding where Intel *and* AMD are concerned), even though Conroe actually uses fewer clock cycles than either Cedar Mill or Smithfield. Does Conroe, like basically every core back to Williamette, use a quad-pumped FSB (333 MHZ quad-pumped) or did they shift to eight-pumped with the die-shrink?

Nothing here that I didn't know, but there are no heat issues with Conroe, one the TDP is 65W, with 266 (1066) Quad Pumped FSB, the pipeline is much shorter on Core Architecture helping reduce heat, and as well Victor Wang was able to run E6600 2.4GHz Conroe on a passive heatsink, so Conroe runs quite cool. The FSB will be (333) 1333 on the Extreme Edition but that has a 95 W TDP and is clocked at 3GHZ or higher. Not to mention Conroe also rund at a voltage of 1.2V, also further helpingf reduce heat dissappation.

The Quad Pumped Bus architecture isn't what was generating that much heat, it was the Intel using the leakiest transistors possible when the 90nm process was very young, to gain as much clockspeed as possible as that is what Prescott likes best clockspeed. The 65nm has improved leakage dramatically, and reduced power usage for all of intel's products including NetBurst. See information about the C1 Stepping Preslers for details, the Extreme Edition of 965 is now fairly close to the Athlon FX 60 in power usage thanks to the new stepping.

 

[Donnie27]

Don't expect 4-issue, Z-Ram, blah blah. These are just wild rumors. Expect AMD to put the same kind of effort and get the same kind of results into quad core as with dual core. Expectig major architectural changes would be expecting too much based on what AMD has been able to do from the first opterons to AM2. Anyway, K8L won't be competing with Conroe. That's what 65nm AM2 will be doing and it will not be doing it well unless it can hit 4 GHz effortlessly....early next year when the market will be saturated with Conroe parts.

There's alot more to be said. I'm still wondering why anyone would expect Donnie to have been saying positive things for AMD back when no one had anything bad to say against X2's except that they were costing an arm and leg. Why would there be any need to add another voice to the cheerleading choir, the horde of penile fencers, or the fanatical mob, when they lacked any kind of opposition???

AMD could easily inform or even brainwash their Fans if they had something to show=P AM2 is an adjustment at most. Even Anand said they rarely saw the 10% performance increase AMD and other experts mentioned. Unlike them, you also don't see me on the AMD side of the forum bad mouthing them there, hehehehe!

X2, I don't know of anyone saying X2's performance wasn't excellent or that it wasn't Kick-ass. I'll always say the cost too frackin much! I had no problems with AMD and buy their products, it's their Fans I don't like. According to their Fans AMD would never Charge what they've been charging since X2 and FX launched.

I did what PC Perspective and Extremetech did when it can to X2. They started looking at Frames Per second to Dollar Ratio. They looked at Seconds saved to Extra Costs ratio and etc..... No, 5% faster for a 50% higher cost is a bad deal IMHO!

It is absolutely LAME that anyone would ask me what's the difference between planning to buy a 2.4GHz Conroe and a 3800+ X2? 3800+ is NOT worth $300, never was IMHO so I didn't spend that much on it. Conroe 2.4GHz? You better buy them quickly because like the Opteron 1xx Sc-939's I missed out on before the prices jumpped I see their prices being higher in this market. Simple supply and demand. AMD will have to cut prices and it will be because of Intel, not because they love their Fans. I'd bet their Fans will be the first to say it was because they love them LOL!

 

[Donnie27]

Nothing here that I didn't know, but there are no heat issues with Conroe, one the TDP is 65W, with 266 (1066) Quad Pumped FSB, the pipeline is much shorter on Core Architecture helping reduce heat, and as well Victor Wang was able to run E6600 2.4GHz Conroe on a passive heatsink, so Conroe runs quite cool. The FSB will be (333) 1333 on the Extreme Edition but that has a 95 W TDP and is clocked at 3GHZ or higher. Not to mention Conroe also rund at a voltage of 1.2V, also further helpingf reduce heat dissappation.

The Quad Pumped Bus architecture isn't what was generating that much heat, it was the Intel using the leakiest transistors possible when the 90nm process was very young, to gain as much clockspeed as possible as that is what Prescott likes best clockspeed. The 65nm has improved leakage dramatically, and reduced power usage for all of intel's products including NetBurst. See information about the C1 Stepping Preslers for details, the Extreme Edition of 965 is now fairly close to the Athlon FX 60 in power usage thanks to the new stepping.

QFT!

