• 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.

tbred = overhyped

T

tritium4ever

Senior member
/me throws out flame bait... 😀

The only difference between Thoroughbred and Palomino is the 0.13 micron process used to make the tbreds. Innately, it doesn't make the tbreds any faster than the current palominos. Sure, the tbreds will run cooler and thus allow for a bit more overclocking room, but at stock speeds a tbred and a palomino will run at exactly the same speed given an identical performance rating. Where's the beef?!
 
The reason that T-bred was so hyped about was because it may catch up with the Northwoods in terms of overclockability, not due to any revisions in the microarchitecture. Currently, the Athlons are trailing behind the Northwoods in terms of overclockability and AMD fanboys and others alike are hoping that t-bred will cut down on some of the lead the NW has.
 


<< Sure, the tbreds will run cooler and thus allow for a bit more overclocking room, but at stock speeds a tbred and a palomino will run at exactly the same speed given an identical performance rating. Where's the beef?! >>



The .18 to .13 switch worked wonders for Intel, as have other die shrinkings. The P3 went from 1.0 - 1.1 Ghz max speed to about 1.5-1.6 GHz max speed; that's around a 50% increase. The P4 has gone from 2 Ghz to 2.7+ GHz in a short amount of time with the same die shrink. Sure AMD chips run cooler, but if the switch is anything like Intel's, it should allow for a LOT more overclocking headroom.
 
Both of you are right, which also goes a long way in explaining why nobody outside of enthusiast/overclocker circles gives a rat's ass about Thoroughbred. What needs to be done now is to have motherboard manufacturers actually put proper hardware support for the thermal diode as well as throttling capability so that the computer doesn't just shut down when the processor overheats. Don't get me wrong here, I'm a big fan of AMD ever since I got an Athlon XP several months back, but does the technological "leap" justify the hype? Well, no.

One thing I forgot to mention is that the die shrink will allow the Athlon architecture to ramp up to higher speeds, which of course is a good thing.
 
but teh overclocking headroom isn't much more than palomino's in actual testing. Take hte overclockers.com.au tests. They used a vapochill unit, and got ~100mhz or so more than their Palomino chip on the same cooling setup.

Bear in mind that this is 100mhz on an extreme cooling setup. Also bear in mind that having a much smaller contact surface with the Tbred definately hurts heat transfer, nad is contributing to its "lack" of northwood type overlocking headroom.

Seeing as how Tbred isn't supposed to be released higher than 2ghz or so (not gonna get into PR, i have no idea what PR rated this chip would be), I doubt we'll see anywhere near 50% overclocks with Tbreds. Plus, 2200+ tbreds have a higher vcore than other Tbred chips, 1.65V, which doesn't make it all that much cooler than XP palominos.



Mike

P.S. At 1.65V Vcore versus 1.75V vcore for a palomino, the tbred runs 150-160% hotter in terms of wattage/sq mm of contact surface. That is why the reduced core size is a problem for cooling.
 
Mikewarrior2, did you read the article completely? The used the exact same setup for both the Palomino and the Thoroughbred, the t-bred ran cooler under the exact same test conditions. This is the second time I have seen an engineering sample report lower cpu temps compared to a palomino. That doesn't sound like a cooling problem to me. There are two other reports out there where the T-bred 1800 hit 1900MHz air-cooled. Seeing how AMD is just getting started on the .13 micron core, I don't think it is to premature to predict higher overclocks after a few months. 50%? No I think not, but then when the 1800 is priced the same as the XP 1800 at $92 and hits 1900MHz air-cooled right now, and most likely 2000-2100 after the yields get better, it doesn't need too. 2GHz for $92 is gonna more than keep AMD delivering the most bang for the least buck. This will beat a 1.6 Northwood clocked at 2.5GHz+ and still cost less.

Better get all your arguing in now, because in a very short time the proof will be running in my motherboard😉

Bear in mind that this is 100mhz on an extreme cooling setup. Also bear in mind that having a much smaller contact surface with the Tbred definately hurts heat transfer, nad is contributing to its "lack" of northwood type overlocking headroom.

Obviously with a vapochill on it, heat isn't the biggest factor limiting the overclock, it is more likely the core itself. But of course, it is common knowledge the thunderbird, XP, t-bred doesn't have to run the same clock speed to get the same performance as the P4.


