AMD 35W/65W dual-core CPU review @ X-Bit Labs

[jaganath]

you need a Conroe system to go into standby as it can’t compete with an X2 at idle due to the higher minimum clock speed of C2D. For a typical desktop system which is at idle ~90% of the time, AMD still offer the lowest power consumption if you exclude Intel’s mobile parts, which outshine everything else.

Do you think we will see MoTD (Mobile on the Desktop) parts for Merom like we saw with Pentium M and Core Duo? Or will it remain an OEM part?

[smilingcrow]

Do you think we will see MoTD (Mobile on the Desktop) parts for Merom like we saw with Pentium M and Core Duo? Or will it remain an OEM part?

Dothan and Yonah CPUs and compatible motherboards are available in retail, so I don’t know what you mean by OEM only in this context!

Merom will surely be the most efficient CPU to use in a high performance energy efficient desktop. The T5500 is due to retail at $209 and is rated at 1.66 GHz, which should make it on average on par with an X2 3800 performance wise. Under-volted, the 1.66 CPU is going to consume about 16-18W. My Yonah laptop which is under-volted, consumes 33W when running dual copies of CPU Burn-in and is tolerable in the noise stakes; I just wish it was silent at idle.

[ziphnor]

You’ve hit it on the head there; you need a Conroe system to go into standby as it can’t compete with an X2 at idle due to the higher minimum clock speed of C2D. For a typical desktop system which is at idle ~90% of the time, AMD still offer the lowest power consumption if you exclude Intel’s mobile parts, which outshine everything else.

I also need an AMD system to go into standby! Do you think im mad enough to leave even a 80W system(my current 1.1V A64 3000+ HTPC) on 24/7?? All systems need to enter standby, the system would have to reach the 10W point before i would leave it on without standby. My HTPC spends 70% time being in standby and 25% time recording/viewing TV or playing music etc, the remaining 5% is when its waiting to enter standby

My point was that for a fairly small premium in wattage you gain a pretty huge advantage in performance(especially concerning media de/encoding). Take a look at the difference in power usage again and compare it to the difference made by the choice of graphics card or just the motherboard.

And as for not going as low when throttling the CPU, it might easily be possible to undervolt, which has far more impact(i actually turned CnQ off on my computer, and instead undervolted it to run at a constant [email protected], as its actually only +1W idle power usage).

[Poodle]

Thanks for the info/link. I guess the 7900GT isnt that bad, but its still a jump up from 7600GT. I just dont think the 7600GT is just the thing for gaming at 1920x1080, which i was hoping to try out The passively cooled MSI 7900 sounds more like it.

Anyway, dont let me pollute this thread anymore, and thanks again for the info.

To get back on topic, i must say the Conroe has altered my HTPC upgrade plans. I had originally planned for the 3800+ 35W, but the price cuts i have seen locally have only applied to the 65+W versions. Furthermore, the pretty huge perfomance/watt increase supplied by the Conroe, combined with the ease of automatic standby in HTPC software has convinced me. But then im planning for a high-end HTPC that can, say, record 2 HD streams, and play a third one using all kinds of fancy filtering etc. My current A64 3000+ cant even handle 1080i MPEG-2 playback in MediaPortal.

This is going to be my first Intel CPU since my Pentium-II 450Mhz

My 7600gt gets me a nice 60fps in BF2 @ 1920*1200 with some aa and stuff. Can't use 16xAA as my "old " X2, SLi setup. But the Psu is MUCH cooler now
I shoucl mention that Cpu is overclocked to 2,6ghz...


Oblivion is not worth mentioning though. But it "sucks" so I don't care.

[ziphnor]

My 7600gt gets me a nice 60fps in BF2 @ 1920*1200 with some aa and stuff. Can't use 16xAA as my "old " X2, SLi setup. But the Psu is MUCH cooler now
I shoucl mention that Cpu is overclocked to 2,6ghz...


Oblivion is not worth mentioning though. But it "sucks" so I don't care.

Thanks for the tip, im still caught somewhere between 7600GT and 7900GT as im unsure as to how much i would use my HTPC for gaming(never tried it).

Btw, Oblivion might have a sucky rule-system but it still looks great But dont worry, i stopped playing it after a few days being too frustrated with the game mechanics.

[frostedflakes]

Would anybody be interested in purchasing an X2 3800+ 35w with me? I know a guy with distributor connections (he's the one who got me Turion64 and Newark long before any online stores were carrying them) who says he could get me this processor. I haven't confirmed it yet, but I don't see why he wouldn't be willing to order a few more for others who are interested. He quoted me $190-200 + PayPal fees and shipping for a single processor, if a few more are interested he could probably get a discount (p.s. he sells these at cost).