 

[sparks]

Donnie you hit it right on the head.
WHY is AMD ====HOW==== moving the delivery date of the AM2 to may?
WHY to get the !!!!!! money before the conroe ships.
GET $300 A PROCESSOR MORE THAN CONROE before they get
their arse kicked and have to lower the price.
IF and I say IF AMD wants my money they need to want less

the less for more makes NO SINCE

just my 2 cents

sparks (puts on his fireproof suit)

 

[Donnie27]

Donnie you hit it right on the head.
WHY is AMD ====HOW==== moving the delivery date of the AM2 to may?
WHY to get the !!!!!! money before the conroe ships.
GET $300 A PROCESSOR MORE THAN CONROE before they get
their arse kicked and have to lower the price.
IF and I say IF AMD wants my money they need to want less

the less for more makes NO SINCE

just my 2 cents

sparks (puts on his fireproof suit)

They don't have to trick their Fans, they love giving AMD extra money Of course they're moving the dates up to get more money before the bottom falls out, makes sense to me. And thanks!

 

[$BangforThe$]

It's because he hasn't praised AMD at all. Do a search on his posts...and you're gonna see that most of them a pro-Intel, pro-Conroe. I searched in "AMD Processors" and nearly all of them are concerning AM2 vs. 9xx, again pro-Intel. I hope someone can prove me wrong, but from those results alone (and the impression I get from this thread), Donnie has always been pro-Intel.

You guys are freaken me out! Donnie does not seem pro intel at all. If anything he is pro tech. The best tech gets his praise. Than the man researches so as to know a little bit about the tech he discuses.

I don't know if your aware of it but AMD has scarped the old K10 and is now planning todo a new core. It will be a time before we see that core.

So we well not see Conroe vs K10 what well see is conroe 2 vs. k10

Donnie keep up the good work .Better yet become Intel fan . In this manner you won't have to stay fair and balanced- Just kidding

 

[robberbaron]

Least its always fun overclocking an intel. AMD and its cheap turn-key overclocks, sheesh. The Opterons were the first open show of that. Intels may be hard to clock but at least its more fun lol. People got a lot out of the 2.4 Pressys. jacked em up to 3.8 and beyond . Wonder how it will change with Conroe...And lets be honest, why spend ungodly amounts of money on something that you can get the same performance for, for less than half price?

Yes, raising the frontside bus and occaisionally raising the vcore is a whole lot of fun.

And I'm not aware of an Intel chip that costs half of what my $270 Opteron 165 but gets the same performance after being overclocked. Which chip is that again?

 

[Emission]

Maybe that was a bland comment due to inexperience. If you've ever actually done it you know it can be hard. And im not talking about some tiny .2 GHz overclock, im talking pushing it as far as it will go.

 

[robberbaron]

Maybe that was a bland comment due to inexperience. If you've ever actually done it you know it can be hard. And im not talking about some tiny .2 GHz overclock, im talking pushing it as far as it will go.

Yeah, I've played with a 920. Got it from 2.8 to 3.73. At that speed it flew, but, there werent nearly as many things to tweak as there are on my DFI/AMD system. Seriously, all it took was frontside bus increases and vcore. Then the motherboard stopped at 267MHz FSB. Yay.
I know the P5WD2-E can do 400MHz FSB, but, its also twice as much a good DFI board for 939. Meh.

 

[$BangforThe$]

Current benchmarks are based upon Engineering Samples. I think Kyle and crew have proven how Engineering Samples, regardless of manufacturer, can provide bubkis to real world. My thoughts are... If it worked as it's supposed to, then why isn't it being released now? Why in a plethora of months? Sure, you have to ramp up production... But something seems... off. I'm not saying it's going to be another Penium III 1.13GHz, but why say the technology is proven, but hold back on everybody?

Typically, a company previews product when it's in its final form and practically ready for release. They don't show their hand 2 quarters ahead of time. Rumors fly about, but that's about it. Why this pre-preview on a machine that they built that is obviously engineered to the hilt?

I've got nothing against Intel. I've owned a few. Without Intel, the Athlon would probably have never existed and AMD wouldn't be the player it is. However... Something seems misexecuted and it looks like Intel is throwing around marking FUD in a desparation move.

I think your right . Intels early showing of conroe was a move of despiration. But a modded up preview not likely. I am a long time Intel fan . If conroe comes out and its not better than the preview . I will go AMD for life. I really don't see that happening however.

Intels preview worked . There are a lot of people now waiting to buy Conroe and if Conroe is for real . Its a smart move to wait.

 

[Emission]

Yeah, I've played with a 920. Got it from 2.8 to 3.73. At that speed it flew, but, there werent nearly as many things to tweak as there are on my DFI/AMD system. Seriously, all it took was frontside bus increases and vcore. Then the motherboard stopped at 267MHz FSB. Yay.
I know the P5WD2-E can do 400MHz FSB, but, its also twice as much a good DFI board for 939. Meh.

Sorry on the bash, but im a [H]ardcore intel fan, its typical of me to not like the situation intel used to be in lol. There are some intel boards that let you tweak your settings as much as a DFI. Just so we both know, I'm a former S754 3000+ Owner, with a DFI nf3 250GB of my own. The massive memory page wasn't much to brag about, I got my chip to 2.93 leaving most all of those settings at auto.

