Sign in with
Sign up | Sign in
Your question

AMD Athlon 64 X2 4800+ 65-nm Brisbane Preview

Tags:
Last response: in CPUs
Share
December 20, 2006 1:46:58 AM

Business as usual. But I can't understand why AMD doesn't produce anything above the X2 5000 at 65nm. Seems logical to me to make the best chips out of the best process.
December 20, 2006 1:49:44 AM

They like to start with their more mainstream models until they can get yields right. Makes sense to me, as I wouldn't want to be throwing away defective FX CPUs...much better to have to chuck a few X2s.

-J
Related resources
December 20, 2006 1:53:38 AM

Those are some wicked-low temps...

Overclocks to about 2.8... not very impressive...

Still handed its ass by the E6600

And why does the X2 4800 lose to the X2 4600 in most benchmarks? Someone explain that to me.
December 20, 2006 2:31:13 AM

Quote:
For one thing, they generate less power than their 90-nm counterparts. In our testing total system power consumption at equal clock speeds with our X2 4800+ was 11W lower under load than the equivalent X2 CPU at 90-nm. 11W doesn’t sound like much, but in practice our X2 4800+ CPU generated significantly less heat: temps were down up to 26 degrees Celsius under load.



Dam. I was hoping 65nm would generate more power. It would have helped cut down on the 4x4 electric bill.

I know, I know, it was a typo, but I couldnt resist
December 20, 2006 2:45:03 AM

makes me feel better with my s939 :D 
December 20, 2006 2:54:12 AM

Well At least you have the liberty of touching yourself inappropriately at your computer because you at bare minimum have a Worthy Dual Core
(Anything better than a Pentium is "worthy"). !Look at my setup. I'm so far behind in CPU technology it takes me 7 Minutes for a Boot-up to Usable state On this POSS.
December 20, 2006 2:54:55 AM

Quote:
Good find, thanks...



Where is Endyen.... did I not predict this :)  ....

Jack


Careful Jack! This sounds kind of BM ish.
December 20, 2006 2:59:32 AM

:lol: 
December 20, 2006 3:03:37 AM

Quote:
Business as usual. But I can't understand why AMD doesn't produce anything above the X2 5000 at 65nm. Seems logical to me to make the best chips out of the best process.


That's the problem, right now 65 nm is not their best process :)  .... something is fishy with what they presented to the world with the 40% improvement headline and what we are seeing.

Also, very oddly --- the die size for the 512KBx2 brisbane core is 112.7 mm^2 compared to ~183 mm^2 for Windsor.... this is not a 50% area scaling that would indicate a true 65 nm shrink.

My suspicion is they did not scale the pitch down as much in order to get the extra power savings, or SOI is not allowing a perfect pitch scaling due to self-heating. Hard to tell...

(Link for die size: http://www.semiconductor.com/mysi/index.asp?destination... requires registration, it is free to get the cover sheet that specifies the die size).

Jack

It sounds that the 112.7mm^2 represents 65nm 1MB L2 x2 dies... :wink:
December 20, 2006 3:14:36 AM

Quick question:

Isn't 10C and 14C kind of low temp, even at idle? I mean, that would be actually cool to the touch, if you could touch the CPU.
December 20, 2006 3:45:38 AM

Quote:
Well At least you have the liberty of touching yourself inappropriately at your computer because you at bare minimum have a Worthy Dual Core
(Anything better than a Pentium is "worthy"). !Look at my setup. I'm so far behind in CPU technology it takes me 7 Minutes for a Boot-up to Usable state On this POSS.


Odd, it takes less then a minute on my PIII 933 mhz, pc133 CL3 computer. Then again, I never shut it off, so I could be off on my boot time. Least you've got the benefit of dual channel with DDR, I would think. All hail my legacy computer. Thank God it kicks it against something out there.
December 20, 2006 3:47:56 AM

Quote:
Well At least you have the liberty of touching yourself inappropriately at your computer because you at bare minimum have a Worthy Dual Core
(Anything better than a Pentium is "worthy"). !Look at my setup. I'm so far behind in CPU technology it takes me 7 Minutes for a Boot-up to Usable state On this POSS.


