Could B3 Phenom be the rebirth of AMD....+1000MHZ OC done!

[Extelleron]

Keeping this progress with CTI,AMD can and probably will bring us to B3 chip@2.8Ghz by Q3(before Shanghai).

Phenom 9950 comes Q3 @ 140W TDP and unless something changes, that is the last of Agena 65nm.

Anandtech in their B3 review said AMD was prioritizing 45nm over continuing 65nm development, and I'm sure that is what they will do. Some B3's could probably be sold @ 2.8GHz but not many, remember that stability for an oc'd system vs. stability for a CPU sold by AMD is a different story. Just because a 9850 can do 2.8GHz @ stock volts doesn't mean that AMD can take that CPU and certify it to work under load 24/7 without error.

But B3 is looking pretty good, if 45nm is an improvement in frequency and not just power/thermals then we will see some good results come Q4.

[Nedjo]

if you take look at Crti$ch and Oliveirda's Phenoms you'll notice that they are build in the last week of February...



obviously sample rate of XS owners compared to the full production of these CPU's is to low to make concrete arguing, but I feel it's safe to bet that with B3 AMD has room to go beyond 2.6 GHz planned for 9950 at default voltage, and bring out another speed bin prior to Daneb.

Like Informal wrote, with CTI in full swing, AMD has a chance to brush B3 revision trough out Q2 (90-100 days is a lots of time) and launch 2.8 GHz model in acceptable TDP (140W for 1.3V fed models)

[Andi64]

Because of the naming scheme I doubt AMD will realease anything beyond a Phenom X4 9950 2.6Ghz on 65nm.

IMHO 2.7 and 2.8 Ghz parts will be on 45nm with a different name... IDK, something like Phenom X4 X550 @ 2.8Ghz :P

[Nedjo]

Because of the naming scheme I doubt AMD will realease anything beyond a Phenom X4 9950 2.6Ghz on 65nm.

IMHO 2.7 and 2.8 Ghz parts will be on 45nm with a different name... IDK, something like Phenom X4 X550 @ 2.8Ghz :P

well between 9950 and 10000 you have lots of space... you have 60, 70, 80, 90

[Andi64]

AMD Phenom X4 9999 3.21099Ghz 2.21M L3?

[Brother Esau]

Phenom 9950 comes Q3 @ 140W TDP and unless something changes, that is the last of Agena 65nm.

Anandtech in their B3 review said AMD was prioritizing 45nm over continuing 65nm development, and I'm sure that is what they will do. Some B3's could probably be sold @ 2.8GHz but not many, remember that stability for an oc'd system vs. stability for a CPU sold by AMD is a different story. Just because a 9850 can do 2.8GHz @ stock volts doesn't mean that AMD can take that CPU and certify it to work under load 24/7 without error.

But B3 is looking pretty good, if 45nm is an improvement in frequency and not just power/thermals then we will see some good results come Q4.

AMD will provide current B3 to earn the approval of the Public and take them out of the Hot seat that they are in with B2 Rev but I should not think they will put anymore effort into B3 65nm just would not make any sense not to focus on the 45nm die shrink and focus exclusively on that.

[triple_A]

I swore not to buy another ASUS board after my M2NPV-VM, but as its the only Nvidia option right now...

Stupid board cant survive 1MHz above stock.

lolz i have that same mobo for my second rig and it does 240mhz on ht bus easily. With 5x HT multiplier that is, because there is no setting for that in bios

[Nedjo]

AMD will provide current B3 to earn the approval of the Public and take them out of the Hot seat that they are in with B2 Rev but I should not think they will put anymore effort into B3 65nm just would not make any sense not to focus on the 45nm die shrink and focus exclusively on that.

Well the fact is that you can't simply overnight swap complete production from 65 to 45nm, and as we all come to learn, CPU's with previous node generation will be available in parallel with the new ones…

Due to fact that AMD is quite agile when it comes to perfecting current technology thanks to CTI, I wouldn't dismiss possibility of further perfection of B3's… they don't need to call them B4 all they need to do is to provide enough high yield on frequencies beyond 2.6 GHz for more 65nm Phenoms.