 

[Donnie27]

You guys are freaken me out! Donnie does not seem pro intel at all. If anything he is pro tech. The best tech gets his praise. Than the man researches so as to know a little bit about the tech he discuses.

I don't know if your aware of it but AMD has scarped the old K10 and is now planning todo a new core. It will be a time before we see that core.

So we well not see Conroe vs K10 what well see is conroe 2 vs. k10

Donnie keep up the good work .Better yet become Intel fan . In this manner you won't have to stay fair and balanced- Just kidding

Thanks Bangfor! I've tried my best to present Pro and Con for and against Intel and AMD.

Here's an example of RAM Latency and speed on AMD AM2 rigs.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2741&p=3

AMD Follower Wes Fink is doing the review.

Since increased memory bandwidth did not translate into similar increases in gaming performance we can only conclude that this iteration of AM2 is not particularly memory starved - a result that was really expected.

Something some of us have said from Day one. Why would it be memory starved when Current ones aren't? It's not like this is a new processor. If the Base/Core FSB isn't increased any, it becomes moot how much memory bandwidth is increased. You have something like adding RAMBUS to a P3. I remember Intel fans flaming me for saying it was a gimmick back then. Change the 128bit controller to a 256Bit one like some video cards and see what happens? But oh wait, that's a real upgrade

AMD will have to incrase the rates on the Crossbar, HT, cache and IMC to see any increased performance and that means a real 1333MHz in each direction and increase to the interprocessor bit widths as well. Then they can add some more IPC magic while they're at it. Then I'd be either here or on the AMD side talking about how Intel got their asses kicked just as I'm talking about Conroe now.

I can't become a real Intel Fan because I truely do Hope AMD's real answer to Conroe and Conroe 2 kicks Intel's ass. Had Intel launched Conroe LAST YEAR and it kicked AMD booty, do you guys reaaallllly think we'd be talking about AM2? AMD would have already added speed, Cache, wider BUS and everything else that might be talked about next year.

 

[$BangforThe$]

Donnie I am assuming (cuts own throat) That Conroe will require greater bandwidth than AMD cpu's. My reasoning is this.

Longer pipes on the conroe and 4 issue core. So I believe DDR2 1066 is going to be a great benefit to Intel conroe . Just guessing but it seems to hold water.

 

[Emission]

As stated:

Regarding Conroe vs AM2:

The E6700 was shown to be 20% faster than a 2.8 X2 at Spring IDF. We also have the other Gelsinger benchmarks of Conroe, Merom vs Yonah, and Victor Wang's benchmarks(the Sandra results were all I was looking for) backing these numbers up. This also means that Conroe is 25% faster per clock than X2 as demoed at Spring IDF. But, wait. Recall that Conroe was running DDR2-667 @ 5-5-5-15. That's pretty sad stuff. Applying the same logic as above, we can conclude that Conroe will be 30-35% faster per clock than AM2 X2s when running equivalent memory.

Even so, DDR2 1066 Modules would be very unrealistic. (1) Conroe dosnt need that much bandwidth to perform. and (2) DDR2 1066 would come slower than 5-5-5-15. Even the latest intel's can't make use of DDR2 1000, its just a lot of bandwidth.

That being said, AMD still wont have any useful cores at the release of AM2, it will just be a DDR2 addition to current cores, and have only slightly lower pricing:
http://www.dailytech.com/article.aspx?newsid=1601

Not much to currently show for on either side. Bottom line is, we just dont know enough. Other than that, 5000+ and FX-62 will be unveiled at Computex, June 06, 06. (6/6/6 Scary isnt it? Funny actually ), in AM2 glory. We just need to wait and see.

 

[sparks]

Emission is that 5000+ ,speed, price or both???

AM2 the latest test are showing a little increase with new bios but nothing earth shaking.
Maybe final will do a little more.

of course same could be said of conroe as well.


sparks

 

[Emission]

Speed...Price, maybe for AMD , haha, just having a laugh

 

[Donnie27]

Donnie I am assuming (cuts own throat) That Conroe will require greater bandwidth than AMD cpu's. My reasoning is this.

Longer pipes on the conroe and 4 issue core. So I believe DDR2 1066 is going to be a great benefit to Intel conroe . Just guessing but it seems to hold water.

Conroe will have greater Processor Bandwidth as its FSB is increased, 6.4, 8.4 and 10.6GB. If it keeps its low or lower latency, it will make better use of its availible bandwidth. With 4MB of smart Cache it is also ease going to System RAM (less). Like A64/X2, Conroe was built to be MultiCore and but unlike X2/A64 it was optimized for DDR2, not have it added as an Afterthought.

