PDA

View Full Version : AM2 numbers show chipset power is out of line



sladesurfer
05-23-2006, 05:19 PM
http://www.theinquirer.net/?article=31898


Here is the problem, with these numbers, massively overstated for what they are, the NVidia chipset draws 50% more power than the ATI solution. If you scale the numbers down to realistic levels, it is not a stretch to imagine the ATI solution taking half the power of the NV one. This is a HUGE difference, even with the power measured directly to avoid PSU conversion losses, you are talking big time heat/wattage differences. To me, this puts the NF5 well into the unacceptable range.

OmegaMerc
05-23-2006, 05:44 PM
Theres alot of rounding off, and assuming, and guessing, and guestimating, and "~"

[XC] leviathan18
05-23-2006, 05:48 PM
Theres alot of rounding off, and assuming, and guessing, and guestimating, and "~"

welcome to the inquierer world :toast:

Thorry
05-23-2006, 05:56 PM
Uhm.... what does this have to do with AM2?????

AMD has stated (just like it's clearly stated in the specs) the chipset design does not have to change to be used with a AM2 CPU.

The only difference is the memory controller and the path between the CPU and the memory. The chipset design does not have to change at all.

Even if the nForce 5 chipsets are using this much power (which I highly doubt) it hasn't anything to do with AM2 and everything to do with nVidia's nForce 5...

Since the nForce 5 would be very simular to the nForce 4 I expect power usage to be near equal.

Too bad Inquirer knows :banana::banana::banana::banana:

ozzimark
05-23-2006, 06:20 PM
http://techreport.com/reviews/2006q2/am2-chipsets/index.x?pg=19
but that's not really a reliable test due to differences between the test rigs.

keiths
05-23-2006, 07:39 PM
Lowering power usage was supposed to be one of the main selling points of DDR2 and AM2 platform; instead it does the opposite.

nn_step
05-23-2006, 07:43 PM
Lowering power usage was supposed to be one of the main selling points of DDR2 and AM2 platform; instead it does the opposite.
It is a nVidia problem not an AMD :stick:

Thorry
05-23-2006, 08:12 PM
Lowering power usage was supposed to be one of the main selling points of DDR2 and AM2 platform; instead it does the opposite.

Please read my post, as nn stated. This is 100% nVidia, I don't know why Inq is bringing AM2 into this.

It's 100% nVidia... I do not see how the nForce5 would use more power as nForce4. Everybody knows nForce4 are hotheads...

The chipset itself is divided into several different parts.

The PCI-E controller (about 80-90% of total power usage depending on the chipset type)

The HyperTransport interface (about 5-10% of total power usage)

Other stuff (SMBus, IDE, 2x SATA etc) also about 5-10%

It's the PCI express that sucks up all the power, especially when using a 32x chipset (dual 16x SLI).

When using a simple chipset power usage would be very low, when using a high-end SLI chipset like the nForce5 series chipset power usage would actually be high. It's not really possible to compare different chipsets since features are rarely the same.

stealth17
05-23-2006, 08:22 PM
heh, Thorry is right.

You guys should know by now the internet is atleast 60% bull too! (That goes for the whole world, not just on the internet)

EDIT: Just noticed this too:


My advice, look to SiS and VIA unless you know you are going to need two GPUs. Both produce high quality and inexpensive parts, what more do you really need? ยต

Overclocking options :fact:

mursaat
05-24-2006, 12:28 AM
And AGP/PCI lock that actually works, not like the ABIT KV8-Pro with K8T800Pro. The lock was erratic, dunno if that was a VIA or ABIT fault.

Any insight (if it has any credibility) of why nf5 is that much power hog than nf4? PCI-E standard hasn't changed, nor HTT SATA IDE SMBus... and the new LinkBoost, Max Overclocking, FirstPacket and TCP/IP Acceleration can't be the problem.

MaxxxRacer
05-24-2006, 01:39 AM
If I had to venture a guess this problems come in with handing the SLI side of things. From what I have read nVidia designed the nf4 (and I'm assuming the nF5) so that it has an abnormally high speed/bandwidth connection between the PCIE slots to facilitate in good SLI performance.

ATI on the other hand has not gone to such lenghts and keep to a more standard PCIE crosstalk setup for CrossFire.

