Desktop Turion compared with Desktop Pentium M...

http://www.gamepc.com/labs/print_conten ... d=turion64

Note they compared a 25 TDP Turion on a DFI 250-GB desktop board
versus
27 TDP Pentium M on DFI's 855GM board.

Good info in there, I only wish they had chosen to use an ASUS board with the CT-479 adapter for the Pentium M, but you can easily extrapolate those numbers by looking at their other article here:

http://www.gamepc.com/labs/view_content ... 479&page=1

I still don't know what to think about the stuff the computer store GamePC writes. Are they reliable? I don't know. Are they less reliable than most regular reviewers who also have their economical reasons for presenting a certain result? I don't know that either.

You hit the nail on the head. There are always pressures/desires on people for one thing or another. But sometimes we forget that there are also pressures/desires to be objective.

I usually look at numbers in reviews since they can always be corroborated by additional data: there is less "wiggle room" for someone to play with.

Having said that, these particular articles seem fine. I also have a hard time seeing how gamepc would benefit by favoring one over the other. In fact they'd be harming themselves if they said something demonstrably wrong.

------------

Most everything we do in this world is based on trust. We take it for granted so often that we forget.

I'm slightly suspicious of GamePC here.

Some of the Pentium M numbers seem higher than in other reviews and likewise the Turion numbers seem a little lower. I also remember that their power consumption figures seemed higher in previous reviews than for some other sites.



The main thing to note is that, although GamePC sell Pentium M machines, motherboards and CPUs, they don't sell anything to do with Socket 754. I wonder, cause for concern about the figures shown?

Except the overall impression of the article was positive for the Turion.

PS. If I were gamepc, I'd start offering Turion/Socket 754 combos... and I'd be surprised if they didn't do that soon [and yes I'd start to wonder about their objectivity]

For those who want to see more critical responses to the GamePC article,

There's a whole thread on anandtech bashing the review:

http://forums.anandtech.com/messageview ... TMP=Linear

EDIT: Although they tested the Turion at 1,35V it nevertheless shows a great potential and lags "only" ~10W AC (~7W DC) behind.

all tests use the same hardware:
2 x Samsung PC-3200 512MB DDR Memory Modules (1GB Total)
nVidia GeForce 6800 GT 256MB (AGP 8x for Pentium-M, PCI Express x16 for Athlon64)
Enermax Noisetaker 2.0 600W Power Supply
Western Digital Raptor 74GB Serial ATA/150 Hard Disk Drive
Plextor PX-712A Ultra ATA DVD ReWritable Drive
Standard 1.44MB Floppy Disk Drive

All wattage figures are AC.

http://www.gamepc.com/labs/view_content ... uso&page=3
P-M 2Ghz AOpen i855GMEm-LFS
77W idle w Speestep
81W idle w/o Speestep
96W Prime95

http://www.gamepc.com/labs/view_content ... 479&page=4
P-M 2Ghz Asus + CT-479
100-110W idle w/o Speedstep
115-125W Prime95

conclusion:
Use of adapter and Asus desktop mobo causes 20-30W more power at idle and full load.

http://www.gamepc.com/labs/view_content ... n64&page=4
Turion on nforce 3
97W idle w Powernow
130W Prime95

So there are two ways of trying to show that a Turion needs only a bit more power than a P-M:
1. A Turion 800Mhz @0.95V at idle should need next to nothing - except IO power. But this one is also present on the P-M platform. So just take the difference at idle between the Turion and the P-M as extra for the desktop mobo.
I suspect the desktop motherboard to need around the same amount of power at idle and at Prime95. So just add the 20W difference to the ~100W of the P-M at Prime95 and you get 120W. That's only 10W less than the Turion.

2. Calculating with the CT-479 figures above.
Full power: P-M with adapter + desktop mobo needs again around 10W less at full power compared to the Turion.
Idle: About the same (97W vs 110/100W minus 4W Speedstep gain)
note: It's impossible for the CT-479 to suck a large amount of power because it has no heatsink.

Please note that the clowns at GamePC tested the Turion on much higher core voltage than what it's rated for. The MT-34 should be fed 1.2V, but GamePC used 1.35V instead, effectively turning the 25W Turion MT-34 into its 35W cousin, the ML-34. Way to go, GamePC! I have notified them of this, but as I expected, they didn't respond.

I'd also say that comparing the Turion on a desktop mobo to the P-M on a mobo with laptop chipset doesn't really make for a fair comparison. Judging by the recent Venice test done by Lost Circuits, the Turion should be very competitive with the P-M in the thermals department. The only good way of comparing CPU power consumption is to take the rest of the system variables out of the mix. It's harder, but the results are much more useful.

The P-M spanks the Turion. Im not sure what the problem is. Good for intel, now we just need non ridiculous prices. Personally I cant wait to ditch AMD, but it wont happen till intel is more price/performance/thermal competitive.

yah the best processor out there, I feel is a 2.x ghz centrino. LIke in general. I roasts games more than anyone figured, its a cold chip on idle, can be undervolted from that even. It just costs like thte price of an opteron setup!

so thats not even in the picture. Also note the wattages tested.... anyone besides me notice how HUGE the wattage draws are on either chip?? I dont think this is possible. It looks like a Prescott review, like a lower ghz prescott. Yeah, the voltages for the chips werent even brought down to a realistic level, thats the whole point of turion, to find the lowest usable voltage setting.

When's mike C going to do it up

Are there any information about the vcore of the Turions? Google was not very revealing and even good old Chris Hare does not have it.

regarding voltage of the MT Turion:

frostedflakes on this very forum had this to say on another thread about his very own MT Turion:


So from the Part Numbering scheme it may imply that it is 1.2v.

Yup, they tested the 25w processor at 35w voltage. Don't know if this was intentional or not, but it definitely throws off the power consumption numbers.

I personally think the review is suspicious.

The Turion 35w chip runs at 1.35V and 1.6 Ghz. I just got a Sempron 2600+ at 1.6 Ghz and 1.4V. It has no SSE3, CoolnQuiet, and AMD64 but this thing runs really cool at stock speed and voltage for only 80 bucks. I was going to undervolt it using CrystalCPUID (my MSI board does not support underclocking in the bios) but this thing runs at 40C with a Panaflo L1A at 5V and Aerocool heatpipe heatsink that never really worked that well. For comparison, the heatsink overheated the chip when i ran a 2000+ with the fan at 5V. I hear rumors that this chip can get to 1.1 to 1 volt at stock speed putting that under the 1.2V 25W Turion for less money. It would be interesting to see a comparison here at SPCR of all the low power chips at once to see what is the best chip out there for passive cooling or very low power high speed chips. Anyways, i'm rambling off topic but if you want to check out the new 90nm sempron and can live without the SSE3, and AMD64 it might be worth a try.

Just got a reply from GamePC about the voltage issue. They're aware of it and are conducting additional tests. Don't know if they'll edit the original article though.

EDIT: I'll take the liberty to post Chris' mail here:

Hi Mickael :

We're looking into a follow-up article with another platform which can
run the Turion at the correct voltage levels.

We've now been able to get the Turion to run on the Asus K8N4-E Deluxe
platform, which will likely be the basis of the next article.

Thanks for the "clowns" comment on SPCR! Appreciated.

- Chris / GamePC

So, they read the SPCR forums! Why not answer directly in this thread?

LOL