 

[GilmourD]

Conroe will have greater Processor Bandwidth as its FSB is increased, 6.4, 8.4 and 10.6GB. If it keeps its low or lower latency, it will make better use of its availible bandwidth. With 4MB of smart Cache it is also ease going to System RAM (less). Like A64/X2, Conroe was built to be MultiCore and but unlike X2/A64 it was optimized for DDR2, not have it added as an Afterthought.

First... The Conroe isn't built for any type of memory. It could really give a rats ass what kind of memory you're using on the board as long as it delivers bandwidth. You can thank Intel chipsets. They could use octuple channel SDR for all the Conroe cares, since it has to go through the chipset first then through the FSB.

Secondly, unlike Intel, AMD isn't going around saying "OMGWTFBBQ! HERE'S A NEW PRODUCT THAT'S TWO QUARTERS AWAY THAT WE HAVEN'T MADE YET!!!1111oneoneone111!!!" The memory controller is built into the Athlon64 series processors, and DDR2 can't be a mere afterthought since it requires a retooling of said memory controller.

 

[(cf)Eclipse]

unlike X2/A64 it was optimized for DDR2, not have it added as an Afterthought.

as gilmour said, the cpu is relatively blind to the ram type, as long as it can get the data it needs. DDR2 has very similar overall latencies to DDR1 (actually a bit faster), and gobs more bandwidth.
the only way you can really "optimize" for that is make a core that is very dependant on the memory subsystem to feed it with data. guess what netburst was

 

[(cf)Eclipse]

Donnie you hit it right on the head.
WHY is AMD ====HOW==== moving the delivery date of the AM2 to may?

nothing was changed. launch for AM2 is still 6/6/06. system builders can get ahold of it starting 5/15/06.. same as i've seen it for a long time now


also, a lot of you are pretending that AM2 is supposed to be the competitor to conroe. well.. i can understand how one would see it that way, but we all know that there's no way in hell that a socket change will ever beat out a new core. take LGA775 for example. did we expect netburst to suddenly beat K8 with a new socket? no. so why should AM2 be seen in the same light? the only plausible reason i can see for this is due to the timing, nothing more

 

[brucedeluxe169]

noone is saying AM2 is AMD's response to Conroe... BUT... it will be what AMD is selling at the time that Conroe is released....

ppl keep talking abt "next year", "K8L", "K10", blah blah, but the fact remains that Intel will have Conroe out for many many many months before any of that materializes... and when I say "many many many months" I mean pushing a whole year, maybe even more....

And even if K8L comes out and it edges Conroe, Intel will have faster Conroes by then anyhow...not to mention that 45nm die shrink next year too....

 

[ToastMaster]

AMD could easily inform or even brainwash their Fans if they had something to show=P AM2 is an adjustment at most. Even Anand said they rarely saw the 10% performance increase AMD and other experts mentioned. Unlike them, you also don't see me on the AMD side of the forum bad mouthing them there, hehehehe!

Apparently I'm a "brainwashed AMD fan", simply because I enjoy the ability of the X2 and was willing to pay, what was to me, a very acceptable $300 on a processor that, when I bought it (late August of 2005) couldn't be beat for its price to performance ratio. Don't yell at early-adopters because there's Intel offerings now that, in terms of price, beats it, but that didn't exist when the 3800+ was released.

Infact, noone ever seemed to see you on the AMD forum, *ever*, commenting about a "job well done", "way to go AMD", "those results are impressive", etc. Why? Because you're an Intel !!!!!!. Just admit it. Nothing wrong with it, I guess.

X2, I don't know of anyone saying X2's performance wasn't excellent or that it wasn't Kick-ass. I'll always say the cost too frackin much! I had no problems with AMD and buy their products, it's their Fans I don't like. According to their Fans AMD would never Charge what they've been charging since X2 and FX launched.

I don't see you whining about the cost of any of Intel's higher-priced offerings. It's only anti-AMD this, anti-AMD that. Hence why people think you're completely Intel-biased.

I did what PC Perspective and Extremetech did when it can to X2. They started looking at Frames Per second to Dollar Ratio. They looked at Seconds saved to Extra Costs ratio and etc..... No, 5% faster for a 50% higher cost is a bad deal IMHO!

I will fully concur, that at this point in time, the X2 isn't a truly affordable dual-core to invest in. Stock for stock, it's a better performing than generally any of the Intel offerings at stock speeds beneath it, but I guess you have to be willing to pay for that performance.

Why is it that people will pay $400 extra for an FX-60, when it's practically identical to the 4800+ X2? Because they choose too.

However, at the time, when the 3800+ debuted, the dual-core offerings from both Intel and AMD were more expensive, and the 3800+ was a relative god-sense due to the price vs. performance that it allowed. Now, 9 months later, I'd hope that we'd have many more affordable dual cores out. That's only common sense.