Its probable that nVidias more agressive approach is leading to the increase power consumption.

keiths
05-24-2006, 02:08 AM
It is a nVidia problem not an AMD :stick:

And ATI? The same chipset(Xpress 3200) used for both platforms; it draws more power with AM2 and DDR2: http://techreport.com/reviews/2006q2/am2-chipsets/index.x?pg=19

Bar81
05-24-2006, 02:45 AM
DAMN. Very interesting. I would have thought that chipset power would be down given the lower voltage memory. Would be nice if someone had an explanation.

The ULI chipset I have runs oh so cool :P

MaxxxRacer
05-24-2006, 04:16 AM
DAMN. Very interesting. I would have thought that chipset power would be down given the lower voltage memory. Would be nice if someone had an explanation.

The ULI chipset I have runs oh so cool :P

the chipset doesnt even see the memory.. has absolutley nothing to do with it. if the chipsets wants system memory it has to talk to the CPU first via the hyper transport bus.

Vapor
05-24-2006, 04:22 AM
It does look like loaded power figures are up for the entire system, however....system power levels from AT: http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2762&p=11

For those that don't like clicking:
Idle:
5000+: 109
FX-62: 113
FX-60: 114
965XE: 130

LOAD:
5000+: 176
FX-60: 188
965XE: 206
FX-62: 207

All figures in watts.


....just looking at system level draw says that SOMETHING is a power hog, and the DDR2 likely isn't, but at the same time the 5000+ does keep power draw down, so maybe it's just an FX-62 thing?

Bar81
05-24-2006, 04:42 AM
the chipset doesnt even see the memory.. has absolutley nothing to do with it. if the chipsets wants system memory it has to talk to the CPU first via the hyper transport bus.

My poor terminology. I was talking about the system power draw which is what he was testing.

perkam
05-24-2006, 05:54 AM
Threads on AM2 news either belong in the official AM2 thread here: http://www.xtremesystems.org/forums/showthread.php?t=100516

OR they belong in the AMD section.

--Thread Moved--

Perkam

keiths
05-25-2006, 01:43 AM
Looks like the thread move effectively killed the thread..

OPPAINTER
05-25-2006, 04:59 AM
Here's a thought,,, If the NF5 isn't using much power then why would a motherboard maker put on such a wild cooling unit? :D

OPP

IvanAndreevich
05-25-2006, 10:01 AM
Need to run the AM2 mobos (ATI vs NV) with the slowest HE CPU forced to run 800 MHz at lowest possible voltage and a PCI 2MB videocard.

Then since all other sources of consumption will be minimized, we'll see if it's the mobo eating all that juice.

keiths
05-25-2006, 11:22 AM
Something is up; check out idle power difference between 667 & 800 DDR2 on the fx-62: http://www.lostcircuits.com/cpu/amd_am2/7.shtml

3.7 Watts! That's the same system, just running differently clocked memory. The commentary says the power measurement is of the memory controller. Nvidia is wasting more than ATI, but from this it's also part AMD. Looks like they all need another revision before power savings over 939/754 become a reality.

Kjaks
05-25-2006, 11:50 AM
Here's a thought,,, If the NF5 isn't using much power then why would a motherboard maker put on such a wild cooling unit? :D

OPP

Hehe, good point..

Jaco
05-25-2006, 12:04 PM
this (possible) extra power consumption will be largely compensated by EE chips, I think. Those EE chips will sell like hot cakes if prices are not too high.

Zeus
05-25-2006, 12:58 PM
Back to watercooling the chipset it seems.

Brings back old NF2 memories. :)

Haltech
05-25-2006, 01:11 PM
Forget that, i wont be placing a restrictive chipset cooler in my loop. I think ATI will get the chipset win this round.

stealth17
05-25-2006, 05:32 PM
From AnandTech...


AMD does have one last trick up its sleeve before the end of the year, and you will hear about it in June. It's not K8L and it's not going to affect the majority of people, but it is an interesting stop gap solution for the high end in 2006...

Interesting. Does not affect the majority of people? Cold bug fix?!?!?!?! :woot:

Doubt it but that would be sick.


AMD does have one last trick up its sleeve before the end of the year, and you will hear about it in June. It's not K8L and it's not going to affect the majority of people, but it is an interesting stop gap solution for the high end in 2006...

wtf does that mean? No more high end?

mursaat
05-26-2006, 02:28 AM
There were rumours AMD would stop their FX line, and focus in "low" X2 and 1, 2,4 and 8 Opteron series. They simply doesn't sell too well, and the "low" chips do perform at least as good. The unlocked multi will be a thing of the past?