As an illustration to how effective CTI is I would remind you that B1 or BA, how they were called for commercial K10's, was out in Q3 2007, and B2 followed in Q4 2007 and B3 was in production in Q1 2008! And if you look back and add in account all necessary debugging, AMD has made tremendous advancement from B1 to B3, basically having new core in three quarters consequently!

[Manicdan]

correct me if im wrong, but once they go past 3.0ghz wernt they gonna call it like C1, C2, C3 ect instead of B4 or whatever they are up to

[Brother Esau]

Well the fact is that you can't simply overnight swap complete production from 65 to 45nm, and as we all come to learn, CPU's with previous node generation will be available in parallel with the new ones…

Due to fact that AMD is quite agile when it comes to perfecting current technology thanks to CTI, I wouldn't dismiss possibility of further perfection of B3's… they don't need to call them B4 all they need to do is to provide enough high yield on frequencies beyond 2.6 GHz for more 65nm Phenoms.

As an illustration to how effective CTI is I would remind you that B1 or BA, how they were called for commercial K10's, was out in Q3 2007, and B2 followed in Q4 2007 and B3 was in production in Q1 2008! And if you look back and add in account all necessary debugging, AMD has made tremendous advancement from B1 to B3, basically having new core in three quarters consequently!

I understand and acknowledge all that you just said Nedjo but the facts are that they have serious cash Flow issues ATM and I know thats what I would be doing if it were my business and I had those problems.

Just my opinion but it makes logical sense regardless

[Extelleron]

correct me if im wrong, but once they go past 3.0ghz wernt they gonna call it like C1, C2, C3 ect instead of B4 or whatever they are up to

Bx or Cx refers to the revision of the chip and really doesn't have anything to do with frequency; a core will typically undergo a number of revisions improving slightly upon the previous one. An example is Intel's Q6600 - it launched w/ B3 stepping and now has moved to G0 stepping, which consumes less power and overclocks more. With Phenom, the move from B2->B3 was primarily to fix the TLB erratum, but it seems to have improved the overclocking ability as well.

AMD's 45nm chips will be revision Cx, ex. C0, C1, etc, while AMD's 65nm Phenoms are Bx; B0, B1, etc. Since AMD's 45nm Phenoms will be the first to go beyond 3GHz (hopefully), yes they will probably be something like C1 or C2 stepping.

[Rammsteiner]

Ugh, I must be a n00b or so.

What's the AMD CPU setting with options p0 and p1 for?

[KTE]

TDP will restrict any higher releases regardless of the process ability and they are not interested in more volume production quads at 65nm now but 45nm

The FX series was always planned and they have the models for a while now. I do not know what is holding them back apart from competition and the lack of market for it. However, I am not AMD, heck even AMD do not know what will happen as Q3 and Q4 approaches, so you could see higher releases which are not currently known officially. FX82 is already tested since February though, and a Feb 9th update report much after the TLB bug being known mentions more releases: http://www.pcadvisor.co.uk/news/inde...sID=12061&pn=5

Although all those notes are what we knew from Jan, K10 will obviously continue in volume production even while AMD 45nm releases, so you will most likely have higher grades in K10 as you have with K8 65nm G2 now. The report mentions upto 3.4GHz K10 by years end from AMD.

The profit volumes are in the low end/mainstream and the server end, wanting energy efficiency, and that demands an increase in performance efficiency per clock rather than high clocks and high TDPs. If AMD will ever directly compete at 45nm, it will have to be with performance per clock as the main weapon and looking at the cards, that will be an very steep uphill battle.

[Sparky]

/looks at B3 overclocks
/looks at broken B2 that isn't even stable at stock

me wants a B3... *cries*

Gonna call AMD soon to get this chip RMAed, who knows, if they have cut off production of the B2s now (I sure hope they have, why make more bugged chips?) I might get a B3 in return, that would be sweet.

[Rammsteiner]

Ugh, I must be a n00b or so.

What's the AMD CPU setting with options p0 and p1 for?

I went to Sapphires site and read their BIOS settings guide. Apparently the AMD CPU stepping handles performance states. However, are those P-states pre-configured? Cause I tried 1.3V with 14x multi and 200 HTT and it froze in loading Windows. Even at 10x250 it didnt even post while everything was after all stock speeds.

