AMD's Smart Strike: Athlon X2 BE-2350

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
angelkiller
Posts: 871
Joined: Fri Jan 05, 2007 11:37 am
Location: North Carolina

AMD's Smart Strike: Athlon X2 BE-2350

Post by angelkiller » Mon Jun 18, 2007 7:40 pm

AMD's Smart Strike: Athlon X2 BE-2350

I don't think this has been posted yet.... if it has, I'm terribly sorry, and a mod can delete this thread. :oops: (And for the record, I did not write this article, I simply linked to it from Tom's Hardware)

But anyways, I think this has a good future for CPU's here at SPCR. Dual core 45W goodness!


What do you guys think about AMD and Intel's low-end dual core offerings?

Discuss.

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

Post by jaganath » Tue Jun 19, 2007 12:45 am

I wouldn't pay $85 for a BE-2300 when X2 3600 (Brisbane) is $59 @Newegg.undervolting will make up the power difference.

angelkiller
Posts: 871
Joined: Fri Jan 05, 2007 11:37 am
Location: North Carolina

Post by angelkiller » Tue Jun 19, 2007 7:47 am

jaganath wrote:I wouldn't pay $85 for a BE-2300 when X2 3600 (Brisbane) is $59 @Newegg.undervolting will make up the power difference.
Two responses.

1) Would undervolting a CPU at stock speeds really reduce the TDP by 20W? If so, why didn't AMD already do this? :?

2) But even though you pay more for for the BE-2300, you can also undervolt it, and using your logic, even further reduce its TDP. Isn't that worth it the extra money? A dual core, sub-45W CPU?

pipperoni
Posts: 218
Joined: Sun Oct 24, 2004 9:10 pm
Location: Toronto
Contact:

Post by pipperoni » Tue Jun 19, 2007 8:37 am

The BE-2300 and X2 3600 (if you get the 65nm part) are based off of the same core. I suspect the X2 3600 is probably pretty close to 45W already since it operates at very similar speeds and voltages to the BE-2300. The 65W class includes higher clocked (and thus more powerful) processors so the 65W designation is less meaningful for the bottom rung processors.

That said, if you want to run in spec or don't have a motherboard that supports undervolting, the BE-23x0 is probably a good choice.

I would guess however that with the BE-23x0 parts, you'll probably get better "quality" chips than with the X2 3600 due to binning.

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

Post by jaganath » Tue Jun 19, 2007 9:25 am

I would guess however that with the BE-23x0 parts, you'll probably get better "quality" chips than with the X2 3600 due to binning.
how exactly would this "quality" manifest itself? more overclockability/undervoltability?

frostedflakes
Posts: 1608
Joined: Tue Jan 04, 2005 4:02 pm
Location: United States

Post by frostedflakes » Tue Jun 19, 2007 11:10 am

I also think the 3600+ is a good bet if you can undervolt. I was able to lower the Vcore on mine to 0.975V (I think another SPCR member was able to achieve the same undervolt, and another could do 1.0V, so 1.0V on these processors seems to be a given). Assuming the max power is 65w (@ 1.3V) at 0.975V power should be about 37w. As others have pointed out 65w max power for the 3600+ is probably exaggerated, I wouldn't be surprised if power consumption was only in the 45w range. With that assumption max power at 0.975V should be roughly 25w.

So IMO the BE-2300 isn't anything special and isn't worth the price premium over a 3600+. I have no experience with undervolting faster clocked Brisbane, though, maybe the BE-2350 offers more impressive undervolting than is available with normal 65w processors.

I'd imagine though that these are really intended for OEM SFFs and applications like that where the system manufacturer can't really tweak and undervolt the processor like we are able to.

prodeous
Posts: 90
Joined: Sun Nov 30, 2003 8:01 pm
Location: Poland
Contact:

Post by prodeous » Wed Jun 20, 2007 1:32 am

I'm going to guess that undervolting sacrifices C&Q?

I know that DFI and most other boards had overvolting with C&Q by having voltage adjustmen like Vcc 104% etc.

It be nice if there is a board that permits undervolting with C&Q.


But from what you are saying, most people easly reach near 1.0V for 3600+? That brings more options to my project.

Steve_Y
Posts: 214
Joined: Tue Jun 21, 2005 2:17 pm

Post by Steve_Y » Wed Jun 20, 2007 3:31 am

prodeous wrote:I'm going to guess that undervolting sacrifices C&Q?
CrystalCPUID basically gives you configurable Cool and Quiet.

Compared with the stock C&Q settings, the undervolting offered by CrystalCPUID reduced my X2 3600+ power consumption by around 10W at idle and around 20W under load.

My older 90nm 3600+ is Prime 95 stable at 1v when running at stock speed (2Ghz), and runs fine undervolted to 0.8v when idle (1Ghz). Although I raised both by 0.025v over that just to make sure, which only added one or two watts to power consumption.