It is absolutely LAME that anyone would ask me what's the difference between planning to buy a 2.4GHz Conroe and a 3800+ X2? 3800+ is NOT worth $300, never was IMHO so I didn't spend that much on it. Conroe 2.4GHz? You better buy them quickly because like the Opteron 1xx Sc-939's I missed out on before the prices jumpped I see their prices being higher in this market. Simple supply and demand. AMD will have to cut prices and it will be because of Intel, not because they love their Fans. I'd bet their Fans will be the first to say it was because they love them LOL!

So basically, even though when it debuted, the 3800+ was clearly a perfect choice to buy, given the performance it allowed, the multitasking, and yet you were too cheap to buy it. That, or you simply keep having to justify to yourself reasons for why AMD apparently "sucks". That's the only way I see it. You can talk about the huge performance difference Conroe may provide once it launches, and how that all-of-a-sudden makes it a great purchase. The fact of the matter is, when the 3800+ X2 debuted, it was easily able to outpace processors from Intel that were a lot more expensive than it was, as well as keeping up with the 4200+ which, at the time, I believe was almost $150 more. So back in September, to me and all the others who invested in one, it looked like a steal. And it was.

Oh, and I might just add: your dislike of AMD fans is truly immature, and reflect as well just how Intel-biased you truly are, as you look at every oppurtunity to bash AMD itself or its fans. There are radical, rabid AMD fans just like there are radical, rabid Intel fans *cough*. I'm an AMD fan, but I won't let that stop me from buying Conroe, because the success of Conroe will drive AMD to make a more powerful product, and in the end, everyone benefits.

 

[ToastMaster]

You guys are freaken me out! Donnie does not seem pro intel at all. If anything he is pro tech. The best tech gets his praise. Than the man researches so as to know a little bit about the tech he discuses.

Funny then how he never has anything but disdain for AMD publicly, until called out, then he'll simply say "omg i only dislike how amd's prices are bad", and yet that's still nothing more than a cope-out.

I don't know if your aware of it but AMD has scarped the old K10 and is now planning todo a new core. It will be a time before we see that core.

Um, where'd you pull that from? While it is known that AMD is working on a true successor to the K8 architecture, the old K10 was the name for the socket AM2 K8s. Now that it's named AM2, there's no need for it (just as K9 was for Dual-Cores). And K8L isn't an entirerly new core. Guess you overlooked that.

So we well not see Conroe vs K10 what well see is conroe 2 vs. k10

You'll probably see quad-core, 65nm K8Ls vs. quad-core "Conroes" (moreso Kentsfields). Eventually though, Intel must ditch the FSB.

 

[Eva_Unit_0]

Eventually though, Intel must ditch the FSB.

see and that's where I don't agree with you. AMD's integrated memory controller is damn good, I'll definitely give them that. But it also hurts them in that they cannot easily move to new memory styles (AM2, anyone?). Furthermore, if the FSB is such a terrible thing how is it that the pentium M's can consistently perform identically (if not a bit faster) clock-for-clock with the A64's? Even on the old single-channel DDR333-only 855GME running on only an "outdated" 100mhz fsb the pentium M's compete with the A64 in pretty much every task (except those where it is simply outclasses, such as FPU-limited tasks).

There's still something to be said about the chipset flexibility that the FSB-style setup gives.

 

[ToastMaster]

see and that's where I don't agree with you. AMD's integrated memory controller is damn good, I'll definitely give them that. But it also hurts them in that they cannot easily move to new memory styles (AM2, anyone?). Furthermore, if the FSB is such a terrible thing how is it that the pentium M's can consistently perform identically (if not a bit faster) clock-for-clock with the A64's? Even on the old single-channel DDR333-only 855GME running on only an "outdated" 100mhz fsb the pentium M's compete with the A64 in pretty much every task (except those where it is simply outclasses, such as FPU-limited tasks).

There's still something to be said about the chipset flexibility that the FSB-style setup gives.

Just to ask, do you have benchmarks of a Pentium M beating out a clock-for-clock Athlon 64, because in all of the Yonah reviews, which is an improvement over the previous Dothan and earlier cores, the Athlon 64 can still outpace it at the same stock clock speeds (although these are generally limited to 1.8Ghz and 2.0Ghz). I've never seen it, and I'd be seriously surprised if a Pentium M could beat a Athlon 64 clock-for-clock.

That said, simply take the current use of DDR2 667 with Conroe:in dual-channel, Conroe's 1066mhz FSB can't even make full use of the available bandwidth. I doubt the planned 1333mhz FSB for the EE will be able to utilize that much more either. Thus, eithe Intel needs to seriously revamp the FSB to higher speeds to allow Conroe to make greater user of the available bandwidth, or they're basically going to be stuck at DDR2 667 for atleast another year after Conroe finally debuts, if not longer. Will it make that much of a difference in performance relation to an Athlon 64? Ofcourse not. But think about how much Intel is holding back Conroe by not simply heavily altering/replacing the FSB.