Those were some rough settings though but just trying out heh. Of course I set NB and HT multi lower, as well RAM.

Then I set P1 state and just decided to try out slowly. Now running small FFT's at 12.5x210, if it's stable for about and hour and a half Ill move HTT a little bit up etc.

But if anyone could comment whether to use P-0 or P-1 for OCíng, that would be awesome

[KTE]

But if anyone could comment whether to use P-0 or P-1 for OCíng, that would be awesome

I don't understand what you're saying properly.. are you saying P-State options are available to you in BIOS?

P-State 0 is maximum performance state, i.e. your stock boot clock without CnQ
P-State 1 is your CnQ clocks/voltages

[Rammsteiner]

Hmmm, thx KTE.

Strange thing though, with CMOS default my PC starts up at 1.05V and about lowest multi lol (with P-0).

But yes, P-states are available in BIOS although Ive CnQ disabled. Also I think there might be a new BIOS for DFI since Sapphire uploaded a new BIOS with some new Phenom codes, however the DFI sites has problems (at least on this PC).

Im running P-1 now with OC'd settings and CnQ disabled so I dont know whether the settings of the P-states are having an effect on anything now anyway.

[KTE]

Run P-State 0, that's default boot. Then oc, it should change along with it.
Check MSR C001_0064, MSR C001_0065 and MSR C001_0070, what all 3 read. [65] should be different to [64] in VID/FID/DID values.

[MonsterMash]

Mine seems to be the only one that don't want to keep the 3ghz. Max freq @ vcore default is 2600mhz. Max OC is 2800mhz, and with this freq i can load windows only with heavy overvolt of 1.54V. And also in this way, it's not RS.

Now i'm selling it, and in 1-2 week i think to by another one, hoping to be much luky.

[Loser777]

lolz i have that same mobo for my second rig and it does 240mhz on ht bus easily. With 5x HT multiplier that is, because there is no setting for that in bios

Processor/BIOS please?
I've heard of it doing 240 on single core 3000+, but not anything else.

[awdrifter]

well between 9950 and 10000 you have lots of space... you have 60, 70, 80, 90

They'll probably go to X950 or HD950.

[justapost]

Hmmm, thx KTE.

Strange thing though, with CMOS default my PC starts up at 1.05V and about lowest multi lol (with P-0).

But yes, P-states are available in BIOS although Ive CnQ disabled. Also I think there might be a new BIOS for DFI since Sapphire uploaded a new BIOS with some new Phenom codes, however the DFI sites has problems (at least on this PC).

Im running P-1 now with OC'd settings and CnQ disabled so I dont know whether the settings of the P-states are having an effect on anything now anyway.

That p-state setting is abit odd. I tried P-1 and checked the msr's. Seemed i was still in p-0 stage.
The field itself is labeld "AMD CPU Stepping" but steppinng and p-states are different things.

[SoloBuggMaster]

this looks really good for AMD

[medjohnson77]

Just wanted to get in on this form about a week ago but they just approved me tonight, so here this goes. I just built a custom build about three months ago going with this set up.

Tony Just want to say thanks for what your are doing, It really helps others make informed choice's. And I love my OCZ Ram BTW!

http://i265.photobucket.com/albums/i...nson77/006.jpg
http://i265.photobucket.com/albums/i...ombuild003.jpg
http://i265.photobucket.com/albums/i...nson77/004.jpg

http://i265.photobucket.com/albums/i...7/015-Copy.jpg




custom AMD 64 x2 6400+BE, 8800GTS 512mb (G92) BFG computerTECH SPECS:



1@ NZXT Alpha Black Steel ATX Mid Tower Computer Case with open Window

1@ Rosewill RX750-D-B ATX12V v2.2 & EPS12V v2.91 750W power supply

1@ ASUS M2N32-SLI Deluxe wireless edition AM2 Nvidia nforce 590 SLI mcp atx amd motherboard

1@ AMD Athlon 64 x2 6400+ Windsor 3.2 GHz Socket AM2 125W Dual-core Processor

1@ Rosewill RCX-Z940-SL92 mm 2 ball CPU cooler

2@ OCZ Gold 2GB STicks 240 pin DDR2 SDRAM DDR2 800(PC2 6400) can go up to 8GB with Mboard. 4GB on the board

1@ Seagate Sata/300 Barracuda 3.5" 500GB 16Mb Cache 3Gbs 7200Rpm

2@ Samsung SH-S203 DVD+R/-R 20x Sata retail ver. with lightsribe

1@ BFG 8800GTS Grafix card 512MB on board DDR3 Ram This is the one that beats the GTX/GT in almost every bench mark! Decided the 8600gt wasn't good enough.

