AMDs neue Roadmap #2

[Bozo Galora]

link

 

[AGodspeed]

I wonder if a 2000+ is really all AMD is going to release in Q1. Although if that 2000+ is a Thoroughbred part, then I can understand what AMD is thinking.

I hope that AMD roadmap is off though, I'd hate to still be at 133MHz FSB. I want a 166MHz FSB!!!!!!!

 

[Demonicon]

The jump from 2000+ to 2200+ on that list hopefully means the end of stupid 66MHZ increases from proc to proc. 132MHZ seems more resonable.

 

[AGodspeed]

The jump from 2000+ to 2200+ on that list hopefully means the end of stupid 66MHZ increases from proc to proc. 132MHZ seems more resonable.

Possibly, but it all depends on how well Thoroughbred yields at higher clock frequencies. I'm guessing it'll stay at 66MHz.

 

[Pederv]

Since the AthlonXP PR number is based on a given reference, I wonder if the AthlonXP +2200 is a 1.75GHz chip on a 166MHz FSB.
You would think:
+1900 = 1.6GHz @ 133MHz FSB
+2000 = 1.667GHz @ 133MHz FSB
+2100 = 1.733GHz @ 133MHz FSB
+2200 = 1.8GHz @ 133MHz FSB

But maybe:
+1900 = 1.6GHz @ 133MHz FSB
+2000 = 1.667GHz @ 133MHz FSB
+2100 = Won't happen
+2200 = 1.75GHz @ 166MHz FSB


EDIT - Corrected my math

 

[Pederv]

Or how about:

+2200 = 1.733GHz @ 133MHz FSB (with the L2 cache at 384KB / L1 cache at 128KB)

 

[AGodspeed]

But maybe:
+1900 = 1.6GHz @ 133MHz FSB
+2000 = 1.667GHz @ 133MHz FSB
+2100 = Won't happen
+2200 = 1.75GHz @ 166MHz FSB

Yeah, that's an interesting way of looking at it. I'd really like to find out soon though. :frown:

 

[BuckNaked]

I think the second table answers your question... It shows the 2000+ as 1667Mhz core on a 266 Bus (133Mhz) and the 2200+ as being 1800Mhz core on a 266 Bus (133Mhz).

Dave

 

[ElFenix]

gotta wonder when thoroughbred will come out... the 2000+ announcement is supposed to happen pretty early in january. also i'd like to know if it is just a die shrink or if there will be added performance enhancements (i.e. 256 bit bus to cache instead of that tiny 64 bit one now, more cache is always nice, also i wonder when SMT will show up at AMD?)

 

[SnowPunk98]

Thats what im waiting for is the .13 hope it will work in my Epox 8KHA+

 

[BuckNaked]

The way I am reading the roadmap, it shows AMD releasing the 2000+ in the 1st quarter with an .18 micron die, and then shrinking to a .13 micron process within the 1st quarter, and then releasing the .13 micron die 2200+ in the 2nd quarter....

Dave

 

[ElFenix]

thats how everyone else is reading the roadmap but only 1 speed bump the whole 3 months of the 1st quarter? thats pretty unbelievable.

 

[Pederv]

I think the reason that there is such a long period between the +2000 and the +2200 is because AMD is waiting for the KT333 that was rumored to be out in March. That way the +2200 (which I think is a 1.75GHz @ 166MHz FSB) has a platform. If you run the article through a translator it states that they can't back everything up, and it looks like they're making the assumption that the +2200 will be 1.8GHz @ 133MHz.

 

[Bozo Galora]

First KT333 Based Mainboard Announced!

Posted 12/14/01 at 2:13 am by Rat

Sending out different press-releases, Lucky Star seems to have got too carried away by this process and by mistake sent us a press release about the new KTVA333 Socket A mainboard based on VIA KT333 chipset.
There are very few details in this announcements, and no mainboard photo is enclosed, that is why we will consider that Lucky Star is only planning this mainboard now, and the chipset samples aren?t available yet. Nevertheless, we can get some valuable info about the chipset specs from this announcement, which we will do our best to do now.
So, here is what we know about the KTVA333 mainboard. It is an ATX solution supporting DDR333 (PC2700) memory, ATA/133, integrated AC?97 sound, AGP 4x slot. The mass supplies are scheduled for March 2002. Judging by this info we dare suppose that VIA KT333 chipset will:

* Differ from KT266A by the support of faster PC2700 DDR SDRAM;
* Be provided with a new VIA VT8233A South Bridge supporting ATA/133;
* Get into mass production in February-March 2002.

