Tolliman X3 gets a name

Page 5 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
We bring up the subject of low yields. Why? Because that's directly tied to the subject of the AMD tri-core. Then Fudrix comes in and starts spreading his lies. We correct him. It's Fudrix that causes these threads to degrade because of his pro-AMD agenda.

But none-the-less, I will attempt to moderate myself and be nicer.
Click to expand...


There's nothing wrong with bringing up the subject of yields. The point is though that it DOESN'T MATTER WHY THEY WILL RELEASE THEM BUT THAT THEY WILL RELEASE THEM.

I'm not sure what lies I spread so perhaps you could enlighten all of us. The naysayers of Brisbane just got a shot to the chin as VR-ZONE announced YESTERDAY that Brisbane will be at 2.9GHz sometime this month.

Brisbane.gif


http://www.vr-zone.com/articles/AMD_Extends_Athlon_X2_"Brisbane"_Life_Span/5338.html


So does this mean none of you know what you're talking about?
 
Let me see... who started the flame? Who started name calling?

What I brought up here is legitimate, because someone asked about the origin of X3. I wonder who still can't accept the fact that Tollimans are defective Barcelonas.

And no, I'm not a troll. I understand that my tone needs to be changed, but that doesn't make me a troll. I have no intention of starting a flame war.

No offense, but he's being like that since day one. There's a reason why he's being constantly "pushed around'. Is it because of his name? no. Is it because of his avatar? No.

What BM does is put up some biased point of view, and repackage them as facts. When being confronted, he simply called others "rabid", and posted some inflammatory comments. If I'm not mistaken, there's a standard procedure he'll take when being confronted... :kaola:
Click to expand...


Yeah, you guys should stop insulting trolls. The word defective is misleading as it doesn't tell you anything. If you were to say they were "Barcelonas that have a core that doesn't clock as well" or "Barcelona's that have a core that isn't up to par" that would be different. But you make it a point to say how horrible yields are and you don't know.

You even posted a slide that says MATURE YIELDS AFTER 1000 WAFER STARTS, which should have been crossed since they have 9 SKUs of Barcelona out and 2 SKUs of Phenom coming next month.

Can you at least admit that they have done 1000 wafers?
 


Actually, it kind of does. What if Intel was releasing a 5,000 core processor that runs at 200mhz. Yeah, it'd be cool, but since most of our apps aren't even made for quad core it would be useless.

My point, AMD is not responding to market demand but is rather creating a product due to LOW YIELDS. Consumers didn't say, "Quad core is great but I want something that's a little slower. Give me a Tri-Core".


I think the tri-core will go over as well as QuadFX did.
 


Quad core is less than 2% of purchases. I'm not sure what the sales of QFX were but I would bet it's not far off from what FX62 was doing. I think TriCore is a great idea whether they are doing it for yield or not. It's a new product whether you like it or not.

Consumers (OEMs) said no one is really buying quad core for desktop and won't be for at least another year. but by placing a core chip in between dual and quad prices, they can make up for the lack of quad core adoption. I wish Intel had thought of it. Then it would be a product and not a mistake.
 
More choice is better, but I just worry that it won't be a profittable product line; unless Dell and HP run with it. It's all pretty cool that they can just shut down a defective core.

Baron, do they have the ability to independently clock the cores? That would be interesting, but very hard to market at this point.
 




The Dells and HPs are who this is for mainly. Sure I think newegg will see its share but if Dell can make $50 more dollars and only pay $20 more that's a pretty good chunk of change at 10,000 systems per month. the same with Newegg.

And yes, it is possible to clock the cores separately, but it's more so a HW function right now. Someone could create a BIOS that would allow cores to run permanently faster than others though that wouldn't be much of a gain since you would then have to set affinity for the application that could use the faster core(s).
Then you would also need OS awareness to prevent threads from being scheduled on the core while the app in question is running.
 


Imagine if they could sell a proc with two 2.8ghz cores and two 2.0ghz cores. (Assuming the two of the cores didn't turn out so good)

That would be interesting.
 
What i find funny is those calling tri-core a defective part......lets say a process top bins at 4 Ghz, does that mean everyone buying one that can only do 3 Ghz is being ripped off? omg they must be but even worse the 4 Ghz buyers are being ripped even more because on paper this process can do 8 GHz!!! why would i give a crap that an x3 came from an x4 with a dodgy core as long as the 3 remaining cores scaled? So for the bigger brained amongst us...does a laser reduced x3 scale as well as a native x4?

Peace

Carod
 
Word!, if it works, it cool with me.
 


It's defective in the sense that part of it is shut off because it doesn't perform well.

