JAYDEEJOHN :
Saying their IMC is superior is something.
I might ask, hows it fair on DDR2?
AMD doesn't have a single IMC for that. Its their current DDR2 IMC with a DDR3 IMC next to it. Not really special. They had to do it because if they had to migrate to AM3 like they did from S939 to AM2 it would have had a backlash on the "upgrade ability" AMD crowd.
Cryslayer80 :
While eating popcorn and watching this war, I occasionally shout: "Go keithlm!".
You don't really care but just want to hate Intel huh? Because if you even read the OP you would understand it....
Cryslayer80 :
He is being a troll? So, what was your thread's attempt? I'll tell you what. It was intended to calm you down and get a 1000 responses agreeing with your bias (later in text, first post was amazingly true) since you got a shoe from people who are not short-sighted.
Or he just felt like venting it a bit. I understand. Nothing wrong with that and still he did it in a nice way that was informative.
Its just some people don't want to take it as they are, and thats facts. They want to argue or try to dispute them.
It was a nice thread until then....
croc :
Og came up with the concept of the wheel. Zog came along, and put two of them together with something in between, which he called the 'axle'.
Maybe a bit too subtle for this forum...
Neither Intel nor AMD 'innovated' (invented) the IMC. DEC and HP were both using the technology long before. The IMC concept had some good features (shorter 'wire' lengths, reduced latency) and some drawbacks as well (change memory type, build new CPU). The concept may very well have been dreamed up in Armonk for all I know, and then dropped (shelved, in IBM terms) for reasons valid to their engineers at the time.
So arguing about who 'innovated' the IMC is pretty much a pointless excercise in futility, only showing the ignorance of those that take sides.
He just means first x86 IMC. Anything outside that tends to not exist here.
ElMoIsEviL :
I don't view it as a "side" issue. The thread was simply started to refute common made allegations by AMD fanbois. It was a response to this post on AMDZone made by BaronMatrix:
http://img.photobucket.com/albums/v51/ElMoIsEviL/AMDZone/BaronMatrix_1.jpg
As for can see I was "TheWaywardWanderer". I refuted his claims and this is where it all started. KeithLM is just another delusional fanboi from that website. They banned me from the website and claimed victory (because by banning me I could no longer counter their claims).
So I came here.

They can't silence the evidence here (no ability to ban me).
There is no reason to pick a side. What matters are the products. But when some folks cling to one side to such an extent that they spread FUD... I take offense and I rectify the situation (I despise fanboism). If this means taking on false claims made by Intel fans, AMD Fans, ATi Fans, nVIDIA fans then so be it. The evidence and the facts know no bias.
Lol. BTW I went there for fun and found this post:
rak:
Ever since Intel lauch of Lynnfield with dual channel DDR-3 IMC, practically every review shows its integrated memory controller is showing more performing when compared to AMD IMC in K10. Sandra score for Intel IMC is around 17 GB/s and for AMD IMC its 13 GB/s, Everest and other memory testing program also follow this trend. I know AMD chips are not bandwidth starved but it is always nice to have some extra BW. I am wondering if dual nature of AMD IMC is limiting its DDR-3 performance as unlike Intel IMC which only support DDR-3, AMD IMC on K10 support both DDR-2 and DDR-3. Or perhaps Intel implement better prefatch mechanism for its IMC than what AMD IMC have. Can we expect better DDR-3 performace with upcoming new C3 stepping. Any comments will be welcomed.
First response:
JF-AMD
If you do the math you see that one is ~33% faster than the other and one has 33% more channels. This is all logical.
The real question, for desktop users, is how does this matter? In benchmarks, where you can fill all the memory channels, you can see a difference.
But how does that impact day to day apps?
My car has a speedometer to that goes to 180. Pretty sure I could get it to 130+ based on some interesting data points that the austin police department might not approve of.
But when I drive to work, I really can't get it above 60.
Are you worried about maximum effifiency or maximum throughput. Efficeincy impacts how your programs run, but on a desktop maximum throughput is only an issue if you are regularly saturating the bus. And most people aren't on their desktops as far as I know
Notice I bolded Lynnfield. He was asking why Lynfield based Intel CPUs get 17GB/s memory bandwidth (Lynnfield is dual channel DDR3) and Phenom II gets only 13GB/s (also dual channel DDR3).
The response was "33% faster 33% more channels". That's a Nehealem based Core i7.
Second was him stating it was useless except in synthetic benchmarks. But wait wasn't it during the Core 2s first days that that was touted by AMD fans?
Sorry but I have to laugh at it. Its funny to watch my main points. Certain fans grasp and jump around the issues and even go as far as to change their ideals to support their preferred company.
jennyh :
I've noticed this also, our new friend elmo here does tend to crack a bit under the weight of irrefutable logic. I get pissed off lol, but that is through sheer frustration vs many intel fanboys.
Here I see you standing up for your views, with elmo getting a lot of support from others...yet he is the one cracking up. I can see why he got banned from AMDZone - although I'm not a fan of people being banned sometimes you have to look at their motives. Why was such an obvious intel fanboy there to start with, for example?
I've read a few others here saying that they went to AMDZone simply to wind up the natives. Of course they should be banned.
You don't know keith. he twists everything. it gets annoying. His views may not be incorrect but he completely changed the subject, went back to the original one and continues to argue that Intel "abandoned" the IMC when evidence shows that Intel used it in multiple mainstream products and has in XScale for years (since 2000) which is a mainstream product. They just tried a different road first (Netbust) and it didn't work so they went back. Happens to everyone.
BTW Elmo here is a post by Keith from AMDzone just so you know what you are getting into:
Get used to the following types of posts: (Here and on other forums.)
"Why are they picking on Intel... it's a free market. The government should leave Intel alone." (As if.)
"But AMD couldn't make enough chips; they were at full capacity... so they couldn't have gained any more market share." (Are people really that stupid? Don't answer that -- it's rhetorical.)
"I don't understand why this is happening to Intel because the i7 is clearly superior to the Phenom." (Well except for when it's not. Which is anytime an application can't benefit from SMT or Dynamic clocks.)
"This won't help AMD. Intel will still have the performance crown." (I guess if they have the performance "crown" they can't be guilty. Besides having the "crown" means about as much as having a zit.)
==========================
Did I forget any?
==========================
BTW: It has come to my attention that if I post here and also post 500% as much on another forum... that means that this is my "home". (Yeah they aren't really known for being logical are they.)
Phenom II 940 (CACVC 0849GPAW) & Zalman 9700, 3xATI4850 in XFire, 2 RaptorX on Highpoint 3120 Raid0, Gigabyte MA790GP-UD4H, Mushkin 8500 2x2G, Corsair HX1000W, Xaser VI
Huh.
Also like how pretty much everyone there has AMD + ATI. Of course before ATI was bought it could have been a mix. Guess they have no real choice but follow AMD.