I'll probably have him put in the order at the end of this week, so if anybody is interested PM me ASAP. This is just for quiet PC nuts who are interested in getting some cool processors, don't be greedy and buy a bunch just to sell on eBay or whatever for $400. To be honest I don't even know how many he can get, I'd assume 5-10 shouldn't be a problem, though.

[vitaminc]

http://techreport.com/reviews/2006q3/e6 ... ex.x?pg=15

Pointless to go EE SFF 3800+ X2 when you can get E6300 at a cheaper price. Similar level of system power draw at the wall.

[Mats]

Pointless to go EE SFF 3800+ X2 when you can get E6300 at a cheaper price. Similar level of system power draw at the wall.

I agree. But there's another thing that makes C2 even more attractive. It's not factory undervolted like the SFF EE is. The E6300 can be overclocked well beyond 3 GHz on stock Vcore, which makes me believe that there is plenty of undervolting headroom for it. I don't think the situation is the same for AMD's SFF EE.

[jaganath]

The E6300 can be overclocked well beyond 3 GHz on stock Vcore, which makes me believe that there is plenty of undervolting headroom for it.

I am not sure great OC ability necessarily translates to great undervoltability; for example I remember with some P4 Northwoods, they could also hit 3GHz at stock volts, but really were not stable at much below standard Vcore. There is a lot of hype surrounding the undervoltability of Conroe, but not much substance yet, AFAIAA. All I'm saying is it's best to reserve judgement until there are uncontestable facts regarding Conroe's undervoltability.

[Mats]

I am not sure great OC ability necessarily translates to great undervoltability; for example I remember with some P4 Northwoods, they could also hit 3GHz at stock volts, but really were not stable at much below standard Vcore. There is a lot of hype surrounding the undervoltability of Conroe, but not much substance yet, AFAIAA. All I'm saying is it's best to reserve judgement until there are uncontestable facts regarding Conroe's undervoltability.

But still, you get my point I guess. An unfair comparisation has become even more unfair. Well in worst case you can undervolt it just as little as you can do with a SFF EE, which still cost a lot more and performs worse. Intel didn't care too much about low power four years ago. Even the PD is a good undervolter! Allendale is pretty much the same as the mobile counterpart Merom, the latter is known for drawing the exact amount of power as a Yonah CPU at same speed and Vcore, even though the TDP says something else.

But, there's still a possibility that Intel have somehow prevented the Vcore from going too low in C2.

Yeah, I'm speculating a bit, but still I can't see any reason to get the SFF EE even if C2 is totally Vcore locked.

[vitaminc]

http://www.gamepc.com/labs/view_content ... 2le&page=4

More review showing how C2D platforms are consuming less power than comparable AM2 systems.

No surprise here. Why would anyone get an "low power" solution on a 90nm process when they have access to 65nm CPUs is beyond me. Fanboism would be the only explanation.

[jaganath]

More review showing how C2D platforms are consuming less power than comparable AM2 systems.

From the review that you linked to:

Keep in mind, though, that our Socket-AM2 processors were tested on an nForce 590 SLI motherboard, which consumes a bit more power compared to the Intel P965 motherboard we used for the Core 2 Duo testing. nVidia’s chipsets typically consume about 10-15W more than competing chipsets, so when that’s factored in, there isn’t a huge difference between the two processor families.

[vfrex]

No surprise here. Why would anyone get an "low power" solution on a 90nm process when they have access to 65nm CPUs is beyond me. Fanboism would be the only explanation.

Wasn't the pentium D 65nm?

[frostedflakes]

I'm getting Athlon64 because I prefer the platform. Geforce 6100 graphics are efficient (built on 90nm) and more powerful than Intel GMA. And although MSRP for the E6300 is lower, you can't find it for less than $230, whereas I can get the 35w 3800+ for ~$200. Not to mention a decent Conroe motherboard would cost me over $100, whereas a solid 6100 board is only around $70.

So does this make me a fanboy?

[vitaminc]

No surprise here. Why would anyone get an "low power" solution on a 90nm process when they have access to 65nm CPUs is beyond me. Fanboism would be the only explanation.

Wasn't the pentium D 65nm?

Think how much power you would save from changing a D 805 to a D 905.

65nm process saves a lot more power than 95nm process.