Imagine if you bought a 5 cylinder car that is actually a V-6, but has a cylinder that doesn't work well enough to keep up with the rest.

I agree with AMD's decision to make the three core part, it makes perfect sense for them, however, they are responding to a manufacturing issue, not a customer demand.
 
...however, they are responding to a manufacturing issue, not a customer demand.
Click to expand...
And that's my only beef with this. They're not the first company to do this and they won't be the last. But if you look at the coverage so far, you would think otherwise.

I think it's a good move for them, just wondering how will this fit in. If they can't get high enough clocks to beat the competition, then I can't see them charging a premium for their quad to make room for the tricore. They could push dualcore to the low end to make room, but then I think the tricore will need to show a significant performance increase to get people away from the dirt cheap dualcores.
 


Nope. This means AMD finally got their 65nm on track to catch up with Intel's 65nm process. It took them about 10 months to gain that 300Mhz. While it represents progress, IMO, its still not enough to fight Intel in the dual core arena. Its scheduled to launch this month, yet I have not seen any report on it, besides VR-Zone.

We still have yet to see Barcelona above 2.5Ghz, and 2.0Ghz on the market. While I have confidence they'll reach 2.5Ghz mass production by the year end, I wouldn't expect them to hit 3.0Ghz until Q208.
 



DELETED


But if you look at the 15% or so that K10 supposedly has over K8 per core, that means that a 2GHz Kuma will be equivalent to a 2.3GHz K8, perhaps more. Anand just did a new CPU comparison with Source and K8 is approximately 15% behind.

As I said before, I think they weren't increasing speed on 65nm because they still had to make 90nm chips at those speeds. As Fab 38 ramps, they will be able to make less 90nm and more 65nm. Fabtech reported a month or so ago that they have some (no actual numbers) 300mm equipment installed. I can only assume that it's with 65nm.

I mean whoever siad billion dollar upgrades were easy?
 

DELETED

But if you look at the 15% or so that K10 supposedly has over K8 per core, that means that a 2GHz Kuma will be equivalent to a 2.3GHz K8, perhaps more. Anand just did a new CPU comparison with Source and K8 is approximately 15% behind.
Click to expand...
And Core 2 is about 20~30% faster than K8. With the debut of Penryn, it will cost Intel a lot less to manufacture Wolfdale, then AMD to manufacture Kuma. I suspect the reason why AMD is launching all these dual core models, is to fight against Intel's dual core on a cost basis.

As I said before, I think they weren't increasing speed on 65nm because they still had to make 90nm chips at those speeds. As Fab 38 ramps, they will be able to make less 90nm and more 65nm. Fabtech reported a month or so ago that they have some (no actual numbers) 300mm equipment installed. I can only assume that it's with 65nm.
Click to expand...
You mean, they aren't increasing the clockspeed on their K10 core, because they still need to clear out equivalent clocked 90nm? :lol: :lol: :lol: :lol:



And I guess logical thinking was never easy.... huh..
 


50+% increase by 2Q08?


al.jpg


"I don't think so, Tim."
 


Oh stop acting so immature, you're making everyone on this forum very sad.
Why are you trying to trashtalk AMD anyways, did they drive over your cat?

To me it just seems as if AMD did something both brave and risky, by changing technology AND architecture in one go. One of the unwritten laws intel always sticked to, was to never do this.
The problems AMD faces now are a direct result of them breaking this law, and to me it seems to far they are doing ok, as Barcelona is not a total failure yet.
This can still happen of course, if it takes them too much time to get speeds up.

I don't get why many are trying to prove AMD is or is not behind Intel in a certain respect. Both companies are behind, depending on how you define "behind". Performance is all that matters, and so far none of them have been completely "behind" ever. The differences are just too small.

 


This is actually quite commonly done in the automotive world. Take for example GM's 4.2 L inline-six engine used in the TrailBlazer. It had one cylinder sliced off to become the 3.5 L inline-five used in the Colorado pickup and the Hummer H3, and the six less two cylinders is a 2.8-liter six used in the Colorado pickup. It allows for the use of many common components (valves, pistons, etc.) between the engines to save money.

Note that you'd always make sure the engine is balanced, though, so you'd only make a five out of an inline 6, not a V6. You can make a V6 out of a V8 (Ford's 4.2 L F-150 engine is 6 of 8 cylinders of the 5.4 L V8) or a four out of a bank of V8 cylinders (IH did this waaay back in the day) but a V5 would be terribly unbalanced.
 


It's an I-5 by design, which supposedly most of the Toliman chips are going to be. AMD has typically only named chips if they needed a separate die than another chip (e.g. AM2 90 nm 2x512 KB X2s are still called "Windsor" as they are disabled 2x1 MB dies.)
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom