AMD's new AM2 platform launched

Want to talk about one of the articles in SPCR? Here's the forum for you.
MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

AMD's new AM2 platform launched

Post by MikeC » Mon May 22, 2006 11:05 pm


winguy
Posts: 447
Joined: Tue Nov 02, 2004 9:31 am

Post by winguy » Tue May 23, 2006 6:34 am

To kick start, do all AM2 Semprons have CnQ?

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Tue May 23, 2006 8:03 am

winguy wrote:To kick start, do all AM2 Semprons have CnQ?
I find it difficult to think of a reason why they would drop it. Maybe if the AM2 Sempron was so cool it wasn't necessary.

tay
Friend of SPCR
Posts: 793
Joined: Sat Dec 06, 2003 5:56 pm
Location: Boston, MA
Contact:

Post by tay » Tue May 23, 2006 8:51 am

*sigh*
Nforce 5 series is interesting, but it runs hotter than the ATI chipset. Wonder if the new to nforce5 features work stably.

The chips are boring other than the low power ones. Intel's choice to leave the MC off chip now seems well advised. All this hoopla about merely changing the memory controller *meh*.

So when is conroe out anyway? 2 weeks?

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Tue May 23, 2006 11:49 am

jaganath wrote:
winguy wrote:To kick start, do all AM2 Semprons have CnQ?
I find it difficult to think of a reason why they would drop it. Maybe if the AM2 Sempron was so cool it wasn't necessary.
Well Semprons S754 below 3000+ had CnQ disabled; so it’s an obvious question to ask.

I almost have to pinch myself when looking at the TDP of the FX62; 125W. Assuming that the fastest non EE Conroe can match FX62, it will do so with roughly half the TDP. That’s a staggering turnaround for Intel.

Am I correct in saying that K8L is not the 65nm shrink of K8 that’s due in roughly January 07? K8L sounds very promising, I’m just wondering what they have for January.

Prozzaks
Posts: 32
Joined: Thu Mar 25, 2004 12:28 pm
Location: Sherbrooke, QC, Canada

AMD Athlon 64 X2 3800+ Energy Efficient Small Form Factor

Post by Prozzaks » Tue May 23, 2006 12:41 pm

If what they say is true (That the AMD Athlon 64 X2 3800+ Energy Efficient Small Form Factor has a TDP of 35W!) in the Anandtech article, I just can't wait to see those new CPU in the SRPC labs!

Paying a 61$USD premium to get a CPU with equal performance but a TDP that is more than half of the regular TPD is totally worth it! However, I wonder if AMD just added some extra tests in their production line to sort out which CPU could be undervolted while remaining stable and now sell them with a price premium. In other words, before you took your chances by buying a processor and if you were lucky you were able to do some interresting undervolting; now, you pay a premium for the CPU that you were lucky to get.

But still, a 35W Athlon 64 X2 3800+ is just increadible! My good old 1000MHz Pentium3 had a TDP of 35W! It's amazing to get so much more processing power while using the same amount of electricity. I hope this will set a trend in the computing industry toward less power hungry components. (We all know what that translates into ;) ) If only graphic card manufacturers could catch that drift...

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Tue May 23, 2006 12:46 pm

Well Semprons S754 below 3000+ had CnQ disabled; so it’s an obvious question to ask.
The new AM2 Sempron range starts at 3000+, so they all have C'n'Q:

wikipedia list of sempron procs
Sempron "Manila" (Socket AM2, 90 nm)
All models support: MMX, SSE, SSE2, SSE3, Enhanced 3DNow!, NX bit, AMD64, Cool'n'Quiet

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Tue May 23, 2006 1:13 pm

jaganath wrote:
Well Semprons S754 below 3000+ had CnQ disabled; so it’s an obvious question to ask.
The new AM2 Sempron range starts at 3000+, so they all have C'n'Q:

wikipedia list of sempron procs
Sempron "Manila" (Socket AM2, 90 nm)
All models support: MMX, SSE, SSE2, SSE3, Enhanced 3DNow!, NX bit, AMD64, Cool'n'Quiet
Well that’s very good news, although your logic is flawed as AMD could simply have disabled CnQ in the slowest AM2 Semprons as they did with S754. The fact that they are all labelled as 3000+ or higher is not really relevant.