Odd, it takes less then a minute on my PIII 933 mhz, pc133 CL3 computer. Then again, I never shut it off, so I could be off on my boot time. Least you've got the benefit of dual channel with DDR, I would think. All hail my legacy computer. Thank God it kicks it against something out there.

Holy sh1t, I have a P3 933 with 512M CL3 RAM too! It boots in 56 seconds, from a 10000rpm SCSI hard drive.
December 20, 2006 4:04:22 AM

Quote:
Well At least you have the liberty of touching yourself inappropriately at your computer because you at bare minimum have a Worthy Dual Core
(Anything better than a Pentium is "worthy"). !Look at my setup. I'm so far behind in CPU technology it takes me 7 Minutes for a Boot-up to Usable state On this POSS.


Odd, it takes less then a minute on my PIII 933 mhz, pc133 CL3 computer. Then again, I never shut it off, so I could be off on my boot time. Least you've got the benefit of dual channel with DDR, I would think. All hail my legacy computer. Thank God it kicks it against something out there.

Holy sh1t, I have a P3 933 with 512M CL3 RAM too! It boots in 56 seconds, from a 10000rpm SCSI hard drive.

Man, I feel like a 'tard now. Decided to shut down and restart. Took 2 minutes and 10 seconds. Still beats out seven minutes though. ATA100 hard drive.

Might I guess that you're using a PCI or PCI-e card for the SCSI connection? Don't know too much about them, but I figured the only way I could get one of those high powered hard drives would be to get an expansion card. Always figured a bigger bottleneck for my system though was being maxed out at 512 megs though.
December 20, 2006 4:26:22 AM

Quote:
Quick question:

Isn't 10C and 14C kind of low temp, even at idle? I mean, that would be actually cool to the touch, if you could touch the CPU.


Yep... 14C translates to 57F, which would be cool to the touch.

I'm not totally conviced those temps are accurate. They're using a well known and accurate tool (Core Temp), but 14C idle seems way out whack. Esp. considering they weren't able to push the CPU beyound 2.8Ghz. I can squeeze 2.8Ghz out of my 4800+ Toledo.

Dunno. Seems weird. But if that temp is accurate, it's nothing less than amazing!
December 20, 2006 4:28:46 AM

I wonder how cool it can go with liquid cooling...
December 20, 2006 4:41:35 AM

Quote:
Well At least you have the liberty of touching yourself inappropriately at your computer because you at bare minimum have a Worthy Dual Core
(Anything better than a Pentium is "worthy"). !Look at my setup. I'm so far behind in CPU technology it takes me 7 Minutes for a Boot-up to Usable state On this POSS.


Odd, it takes less then a minute on my PIII 933 mhz, pc133 CL3 computer. Then again, I never shut it off, so I could be off on my boot time. Least you've got the benefit of dual channel with DDR, I would think. All hail my legacy computer. Thank God it kicks it against something out there.

Holy sh1t, I have a P3 933 with 512M CL3 RAM too! It boots in 56 seconds, from a 10000rpm SCSI hard drive.

Man, I feel like a 'tard now. Decided to shut down and restart. Took 2 minutes and 10 seconds. Still beats out seven minutes though. ATA100 hard drive.

Might I guess that you're using a PCI or PCI-e card for the SCSI connection? Don't know too much about them, but I figured the only way I could get one of those high powered hard drives would be to get an expansion card. Always figured a bigger bottleneck for my system though was being maxed out at 512 megs though.

I've Got no such luxury I've got a DDR-333 RAM Motherboard Running 2 DDR400 RAm Without any Form of Dual-channel. I've got every thing Over clocked a little and a Previous generation Pentium 3 With 1/2 the speed of ram still can beat me to a Boot-up It maybe my hard drive or it may be that i've got 14 Gigs of programs loaded on this Harddrive which i doubt is any faster than 1Mb per second( correction maximum write is 1.6Mb/s and read is 4Mb/s :p  Is this anywhere near the speed of 2 RAID-0 RAPTORs?