[vitaminc]

I'm getting Athlon64 because I prefer the platform. Geforce 6100 graphics are efficient (built on 90nm) and more powerful than Intel GMA. And although MSRP for the E6300 is lower, you can't find it for less than $230, whereas I can get the 35w 3800+ for ~$200. Not to mention a decent Conroe motherboard would cost me over $100, whereas a solid 6100 board is only around $70.

So does this make me a fanboy?

NVIDIA's integrated graphics chipsets consumes more power than ATI's solution. ATI's integrated graphics chipsets are low power by design with its expertise and leadership in notebook chipset designs.

And btw, find me a site for 35W 3800+ for $200. The closest one I can find is $198 for 65W 3800+. I am sure you already secured some real source for your EE SFF 3800 X2. Or you are just wet dreaming?

[frostedflakes]

Why are you arguing with me? It's not like my processor buying decisions will have any effect on your life.

Power consumption of X1900XTX has turned me off from ATi. Their IGP may very well consume less power, but I'm more comfortable buying 6100. Fanboyism? Guess it depends on how you look at it. I've been very pleased with the efficiency of previous nVidia cards, so I'm sticking w/them I guess.

There is no site for 35w 3800+. I can get it for $200, and therefore any member interested can get it for $200 through the guy I know.

[jaganath]

And btw, find me a site for 35W 3800+ for $200. The closest one I can find is $198 for 65W 3800+. I am sure you already secured some real source for your EE SFF 3800 X2. Or you are just wet dreaming?

On a totally unrelated point, your conversational style and aggressive tone is not only unwelcome on SPCR, it is also very unproductive. I would suggest that a less confrontational style will yield better results both in the way people react to you and your own self-confidence.

[vitaminc]

Why are you arguing with me? It's not like my processor buying decisions will have any effect on your life.

Power consumption of X1900XTX has turned me off from ATi. Their IGP may very well consume less power, but I'm more comfortable buying 6100. Fanboyism? Guess it depends on how you look at it. I've been very pleased with the efficiency of previous nVidia cards, so I'm sticking w/them I guess.

There is no site for 35w 3800+. I can get it for $200, and therefore any member interested can get it for $200 through the guy I know.

ZZF has E6300 for $219 and its not exactly a small no-name e-tailer. The $230 number is probably from price gouging e-tailer like Screwegg.

I wouldn't go NVIDIA's IGP if they don't make it less power hungry but thats just me. Graphics cards are a different story though.

And that's OEM version of EE SFF 3800 X2 I assume? $200 + S&H? Does he have any of the Sempron 3000+ EE SFF in stock? And how would I get to buy one of the EE SFF 3800 X2?

[vitaminc]

And btw, find me a site for 35W 3800+ for $200. The closest one I can find is $198 for 65W 3800+. I am sure you already secured some real source for your EE SFF 3800 X2. Or you are just wet dreaming?

On a totally unrelated point, your conversational style and aggressive tone is not only unwelcome on SPCR, it is also very unproductive. I would suggest that a less confrontational style will yield better results both in the way people react to you and your own self-confidence.

You do know using this kind of tone to confront someone that's aggressive and confrontational in nature will not be very productive, aight?

[Mats]

Frostedflakes: Yes, you are a fanboy. Nothing wrong with that. I think I'm pretty much a fanboy too. But I have to say that C2 looks great.

The nVidia 6100 chipsets do run hot though, even the southbridge. The SB are sometimes placed in a way that makes it hard to put a heatsink on it. ATI have better chipsets now compared to a year ago, and soon they will introduce a new IGP chipset. It's a inquirer post, I know, but they're not the only one writing about it.

I said earlier that the reviews are unfair and yeah, it's no surprise. It's not really an apples to apples (no pun intended, I mean the fruit) comparisation. First it was old versus new, K8 showed up in April 2003. 90 nm versus 65 nm. Then suddently AMD gives us the SFF CPU and now we have one CPU with stock Vcore and one factory undervolted CPU. As jaganath said, we haven't seen how much you can undervolt a C2, and that's the most surprising. Have anyone tried? Any links? I want to see both undervolted as much as possible and then having the power consumption measured. It's still possible that AMD will still have the lowest idle power consumption though. Maybe SPCR will give the answer soon.

[Mats]

More review showing how C2D platforms are consuming less power than comparable AM2 systems.