1@ MI keyboard USB 2.0 plug and play, with 15 hot keys 1@ Reaper gaming mouse, with 5 hot buttons and 1600 DIP,USB.

3@ 120mm case fans with blue lcd lights

2@ 80mm fans in the RX-750watt power source with blue lcd lights

1@SYTRIN 5.25 drive bay cooler with 2x40mm fans.

1 OCZ ram cooler with 2x60m fans.

Bios version 1802/date 1/18/08

O.S. Windows Vista 64bit version


I figured I could update this custom build with putting in this new Phenom 9850. It didn't work out for me on my ASUS M2N32-sli deluxe. Temps were very high on this processor from the get go. It all looked good when I first dropped it in after updating from 1402 up to 1802 the 9600Be b2 stepping flash. I got it to overclock to 2.9 stable with a very small volt increase. upto 1.38 I think. I didn't clear my cmos when I first updated the bios and it proved to affect the performance, even though at that time it had not acted unstable. In benchmark06 it got a 7300 stock score no overclock on the CPU or the 8800GTS. When I started overclocking both up to max 2.9 on the Phenom and maxed the 8800GTS I got like 7985. Not that it really matters but I got a 5.3 in vista when I first put it in. My 6400+BE got a 5.6 on vista stock. It wouldn't let me touch the multi at all but would let me adjust the cpu mhz. I got on toms and posted what was going on under a thread there called "got the phenom 9850 today!!" Someone recomended that I reset the cmos and I did and sure enough it worked. I put everything back to stock settings before I cleared it and started running benchmarks at stock speeds. Sure enough it worked and got a 11,865 on stock scores with the new Phenom 9850. My 6400+ got 11,356 so all looked good with it showing that gain at 2.5Ghz. I was able to adjust the multi after I reset the cmos but did not apply any changes to it. I did the same as I did before I reset the cmos and tried to adjust the CPU mhz and that was the end of the Phenom. If you want you can go to toms under CPU and look up my thread and check out what i did, it ran some what stable when I bumped the multi down to 12, and increased the NB voltage to 1.25 but I could not overclock at all, and it started blue screening and saying that a clock was not recieved to the 2nd processor in time and shut down with a system dump. I have RMa it back to the egg. And not sure if I will have them send me a new one and upgrade my board to the Asus am2+ for 219 on the egg or just wait for the 45's to come out. This chip looked like it had promise, but not sure what I'm gonna do. I just spent about $1500 on my system I just built and was hoping that I could run this new Phenom on my setup, but I had no luck. Anyone with this board have luck??? If so let me know your system specs. I have tore down my whole rig last weekend and Painted the inside metalic blue and rewired all of the wires to get better air flow, I did a pretty good job when I first built it but this time I did it right and there are no wires that are not clipped tucked and out of the way! I will post some pics of the results in a couple days, it looks really nice! I had to clean install vista 64 after all this happened, at first I could drop my 6400+ back in with no problems but it started freezing up so I got rid of everything and Installin everything clean again. I had N tune on my computer because when I first built this Rivatune would not work with my 8800GTS. During the process of puttin this Phenom in and out I got rid of N tune and upgraded to the new Nvidia program tools but later found out it is not supported on my board. It let me use it but wouldn't show all the options. After the clean install I have not installed anything from Nvidia except my drivers and I got Rivatuner to work now so going thru the process of install all my Warhammer games and other programs.

Sorry for the long post guys,

Keep up the good work cuz I might just build a new rig in a couple of weeks with the new board and the 9850BE This is getting to be a very expensive hobby!

[biohead]

^ thread hijack imo. please start a new thread...or something.