 
A shrink in transister size does not neccessarily mean a significant increase in yield. There are many other factors asside from transister size and heat that allows for higher yields. In a 10-stage pipelined processor, each stage has pretty complicated gates compared to a 20-stage processor. You can only operate something that complex at certain clockspeeds before the delays in signaling become a problem. This is one of the advantages to shifting to simpler stages, with the .18 micron P4, the heat was most likely the cause of the limit in scalability, which is why the shrink to .13 micron helped so much (because of the reduced heat). Whereas, I'm guessing, the K7 architecture is reaching the end of its scalability, not just due to the heat but due to the very nature of how complicated the stages are. I expect SOI (if it is to be used later in the K7 line) to increase scalability somewhat and AMD to improve yields somewhat as well, but I wouldn't expect the shift to .13 micron to help the K7 as much as it did the P4.
 
The P3 coppermine was actually using Al interconnects. And so was the P4 Willamette. Now, going back to the Athlon, the Thunderbirds switched to Cu on 1GHz, then *BOOM* 1.5GHz overclocks suddenly started happening whilst the Durons on Al were pegged at 1GHz until a relaunch.
Intel relaunched the P4 on a copper process.
Expect 25-30% higher clockspeeds on Thoroughbred, and being limited by the system bus.
 
DDR 400 is coming soon, and with a FSB of only 266, the Athlons can't even use the full potential of DDR333. I would think that the increased clock speed will really put a hurting on that FSB.
 
Rogue1979,

Cooler at STOCK VOLTAGE. NOte that overclockers.com.au never said anything about it beign cooler when they cranked the voltage to 2.0+V. Maybe you didnt' read the whole article.

AT Equal voltages, which from all overclocking "reports" is necessary for overclocking the Tbred, the Tbred will be "hotter" running than an equivalent voltage Tbred. Period. No questions at all about that.

SEcondly, Overclockers.com.au used a) an external measurement source and b)I never said at stock voltage that tbred wouldn't be cooler. Maybe you should read my whole post, I said

which doesn't make it all that much cooler than XP palominos.
Click to expand...


Mike

P.S. I'm not arguing price/performance ratios betwene intel or AMd, or one's better value over the other. Purely stating the thermal concerns about Tbred versus Palomino, and the known test data out there. if you want to buy a tbred, that's fine, but for the other posts thinking that tbred will overclock similiar to northwoods simply because of the .13 micron switch.

P.S.2 Dont' bother post socket-thermistor or external thermistor temp readings with your tbred. They're not accurate by any means. hell, there's a motherboard (Soltek DRV5), that reads teh diode but gives you no way to read it via software or bios... why? probably because internal diode is significantly higher than the crap-ass socket-thermistor they are using.
 
Originally posted by: imgod2u
A shrink in transister size does not neccessarily mean a significant increase in yield. There are many other factors asside from transister size and heat that allows for higher yields. In a 10-stage pipelined processor, each stage has pretty complicated gates compared to a 20-stage processor. You can only operate something that complex at certain clockspeeds before the delays in signaling become a problem. This is one of the advantages to shifting to simpler stages, with the .18 micron P4, the heat was most likely the cause of the limit in scalability, which is why the shrink to .13 micron helped so much (because of the reduced heat). Whereas, I'm guessing, the K7 architecture is reaching the end of its scalability, not just due to the heat but due to the very nature of how complicated the stages are. I expect SOI (if it is to be used later in the K7 line) to increase scalability somewhat and AMD to improve yields somewhat as well, but I wouldn't expect the shift to .13 micron to help the K7 as much as it did the P4.
Click to expand...

I learned something today 😉 So from .18um the palomino tops at 1.73Ghz and the .13um will only max at around 2.0GHz .... thats not that amazing .... I guess what you just said explains that 😀 I also looked at the TBred as just hype from all the amd zelots (hehe) that got jelious from our 800-1000MHz overclocks 😀 Alot of sites have the 1.533GHz Tbreds and they didnt oc that great .... about 366MHz (up to 1.9Ghz stable) at 1.9vcore .... And thats with unlocking too .... Ive seen C3's do about 500Mhz oces .... Still a better overclocker then the palomino but nothing compaired to the Woodies OCes 😀 Too bad the hammer might not overclock .... If it did wouldnt it mess with the memory controller?

SSXeon
 
Overclocking:
It would be a fair guess that you're reading this because you want to know how it overclocked. Well, I can't disclose that due to.. only kidding! The first thing that struck me was the drop in core temperatures on the Vapochill at default voltage. Usually the Vapo will show a LCD temp display of around -16C at idle, with the the lower voltage of the Tbred the idle temperature reading dropped to -20C. Recording accurate temperatures when dealing with CPU's is beyond any equipment I have here, so let me just say that there is, as expected, a definite drop in CPU temperature with this new core. This I assume is directly related to the lower voltage of the Tbred. Now that is always good news, whether you?re using a FOP38 or a Vapochill!

You were right, they said at default voltage. But that is still 4c lower temps for at 1.65v compared to a Palamino at 1.75v. That tells us right there that at equal voltages the core temps are gonna be basically equal.

