Sidewinder TC-4 vs Geminicool spiral block

[schoolie]

I've heard estimates of only 15%-20% heat reduction for the t-breds. Also, the cores will be smaller which will reduce the heat transfer to the block.

[jaydee]

Originally posted by schoolie
I've heard estimates of only 15%-20% heat reduction for the t-breds. Also, the cores will be smaller which will reduce the heat transfer to the block.

That is a considerable amount. Takes a 80watt CPU and drops it to 64watts back into the Duron Morgan level. Thats huge.

[schoolie]

Hi Jaydee!

It is a big decrease in power. I guess my point is that they will produce about the same amount of heat as the current models because they will be clocked higher.


--Rob

[jaydee]

Originally posted by schoolie
Hi Jaydee!

It is a big decrease in power. I guess my point is that they will produce about the same amount of heat as the current models because they will be clocked higher.


--Rob

If they use less power (which I heard 1.5V) then even higher clocked will not get as hot untill it reaches the end of it clock limit(which probably will not be to long the way it goes now days). So yes it should still be good. Cooler running chips is a good thing, that means the water coolers can provide even more cooling power to push the CPU that much harder. I think....

[schoolie]

I'm such a good thread hijacker I think the t-breds will run at 1.65V. What do you think about the smaller core size? I would think it would be a problem for heat transfer, but I guess they will produce less heat.

[redleader]

If they use less power (which I heard 1.5V) then even higher clocked will not get as hot untill it reaches the end of it clock limit(which probably will not be to long the way it goes now days).

Hey Jaydee, guess what the first Socket A 180nm chips ran at . . . Thats right 1.5v (typing on one now actually). They sure as hell didn't stay there.

Somehow I doubt TBred will either.

That is a considerable amount. Takes a 80watt CPU and drops it to 64watts back into the Duron Morgan level. Thats huge.

Yes but TBred is about 80mm^2 (IIRC) verses 128mm^2 for Paly. So while heat maybe 20% less, you have only ~63% of the surface area to conduct it with. That means that the thermal density is actually increasing, not droping.

[Brad]

when we changed to .18 from .25 there was a big reduction, and everyone was happy for a while, then with TBirds above 1ghz we hit the same thing again, especially with the 1.4ghz. The the AXP came out with reduced power, everyone became happy again, then 2000+ and 2100+ came out, making it hotter again. Now we are all seeing .13u as the saviour. It isn't really that great. sure it will use lower voltage, and lower heat at first, but by the end of the model line we will be above what the current 2100+ is doing

[Psychotic]

Blah... screw this .13µ stuff. I'm waiting to see Prestonia at .09µ.

[Haddy]

Originally posted by Brad
when we changed to .18 from .25 there was a big reduction, and everyone was happy for a while, then with TBirds above 1ghz we hit the same thing again, especially with the 1.4ghz. The the AXP came out with reduced power, everyone became happy again, then 2000+ and 2100+ came out, making it hotter again. Now we are all seeing .13u as the saviour. It isn't really that great. sure it will use lower voltage, and lower heat at first, but by the end of the model line we will be above what the current 2100+ is doing

yea but wuznt the t-bird the same voltage as the xp...1.75? i thought the new line would be lower...around 1.6 i herd

[Psychotic]

Originally posted by Haddy
yea but wuznt the t-bird the same voltage as the xp...1.75? i thought the new line would be lower...around 1.6 i herd

Methinks you are correct Haddy.

[EMC2]

Smaller feature size - definitely.
Lower Vcore (stock) - probably.
Smaller die size - probably.
Higher transistor count - {maybe}
Higher frequency of operation - definitely.

Lower Vcore helps reduce power.
Smaller Die size makes it harder to conduct the thermal energy to the block.
Higher transistor count means more power.
Higher frequency of operation means more power.

End result - impossible to call at this point and speculative.

{Another big unknown - the "heat spreader". If done right it can help - if done poorly (P4 is an example) it hurts.}

{editted - was thinking claw, so transistor count and spreader comments don't apply to the breads, everything else still applicable}

[schoolie]

Interesting EMC2. I thought the t-bred had exactly the same design as the Paly, only differing in scale. Maybe they will need to modify the design to compensate for the smaller element size. The engineering sample I saw didn't use heat spreaders---and yes it's pure speculation, but it sure beats working.

Well, I might as well take this thread to Cuba Think the t-breds will be more easily damaged by overvolting?

[jaydee]

Finally found a good quote:

So the 2600+, 2400+, 2200+ and the 2000+ at .13 micron will have a core voltage of 1.60 volts, use a maximum power of less than 72 watts, and all will use the famous "organic package" we spoke of before.

AMD warns that changes for Thoroughbred, given Palomino as the starting point, mean system integrators should be aware the CPUID will be 680h, voltage 1.60V, and PowerNow frequency switching may be turned on. Passive components will now go back on the top of the CPU package.

It will have a 80 millimetre square die and use the .13 micron process that is currently sampling.

Going to be interesting!!!

[schoolie]

Good information! Thanks JD

[jaydee]

I wonder what "Power Now Frequency Switching" is and what it does? man can't wait to see some results from one of these.

[schoolie]

I think PowerNow is the technology that reduces power consumption by throttling the processor frequency when not heavily loaded. It was designed for the mobile market.

[Brad]

it changes the multiplier and voltage according to cpu demand

[jaydee]

Originally posted by Brad
it changes the multiplier and voltage according to cpu demand

Hopefully we can turn that off? Sounds like a overclocking nightmare.

[EMC2]

Well, I might as well take this thread to Cuba Think the t-breds will be more easily damaged by overvolting?

As long as the relative (percent) increase is the same and we stick to short term effects, no. Could they safely handle the voltage levels some people use on XPs, I wouldn't try it. Without seeing their final specs, I would say 1.9V max. (10% under absolute max with proper cooling is what I consider a safe limit)

[Brad]

http://www.muropaketti.com/artikkelit/cpu/tbred_sample/


oc's quite poorly

[EMC2]

Ummm... one person, high multiplier, air cooled, low voltage, early sample

[jtroutma]

Late comer to this thread.......

I purchased a TC-4 about 3 months ago and have been very happy with it. I specifically bought it for the design, mounting, and style. I cant say that it is the best performer due to the fact that my radiator is what is holding my system back. I am looking into replacing that radiator with something better or adding another radiator to the mix so I can get better results. But so far, it has been worth the money I put out in getting it.

That's my 2 cents....

[DodgeViper]

I removed the tubulators in my TC4 and dropped my temps 2C using an Eheim 1250 pump.

[bigben2k]

Originally posted by DodgeViper
I removed the tubulators in my TC4 and dropped my temps 2C using an Eheim 1250 pump.

You know, that's been bothering me, ever since you posted it.

Without the turbulators, the TC-4 is nothing else but a Maze 1 type block.

Thinking about it, and from other threads in here, I figured that you could replace the turbulators with a chain of inverted cones. They would create the type of turbulence required to increase performance.

Twisting the water around won't do anything but add heat.

If you're interested in experimenting wiith this, let me know: I'll see how it could be put together.

[bigben2k]

Here's a pic.