The sad thing is, I doubt it'd be that difficult for Intel to integrate a memory controller, and given Intel's size and available resources, they could update it and change it far easier than AMD can. Oh well though.

 

[DejaWiz]

But it also hurts them in that they cannot easily move to new memory styles (AM2, anyone?).

There's still something to be said about the chipset flexibility that the FSB-style setup gives.

However, intel seems to be in a habit of keeping their cpu architecture mostly unchanged and just changes sockets instead.

 

[coldpower27]

Just to ask, do you have benchmarks of a Pentium M beating out a clock-for-clock Athlon 64, because in all of the Yonah reviews, which is an improvement over the previous Dothan and earlier cores, the Athlon 64 can still outpace it at the same stock clock speeds (although these are generally limited to 1.8Ghz and 2.0Ghz). I've never seen it, and I'd be seriously surprised if a Pentium M could beat a Athlon 64 clock-for-clock.

That said, simply take the current use of DDR2 667 with Conroe:in dual-channel, Conroe's 1066mhz FSB can't even make full use of the available bandwidth. I doubt the planned 1333mhz FSB for the EE will be able to utilize that much more either. Thus, eithe Intel needs to seriously revamp the FSB to higher speeds to allow Conroe to make greater user of the available bandwidth, or they're basically going to be stuck at DDR2 667 for atleast another year after Conroe finally debuts, if not longer. Will it make that much of a difference in performance relation to an Athlon 64? Ofcourse not. But think about how much Intel is holding back Conroe by not simply heavily altering/replacing the FSB.

The sad thing is, I doubt it'd be that difficult for Intel to integrate a memory controller, and given Intel's size and available resources, they could update it and change it far easier than AMD can. Oh well though.

It depends on where you look, not all cases mind you. the Pentium M is competitive in gaming agains the Athlon 64's.
http://www.gamepc.com/labs/view_content.asp?id=pm780&page=7

As well, regarding the FSB, there will be support for DDR2-800 in the i965 Series Chipsets, even though the bus cannot take full advanatge of it placing Dual Channel DDR2-800 helps the 1066FSB reach closer to it's theoretical limits. As the FSB bandwidth is nowhere near 100% efficiency. More like 50-70 range..

There are also other reasons for not integrating memroy technology onto the processors themselves, Intel has a chipset business to worry about as well.

 

[ToastMaster]

It depends on where you look, not all cases mind you. the Pentium M is competitive in gaming agains the Athlon 64's.
http://www.gamepc.com/labs/view_content.asp?id=pm780&page=7

As well, regarding the FSB, there will be support for DDR2-800 in the i965 Series Chipsets, even though the bus cannot take full advanatge of it placing Dual Channel DDR2-800 helps the 1066FSB reach closer to it's theoretical limits. As the FSB bandwidth is nowhere near 100% efficiency. More like 50-70 range..

There are also other reasons for not integrating memroy technology onto the processors themselves, Intel has a chipset business to worry about as well.

Ah, thanks for the benchmarks. I hope I didn't imply that the Pentium M series was a underachiever, because I've been a big fan of it from the beginning (and this is one situation where I think we can all agree that AMD's Turion series is pretty much trash). As for the switch to DDR2-800 eventually, it may help it get closer, but I have a feeling it won't help very much. If the FSB is already using as much bandwidth as it essentially can from dual-channel DDR2-667, I still can't see faster DDR2 having much of an effect. If it does though, it'd be a pleasent surprise to me.

And good point about the reason for why they shouldn't integrate a IMC. As blatantly obvoius as it is that Intel makes chipsets, I tend to always forget, even though I know about them. Then again, that end-of-the-semester finals college time is having not-so-good effects on me.

Conroe does excite me a lot though. I'll have to figure out a way when it comes out to justify replacing my 3800+

 

[duby229]

Sorry folks but I gotta say this.

There is absolutely no reason why Intel couldn't adopt a better bus design and still keep the memory controller on the chipset.

Hell with a few minor tweaks even PCI-e could be revamped and used as a p2p bus, that would blow GTL+ out of the water. Lower the latency and crank the clock, and ding we have a winner.

 

[Bao01]

All I was trying to do, was show that that viewpoint, that Intel would have nothing truly competitive to offer in 2006, is similar to how many people here view the idea that AMD won't have anything competitive to offer in 2007. Noone had seen Conroe yet, and yet people still bought it. Noone here has actually seen what AMD has planned for 2007, and thus to be making all of the ignorant statements that we are now seeing concerning AMD not having anything for a few years, etc. etc., is just ignorant at best.

