PDA

View Full Version : List of AMD's K8 cores on rev. G2



Nedjo
10-21-2007, 11:09 AM
According to AMD these are the CPU's that are using new G2 revision of K8 core.

Starting with model LE-1250, Sempron has this core:

http://img132.imageshack.us/img132/1043/amdprocessorsfordesktopej5.th.png (http://img132.imageshack.us/my.php?image=amdprocessorsfordesktopej5.png)

When it comes to Athlon X2 family G2 revision has arrived with new BE-2400:

http://img107.imageshack.us/img107/1446/amdprocessorsfordesktopem7.th.png (http://img107.imageshack.us/my.php?image=amdprocessorsfordesktopem7.png)

For G2 based Athlon 64 X2 5000 we've all heard - unlocked Black edition:

http://img150.imageshack.us/img150/2135/amdprocessorsfordesktopac3.th.png (http://img150.imageshack.us/my.php?image=amdprocessorsfordesktopac3.png)

But I'm not positive that many know about A64 X2 5200 based on G2:

http://img138.imageshack.us/img138/3931/amdprocessorsfordesktopjz8.th.png (http://img138.imageshack.us/my.php?image=amdprocessorsfordesktopjz8.png)

It's obvious that AMD has made tremendous improvement with G2 revision of their 65nm production process. Experiments with 5000 Black Edition are showing that CPU is capable of reaching 3.2GHz on default voltage. That was unimaginable with G1 revision of Brisbane core!

Also it's clear that this breakthrough that's achieved with rev. G2 for K8 is behind B2 revision of K10 core.

If you take in account that AMD had SIX revisions of 90nm manufacturing technology: D0, E3, E4, E6, F2, F3 it's reasonable to believe that we'll have at least two more revisions of 65nm tech, but primarily for K10. When it comes to K8, G2 is the peak, and there is no reason for K8 to get any new cores.

I hope that this list will help someone in a search for overclockable K8 CPU.

mad_skills
10-21-2007, 11:24 AM
well done Nedjo, tnx ;)

funnyperson1
10-21-2007, 11:41 AM
Good work, that LE 1250 is very appealing for a chip to play with, should be a crazy overclocker.

Also the BE 2400 should offer a great deal considering its 20 bucks less than the 5000+ Black Edition.

woppy101
10-21-2007, 11:45 AM
any one else notice that x2 5200 has 2mb of L2 and not the normal 1mb of 65nm cores

yichen382
10-21-2007, 01:29 PM
G2 vs. F3, which OCs higher?

informal
10-21-2007, 01:31 PM
The 5200+ G2 has 2x512KB L2,it's a typo on amd website(probably the newest entry ,they will correct this soon)

The important thing about G2,aside the much better OC potential ,is the fact that AMD did make some changes to mem. controller and L2 latencies!L2 latencies went down while mem.bandwidth went up considerably when compared with G1!
Also mark03's CPU test(obviously CPU bound test,doh :)) shows nice improvement of 6% going from G1 to G2 5000+,thanks to mem. contrl. improvement.

Look here for details:
http://translate.google.com/translate?u=http://www.planet3dnow.de/vbulletin/showthread.php%3Ft%3D321677%255B/u%255D


Who but a look at the rest of latency raises, the addition of the L2 cache, will immediately notice that the G2-Stepping here up to 30 cycles is lower than the G1. Daher prüfen wir doch gleich einmal, ob das BIOS die Speicherlatenzen identisch eingestellt hat: Therefore, we consider it equally even if the BIOS to set Speicherlatenzen identical

Obviously AMD has also fine-tuning here at the Memory Controller
The Science Mark V2 attaches Speicherlatenzen both, as well as the memory bandwidth. Auch hier kann man deutlich sehen, dass das G2-Stepping einen beachtlichen Vorsprung auf seinen Vorfahren herausarbeiten kann. Here you can clearly see that the G2-Stepping a significant lead over his ancestors out.

Morais
10-21-2007, 05:14 PM
Good post nedjo, thanks for the information

cky2k6
10-21-2007, 05:24 PM
The 5200+ G2 has 2x512KB L2,it's a typo on amd website(probably the newest entry ,they will correct this soon)

The important thing about G2,aside the much better OC potential ,is the fact that AMD did make some changes to mem. controller and L2 latencies!L2 latencies went down while mem.bandwidth went up considerably when compared with G1!
Also mark03's CPU test(obviously CPU bound test,doh :)) shows nice improvement of 6% going from G1 to G2 5000+,thanks to mem. contrl. improvement.

Look here for details:
http://translate.google.com/translate?u=http://www.planet3dnow.de/vbulletin/showthread.php%3Ft%3D321677%255B/u%255D

in other words, g2 can now compete with 90nm clock for clock... lol

PhilDoc
10-21-2007, 05:52 PM
I don't know to many 2.6G stock 90nm cpus going 3.7G on air. Irrc, the wr for my 185 2.6G opty is 3.8G and I'm sure thats with extreme cooling.

informal
10-21-2007, 06:48 PM
in other words, g2 can now compete with 90nm clock for clock... lol

Yeah,basically it's like having a 65nm Brisbane CPU with CZ RevF overclocking abilities(or better) and Windsor-512 performance :)

Wiggy McShades
10-21-2007, 06:51 PM
3.2 without touching volts if that ever translated over to phenom that'd be sick.

cky2k6
10-21-2007, 09:14 PM
scary part? it still won't be enough for penryn... i don't care though, i would like to see amd survive, so i'll still buy their cpu...

LIKMARK
10-21-2007, 10:25 PM
The 5200+ with 2x1024 L2 cashe. Isn't there even a 5600+ 65nm version coming?

informal
10-22-2007, 05:56 AM
The 5200+ with 2x1024 L2 cashe. Isn't there even a 5600+ 65nm version coming?

The 5200+ G2 has 2x512KB L2,it's a typo on amd website