Thoroughbred to Require 45 Amps

[NOX]

<< AMD is warning that to support 2.0GHz operations, the onboard regulator must be capable of yielding 45 amps. >>

Even that is unreasonable. My current 350w PSU will run a 2.4 P4, but I would need a 551w (enermax) to be able to run this proc. Not even the Antec True Power 480 would be able to support that. In fact I don't think the 550 can do it also, will check on that. This has got to be a mistake!

 

[Swanny]

I think that's a typo. No way they would be that crazy.

 

[microAmp]

Hope that is a typo :Q maybe the 3.3V and 5V rail combined? to achieve the 45 AMPs.

AMD's site on the Athlon XP even says 3.3V and 5V rail at this link.

Please be a typo, don't want to buy a new PS.

 

[RagingGuardian]

It has to be a mistake.

 

[robg1701]

I think everyone is misinterpretting this....its meant to be 45A for the regulators, not 45A from the 5V rail.......the core doesnt run on 3.3 or 5v, its pulled down from the PSU rails...and if they work basically like a transformer this is gonna mean an increase the current, right ? As has been said (and i actually typed this myself in a fairly long reply, but got cut off whilst posting and IE deleted my text when i hit back...couldnt be assed retyping).....the specs for Palomino called for 40A on the power.....that was certainly not 40A from the 5v rail (otherwise every would already have had to replace their PSU), it was 40A from the regulators and is why everyone went 3-phase power nutz.....

Oh, and to the person saying it cant be 45 cause pally was only 40....well, the Tbird was less than pally wanst it ? Processors are using more power than ever, but lower voltages than ever...that means higher currents than ever.

 

[Budman]

My power supply can do 47amps on my +5.

But can you imagive the heat being produced by a cpu that sucks 45 amps!!!!

I am glad I went the P4 northwood route.

 

[Swanny]

<< I think everyone is misinterpretting this....its meant to be 45A for the regulators, not 45A from the 5V rail.......the core doesnt run on 3.3 or 5v, its pulled down from the PSU rails...and if they work basically like a transformer this is gonna mean an increase the current, right ? As has been said (and i actually typed this myself in a fairly long reply, but got cut off whilst posting and IE deleted my text when i hit back...couldnt be assed retyping).....the specs for Palomino called for 40A on the power.....that was certainly not 40A from the 5v rail (otherwise every would already have had to replace their PSU), it was 40A from the regulators and is why everyone went 3-phase power nutz.....

Oh, and to the person saying it cant be 45 cause pally was only 40....well, the Tbird was less than pally wanst it ? Processors are using more power than ever, but lower voltages than ever...that means higher currents than ever. >>



Ah, now that answer actually makes sense! Thank you, robg1701

 

[Bozo Galora]

ya, the key words are "regulator" and "handle'. The regulator is like a railroad roundhouse dishing out
power - all power. But it does seem to mean your 300/350W PSU is no longer viable for reliability.

 

[BD231]

Well 300 watt PSU's are becoming kind of weak by today?s standards. Most people are using 2 or more HD's these days, 2 optical drives, 3 to 4 80mm fans. If you 7volt mod your fans, that limits the ability to use the extra connector to only other fans. I would say that 350 and 450 watt PSU?s are becoming the standard. It?s better to have headroom for extra devices anyway. I?m currently stuck with a 300 watt PSU and I already feel its limits.

 

[Sid03]

3-4 80mm fans? actually, MOST people operate OEM systems which have either 0 or 1 case fan.

do most enthusiasts have that many case fans? what for??? do they like all that noise? no thanks, none of my systems have more than one.

 

[mastertech01]

My AMD approved MP PSU, NMB, only has 27A 5V... and it runs dual MP processors.. so this is obviously a misunderstanding?mis-statement

 

[LordOfAll]

Well, for one thing there is no way the 5 volt lines from the psu to the atx connector will handle the 225 watts this would call for. I think they are only 20 gauge for chris' sake. They would catch fire in no time. This is why Intel added the 12v connector. The 12v lines to the normal atx connector couldnt carry that much power safely. gotta be a misprint.

 

[zemus]

Swanny is correct, When the voltage is bumped down to core voltage, it leaves alot of room for higher current


Here is something a little fun to think about for those who might not beleive that you can get more current than one puts in, your power socket in the average north american home is only 15 AMPS , yet your PSU put alots alot more than that .... by dropping the voltage of course. Well, the PSU does this for us, and as swanny says, your mobo regulators do this yet again.


And 45 Amps is very easy to gain on the mobo btw

 

[AGodspeed]

Am I the only one questionig the ignorance of the individual who wrote up this stuff? Look at what HardOCP wrote in the paragraph just above the one that talks about the 45amp requirement:

It was discussed that the Barton (SOI core technology) would have 384K on-die L2 cache as well.