I guess anybody with anything faster than a 2.4Ghz Pentium 4 or Athlon Xp 2400+ o/b now can have "sexytime" with them selfs. Well there are quite a few people here that have Much Slower computers than mine but those are very few. Most of the people here have Athlon64 or Pentium D and better
December 20, 2006 4:59:08 AM

Don't know what it could be. For myself, I suppose a little bit could have to do with some kind of "Get the thing booted quick" mentality that could have been behind the design of the board. Read somewhere that under the proper settings, this thing could get booted in 8 seconds with Win2K.
December 20, 2006 4:59:17 AM

what's also strange to me is the article makes no mention (except for the caption) on how cool it runs. are they intentionally misleading people? you'd think the real story would be how cool the CPU runs.

but that's just me...
December 20, 2006 9:15:26 AM

Quote:
Well At least you have the liberty of touching yourself inappropriately at your computer because you at bare minimum have a Worthy Dual Core
(Anything better than a Pentium is "worthy"). !Look at my setup. I'm so far behind in CPU technology it takes me 7 Minutes for a Boot-up to Usable state On this POSS.


Odd, it takes less then a minute on my PIII 933 mhz, pc133 CL3 computer. Then again, I never shut it off, so I could be off on my boot time. Least you've got the benefit of dual channel with DDR, I would think. All hail my legacy computer. Thank God it kicks it against something out there.

Holy sh1t, I have a P3 933 with 512M CL3 RAM too! It boots in 56 seconds, from a 10000rpm SCSI hard drive.

Man, I feel like a 'tard now. Decided to shut down and restart. Took 2 minutes and 10 seconds. Still beats out seven minutes though. ATA100 hard drive.

Might I guess that you're using a PCI or PCI-e card for the SCSI connection? Don't know too much about them, but I figured the only way I could get one of those high powered hard drives would be to get an expansion card. Always figured a bigger bottleneck for my system though was being maxed out at 512 megs though.

I've Got no such luxury I've got a DDR-333 RAM Motherboard Running 2 DDR400 RAm Without any Form of Dual-channel. I've got every thing Over clocked a little and a Previous generation Pentium 3 With 1/2 the speed of ram still can beat me to a Boot-up It maybe my hard drive or it may be that i've got 14 Gigs of programs loaded on this Harddrive which i doubt is any faster than 1Mb per second( correction maximum write is 1.6Mb/s and read is 4Mb/s :p  Is this anywhere near the speed of 2 RAID-0 RAPTORs?

I guess anybody with anything faster than a 2.4Ghz Pentium 4 or Athlon Xp 2400+ o/b now can have "sexytime" with them selfs. Well there are quite a few people here that have Much Slower computers than mine but those are very few. Most of the people here have Athlon64 or Pentium D and betterSounds like you're running in PIO mode. Check and make sure that DMA is enabled.
December 20, 2006 10:25:36 AM

Oh Thank god, I've just done a proper Si soft Sandra bench Mark My Read work is are 27MB/s and My Write work is 25MB/s At least that's much better than what i previously thought.
December 20, 2006 10:45:07 AM

Quote:
what's also strange to me is the article makes no mention (except for the caption) on how cool it runs. are they intentionally misleading people? you'd think the real story would be how cool the CPU runs.

but that's just me...


The data as it is unfolding is clear that AMD is running about 8-12 watts lower in power during idle than Intel --- 3 sites now, different chipsets and the results appear consistent.

What I am curious about is how much is due to SOI/65 nm and how much is due to C&Q (i.e. what speed - clock speed, is it throttling to in idle mode). Honestly, it is beside the point -- 12 watts is nothing -- it will be a big braggin' point I am sure, but 12 watts at $0.12 per KWH amounts to about $1.00 per month more in the electric bill.

At load, it is a wash --- in some loads Intel is lower by 3-5 watts, in others AMD is lower by 3-5 watts. Close enough to call a draw in that scenario.

I do not trust any of the temperatures reported just yet, these are varying wildly including.... 14 deg C at idle --- this is not possible unless they are testing outside in the fall/winter or inside a walk in refrig. :)  ....

It is fun to watch the site paint a positive light on this.... TR and FSQD are both dwelling on the positive low idle power. The OC results are disappointing though.

I think the idle power difference is because of C&Q. This is from the Anand article:

"Note that Cool 'n Quiet and EIST were enabled for all tests, but running at 1GHz the AMD CPUs at idle are able to draw much less power than the Intel system (which runs at an idle clock speed of 1.6GHz)."

