Why all the AMD hate, guys?

[werxen]

Let us stop the fanboyism and acknowledge AMD's market strategy - the AVERAGE consumer not troubled by synthetic benchmark scores or maxing our Crysis. I see too much hate going on from both camps - let us make love not war. But in reality anyone who 'hates' AMD needs a friggin reality check. Intel will not be on top forever - anyone remember the Athlon series? People were sure AMD would lead the way forever after that.

 

[JAYDEEJOHN]

Never say never, as Im sure the cpus wont be too early, but the gpus may surprise, if chartered is compatable.
http://www.charteredsemi.com/technology/32nm_available.aspx

While this is not liekly, it could happen until GF picked it up, and then everything for gpus goes 28nm, while the cpus goes 32nm

 

[yomamafor1]

AMDs current is more than 2x, this is a huge perf jump, and try to remember, its not discrete, so no pcb, no ram, no fan and itll fit within current IGP TDPs at the smaller process and HKMG.

That's great. Nvidia's Ion is also light year's ahead of Intel's IGP. No one can deny those facts. Has Intel's market share decreased? Not at all. With all due respect, I don't think you fully grasp the concept of IGP, at all.

And also, you're not trying to fit in the TDP of the current IGP. AMD plans to fit it into the TDP of the CPU, along with 4 more x86 cores.

Remember, some are thinking 480 shaders, such as Hans Devries, so 320 certainly seems doable, as this is simply not just a IGP functional unit , but the APU side of that particular change, doing all the FP work, which also eliminates the dies space on the cpu, so a trade off, allowing for overall TDP distributions.

Again, I believe you're absolutely incorrect. If APU does the work for FP, then why does CPU needs to include a FP execution unit? Even on BD, there's a FP execution engine in every core. If APU will take over ALL of the FP work, why would AMD design FP unit inside BD? A more logical explanation is that the first Fusion is simply an inclusion of the IGP onto the die. As Fusion moves to its ultimate form (BD / Liano was the 1st stage of Fusion), perhaps the IGP will become the co-processor.

But again, most programs nowadays don't even take advantage of GPU computing. It is starting to gain some traction, but its still a long way off before the inclusion of a math co-processor is worth it. AMD knows that.

As for 480 / 320 shaders, its still a wild speculation. Before 4800 series was launched, some people speculated that it would have up to 1000~1200 shaders. Some people also speculated that 5800 series had up to 2000 shaders. Now we all know how those turned out. Not saying they're absolutely wrong, but before AMD announces anything, its wise to take those figures with a grain of salt.

So, theres no doubt it will be hugely better, and its the primary reason Intel is doing this, as LRB was set to do gpgpu functionality, discretely for this particular use, which will be combined for lessor use on chip with BD, and still discretely with future ATI and fermi designs

PS Later on, once LRB is functioning, this too will be Intels approach on Haswell, I believe, using LRB on die

Again, AMD can develop an IGP that's 10x better than Intel's version, and it will hardly make a dent in Intel's market share. I don't believe Larabee was designed for IGP. We've talked about this.

 

[noob2222]

spintelligence - The deliberate manipulation of intelligence reports to suit one's desired goals. Includes deliberate manipulation, exaggeration, or omission.

Basically its Intel's business practice to the tee, hence Spintel.

This thread has gone way beyone the topic, should be locked. Its just an insult war right now.

 

[JAYDEEJOHN]

I didnt say LRB was ever designed for IGP, but will be used in Intel version of fusion, which apparently isnt needed?
I think you dont see the upcoming changes Im putting forth, and denying it, and is what we'll soon have, and is when we will see if it makes differences in the growing mobile market vs the shrinking DT one.

As to the FP point, my mention of 32 vs 28 is, one would be on die while the other wouldnt be, and i should have clarified this.

As for the gpu speculation, leave that to me, and youre way off here,
The 4xxx series was expected to be 640 SP, and nVidia and everyone was taken by surprise when it arrived at 800, so in reality, you have it backwards, I do not, you said more was expected which wasnt true, Im saying hans et all are saying 480, Im saying less, 320.
So, disagree all you want,and as ive been saying all along here, if it werent important, Intel wouldnt spend money needlessly, as you claim, but since they are doing this, your claims are wrong, so good enough apparently isnt good enough.

And again, time will tell as ive been saying, whether IGPs will have decent impacts in the ever growing mobile markets, it would seem by Intels reaction, it will, by the very fact theyve made decebt changes in the coming IGPs..When power levels and solutions become more efficient, all this will be common day, and therell be no excuse, and as average Joe migrates from DT to mobile, and wants an everything PC, Intel would be left behind save for their changes that you claim arent needed, since AMD has that alternative, where power becomes a lessor worry.