There's an article describing how CrystalCPUID is used here: http://www.silentpcreview.com/article231-page1.html

prodeous
Posts: 90
Joined: Sun Nov 30, 2003 8:01 pm
Location: Poland
Contact:

Post by prodeous » Wed Jun 20, 2007 9:40 pm

Thanks for the info. So that covers the Window machine.

But my project will involve 4 nodes running on Linux on a matx board (not sure which one yet) and 1 master node, running windows and linux (ubuntu to be precice). Anything equivalent for the linux environment?

And again thanks for the info.

So getting back to the general subject, it looks like these BE-s are just undervolted Brisbane. Generaly it solves the issue of using the extra software to configure C&Q. So on boards that will run non windows environment it still might be a good cpu. Additionally for business systems along where they dont' want' to play with additional software, they rather just get these cpus.


Still combined with DTX runnign AMD 690g/v chipset. its a killer power system (performance per watt).

Either way cudos for AMD's marketing team ;)

likwidsh0k
Posts: 33
Joined: Wed Jan 24, 2007 2:28 pm
Location: Brisbane, Australia

Post by likwidsh0k » Mon Dec 10, 2007 9:39 am

Hi,

I am building a new box that will become a htpc with a 690G chipset mobo (GIGABYTE GA-MA69GM-S2H @ newegg) and was all set to get the Athlon 4800 until I saw this chip.

Can someone please explain how a 45W chip will make the PC quieter? I have read a couple of reviews that talk about how it is great for this but none explain how. Apologies for the newb question ;)

Thanks

doodah
Posts: 48
Joined: Tue May 23, 2006 10:31 am
Location: CA, USA

Post by doodah » Mon Dec 10, 2007 11:58 am

likwidsh0k wrote:Hi,

I am building a new box that will become a htpc with a 690G chipset mobo (GIGABYTE GA-MA69GM-S2H @ newegg) and was all set to get the Athlon 4800 until I saw this chip.

Can someone please explain how a 45W chip will make the PC quieter? I have read a couple of reviews that talk about how it is great for this but none explain how. Apologies for the newb question ;)

Thanks
Hi likwidsh0k,

Here's the quick version: Lower wattage (45W) = less heat so the fans don't have to work as hard.

On the forums, several members talk about undervolting the less expensive X2 processors--like your 4800--to get the same results as the BE's.

Just so you know, a BE2400 was released a while back. It's mentioned in a few threads.

I hope that helps. I'm confident others will be able to give you much more info than I can at the moment.

Good luck,
Jan

fri2219
Posts: 222
Joined: Mon Feb 05, 2007 4:14 pm
Location: Forkbomb, New South Wales

Post by fri2219 » Mon Dec 10, 2007 12:09 pm

likwidsh0k wrote:Can someone please explain how a 45W chip will make the PC quieter? I have read a couple of reviews that talk about how it is great for this but none explain how.
The chain of logic is:
1) Reducing voltage at a constant current results in lower power. Power = (Voltage)(Current). Power can be expressed in a number of different units, one of them being watts.
2) A certain percentage of power goes to doing useful work, like lifting a horse out of a train moving eastbound at 72 meters/century or computing the amount of red needed to properly color an LCD pixel representing a head exploding in blood after being shot with a zortilizing gravnetic fart-thrower. The part that doesn't do fantastically useful work like that goes to waste. Waste in this circumstance is also known as heat.
3) Heat accumulates to the point where computer components no longer work well, or work at all. Frequently, that heat also goes into the air bathing the computer components. Unfortunately, the air will eventually refuse to absorb any more heat in the current set of conditions, and the components will take the brunt of it.
4) If you swap fresh cold air for hot air, one common way to remove heat is by moving air that has absorbed the heat out of the case so that the cooler air can absorb the heat. Any way you try to make it happen, moving around air (or water, liquid sodium, mercury vapor, or liquid oxygen) will make noise. You can reduce that noise by moving less of the air around.

Less voltage -> less power available to waste -> less power wasted -> less heat -> less air to be moved -> less noise from moving the air

A perfectly efficient process wouldn't make any heat, so all's good as long as you can make it happen. However, it will take an infinite number of steps taken in an infinite length of time, so eliminating all inefficiency (waste aka heat) isn't an option and significantly reducing it isn't economically viable until new technology steps in (like moving from a photolithographic to a molecular-beam epitaxy process that allows you to shrink circuit width from 130nm to 15mm on silicon chips). Reducing the amount of power available to waste is frequently the easiest option.

P.S. If you really want to impress chicks and frighten bullies at the beach, take up classical thermodynamics as a hobby. This Wikipedia article is a good start and explains some of the mathematics behind it. Not only does it make you fun at parties, just talking about it can get you your own seat on airplanes, busses, and trains.
Last edited by fri2219 on Mon Dec 10, 2007 12:27 pm, edited 3 times in total.