Those who held the first view ignored the differences between Yonah and Conroe. The evidence had been around for almost a year. They ignored the evidence. Those who hold the second view do not ignore the evidence. The similarity is not between these two views.

The real similarity is with those who held the first view, who are unable to critically reason, and those who believe KateL, Revision G, or whatever you want to call yet another little change to the K8 architecture is going to somehow make AMD competitive against Conroe and its derivatives. They are the same people. They had doubts about Conroe until very recently. They reason based on faith in one company and lack of faith in another.

You're correct. You also however forget that Anandtech readjusted the system for 4-4-4 timings, in their Conroe follow-up: http://www.anandtech.com/tradeshows...aspx?i=2716&p=5

Ultimately, as they themselves put it, the timings made very little difference, a few fps, and certainly not the 10%+ difference you believe the difference in timing will have. I think it's also important to add that Conroe will no reason to support anything higher than DDR2-667, as dual-channel DDR2-667 offers more bandwidth than Conroe's FSB can utiliize.

As long as I am correct, you're just picking nits Remember, I didn’t claim a 10%+ difference. I made an assumption *for the sake of argument* and it was a range because the difference a user will see would depend of their usage patterns. Generally, we're talking about those benchmarks. Are you not ok with 30% Conroe over X2? Now, I’m basing this off the 20+% gain over 2.8 X2, 5% difference in clockspeed, and a couple of % for tighter timings – forgetting trying to run at 1T command rate like the X2 at IDF, that may not happen with Intel chipsets for Conroe.

Ok so, 30% per clock and a 19% clock gain between the 3.33XE and the FX-62, the fastest X2 on AMD roadmaps until next year. That’s a 54.5% difference in performance, generally speaking, more on this later. Yea, I know, you’re not talking about ’06 or even AM2. I’m just rubbing it in.

Actually, we truly don't know anything really about K8L, except that it is seemingly based off of the K8, as given the K8 in its name. However, stating that it is simply a probable name for quad-cores is just as much of a speculative statement as anything else. Also, I find it amusing that everyone assumes that AM2 and 65nm are AMD's only response to Conroe for next year. Once again, as I've said before, unless you're a design engineer at AMD, you don't know what AMD's true plans are. AM2 was simply a change of memory, which AMD had to do sooner or later, just as 65nm is another step in the process to reduce power consumption, costs, etc. I don't think Intel moved to 65 nm simply to compete against AMD, and it shouldn't be viewed as AMD's answer to Intel either

This is called an argument out of ignorance. You’ve also injected some AMD-faith into it. K8L is K8 with an L, an internal designation. It’s a revision, just like all the other revisions up to Rev. F. None of them made any great gains in performance per clock with the exception of Rev. E, dual core. So, just as it says in this article,

http://www.tgdaily.com/2006/03/27/amd_quad_core/,

K8L will be quad core. It will bring great gains in performance, as long as it is a quad core being compared to dual core.

Anyway, that’s neither here nor there. If you can rationalize in some way AMD making up most of the 54.5% performance deficit in 2007 with clockspeed and IPC, good for you. I’d like to see it. But, if you’re going to make the claim that K8L exceed that 54.5% and be better than Conroe, I want to know what you're smoking.

 

[Bao01]

I did what PC Perspective and Extremetech did when it can to X2. They started looking at Frames Per second to Dollar Ratio. They looked at Seconds saved to Extra Costs ratio and etc..... No, 5% faster for a 50% higher cost is a bad deal IMHO!

You look at it from a value standpoint. I totally agree.

It is absolutely LAME that anyone would ask me what's the difference between planning to buy a 2.4GHz Conroe and a 3800+ X2? 3800+ is NOT worth $300, never was IMHO so I didn't spend that much on it.

Same price, more than 50% the performance difference.

I understand your decision to get an A64. I got a Venice for my sister some time before I went with the Opterons. I liked how the AMD chips ran cooler. I didn't game much and don't game anymore, so mine wasn't really meant to be a gaming rig. I just figured that if I could clock my AMD chips high enough, I could do without hyperthreading. Last time, I spent almost 300 bucks on an opteron. I'm afraid I'm not shelling out 300 bucks for another chip anytime soon.

 

[Bao01]

Even so, DDR2 1066 Modules would be very unrealistic. (1) Conroe dosnt need that much bandwidth to perform. and (2) DDR2 1066 would come slower than 5-5-5-15. Even the latest intel's can't make use of DDR2 1000, its just a lot of bandwidth.

(1) Who's talking about bandwidth?

(2) Who's talking about ddr2 1066?

I'm talking about lowering memory access times for a low latency architecture, not increasing bandwidth.

 

[Emission]

(1) Someone referenced bandwidth.
(2) Someone refrenced DDR2 1066

Read the thread beforehand....