Notice that the AMD cpu was 1GHz at idle and the Intel one is at 1.6GHz... also, since AMD has the memory controller on-die it also may clock the ram lower during idle. I am pretty sure i read that somewhere, ill look for a link.

EDIT: I loled at those Temps. though... there are people that actually believe those are correct??
December 20, 2006 10:54:53 AM

Quote:
Business as usual. But I can't understand why AMD doesn't produce anything above the X2 5000 at 65nm. Seems logical to me to make the best chips out of the best process.


That's the problem, right now 65 nm is not their best process :)  .... something is fishy with what they presented to the world with the 40% improvement headline and what we are seeing.

Also, very oddly --- the die size for the 512KBx2 brisbane core is 112.7 mm^2 compared to ~183 mm^2 for Windsor.... this is not a 50% area scaling that would indicate a true 65 nm shrink.

My suspicion is they did not scale the pitch down as much in order to get the extra power savings, or SOI is not allowing a perfect pitch scaling due to self-heating. Hard to tell...

(Link for die size: http://www.semiconductor.com/mysi/index.asp?destination... requires registration, it is free to get the cover sheet that specifies the die size).

Jack

It sounds that the 112.7mm^2 represents 65nm 1MB L2 x2 dies... :wink:

Actually, this is a good point --- I forgot they were re-implementing 1MBx2, this would scale correctly then.

However, why on earth is Anand and Firingsquad reporting this:
Quote:
With the smaller process, die size is also reduced, down from 183mm2 to 126mm2


I think those reports are probably just inaccurate... I could be wrong though. Its also possible that it isn't a linear decrease in size... although like I said I could be wrong :) 
December 20, 2006 11:59:07 AM

Quote:
Quick question:

Isn't 10C and 14C kind of low temp, even at idle? I mean, that would be actually cool to the touch, if you could touch the CPU.


Yep... 14C translates to 57F, which would be cool to the touch.

I'm not totally conviced those temps are accurate. They're using a well known and accurate tool (Core Temp), but 14C idle seems way out whack. Esp. considering they weren't able to push the CPU beyound 2.8Ghz. I can squeeze 2.8Ghz out of my 4800+ Toledo.

Dunno. Seems weird. But if that temp is accurate, it's nothing less than amazing!

Actually, if AMD did not spec out their ODTD correctly and the current temp reporting software is not calibrated to give accurate temperatures. This also happened on C2D, lots of debate on accurate temps.

14 degrees idle is impossible unless they did their benching in a walking refrigerator, else the laws of physics cease to exist in their AMD computer. :lol: 

Naah, maybe they just saving on their heating bills? They did not mention temperature in the room. :) 
Anybody knows where they are located?
December 20, 2006 12:07:58 PM

Quote:
Quick question:

Isn't 10C and 14C kind of low temp, even at idle? I mean, that would be actually cool to the touch, if you could touch the CPU.


Yep... 14C translates to 57F, which would be cool to the touch.

I'm not totally conviced those temps are accurate. They're using a well known and accurate tool (Core Temp), but 14C idle seems way out whack. Esp. considering they weren't able to push the CPU beyound 2.8Ghz. I can squeeze 2.8Ghz out of my 4800+ Toledo.

Dunno. Seems weird. But if that temp is accurate, it's nothing less than amazing!

Actually, if AMD did not spec out their ODTD correctly and the current temp reporting software is not calibrated to give accurate temperatures. This also happened on C2D, lots of debate on accurate temps.

14 degrees idle is impossible unless they did their benching in a walking refrigerator, else the laws of physics cease to exist in their AMD computer. :lol: 

Naah, maybe they just saving on their heating bills? They did not mention temperature in the room. :) 
Anybody knows where they are located?North Pole. :) 
a c 99 à CPUs
December 20, 2006 1:28:00 PM

Boot speed depends on a ton of things, the processor speed being the least influential of them. My Athlon 64 X2 4200+ box on a 10,000 rpm SATA drive takes about 2 minutes to go from switch on to desktop. That is actually a tad slower than my Pentium 4-M laptop on a 5400 rpm ATA/100 drive. Most of it is my desktop's BIOS taking about 30-45 seconds to probe devices and hand off the boot order to GRUB and let the OS start while the laptop's BIOS immediately hands off to the bootloader. But the OS itself (Gentoo 2006.1 AMD64) is not all that quick of a booter as I have an initrd and a fully modularized kernel to be more flexible with hardware. I could be very quick in booting the OS if everything was statically compiled into the kernel and I didn't use an initrd, but an extra minute every other month is not worth it. That's about how often I reboot and how much I care about boot speed.
December 20, 2006 2:03:21 PM