This is totally and completely off. First off:

1. Barton will NOT incorporate SOI technology.
2. Barton will NOT have 384K of L2 cache (I think they meant 384K of total on-die cache, which brings me to # 3).
3. Barton will NOT have 384K of total on-die cache, it will have 640K of total on-die cache (128K L1, 512K L2).

Now why would anyone take this "45AMPS" info seriously if HardOCP can't even get the facts straight about something that has been known for weeks now (that is, it was revealed a few weeks ago that Barton would not be SOI and would have 512K L2 instead of 256K L2 like the current Palominos and Tbreds).

Btw, 45AMPS on a 5V rail would require 225 watts of power. Do the math, it doesn't work. 225 watts is about as much power as Intel's massive Itanium die requires, which is roughly 4 times as large and has many times the amount of on-die cache as the Athlon.

 

[robg1701]

zemus


<< Swanny is correct, When the voltage is bumped down to core voltage, it leaves alot of room for higher current


Here is something a little fun to think about for those who might not beleive that you can get more current than one puts in, your power socket in the average north american home is only 15 AMPS , yet your PSU put alots alot more than that .... by dropping the voltage of course. Well, the PSU does this for us, and as swanny says, your mobo regulators do this yet again.


And 45 Amps is very easy to gain on the mobo btw >>



cough

Ill take that to mean you think I was right ?

 

[WarCon]

As I stated above, my T-bird was running somewhere around 49 amps of vcore current (and my T-bird wasn't even running that high of an overclock). Simple logic implies that the lower the vcore that you go the higher the current requirements are if the power requirements remain constant. Less semiconductor material (.18u versus .13u) implies a lessening of junction resistance which will cause a greater current draw for the same voltage. I think mobo manufacturers will probably have to design a whole new power setup for processors when they start hitting 5-10 gig, because voltages will be so low that a slight change in the voltage will be a huge change in available power and will end up causing instability.

I vaguely remember 2 phase regulator circuit on one of the AMD boards that was causing issues for people and making them get higher wattage supplies in order to overclock using it.

 

[joohang]

<< It has to be a mistake. >>

 

[arcas]

Even if the requirement is that the onboard regulators be able to deliver 45 amps...That's quite alot of current. Won't be long and we'll
be able to use our computers as makeshift arc-welders.

 

[McCarthy]

Good to see some are finally thinking in terms of amps and volts. Seems people forget one without the other doesn't mean much. I'm hoping it's 45 amps draw at 1.65v and they mistakening took that to mean 45amps on the 5v line.

Now if the T-bred was a 5v CPU drawing 45amps we'd have some big problems

Still, have to wait and see what this does end up meaning for PS requirements and heat output.

BTW, about 300w PS's not being enough anymore. Eh? Jeesh, we sure blew through them fast then. 200w PS's were the standard forever, with OEMs doing 145's. Heck, think I even saw a HP with a 145 powering a PIII last year. I've had that many drives attached for years, only thing with more draw in my system than in the 200w AT days is the video card and CPU. Everything else (HD's, CDs, etc) are drawing the same or less than their older brothers.

--Mc

 

[salman327]

I think 300 psu's should be alright, as long as they provide a current stream. Not ups and downs.

 

[MadRat]

<<According to that (here ) site are the voltage and power consumption figures:

Athlon XP(Palomino)
1700+(1467MHz) - 1.75V - 64.0W
1800+(1533MHz) - 1.75V - 66.0W
1900+(1600MHz) - 1.75V - 68.0W
2000+(1667MHz) - 1.75V - 70.0W
2100+(1733MHz) - 1.75V - 72.0W

Thoroughbred
1700+(1467MHz) - 1.60V - 55.2W
1800+(1533MHz) - 1.60V - 56.9W
1900+(1600MHz) - 1.60V - 58.6W
2000+(1667MHz) - 1.60V - 60.3W
2100+(1733MHz) - 1.60V - 62.1W

-Goi>>

Athlon XP(Palomino)
2100+(1733MHz) - 1.75V - 72.0W - 41.1A

Thoroughbred
1700+(1467MHz) - 1.60V - 55.2W - 34.5A
1800+(1533MHz) - 1.60V - 56.9W - 35.6A
1900+(1600MHz) - 1.60V - 58.6W - 36.6A
2000+(1667MHz) - 1.60V - 60.3W - 37.7A
2100+(1733MHz) - 1.60V - 62.1W - 38.8A

We'd roughly need to scale T-bred to 2600+ @1.65v before the 45A is hit according to those numbers.

 

[NFS4]

FALSE

http://www.hardocp.com/articles/amdtektour02follup.html

 

[NOX]

<< FALSE

http://www.hardocp.com/articles/amdtektour02follup.html >>

I was very much hoping so, knowing that many of the high ends PSU were not capable of producing 45A.

 

[MadRat]

Sheesh, NFS4, you had me wondering if you were calling my claims FALSE!

I didn't bother to double check the math, but I'm pretty sure on those numbers. hehe