• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

A Little More Info On Thoroughbred.

A

AGodspeed

Diamond Member
If you go to Planet3DNow, you'll see a chart listing the model ratings, Vcore, and a bunch of other stuff about Thoroughbred.

Most of this stuff is already pretty well known around here, but what might not be so well known is the voltage of Tbred, which seems to be 1.6 instead of 1.65, which several other web sites have claimed as the true voltage. Whether this means Tbred will be cooler than Palomino by the same amount Palomino was cooler than T-bird is unknown though.

Please, discuss. 🙂

EDIT: Also note that Tbred will be available in model ratings spanning from 1800+ to 2600+, which means some huge value overclocking for a lot of people out there.
 
the voltage drop will be nice.....it corresponds to 17% less power drawn than the palomino at the same frequency.....so, a T-Bred running at 2073 MHz will draw around the same power as a Palomino at 1733 MHz...provided that there are no tweaks to the core, a 2073 MHz T-Bred will be around a 2610+ model number Athlon, which is almost exactly the model number that the T-Breds top out at. therefore, any cooler that is spec'ed to cool a 2100+ Palomino should also be able to cool a 2600+ T-Bred

calculations (for others to check)

since the power draw is proportional to frequency*voltage*voltage, we get:

1.6*1.6*frequency (T-Bred) = 1.75*1.75*1733 MHz (frequency of the highest speed Palomino)
solving for frequency (T-Bred) = 2073 MHz

model number of a 2073 MHz T-Bred = 2100 + 1.5(2073 - 1733) = 1733 + 1.5(340) = 2100 + 510 = 2610

one should note, however, that because the core of the T-Bred is so effin tiny (80mm^2), that coolers will have a bitch of a time pulling the heat away from the core.....therefore, it may almost be necessary that coolers spec'ed to cool a 2600+ T-Bred have a copper insert. i'm not sure an all-aluminum heatsink will be able to pull the heat away fast enough

your guys input to my ideas is appreciated
 
No word on the Level 2 cache or FSB of the Tbred? There is a hardware site claiming that the Athlon Tbred will have 333fsb and coincide with the launch of the KT400 which will support the 333fsb! I forget the site, but that would be nice to see the Athlon with a higher fsb and more cache! I dont think it will actually happen though from a cost standpoint!

The overclocking headroom should be much higher due to lower voltage mostly! Where is the Duron .13micron core? I know at 1400mhz or so they are suppose to go .13micron and 266 fsb which will be most welcome!
 
> No word on the Level 2 cache or FSB of the Tbred?

In Planet3dnow.de's interview, AMD confirmed that Tbred will have 256K L2 and 266FSB.
 
acejj26 wrote:

"the voltage drop will be nice.....it corresponds to 17% less power drawn than the palomino at the same frequency.....so, a T-Bred running at 2073 MHz will draw around the same power as a Palomino at 1733 MHz...provided that there are no tweaks to the core, a 2073 MHz T-Bred will be around a 2610+ model number Athlon, which is almost exactly the model number that the T-Breds top out at. therefore, any cooler that is spec'ed to cool a 2100+ Palomino should also be able to cool a 2600+ T-Bred"

Just a couple of points. First of all, the AMD representative I spoke to last week told me Thoroughbred's vCore will be 1.65, not the 1.60 figure now floating around. I'm inclined to believe that.

Second, a cooler which is sufficient for the hotter-than-shannon-elizabeth XP 2100+ will likely not suffice for a 2600+ Thoroughbred. Remember, the 80mm die size is incredibly tiny. It's like a 75W light bulb element in the size of a piece of Dentyne gum. This means more heat over a much smaller die. Which means more efficient cooling will (likely) be necessary.

What is most disturbing about Thoroughbred is the fact that AMD seems to have completely neglected thermal management. Continuing to rely on motherboard and chipset manufacturers' to implement thermal monitoring and management is ridiculous, and it will be even more criticial with Thoroughbred. Thus far, we've not seen any indications that AMD has improved upon Palomino's capabilities (or lack thereof, to be more accurate.) This is one area in particular where AMD remains behind the times. I suspect we'll see no shortage of "Honey, I fried the Thoroughbred" posts. 😀
 


<< I suspect we'll see no shortage of "Honey, I fried the Thoroughbred" posts >>



Yes I expect to see a lot of those too.:disgust:
 
If we ask nice, will they put an integrated heat spreader on that for us. I know its not the most effiecient cooling possible but it allows for so much high clamp pressures and increased surface area for the thermal interface that the consumer has to deal with that they would be silly not to include one. I personally think duron's were much easy to chip than athlons simply because of the reduced size of the core (easier to tilt the sink). Now they want to dump an Athlon's heat through a duron's interface.

Otherwise they need to come with some kind of new heatsink mounting bracket like the pentium 4, that doesn't allow shifting left/right back/forward. Still going to be a job taking all that heat out of that size core. I see a potential problem right there for overclocking.
 


<< No word on the Level 2 cache or FSB of the Tbred? There is a hardware site claiming that the Athlon Tbred will have 333fsb and coincide with the launch of the KT400 which will support the 333fsb! I forget the site, but that would be nice to see the Athlon with a higher fsb and more cache! I dont think it will actually happen though from a cost standpoint! >>



It has been AMD's official stance for well over a year now that the Thoroughbred core will continue to utilize the current 266MHz FSB, and will not see any micro-architectural advancements beyond that already seen in the Palomino. Accroding to AMD, this will be a dumb shrink of the existing Palomino core, nothing more.
Needless to say rumours to the contrary run rampant however, despite AMD's continued stance that the TBred will not see any modifications relative to the current Palomino core.