P.S. At 1.65V Vcore versus 1.75V vcore for a palomino, the tbred runs 150-160% hotter in terms of wattage/sq mm of contact surface. That is why the reduced core size is a problem for cooling.

Where is this cooling problem? While mathematics tells us your calculations are correct, it just isn't happening that way in the actual tests. You said in some of your earlier posts on the subject that t-bred was gonna be hotter than the original thunderbird. This just isn't true from the info we have seen so far. No, the t-bred isn't going to overclock as well as a Northwood. But like I said earlier, it doesn't have to! It is a incredible deal for under $100 to compete with the more expensive 1.6 and 1.8 Northwoods. And true to AMD value, you won't have to invest on a new motherboard. I noticed almost every person switching to a Northwood set up also had to buy a new motherboard.

The Northwood is a fine cpu and overclocks well, I wouldn't complain if someone gave me a 1.6@2.5GHz set up. But my cash will go for AMD because you still get more computing power for the money at any price level.
 
Rogue1979,

I said hotter per square mm and harder to cool than palominos. Sure, at default voltages adn default clocks, Tbred will be slighly cooler. And don't say that its more than that, because socket-thermistors aren't going to accurately show this. NEither are external thermistors. Until itnernal diode tbred temps are shown, we wont' know the exact amount of how much cooler an Tbred chip is versus an XP chip.

AT the same voltage as an XP chip, the tbred will be hotter. No one has posted temps of any type with their "overclocked" tbred. They only show the temps at default voltages/clockspeeds... I wonder why?



mike
 
Originally posted by: imgod2u
I feel like I just fueled the fire for something horrid.......
Click to expand...

nah ... I like thease posts .... because if my info is wrong I will be told ect. And I wasnt picking on the TBred Im pissed that amd is delaying the tbred .... as much as I love this 800Mhz lead by intel .... i want to see what the TBred can do ..... I think little-medium flame wars are healthy .... like fighting in a relationship 😀

SSXeon
 
AT the same voltage as an XP chip, the tbred will be hotter. No one has posted temps of any type with their "overclocked" tbred. They only show the temps at default voltages/clockspeeds... I wonder why?

Exactly, and according to those stock voltages the t-bred was 4c cooler. Now you run a XP set up yourself, and it is obvious that if at 1.65v for the t-bred vs 1.75v for the palomino resulted in a 4c drop in temps, then adding that .1v to the t-bred core is gonna make the temps real close. So again I say, where is the cooling problem? While thermistor temps are not accurate, using the exact same set up to test two different cpu's will have enough accuracy to to show that the t-bred runs way hotter than the palomino if that was true. Do I really need to use search to bring up all your former posts where you predicted the t-bred would run hotter than the original thunderbird because of your die size vs. thermal output calculations? You were wrong then and you're wrong now. Go ahead and admit, it won't hurt too much😀

So this doesn't take a wrong turn here, I am enjoying this debate with you, don't take it too personal😉 If you can prove I am wrong I will yield, but you've got some work to do!
 
"I expect SOI (if it is to be used later in the K7 line) to increase scalability somewhat and AMD to improve yields somewhat as well"

SOI probably will improve scalability a bit. On the down side - it will probably hurt yields. As I understand it, the type of SOI chosen by AMD requires extremely tight process control to get things right. And SOI wafers cost about 35% more, so AMD will almost certainly pass along the added cost.:Q
 
Rogue1979,

Please find those links.>> i have only found links where I say hotter per square mm than XP chips and not any where I said htat default clock versus default clock that the Tbred would be hotter than an XP chip.

IF we take a non-agressive cooling (ie, not a vapo-chill setup), where teh contact base is of questionable quality or aluminum contact material, then the contact surface of the tbred will be a hindrance for cooling. Its a question of concentration of heat per contact surface and the interface material. The smaller contact surface of a tbred shoudn't be a problem for those with agressive cooling setups, because aggressive cooling setups often use good quality thermal interfaces and high quality heatsink/waterblock/contact bases.

With a standard heatsink, like an AL coolermaster, the smaller contact surface hurts cooling.


Mike

P.S. Some links where I have posted Tbred will be hotter per square mm than Xp Palominos, not HOtter than palominos. There is a very clear distinction between the two.

Link 1

Link 2

Link 3

Note none of them do I say that Tbred will be hotter, rather specifically that it will be hotter in terms of heat/sq mm.

P.S.2. I am done with this thread. I do not tolerate people selectively reading my posts, and trying to interpret them their own way... I have linked the 3 search profiles iwth my name regarding tbred posts for anyone who wants to see them.... Again, nowhere did I say hotter, but I did say hotter per square mm. If you can't see that there is a difference between the two, then I can't help you.
 