In other words, all we know now doesn?t reveal anything totally new about KT333 chipset: so far it looks like another pseudo-new solution from VIA. We hope that we are wrong about it...

 

[Pabster]

SiS 745 will likely force VIA to rush production of KT333.

I do believe the Athlon market is heading for 166MHz FSB, thus the plethora of PC2700 capable chipsets.

 

[MadRat]

If they move to 166fsb then it'll go:

Palomino (.18)
MP/1GHz = 1.333GHz @ 133MHz FSB (7.5X)
MP/1.2GHz = 1.333GHz @ 133MHz FSB (9X)
1500+ = 1.333GHz @ 133MHz FSB (10X)
1600+ = 1.4GHz @ 133MHz FSB (10.5X)
1700+ = 1.467GHz @ 133MHz FSB (11X)
1800+ = 1.533GHz @ 133MHz FSB (11.5X)
1900+ = 1.6GHz @ 133MHz FSB (12X)
2000+ = 1.667GHz @ 133MHz FSB (12.5X)

Scores increase by 100 but true speeds done in 66MHz increments

Thoroughbred (.13)
2000+ = 1.667GHz @ 166MHz FSB (10X)
2200+ = 1.73GHz @ 166MHz FSB (10.5X)
2333+ = 1.83GHz @ 166MHz FSB (11X)
2466+ = 1.90GHz @ 166MHz FSB (11.5X)
2600+ = 1.99GHz @ 166MHz FSB (12X)

Scores increase by 133 but true speeds done in 83MHz increments(?)

 

[SexyK]

I'd be willing to bet my bottom dollar you won't see .13u AMD chips in the first quarter next year, we'll see though, maybe they can pull it off.

Kramer

 

[bjc112]

willl that .13 work in Most motherboards now, like my Epox 8k7a

 

[bjc112]

?

 

[Demonicon]

I think/hope MadRat has got it.

 

[FlippyBoy]

i, too, am curious to know if there will be any core enhancements, or if this is just gonna be a die shrink?

 

[BeauJangles]

I want my Hammer Core!!

mmmm..... Hammer..... so fast... so cool

The next Athlon, it's gonna beat the pants of Intel. I can't wait

 

[Pederv]

bjc112:
I don't think the Thoroughbred will work on the 8K7A, there isn't the voltage support for it. Either the (Default) voltage value is set by the CPU and the jumpers add to that value or the motherboard sets the default to 1.75 and the jumpers add to that. A way to check which way the CPU core voltage is determined, would be to put an older Duron (lower core voltage) in the board and see what the voltage is.
If the CPU sets the voltage and there are Thoroughbreds at 133(266)MHz FSB and you set the multiplier to auto and there is bios support, then yes it will run on the 8K7A.

You know readin back through this I realize 2 things:
1) Of course the CPU sets the default voltage value, and there should be voltage support for the Thoroughbred core on the 8K7A(I'd wait for official word though). It should be mainly a matter of what FSB the Thoroughbred will run at.
2) I should back away from the computer and go to bed.

 

[MadRat]

I rehashed my numbers to show for the differences between Palominoes and Thoroughbreds. After considering that .13 Thoroughbred will take a new socket, plus it may not be compatible with Hammers, I am leary of its release. Sure the speeds of it will be higher than Palomino, but the cost may make it the equivalent of the .13 P!!!. Great idea too late for its time.

 

[Pabster]

Of course Thoroughbred won't be compatible with Hammer.

That said, I see no reason to assume Thoroughbred will require a new socket. AMD has planned to stick with Socket A past Thoroughbred even. Thoroughbred will use a lower core voltage, and some boards might require a BIOS update to allow the lower default voltage. Other than that, I don't know where you're getting the idea that a 'new socket' will be required.