True, but as I said in another thread if your a business you might care about idle power more than load power (where they are both equal anyway). I know at my Uni most of the computers were sitting idle a lot of the time on the logon screen (can you believe it took them until the end of my course to have the sense to auto-shutdown if the machine wasn't being used after 6pm :roll:) . It's not a huge thing but it is a reason to go for these chips over C2D if your on a budget and performance isn't of paramount concern.

I'd be interested what your predictions are Jack about the next rev of 65nm that AMD has planned - do you think this will yield further power savings or will only allow for higher clock speeds/better yields?
December 20, 2006 3:31:28 PM

i think there talking about 14c delta over ambient temp....
which would be only slightly worse than my w/c system which never goes over 90'f since i live someplace it never gets over 90'f.
December 20, 2006 6:07:07 PM



i'm liking the look of the temperatures...but the overclocking capability is disappointing! Compare it to the Intel switch from 90nm - 65nm and it looks as if AMD's transfer is a failure in comparison!

Look early 90nm Athlons. AMD process is like wine... 8)
Athlon x2 3600+ is nice to overclock and can beat stock processors (x2 5000) when overclocked to their level.
x2 3600+ overclock

ScienceMark 2.0 results are crazy.
December 20, 2006 8:30:08 PM

Quote:
what's also strange to me is the article makes no mention (except for the caption) on how cool it runs. are they intentionally misleading people? you'd think the real story would be how cool the CPU runs.

but that's just me...


The data as it is unfolding is clear that AMD is running about 8-12 watts lower in power during idle than Intel --- 3 sites now, different chipsets and the results appear consistent.

What I am curious about is how much is due to SOI/65 nm and how much is due to C&Q (i.e. what speed - clock speed, is it throttling to in idle mode). Honestly, it is beside the point -- 12 watts is nothing -- it will be a big braggin' point I am sure, but 12 watts at $0.12 per KWH amounts to about $1.00 per month more in the electric bill.

At load, it is a wash --- in some loads Intel is lower by 3-5 watts, in others AMD is lower by 3-5 watts. Close enough to call a draw in that scenario.

I do not trust any of the temperatures reported just yet, these are varying wildly including.... 14 deg C at idle --- this is not possible unless they are testing outside in the fall/winter or inside a walk in refrig. :)  ....

It is fun to watch the site paint a positive light on this.... TR and FSQD are both dwelling on the positive low idle power. The OC results are disappointing though.

I think the idle power difference is because of C&Q. This is from the Anand article:

"Note that Cool 'n Quiet and EIST were enabled for all tests, but running at 1GHz the AMD CPUs at idle are able to draw much less power than the Intel system (which runs at an idle clock speed of 1.6GHz)."

Notice that the AMD cpu was 1GHz at idle and the Intel one is at 1.6GHz... also, since AMD has the memory controller on-die it also may clock the ram lower during idle. I am pretty sure i read that somewhere, ill look for a link.

EDIT: I loled at those Temps. though... there are people that actually believe those are correct??

AMD stated before that their 65nm chips run cooler than intels. If I'm not wrong, there's an article that states that one of the lower clocked 65nm chip from AMD can consume 7W of power when iddle (I guess it runs at 2.1GHz). At this moment, AMD has the power consumption crown; all we have to wait if for the second revision of their 65nm process for enthusiast who want to overclock. I guess they are just "playing around" with this process untill they can release a killer revision. Pair this process up with AMD's upcoming Barcelona or the new mobile core and once again they'll have the performance/watt crown.
December 20, 2006 8:44:19 PM

Quote:

At this moment, AMD has the power consumption crown; all we have to wait if for the second revision of their 65nm process for enthusiast who want to overclock.

That's debateable given that in all three previews, the C2D was paired with the hotter chipset. Despite this, the E6600 system used less power in the Anandtech preview, an E6400 system uses less power in the Techreport preview and practically drew in the Firingsquad preview.
!