What is this!? New Duron or something!?

[Helgaiden]

http://www.pcsupplysource.com/Itemdesc.asp?CartId=8980-ACCWARE-557XMZYR920&ic=ZCPU%2DKD%2DOF1%2E6&tpc=

look! Duron 1.6ghz with 266FSB

i thought durons stopped at 1.3 ghz with 200mhzFSB.

Is that for real!?

[c627627]

Wake up:
http://www.c627627.com/AMD/AthlonXP/

...and don't be too shaken up when Hammers show up next week.

Thanks for posting where they can be bought though.
I expect they'll do 2.3 GHz, just like T-Bred B's and Bartons.
Only 64KB cache though.

[Helgaiden]

Well i didnt know they would be out this soon already. *shrugs*

so its a barton with a smaller FSB and 1/8th the L2 cache?

[c627627]

Yes. They're just out.
I would think low cache would criple it but people say not.

Soon there'll be lots more talk about its performance.

You're the first person to post wher it can be bought that I've seen.
It's not free shipping so it's probably more expensive than the guaranteed $49 1700+ T-Bred B.

[Amorphus]

well, a barton is just a thoroughbred with a larger l2 cache and some core optimizations, so its not quite right to say its a simplified version of an advanced version. =P performance is gauged relative to the baseline, which would be the Tbreds.

[Damian]

Barton doesn't have any core optimizations... only difference between it and the TBred is the extra 256kb cache.

[c627627]

...many threads on why Barton is not "just a T-Bred B with double cache".

Key words in concluding arguments tend to be: "Barton grade processor modules..."

Here are two links for Damian. Post if you don't agree with conclusions.
http://forum.oc-forums.com/vb/showthread.php?s=&threadid=213004
http://forum.oc-forums.com/vb/showthread.php?s=&threadid=204995

[Gautam]

Originally posted by OC Detective
The Barton is a Tbred but with more cache - there is no redesign. Go back to 256 L2 cache and its a Tbred again. Either read the tech specs on the AMD site for model 8 and model 10 and try and spot the (lack of) difference or read here.
http://anandtech.com/cpu/showdoc.html?i=1783&p=2
or here
http://www17.tomshardware.com/cpu/20030210/barton-02.html

The core is miraculously larger for one, and only one obvious reason- that the extra space is where the cache goes. That's all; no more. I don't quite know for sure what makes a "Barton grade" a Barton grade; probably something subtle and insignificant in the manufacturing process. The Thortons and Bartons do have their own type of stepping code. But until anyone can show any definitive differences between the Barton and T-Bred aside from cache, I think its safe to say the only difference is the cache.

[c627627]

Gautam's in town to set everyone straight. ;)

Back in the days when we thought Thor(n)ton was going to be Applebred:

Originally posted by Graphic67
I had an opportunity to see the reported AMD roadmap in the short time it was posted on x86-secrets and it showed a list of modules within the three cores and which ones were overlaps from one core to another. The 'thorton' had significantly more than half of the modules in common with the barton. (I do not recall which and the info is no longer available.)

My personal opinion is that this IS a design refinement and the new core could have been named the T-bred C instead. Better a new name to avoid confusion...

Originally posted by Graphic67


Looks like I need to explain myself in more detail...

The T-bred A was essentially a die shrink (from 180nm to 130nm) of the Palomino Core.

The T-bred B core has an extra layer which was used along with rearranging components to change some signal paths from the T-bred A. This was done to reduce noise within the circuits as clock speeds increased. The T-bred B also moved the L2 cache into a location where the size of the cache could be changed easily.

The Barton is ALMOST a T-bred B with more cache. Design refinements were made, but no extra layers of silicone were added.

Many of the modules (I don't recall the exact list, but things like the in the FP Scheduler, Integer Unit, Instruction Control Unit) from the Barton design are to appear in the Thorton, but not the complete L2 cache.

This would make the Thorton out to be a T-bred B built with Barton grade processor modules and the original T-bred size L2 cache. So the logic portion of the chip will be mostly barton in heritage while the cache will follow from the T-bred B design. The thorton would therefore be very similar in size to the T-bred B but have some of the improved circuitry of a barton as well.

To make a crippled Barton (one with some cache disabled) would be a waste of production wafer space, reduce output from the wafer, and make a Thorton just as expensive to produce as a full Barton. The microprocessor industry is all about improving yield percentages and wafer efficiency.

Cache memory on the die is probably the least likely portion to fail in the production process, so the likelyhood of flawed cache chips is very small...too small to market the mistakes as another core.

Since the Thorton appears to be positioned to be a direct replacement for the T-bred B, which is a 266 fsb part, it too will be a 266 fsb part in its initial release. Due to the inclusion of barton circuitry refinements which have scaled up to 400 fsb, it is quite likely that we will see some or all fsb flavors become available eventually.

I hope this answered a few of your questions.

[gamefoo21]

so what i should get this and walk away from a 1700+ that can do 2.2ghz on avg?

[cherryp00t]

Indeed barton and tbred are differnt beyond the cache..

Right now amd has 3 diff cpus, XP/Barton/A64. As of now they only have 1 fab to make the A64s. They're gonna need more next year. So i belive they're prob trying to cut down on the XP production and move it to the barton line. Since bartons are gonna be the future XPs they're gonna need a Future Duron. So they're taking the lower end batches of the barton, gutting out half the cache and sell it as a applebred. This way they can save money by not having so many diff fabs