 

[Bao01]

(1) Someone referenced bandwidth.
(2) Someone refrenced DDR2 1066

Read the thread beforehand....

But, you quoted me...I see what you meant, now. I had that user on mental ignore. Nvm!

 

[sparks]

Hey Toastmaster I have owned amd and intel
but this????

So basically, even though when it debuted, the 3800+ was clearly a perfect choice to buy, given the performance it allowed, the multitasking, and yet you were too cheap to buy it. That, or you simply keep having to justify to yourself reasons for why AMD apparently "sucks". That's the only way I see it. You can talk about the huge performance difference Conroe may provide once it launches, and how that all-of-a-sudden makes it a great purchase. The fact of the matter is, when the 3800+ X2 debuted, it was easily able to outpace processors from Intel that were a lot more expensive than it was, as well as keeping up with the 4200+ which, at the time, I believe was almost $150 more. So back in September, to me and all the others who invested in one, it looked like a steal. And it was.

You say the 3800 was cheaper and outperformes. SO it was the best buy...YEP VERY TRUE

That is EXACTLY what they are saying about conroe its cheaper and it outperformes.

BUT these people by your statements are INTEL FB for saying the exact same thing about the conroe processor that you said about the AMD 3800?????????????????????

NOT FB's just consumers that want their moneys worth..
and on the next go round AMD aint it.

sparks

 

[Donnie27]

First... The Conroe isn't built for any type of memory. It could really give a rats ass what kind of memory you're using on the board as long as it delivers bandwidth. You can thank Intel chipsets. They could use octuple channel SDR for all the Conroe cares, since it has to go through the chipset first then through the FSB.

Intel moved to DDR2 and built it processors with that RAM in mind=P Saying the RAM doesn't matter on Conroe is a a little Crazy.

No one gives a Rats what you or I think about any kind of RAM as well. AMD is moving to it and it's giving them little to show for it. It's not an Upgrade for anyone who already owns a Sc-939 and in the end that will hurt AMD's sales. People with Intel and AMD system will jump to Conroe though.

How The Processor talks to the memory controller isn't as important as how much availible bandwidth the Processor and SYSTEM can make use of. 12.8GB? You can give AM2 25.6GB and if it can only make use of 7GB, you've still wasted a lot of time and money.

Secondly, unlike Intel, AMD isn't going around saying "OMGWTFBBQ! HERE'S A NEW PRODUCT THAT'S TWO QUARTERS AWAY THAT WE HAVEN'T MADE YET!!!1111oneoneone111!!!" The memory controller is built into the Athlon64 series processors, and DDR2 can't be a mere afterthought since it requires a retooling of said memory controller.

Then the only folks I've heard say things like " "OMGWTFBBQ" are folks like you This version of AMD's Integrated memory controller is New for AMD. There will be many tweaks before they get it right=P

 

[GilmourD]

Intel moved to DDR2 and built it processors with that RAM in mind=P Saying the RAM doesn't matter on Conroe is a a little Crazy.

That's the same as everybody saying that the Pentium 4 was built for Rambus. Funny... Worked perfectly fine with daul channel DDR. The processor was built with a bandwidth expectation in mind, not a specific technology.

No one gives a Rats what you or I think about any kind of RAM as well. AMD is moving to it and it's giving them little to show for it. It's not an Upgrade for anyone who already owns a Sc-939 and in the end that will hurt AMD's sales. People with Intel and AMD system will jump to Conroe though.

Well, two things... Nobody ever said it was supposed to be a Socket939 killer, even the fantarts. Also, how 'bout you let a product get released first before you declare yourself an expert on it. That applies to both AM2 and Conroe. And... I'm sticking with AMD.

How The Processor talks to the memory controller isn't as important as how much availible bandwidth the Processor and SYSTEM can make use of. 12.8GB? You can give AM2 25.6GB and if it can only make use of 7GB, you've still wasted a lot of time and money.

Well... It's funny... The FSB for Conroe isn't going to be able to take all the available bandwidth of dual channel DDR2-800... And where are you getting your figures from?

Then the only folks I've heard say things like " "OMGWTFBBQ" are folks like you This version of AMD's Integrated memory controller is New for AMD. There will be many tweaks before they get it right=P

That was in jest, and if you don't realize that... I dunno. Granted, everything gets tweaks, Intel or AMD. If you think Conroe's not going to have as many tweaks and errata as the AM2 DDR2 memory controller, you're fooling yourself.

What I love is how Conroe is farther out from release than AM2, yet you've declared AM2 dead in the water and Conroe champ. As many have said, AM2 isn't a new core or architecture. It's more of a way of keeping people from plugging chips that only support DDR into boards meant for DDR2. And if AMD's done anything right, it's been the onboard memory controller. I don't remember people complaining about the very first onboard DDR controller, so why should the new DDR2 version be any different?