The fact that the .13 athlon chip only resulted in .1v drop in vcore in itself is a bit disappointing....couple that with the fact that 4c drop in temp versus an already extremely hot athlon xp chip and I question how much scalability they will irk out of this....

I think overclocking will be an extreme disappointment compared to intel northwoods....I bet 20 percent will be difficult with air cooling alone....

Look at thugsrook...he has taken a chip .13 1.6ghz northwood and hit 2800mhz with a volatge of 1.8v which is still less then the default 1.85v of the .18 willamette cores....I hit 600mhz overclock with 1.65v and 720mhz overclock with 1.775v....this will not happen with the amd chips...

I think this is all mute anyways as the tbred is short lived until the hammer arrives....a lame duck if you wish....

 
I don't understand AMD's thought behind the core shrinkage? I understand the die shrinkage, but knowing how comparatively hard to cool their chips are, and the number of complaints about cracked/chipped cores you would think they would have at least increased the core's packaging size compared to the XP (or at least left it the same). BTW, unless they changed the packagings material it is going to be easier to chip a smaller core. A smaller fulcrum makes tilting of the heatsink easier.
 
Here is one I found pretty fast link

However, the voltage is not yet solid, there are rumors of 1.6V. However, the DIE size is known. ~80 sq mm. IN order for a tbred to match an XP chip in terms of Heat/Sq mm it would need to run roughly 38% cooler clock for clock. Which just isn't going to happen.



Mike

P.S. A T-bird @110W = .91 W/Sq mm. A tbred @ 70W = .875. HOwever, that is a T-bred at stock voltage/clock(1867). Now, let's estimate what a 2ghz Tbred @ 1.8V would put out: ~84W or so. Which would put heat/sq mm = 1.05. Far higher than the T-bird you have posted for example. now, if you push to 2.5ghz @ 1.8V, you may be looking at 100W of heat, with only 80 sq mm of contact.

Now, maybe you can help me with my comprehension, but this sounds like you are saying the t-bred is gonna be real hot compared to an XP.
 
One last god damn post in this thread because Rogue1979 is once again using selective quoting and reading of my posts.

Please read the quote again, dumb@ss. I'm sorry, but I have to call you a dumb@ss because you keep attempting to misinterpret my posts.


Let me break down your quote step by step for your feeble mind.

P.S. A T-bird @110W = .91 W/Sq mm. A tbred @ 70W = .875.
Click to expand...

that is in terms of Heat/Sq MM.

HOwever, that is a T-bred at stock voltage/clock(1867). Now, let's estimate what a 2ghz Tbred @ 1.8V would put out: ~84W or so. Which would put heat/sq mm = 1.05. Far higher than the T-bird you have posted for example.
Click to expand...

Once again, you misread my post. Just like you selectively misread my posts regarding pretty much anything. once again, I will have to break down the post. The Far higher than a t-bird you have posted for example statement is in relation to, once again, HEAT PER SQUARE MM of Contact surface. If you want to selectively take words and prases out of context, then do so without posting it.

There, once again, i've had to clarify your posts because of your selective reading/memory, or you just being a jackass. Please read the whole quote again, and not selectively read a single sentence. A single sentence isn't complete without the rest of my quote above.

People like you make me want to quite this forum, because of how capable you seem of bending a few words to suit your own needs. That is truly lame.


Mike
 
I am enjoying this debate with you, don't take it too personal😉 If you can prove I am wrong I will yield, but you've got some work to do!
Click to expand...

Funny, there are a lot of people in this forum that think they know it all. Only wise people know that they have still a lot to learn. That is the reason we have to keep our ears and minds open when reading threads in forums...

😉
 
I guess I was wrong again. I thought you were a worthy opponent Mike, but I just broke you down to lowly name calling.

P.S.2. I am done with this thread. I do not tolerate people selectively reading my posts, and trying to interpret them their own way... I have linked the 3 search profiles iwth my name regarding tbred posts for anyone who wants to see them.... Again, nowhere did I say hotter, but I did say hotter per square mm. If you can't see that there is a difference between the two, then I can't help you.

Should have quit while you were behind!

As far as selectively posting your threads, I put the whole link there first so everybody could get the whole context of your post. I pasted the important parts, either way your meaning is clear for anybody to see. Boy, talk about selectively quoting, you are desperately quoting your heat output vs. die size comparisons only, leaving out what you said before and after.

Learn how to lose gracefully!

I really I hope you don't leave this forum, you do have some good input. Besides, when my t-bred is on my motherboard running as cool as any XP, you'll miss the results when I post them😉
 
Who cares if the TBred is hotter per square millimeter if it runs cooler overall? It sounds like the TBred will still run cooler than the Palimino at the same speed.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top