Think of it this way, just as cpus have somewhat become good enough for average Joe, so too will power vs perf in gpus, or, IGPs. Thats what youre missing here, and yes, I quite well understand what a IGP means

 

[BadTrip]

Never say never, as Im sure the cpus wont be too early, but the gpus may surprise, if chartered is compatable.
http://www.charteredsemi.com/technology/32nm_available.aspx

While this is not liekly, it could happen until GF picked it up, and then everything for gpus goes 28nm, while the cpus goes 32nm

Although in the realm of possiblility, it is extremely unlikely. I was not talking about gpu though, only cpu. I can be quoted on this "AMD will NOT have 32nm cpus until Q1/11 at very best." Totally new process, totally new arch at same time wouldnt be easy even for Intel.

 

[JAYDEEJOHN]

Only a very slim chance of either happening, less so with the cpus, and I wasnt disagreeing with you, but things change for the better sometimes as well, and itd be a huge change, but could happen, since tape out to full ramp is alot quicker on gpu than cpu, and thats ONLY if things go extremely well

 

[yomamafor1]

I didnt say LRB was ever designed for IGP, but will be used in Intel version of fusion, which apparently isnt needed?
I think you dont see the upcoming changes Im putting forth, and denying it, and is what we'll soon have, and is when we will see if it makes differences in the growing mobile market vs the shrinking DT one.

No. I just looked at the history trend of the IGP, and gave it a logical conclusion about it. You can wave your "OMG IGP MATTERS!" flag all you want, but that's not going to change a thing: people who buys IGP mostly don't give a crap about performance.

Let's see,

AMD 690G vs. Intel G965

AMD vs. Intel graphical marketshare:

Now,

AMD 780G vs. Intel G45

AMD vs. Intel graphical marketshare:
http://jonpeddie.com/press-releases/details/amd-soars-in-q209-intel-and-nvidia-also-show-great-gains/

So AMD's IGP performance increased nearly 5x fold, and Intel's GPU marketshare went up 10%.

Yeh, IGP performance does matter.

As to the FP point, my mention of 32 vs 28 is, one would be on die while the other wouldnt be, and i should have clarified this.

As for the gpu speculation, leave that to me, and youre way off here,
The 4xxx series was expected to be 640 SP, and nVidia and everyone was taken by surprise when it arrived at 800, so in reality, you have it backwards, I do not, you said more was expected which wasnt true, Im saying hans et all are saying 480, Im saying less, 320.
So, disagree all you want,and as ive been saying all along here, if it werent important, Intel wouldnt spend money needlessly, as you claim, but since they are doing this, your claims are wrong, so good enough apparently isnt good enough.

Similar to what you said about i5's overclockability? About i7's cache relating to gaming performance? About Phenom I being as good as C2Q?

I'm not saying your numbers, or others' speculations could be wrong. I'm simply pointing out that from TDP point of view, its not likely to contain 320 stream processors. I could be wrong, and so can you.

And again, time will tell as ive been saying, whether IGPs will have decent impacts in the ever growing mobile markets, it would seem by Intels reaction, it will, by the very fact theyve made decebt changes in the coming IGPs..When power levels and solutions become more efficient, all this will be common day, and therell be no excuse, and as average Joe migrates from DT to mobile, and wants an everything PC, Intel would be left behind save for their changes that you claim arent needed, since AMD has that alternative, where power becomes a lessor worry.

Think of it this way, just as cpus have somewhat become good enough for average Joe, so too will power vs perf in gpus, or, IGPs. Thats what youre missing here, and yes, I quite well understand what a IGP means

yawn...

Please wave your flag all you want. Its still not going to change a thing.

 

[JAYDEEJOHN]

So, where in mobile, keeping the OEMs and promotion in mind does all this fit?
You didnt mention that, since its the fastest growing segment, and calls for IGPs the most, due to compatability alone, and cornering the do all is more important, as mobiles arent add in friendly.

The Intel IGP is half that of AMDs, and since neither of us know how much is discrete and IGP, Peddies numbers dont apply here, and overall marketshare in mobile, since its the fastest growing, and since theres much more IGPs found in mobiles than in GT % wise, then we will have real numbers.
Rome wasnt built overnight.
And whos to say, much like turbo, these numbers may never come to light, as sometimes itll have it on every chip, and sometimes itll have it and discrete, so then Intels share will show a huge drop, no? Or we will have real numbers for discrete only, since currently, Intel has no discrete, IGP number are used im the overall.

So, to prove Intels IGPs market position is fictisious at best, let alone margins.

As for your marketshare graphs, Ive seen the latest ones, showing ATI gaining 16%, not something from 2007 which predates the 4xxx series, and all the changes weve seen since, and how much all this will soon change which is to my point, and you still ignore as if it wont happen.

As for 320 SPs, Ive always held this POV, hope the best, expect the worst, and this is hope, but anything less than 160 is the worse, which puts it 4x better than current solutions.

If these future solutions cost no more to make, use no more power, even less, and are put into mobile, its a huge edge, and it seems noth AMD and Intel are heading this way, as the newer Intel IGPs perform better, much better

As for P1 perf vs C2Q, clock for clock, how far off is it, or should i mention that many here said theres no difference between the P1 and the P2? Including you?
i7s cache does play a huge role, so your point?
And yes, you got me, and you and everyone else has brought up why Id listen to a former Intel engineer making such claims, oh well, so all I can say is, the LRB knowledge was mine, the 800 SPs were a surprise I wasnt expecting, and no one else as well, and itd do you good to just ask someone else this, as youre completely wrong here.

