Three New, 95 W AMD FX Series Processors Coming Up

[Thefumigator]

we shouldn't forget these are all 95watts processors, if pricing is ok then power consumption will drop to competitive levels. Performance won't be improved tho

 

[Inceptor]

This is just speculation, but it just occurred to me:

B2 Bulldozers are numbered FX-x1xx.
Piledrivers are supposedly FX-x3xx.
So, B3 Bulldozers will be FX-x2xx?

It seems reasonable but reasonableness on this topic doesn't give me much confidence.
Any opinions or scraps of info?

 

[Super XP]

This is just speculation, but it just occurred to me:

B2 Bulldozers are numbered FX-x1xx.
Piledrivers are supposedly FX-x3xx.
So, B3 Bulldozers will be FX-x2xx?

It seems reasonable but reasonableness on this topic doesn't give me much confidence.
Any opinions or scraps of info?

Great catch, makes sense. Piledriver will either be B4 or C2/3 stepping IMO. I know for a fact AMD modified Bulldozer's original design many months before Bulldozer's release and named it Code Name Piledriver. So they've been messing around with Piledriver for some time now. That so called 20% better performance (clock 4 clock) for Piledriver over today's Bulldozer may very well grow to over 20% by the time it's released sometime in Q3 2012.

 

[jpierce55]

we shouldn't forget these are all 95watts processors, if pricing is ok then power consumption will drop to competitive levels. Performance won't be improved tho

It is a surprising move in the correct direction at least.

 

[jihadjoe]

Past gens of AMD chips I'd consider more as lacking in enough cache.

AMD knows exactly what cache can do, and has known for a long time. The main reason the K6-III was so good was the large amount of on-die L2 cache running at full CPU speed. At the time of release, the K6-III 450 was one of the fastest x86 CPU around, handily beating out the K6-II 400 and Pentium II 450 in integer, or when running stuff that made good use of 3DNow.

 

[Wile E]

I know this has probably been asked before, but never really answered to my satisfaction. What would the equivalent power envelope be in an Intel cpu? Would an Intel that consumes this much power also be rated at 95w? Higher? Lower? What?

 

[TheGuruStud]

I know this has probably been asked before, but never really answered to my satisfaction. What would the equivalent power envelope be in an Intel cpu? Would an Intel that consumes this much power also be rated at 95w? Higher? Lower? What?

Lower.

Neither really calculate TDP the same, but intel doesn't include non-core into the figures (aka lying).

Unfortunately, testing this is very hard since intel also draws power besides the 12v rail (really convenient). Intel finally admitted to it and IIRC non-core can consume up to 28 watts as per their documentation (I'm skeptical of anything they say, I don't trust crooks, but it does seem realistic).

You would think total system power consumption could show a clear picture, but it doesn't. With the new FX, power figures are varying greatly, b/c of crap bioses (not sure how much has been fixed). AMD boards seem to be designed a lot different, regardless. The heat produced isn't matching the power figures. If an X6 was really drawing more power than intel's, then how come it runs pretty cool even when OCed? Nothing adds up right. You can't claim an AMD chip is using 250 watts OCed if it's running cool haha. I doubt that the design of the chip can account for using that much power, but not turning it into heat.

 

[Wile E]

Any independent links on this? Somebody had to test it at some point.

 

[Mathragh]

Any independent links on this? Somebody had to test it at some point.

Thats part of the problem, because of the way processors use current from different sources(like the both the motherboard, and the CPU 12V socket).

If both processors are not comparable in how they use the current provided to it by both the mobo and PSU directly, you cannot compare the power draw, because measuring the powerdraw from the motherboard to the cpu directly cannot be done in any easy way.

 

[Thefumigator]

Any independent links on this? Somebody had to test it at some point.

All I can say is that I have an amp clamp multimeter to measure power consumption
and I've measured power consumption of things (usually entire systems like TV, ovens, computers, but not the individual parts that conform them) on several scenarios. I find impressive how energy consumption can drop when using a "low power" energy profile, no matter which system while modern enough (post 2006 I imagine). Its funny how the readings in the clamp meter drop instantly with just a mouse click over energy saving profile, (that if cool n quiet -or the intel equivalent- is enabled of course)

Maybe I should make some tests again and document them. I also did (a long time ago) 80plus vs non-80plus test on the same system and I can't remember a thing.