TheQuietType
Posts: 7
Joined: Sun Jan 08, 2006 1:31 am

Re: AMD Athlon 64 X2 3800+ Energy Efficient Small Form Facto

Post by TheQuietType » Tue May 23, 2006 3:46 pm

Prozzaks wrote:However, I wonder if AMD just added some extra tests in their production line to sort out which CPU could be undervolted while remaining stable and now sell them with a price premium.
DailyTech wrote "By utilizing aggressive power state timings settings and processors binned for lower voltages, AMD has managed to take a select few processor cores and modify them to use less energy during typical operation." (http://www.dailytech.com/article.aspx?newsid=2368) So I'd say you are right.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Re: AMD Athlon 64 X2 3800+ Energy Efficient Small Form Facto

Post by MikeC » Tue May 23, 2006 4:05 pm

TheQuietType wrote:
Prozzaks wrote:However, I wonder if AMD just added some extra tests in their production line to sort out which CPU could be undervolted while remaining stable and now sell them with a price premium.
DailyTech wrote "By utilizing aggressive power state timings settings and processors binned for lower voltages, AMD has managed to take a select few processor cores and modify them to use less energy during typical operation." (http://www.dailytech.com/article.aspx?newsid=2368) So I'd say you are right.
Most "cream of the crop" products off any assembly line are the same -- a certain portion always turn out to be better in one way or another. The work is in sorting them, and once sorted, some people are willing to pay more for them. (Diamonds are one example. Ditto high end speaker drivers.) For the longest time, the only difference between higher and lower priced CPUs from Intel and AMD was clock speed capability (and post-binning multiplier lock).

dragmor
Posts: 301
Joined: Sun Jul 10, 2005 7:54 pm
Location: Oz

Post by dragmor » Tue May 23, 2006 7:57 pm

When you review the AM2's can you make sure to

a) Also compare the Nvidia and ATI chipsets (the power difference is supposed to be huge).
b) Give us some numbers for generic DDR2 (1.6v) and performance DDR2 (2.2v) as well as the effect on performance.

merlin
Friend of SPCR
Posts: 717
Joined: Mon Jun 13, 2005 6:48 am
Location: San Francisco, CA

Post by merlin » Tue May 23, 2006 8:07 pm

tay wrote:*sigh*
Nforce 5 series is interesting, but it runs hotter than the ATI chipset. Wonder if the new to nforce5 features work stably.

The chips are boring other than the low power ones. Intel's choice to leave the MC off chip now seems well advised. All this hoopla about merely changing the memory controller *meh*.

So when is conroe out anyway? 2 weeks?
I'm agreed on the rather boring articles and whatnot for the AM2 launch, but underneath I do like the extra features on the motherboards, the lower power, and the one socket architecture for desktop processors that we're moving into. Not to mention DDR2 is lower power!

The main thing I want to note though, basically every power comparison so far has compared the Nforce 590 SLI against ATI's chipsets. It's seems pretty apparent that their 2 chip architecture is very power inefficient not to mention it seems likely the voltages are higher for the built-in overclocking features. The heatpipes say it all.

In very important comparison, the non-sli 570 and 550 seem to just use simple small heatsinks on the single chip. I suspect their power usage is comparable to ATI and possibly even less. Those are definitely the chipsets I'm drooling over...the 590 is completely useless for a low power silent pc lover :)

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Wed May 24, 2006 1:04 am

Well that’s very good news, although your logic is flawed as AMD could simply have disabled CnQ in the slowest AM2 Semprons as they did with S754. The fact that they are all labelled as 3000+ or higher is not really relevant.
I took the time and trouble to find the answer to your question, so it is disappointing that you choose to nitpick and criticise. I almost certainly won't bother next time.

Envy007
Posts: 13
Joined: Sun Mar 12, 2006 5:21 am
Location: NL

Re: AMD Athlon 64 X2 3800+ Energy Efficient Small Form Facto

Post by Envy007 » Wed May 24, 2006 2:20 am

DailyTech wrote "By utilizing aggressive power state timings settings and processors binned for lower voltages, AMD has managed to take a select few processor cores and modify them to use less energy during typical operation." (http://www.dailytech.com/article.aspx?newsid=2368) So I'd say you are right.[/quote]

Just my thought, isn´t it possible AMD just took a TL-60 and put it on a AM2 socket? Specifications are exactly the same.

2 x 2,0 GHz L1:2 x 128 kB L2:2 x 512 kB TDP:35 W

Or are there things in a Turion that cannot work on a desktop board?