 

[JAYDEEJOHN]

I know i shouldnt be listening to Intel employees, but just cant help myself:
Let's be clear, if you want to play high end game, use a GFX card ... but keep in mind that only 8% of the PC sold are used for games, and only 10% of the 8% play high end games ... Intel iGFx is getting more capable, this is a good new, you ll probably surprise when you OC it
This iGFx will be good enough for most of the people, and if you want to play high end game, buy a GFX card. The principal goal of intel is to provide PCs for every body on this planete, and to do so, we can't have every body running 200Watt GFX cards. a 200watts time the number of PCs needed for everybody on earth would be an ecological catastrophie. 0.8% of the people do what you people do here, so, please try to see this side of the story too.
(I expect to be flamed, as usual, but it is good to give the explanation for those interested into the Math behind all of this)

Francois

I think you will be surprise on this ... not much longer to wait

Francois
http://www.xtremesystems.org/forums/showthread.php?t=242319#post4181082

So, I decided to let everyone decide on their own if good enough is good enough, or Intel is heading in this direction?
So, why do this? Its certainly not needed, but 1 out of every 8 PCs are used for games, and thats a market too large to just give away, without having some presence

 

[muratgemici2003]

I don't hate AMD. I just never have owned one. AMD users seem to like the chips alot. Intel makes a lot of money grubbing decisions about their chips like the 920 I7 vs the 940. Paying $200 more for a couple of MHZ is lame! At least they could add some extra cache or QPI to make it worth while.

 

[JAYDEEJOHN]

Its simply the old pricing strategy, as both were doing this awhiles back, and AMD has changed it up, on the gpus as well, when they dont have the lead on theuir gpus, its a price perf deal, when they do have the lead, its one nice deal, since their highend is the lowest priced highend gpu since 2005 or so

 

[Chad Boga]

There is no way you will see production 32nm from AMD before next christmas. Even the bozos at the zone know that it is not possible.

One clown there was predicting June/July 2010, before others came to his aid. :lol:

 

[yannifb]

One clown there was predicting June/July 2010, before others came to his aid. :lol:

Can you tell me again where i said AMD will have 32nm parts in production in 2010?

 

[BadTrip]

I mean seriously, with 45nm AMD was behind Intel by a year (more or less) and now with 32nm they will only be about 4 months behind.

Being that Intel is releasing 32nm Thursday, four months would definitely be in 2010. That is not going to happen.

 

[Chad Boga]

Can you tell me again where i said AMD will have 32nm parts in production in 2010?

What makes you think I was referring to you?

I wasn't.

 

[yannifb]

What makes you think I was referring to you?

I wasn't.

Not you... I thought it would also quote Bad Trip who was referring to me...

 

[amdfangirl]

YOU DON'T KNOW ME

EVERYTHING YOU SAY DOES NOT DESCRIBE ME

EVERYTHING YOU SAY IS IDIOT INSULT

GET THE F OFF MY CASE MORON.


I AM DONE WITH YOU

DO NOT SPEAK TO ME.

========
this is like a half dozen posts all same kind of useless junk from some mentally deficient persistent troll

well if you took your meds everything would be rosy - such originality.
=======================

You know mister.

Capital letters is shouting.

Shouting like this is mean .

 

[Chad Boga]

Actually this link below is a great testament to the irrational unbalanced stupidity that is AMDZone, where long time contributor to that site(Kaa) talks about the moronic behaviour he had to deal with. In fact that whole thread is a great read.

http://www.xtremesystems.org/forums/showpost.php?p=4181391&postcount=236

 

[Reynod]

Hey we can be much more unbalanced and stupid here buddy ...

I won't personaly be outdone by those rogues over there ...

/wanders outside to talk to the mailbox.

 

[BadTrip]

I see that randomizer has defected. lol

 

[Chad Boga]

Hey we can be much more unbalanced and stupid here buddy ...

I really doubt that.

I won't personaly be outdone by those rogues over there ...

Sometimes you just have to accept that there are mountains you cannot climb.

 

[Upendra09]

Here's the current flow sim I'm doing - it's a rocket at mach 2.8 (note the CPU time readout):

http://i27.photobucket.com/albums/c183/chris_lapanse/random crap/Supersonicflowsim.png

And the CPU usage:

http://i27.photobucket.com/albums/c183/chris_lapanse/random crap/Solidworksflowsimulation.png

That's nice, isn't it true that a tear drop shaped plane experiences less drag?

Chad: if you need to link to x c p u s, you could always use tinyurl. That works fine, and avoids the issue you discovered with your links there.

You can also use the xtremecpu.com domain. It was never added to the profanity filter.

why don't you just link the web address?

 

[Chad Boga]

why don't you just link the web address?

The forum software was censoring it.

 

[Upendra09]

why?

 

[Chad Boga]

why?

Dunno