Regrading HeatSpreaders, AMD has also confirmed we will not see such used on their mainstream desktop processors until the ClawHammer debute.

Whether one chooses to believe AMD's official stance is another matter altogether, though personally I am inclined to believe they are being arrurate in their evealed intentions regarding the TBred, and the future Barton cores.
 
Any idea as to when the NDA's will come off for Thoroughbred? It seems the cloud of speculation that encircled the Palomino is just as thick with the Thoroughbred. I for one look forward to Anand's article on this new chip...
 
if the vcore is 1.6 or 1.65 isn't going to be a huge deal, since the heat drop necessary to make wattage/sq mm match an XP chip would have to be over 30%. Neither 1.65 nor 1.6 satisfy those requirements.

So, you're looking at a chip that is drastically hotter per sq mm as others in this thread have said. Not only is cooling going to be quite tricky. In addition, the failure to implement any kind of onboard thermal management is a folly (as Pabster posted above).

Even with XP chips and Morgan chips being on the market now for some 6 months, you can count the number of motherboards with effective thermal protection on two hands....



Mike
 
I smwell a wat.

Those duwons are wated way too high!

Even that fancy new bus couldn't save them fwom being cwap.

Too little cache..
 


<<
So, you're looking at a chip that is drastically hotter per sq mm as others in this thread have said. Not only is cooling going to be quite tricky. In addition, the failure to implement any kind of onboard thermal management is a folly (as Pabster posted above). >>



and that is PRECISELY why i purchased a Northwood P4 1.6 about 1 month ago and will choose to stay away from AMD.

I don't think that we should consider that a small thing. It's like making a car w/o breaks saying that wheel manufacturers or tire manufacturers should be responsible. AMD is really dropping the ball on this one.
 
Well, it might just be speculation, but one would suspect AMD has to have some sort of handle on the heat generated by a T-Bred. It'd be plain suicide to release a chip that ran too hot for the majority of today's heatsinks without a real marked performace gain, as expressed by Mikewarrior2, earlier. I doubt the T-Bred's gonna be a cool running chip (AMD.. cool.. huh?!) but I have my doubts about the thing being too hot to handle for the majority of non water cooling users, or something drastic. I plan on grabbing one to replace my Athlon 1.4 (not willing to go P4 since I'm not really into o/c'ing anymore..) and if my 8045 is not working.. well, I have my doubts, but we'll see. IMO, a lot matters on the chip's tolerance for heat, and stability. If a processor runs at 55-59C but has a general tolerance of 80C, then there's nothing to worry about.. your room might just be a little toastier. I'm no scientist or expert, so correct me if I'm a little off base. And yes, I know that mobo temp readings aren't 100% accurate.

Well, I dunno.. I guess if AMD really released a processor that required a heatsink upgrade for every single user, I've got a good feeling it'd be a flop..

I do agree with the fact that AMD should be ashamed that their thermal protection has yet to be integrated except into a few, rare motherboards. I also agree that there oughta be a new HS retention system, aside from clips or screws. The most dangerous part of building a computer shouldn't be putting one of the most important components. But until the P4 really starts impressing me with anything other than its overclocking capabilities, I'll be sticking with AMD for a little while.

Oh well. Just my 2 cents. 🙂
 
MistaCakes,

The Thermal numbers are not pulled otu of thin air... Google Translated ZDnet Germany site with THorougbred heat numbers.

Thus far, its several weeks old, with no site nor person coming out to challenge this information. Which leads me to believe that it is fairly accurate.

Look at the XP2500. 74.1/80sq mm contact surface. .925 Watts/Sq mm. That's an extremely high number.

Now lets look at the .18 micron XP chips. Let's take hte Xp2100+, which outputs roughly 72W, but wtih a 128 sq mm contact surface. Heat/Sq mm = .56Watts/Sq mm. That means the thoroughbred has a 60+% higher wattage/sq mm than an athlon XP chip.

Until the zdnet germany site is challeneged, I would definately believe the article as being accurate. In addition, the Vcore drop to 1.65 versus 1.75 doesn't indicate that the temperature drop will be much more than ~10%.

IN order to match XP heat/sq mm, the thorougbreds need a 38% heat output drop, which isn't going to happen.

That being said, current higher end heatsinks should have no major issues with thorougbreds. The retail units, all alu units, etc, especially ones with questionable base qualitys, may very likely experience issues with them.


Mike
 
even if they don't change the cache size if they change the bus width we'd have a pretty good performance increase for relativly low cost in terms of transistors.
 
Mikewarrior.. I stand corrected. 🙂

I thought your comment was directed at a broader amount of the heatsink and cooling solutions out there and was a lil more...extreme, I guess. My bad. 😉
 
<<even if they don't change the cache size if they change the bus width we'd have a pretty good performance increase for relativly low cost in terms of transistors.>>

While sub-1GHz Athlons with half-speed L2 cache had no benefit of more than a 64-bit pathway from the L2 cache, I'd think a 128-bit pathway is probably long overdue. The most efficient way to do it at this point would probably be to just forget it and move ahead with Hammer technology. The Hammer family will make revisions to the Athlon cores obsolete before they'd even be completely through the design phase.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top