Envy007
Posts: 13
Joined: Sun Mar 12, 2006 5:21 am
Location: NL

Re: AMD Athlon 64 X2 3800+ Energy Efficient Small Form Facto

Post by Envy007 » Wed May 24, 2006 2:22 am

DailyTech wrote "By utilizing aggressive power state timings settings and processors binned for lower voltages, AMD has managed to take a select few processor cores and modify them to use less energy during typical operation." (http://www.dailytech.com/article.aspx?newsid=2368) So I'd say you are right.
Just my thought, isn´t it possible AMD just took a TL-60 and put it on a AM2 socket? Specifications are exactly the same.

2 x 2,0 GHz L1:2 x 128 kB L2:2 x 512 kB TDP:35 W

Or are there things in a Turion that cannot work on a desktop board?

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Wed May 24, 2006 2:39 am

jaganath wrote:I took the time and trouble to find the answer to your question, so it is disappointing that you choose to nitpick and criticise. I almost certainly won't bother next time.
I’m sorry to hear you say that as it wasn’t my intention to nitpick or criticise. Rightly or wrongly, I deduced from your posts that you hadn’t been aware that AMD had disabled CnQ in lower clocked Sempron S754 and also not fully understood the nature of this process, so I was attempting to clarify the situation for you. To repeat, it seems highly likely that the disabling of CnQ was a marketing based decision and not a technical one, which is why I suggested that the cut-off point of 3000+ was arbitrary.

Please don’t take this as a criticism, but Wikipedia is not a definitive source of information and personally I don’t treat it as such. I find it useful, but it does contain errors and even misinformation, so I prefer to keep that it mind.

Jaganath, it’s in the spirit of forums in general that we question and challenge each other, so that we all may come closer to the truth. This is a separate thing from criticism! As for nitpicking, picking people up for grammatical or spelling errors seems like nitpicking to me.

tay
Friend of SPCR
Posts: 793
Joined: Sat Dec 06, 2003 5:56 pm
Location: Boston, MA
Contact:

Post by tay » Wed May 24, 2006 9:24 am

merlin wrote: The main thing I want to note though, basically every power comparison so far has compared the Nforce 590 SLI against ATI's chipsets. It's seems pretty apparent that their 2 chip architecture is very power inefficient not to mention it seems likely the voltages are higher for the built-in overclocking features. The heatpipes say it all.

In very important comparison, the non-sli 570 and 550 seem to just use simple small heatsinks on the single chip. I suspect their power usage is comparable to ATI and possibly even less. Those are definitely the chipsets I'm drooling over...the 590 is completely useless for a low power silent pc lover :)
Good points, I completely forgot about that. On another positive note, AT claims that the new nf5 features are working reliably. So the platform is already an improvement I think. The CPU, not so much.

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Wed May 24, 2006 9:48 am

Hothardware have a review which shows nForce 590 SLI consuming ~40W more than CrossFire Xpress 3200. It’s hard to know if any of the difference is down to VRM efficiency differences between the two motherboards, but that’s still a hell of a big difference.

merlin
Friend of SPCR
Posts: 717
Joined: Mon Jun 13, 2005 6:48 am
Location: San Francisco, CA

Post by merlin » Wed May 24, 2006 10:02 am

tay wrote: Good points, I completely forgot about that. On another positive note, AT claims that the new nf5 features are working reliably. So the platform is already an improvement I think. The CPU, not so much.
I'm looking forward to MikeC's review of the Nforce 5xx platform... hopefully we'll see some pretty good power numbers on the 550/570... my friend's very likely to build a 570 ultra Shuttle very soon too!

tay
Friend of SPCR
Posts: 793
Joined: Sat Dec 06, 2003 5:56 pm
Location: Boston, MA
Contact:

Post by tay » Wed May 24, 2006 10:08 am

smilingcrow wrote:Hothardware have a review which shows nForce 590 SLI consuming ~40W more than CrossFire Xpress 3200. It’s hard to know if any of the difference is down to VRM efficiency differences between the two motherboards, but that’s still a hell of a big difference.
You know I noticed that and thought that maybe it was because they used an 1900XT vs a 7900GTX in those comparisons because the difference was so huge. The test setup page isnt clear enough and as you said there are likely other factors such as VRMs at play.

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Wed May 24, 2006 10:20 am

tay wrote:You know I noticed that and thought that maybe it was because they used an 1900XT vs a 7900GTX in those comparisons because the difference was so huge. The test setup page isnt clear enough and as you said there are likely other factors such as VRMs at play.
The Power consumption test page does state that they all used the 7900GTX and 1GB of RAM, at least for that comparison; I didn’t read the rest of the review. The difference of 35W at idle is the thing that stands out for me, as that can’t really be explained by VRM efficiency differences!