From the review that you linked to:

Keep in mind, though, that our Socket-AM2 processors were tested on an nForce 590 SLI motherboard, which consumes a bit more power compared to the Intel P965 motherboard we used for the Core 2 Duo testing. nVidia’s chipsets typically consume about 10-15W more than competing chipsets, so when that’s factored in, there isn’t a huge difference between the two processor families.

Maybe you should read the numbers and not only the conclusions. Even the overclocked 2.7 GHz C2 uses 21 W less than the stock speed X2 3800+, and they are not even comparable when it comes to performance. There's a 43 W difference between the E6300 and the X2 4200+, the E6300 runs cooler and performs better.

nVidia’s chipsets typically consume about 10-15W more than competing chipsets,

Typically? Not really scientifical to me. They don't even specify which chipsets, only talks about "nVidia chipsets" in general.
The P965 draws more power than 975X for instance.

[jaganath]

I don't want to get into the 800 millionth "Conroe is better than AMD" argument. I know Conroe has better performance than AM2 X2, but when it comes to power efficiency the difference is much less clear-cut; for example the X2 4600 (65W version) produces over 10W less than the C2D E6300 in idle (power-saving tech on) and for the 35W X2 3800 the difference is even more pronounced.

Typically? Not really scientifical to me.

You mean scientific.

[smilingcrow]

If you make a comparison of the X2 3800 power consumption from the SPCR Desktop review against my C2D review the following stands out:

X2 3800+ (undervolted)
100GB 2.5â€

[vitaminc]

[quote="smilingcrow"]If you make a comparison of the X2 3800 power consumption from the SPCR Desktop review against my C2D review the following stands out:

X2 3800+ (undervolted)
100GB 2.5â€

[smilingcrow]

Don't you think your comparison is pretty flawed?

100GB 2.5" Notebook HDD (AMD) vs. 250GB 3.5" SATA2 (Intel)
1GB PC3200 (AMD) vs. 2GB PC2-5300 (Intel)
6150 IGP chipset (AMD) vs. 6200 discrete GPU (Intel)

It doesn't take a genious to figure out that notebook drive, 1GB DDR stick(s), and integrated chipset will always consume less power than a regular HDD, 2GB DDR2 stick(s), and discrete graphics.

The comparison shows that even when hindered with much more power hungry components, that a C2D system still beats an X2 system in the performance per watt stakes under load, when the CPU clock speed starts to rise. I’m not sure where the flaw is in that!

I see the comparison itself as being neutral; the only flaws would be the ones that could be made from analyzing the comparison erroneously.

[zenboy]

The Tom's Hardware "Tuned" C2Extreme vs FX-62 article seems to back up some of these findings, specifically that at idle, an AMD system has a significant advantage, power wise, but that things tend to even up (and eventually reverse) as load rises. Apparently either current chipset implementations or the physical hardware isn't capable of stepping down quite as heftily on the Intel side of things.

[smilingcrow]

I think there’s more to come with C2D at idle with a different motherboard. 975X based boards tend to be fully loaded with Crossfire support, an extra IDE chip for multiple PATA ports & extra SATA ports, Dual Ethernet, Firewire etc.

For a modern chipset that supports C2D with low power consumption, we might have to wait for ATI & possibly even SIS to bring out a product. Not that I know whether SIS even intend bringing out a C2D chipset!
I’m not sure Nvidia are the answer either as they seem quite power hungry. I seem to recall that VIA no longer has a license for the current Intel buss. Is that correct?

[grosskur]

I think there’s more to come with C2D at idle with a different motherboard. 975X based boards tend to be fully loaded with Crossfire support, an extra IDE chip for multiple PATA ports & extra SATA ports, Dual Ethernet, Firewire etc.

Has anyone seen power consumption figures for Conroe on a micro-ATX 945G Express board (like the recently announced Gigabyte GA-945GM-S2)? I think it's safe to assume power consumption would be lower than for Conroe on a 975X or P965 board; the question is how much lower?

[Devonavar]

We've done some preliminary testing with Core 2 Duo and AMD's EE processors, and I can confirm that AMD's chips (even regular ones) beat the Core 2 Duo in idle. The advantage is as much as triple for the coolest EE chips. This is power measured at the +12V AUX socket, so it isolates the CPU from the rest of the system. The only variable is the power circuitry on the MB, and we've tested several systems to be sure...

I would second Xbitlabs' comment on using CPUBurn for load measurements. One of the delays with the Core 2 Duo results has been just that; we haven't been able to drive power consumption for our sample above 35W, which is just unbelievable for a chip specified at 65W. Could you point me towards the article you mentioned? I would be helpful for us to read...