HueyCobra
Posts: 217
Joined: Tue Feb 14, 2006 12:16 am
Location: Australia

Post by HueyCobra » Mon Dec 10, 2007 12:23 pm

AMD Resurrects K8 Architecture for 2008 Roadmap
DailyTech.com wrote:AMD will also update its "Energy Efficient" series and will release three new chips, the AMD Athlon 4850e, Athlon 4450e, and Athlon 4050e in Q2 2008. All of the new offerings will be based on AMD's Brisbane core and will feature a 45-Watt thermal envelope. AMD's current energy efficient "BE-2xxx" series will be phased out at that time.

likwidsh0k
Posts: 33
Joined: Wed Jan 24, 2007 2:28 pm
Location: Brisbane, Australia

Post by likwidsh0k » Tue Dec 11, 2007 8:08 pm

fri2219 wrote:P.S. If you really want to impress chicks and frighten bullies at the beach, take up classical thermodynamics as a hobby. This Wikipedia article is a good start and explains some of the mathematics behind it. Not only does it make you fun at parties, just talking about it can get you your own seat on airplanes, busses, and trains.

Thanks alot for all the info and a great read fri2219 :lol:

I went ahead and bought the GA-MA69GM-S2H and the BE-2350, with 2 gig of A-Data 800Mhz ram and a 500gig Samsung SATA drive. All fitted in nicely in my old case for the moment with my Seasonic S12 PSU. Pretty happy with the setup.

I found this great thread on the mobo - Unofficial Gigabyte GA-MA69GM-S2H Mboard HTPC Thread - where the consensus is that to play 1080p you need a 3 Gig chip. So this machine is fine for the current use and if I need to I'll grab one of the newer BE chips that should have the horsepower I need for 1080p when it becomes a HTPC. But I will probably buy a dedicated HD/BluRay player anyway and use this for HDTV, movies and music with HDTV here being only 1080i which the IGP has covered no problemo.

Esben
Posts: 83
Joined: Sun Nov 11, 2007 11:35 am
Location: Denmark

Post by Esben » Sun Feb 03, 2008 6:06 am

I have no problems playing 1080p H264 MKV-files with my 2.4 GHz Brisbane on the same motherboard. Don't write it off yet.

juamez
Posts: 45
Joined: Wed Feb 06, 2008 2:17 pm
Location: Belgium

Post by juamez » Wed Feb 06, 2008 2:47 pm

pipperoni wrote:The BE-2300 and X2 3600 (if you get the 65nm part) are based off of the same core. I suspect the X2 3600 is probably pretty close to 45W already since it operates at very similar speeds and voltages to the BE-2300. The 65W class includes higher clocked (and thus more powerful) processors so the 65W designation is less meaningful for the bottom rung processors.
Not quite true.

Since CPU's are designed and fabricated at miniature scales like 65nm, leak-current is the biggest culprit concerning total dissapated power. At 130nm and bigger, leaking current was much more insignificant compared to the power dissapated by the electrical resistance of the processor.

But since the smaller fabrication scales, the total power dissapated (I VERY MUCH like using the word as you can see - ha!) from the working parts of the cpu - the power that is totally lineair to the clock at which the cpu is working - is much lower in relation to the leaked current.

So basically I said the same in 2 alinea's. The point actually is that in the past cpu's were labeled according to their "quality". Quality is roughly based on the highest clockspeed the chip could reach at the preset voltage. Since regular powerdissapation is directly based on the clockspeed, cpu's rised equally much in powerdissapation as they did in clockspeed.

BUT now with the recent developments in CPU land, cpu's are still labeled according to their quality. Indeed, the quality is based on the same thing as before: possible clockspeeds at a given voltagelevel. The difference to what happened in the past is that lower quality cpu's now suffer from more leaking current than the better ones. Since leaking current is a big factor these days, cpu manufacturers tend to sell their cpu's with a dissapated power plus the leaking current dissapation for a total that is given by the thermal dissapation specified by the cpu manufacturer.

This in fact partly destroyes the principles of buying a low priced budget model cpu and overclocking it to get flagship performance but at no penalty.

The penalty with recent cpu's is mostly the lower performance/watt ratio, no matter what voltage you give the cpu or at what clockspeed you set the cpu. That is indeed bummer, BUT as you know, BE models at 45W per dualcore cpu go out for less than €100 nowadays. Those cpu's are among the best when looking at the performance/watt ratio.

Here is a bunch of (distillable) information about the best cores for AMD cpu's in terms of performance/watt:
http://www.xtremesystems.org/forums/sho ... p?t=162738

Seems that the G2 stepping cpu's are among the best available between A64 X2.

edit: mind you, this is just a theory, I can't really back it up with numbers. Maybe someone else could do that for me or for the sake of justice. ;)

Post Reply