Ackelind
Posts: 467
Joined: Sun Mar 20, 2005 1:18 pm
Location: Umea, Sweden.

Post by Ackelind » Wed May 24, 2006 12:54 pm

The nForce chips are becoming more and more advanced microprocessors, incorporationg lots of features, and creating large amounts of heat.

Strange that they keep using tiny active or not so well working passive solutions, when the chip probably puts out a lot more than a P-M for example, or even more than a lower powered A64!

tempeteduson
Posts: 189
Joined: Sat Jun 05, 2004 12:56 pm
Location: Houston, Texas
Contact:

Post by tempeteduson » Wed May 24, 2006 11:47 pm

According to this translated HKEPC page of the first testing of AM2 Semprons, all these new Manila-based Semprons have CnQ except for the 3000+ (both "regular" and Energy Efficient versions). It looks like Wikipedia was wrong.

That's a pity, as I really liked the 3000+ EE, what with its low power consumption, decent 256KB L2 cache, and even multiplier (enabling it to run with memory at true DDR2-800 speeds). Oh well, it looks like the 3400+ is the Sempron to get (odd multiplier, though).

okay
Posts: 23
Joined: Mon May 08, 2006 1:50 pm

Post by okay » Thu May 25, 2006 1:56 am

Fifteen processors were launched on the new AM2 platform by AMD today, ranging from the new flagship FX-62 at $1031 all the way down to a $77 Sempron 3000+
Sempron 2800+

http://prisguide.hardware.no/product.ph ... ctId=45350

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Thu May 25, 2006 2:30 am

According to that link even the 2800+ has Cool'n'Quiet. Anyone got any more conflicting information? :lol: :lol:

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Thu May 25, 2006 2:38 am

When I built systems for my family using Sempron S754, the situation with regard CnQ still wasn’t clear, so I took the radical step of picking up the phone and phoning AMD. They guy was very helpful and clarified the situation for me; it was also a lot quicker than trawling the web for conflicting advice. If only AMD would update their online spec sheets on more than seemingly an annual basis :(

The moral is: the horse’s mouth is better than a donkey’s arse. :lol:

winguy
Posts: 447
Joined: Tue Nov 02, 2004 9:31 am

Post by winguy » Thu May 25, 2006 4:40 am

tempeteduson wrote:According to this translated HKEPC page of the first testing of AM2 Semprons, all these new Manila-based Semprons have CnQ except for the 3000+ (both "regular" and Energy Efficient versions).
It seems AMD has continued to disable CnQ on processors <1.8GHz.

merlin
Friend of SPCR
Posts: 717
Joined: Mon Jun 13, 2005 6:48 am
Location: San Francisco, CA

Post by merlin » Thu May 25, 2006 6:39 am

Ackelind wrote:The nForce chips are becoming more and more advanced microprocessors, incorporationg lots of features, and creating large amounts of heat.

Strange that they keep using tiny active or not so well working passive solutions, when the chip probably puts out a lot more than a P-M for example, or even more than a lower powered A64!
For the ones with many PCI-E lanes, most certainly they are creating tons of heat. But the standard non-sli ones are not. I believe the chipset went through a minor die shrink from 150nm to 130nm compared to the Nforce4(Can anyone confirm these are 130nm?), and as it's basically the same chipset with a few bonus features, it probably gained the same power advantages as a die shrink of a cpu. This is the likely reason why we see small aluminum heatsinks as enough to cool the nforce 570/550.

QuietOC
Posts: 1407
Joined: Tue Dec 13, 2005 1:08 pm
Location: Michigan
Contact:

Post by QuietOC » Thu May 25, 2006 7:14 am

Image

Didn't anyone else notice that AM2 uses more power than S939? (And S939 requires more power than S754. S754 is still the way to go if you are interested in low heat.) And the *new* low power CPUs for AM2 are nothing that you can't duplicate with existing S939 CPU's.

So, AM2 is slower and hotter than S939. DDR2 is just not an improvement.

rpsgc
Friend of SPCR
Posts: 1630
Joined: Tue Oct 05, 2004 1:59 am
Location: Portugal

Post by rpsgc » Thu May 25, 2006 8:35 am

You do notice that that chart shows the new nVIDIA SLI X16 chipset consuming LESS power than the current one?

Post Reply