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

[flippin_waffles]

I wont take this thread off topic, mine should be here by this weekend.

Just understand that I am more apt to showing the wins on Intel.

I don't think anyone will be surprised. It's quite reminiscent of the Pee4 days.

[edit]

As for SPI, why not benchmark Pacman, or Space Invaders. It is likely as useful with todays code. Here's a quote from Scientia.

I looked over the Super Pi executable. I didn't find any SIMD instructions higher than SSE, no SSE2, SSE3, or SSE4. There are also no MMX instructions that I could find. The SSE instructions that are present seem to be moves and conversion. I did find a single MULPS instruction but I don't believe this is in the main loop.

As far as I could tell the main calculation section uses x87 instructions: FADD, FSUB, FMUL, FDIV, FSQRT.
A quick look at the Optimization Guides for both Intel and AMD shows that x87 instructions are faster on Penryn than they are on K10. In particular the FDIV instruction is quite a bit faster. So, this would make sense that if SuperPi uses these then it should be faster.

However, x87 instructions have been considered obsolete for the past five years. The value of testing these on modern hardware is basically zero. In other words, bragging that your computer is faster with SuperPi is only slightly better than bragging that it is faster with real mode MSDOS applications.

It's like running a 429 cobra jet with diesel.

[Cooper]

Another 3.2GHz run. Cooled with cheap Akasa heatsink and ~15C ambient:



Max OC on same setup. Setting multi to x17 crashes the system no matter what voltage on CPU




Results done on Asus M3A MVP Deluxe with AOD only. Only Vcore and multi were changed.

[KTE]

Wow, that's pretty crazy if a Phenom beats a Kentsfield/Yorkfield clock for clock.

I think the 3dmark run there could have been a glitch... I found this other run...

True, I have those both and difference is an OS one. 3DMark'05, not much different.

Does anyone actually know where the cold bug is on these chips? I don't believe I've seen anyone trying them under ice yet.

-40C without a problem on 5x B2 chips, not tried B3. However they hated voltage under 10C, 1.34v was all that would work fine, so air MHz were better

I wont take this thread off topic, mine should be here by this weekend.

Let us know how it goes under cold Charles

Anyone tested this with an AX78, I'm looking to get one for the next build and want to drop a Phenom in, need to know if it clocks well.

Yes, it never booted

Another 3.2GHz run. Cooled with cheap Akasa heatsink and ~15C ambient:

Promising.
1.4v? Can you not try 1.58v Cooper, it seems you may have a 3.5G chip there

Can anyone of you with +3.0G chips test on MSI 790FX please? It'll sav me the hassle of moving to those boards just to see if my PWM area and 4-pin 12V is what holds me back

[BrokenWall]

I will be purchasing on and showing results in a ASUS 790fx

[westk]

3125MHZ was fully stable at default vcore, no pics as im pushing 3250 now with a small bump in voltage.

So looks like I was a little bullish to set 1.4V at 3200MHZ as im above that now with around 1.35V, this is not to shabby

got part1 of a guide up here will continue testing and post here also with voltage and performance tests.

3250MHZ 33C full load, water temp is 12C

Impressive, really impressive.

[gondezz]

Huge thanks for Tony. Great work, Sir

And one silly question, does the problem of dead processor because of high Vdimm still happen? No no, you dont have to try that

[Cooper]

I'm waiting for TRUE to arrive with 133CFM fan for some proper action

Can you not try 1.58v Cooper, it seems you may have a 3.5G chip there

Guess I can just kill one to try this out

I'll test the MSI board and maybe DFI too tomorrow

[tvdang7]

Impressive, really impressive.

dang reviewers like firing squad didnt do so well. like 2.8ghz with 1.4 v maybe they dont know what they are doing or something.

[Nedjo]

This certainly looks promising. Bulldozer should really make things interesting.

what's Bulldozer?

[Nedjo]

I wont take this thread off topic, mine should be here by this weekend.

Just understand that I am more apt to showing the wins on Intel.

It was all over the Net



looking forward for your results, especially on NB clock front!!

[RFC Rudel]

is 64 bits oc problem present on the b3?

[Undersea]

is 64 bits oc problem present on the b3?

I think Tony mentioned using it

[Tony]

sorry no action today, had emails and stuff to do...more tomorrow though if I can.

GBT UK will flash my board back to life (yippee) so i should be back clocking with this also.

Asked Oskar for a few fixes on the M2R, hopefully this makes it a little better.

[tbone8ty]

awesome cooper!

i waiting for my 9850BE phenom and asus m3a32-mvp to come in the mail and ill join you in testing. glad to see some people using the asus board.

what ddr2 u using?

judging by looking at the asus forums, i was about ready to send mine back for a DFI. hopeully asus will update bios soon.

tyler

Another 3.2GHz run. Cooled with cheap Akasa heatsink and ~15C ambient:



Max OC on same setup. Setting multi to x17 crashes the system no matter what voltage on CPU




Results done on Asus M3A MVP Deluxe with AOD only. Only Vcore and multi were changed.

[Brother Esau]

sorry no action today, had emails and stuff to do...more tomorrow though if I can.

GBT UK will flash my board back to life (yippee) so i should be back clocking with this also.

Asked Oskar for a few fixes on the M2R, hopefully this makes it a little better.

You are my Hero! I responded to your post in DFI thread please check that situation that I mention with the Bios Re-programming on you're end please and make Taiwan aware of this they really need to address that issue and soon.

[wiak]

Tony am hijacking this thread a little bit
you dont mind do you?

some photos of Gigabyte GA-MA790FX-DS5 and Phenom 9850 Blacky Edition

Lovely, Boxes :p


Motherboard packed out


Motherboard


Stock Cooler upsidedown


Phenom 9850 Blacky Edition with Arctic Cooling MX-2 Thermalpaste


Cooler in place


Everything in case, bling bling ;-)




ps: will be changing psu soon

[Brother Esau]

Have fun with your new system Sheesh Gigabyte makes some Ghastly looking boards with them color combo's

[tbone8ty]

and???

[Dagalidis]

Nice setup WIAK,,,
But what about your CPU ?
Did you reach the 3Ghz ?

[Blacklash]

I believe Tony was testing under Vista x64.

I dropped my rig to 378x9 and ran the CPU portion of 06 with 5-5-5-15 timings and no tweaks. I got 4890. So the 9850 @ 10x340 is slightly faster than what I have here at near the same speed. Vista definitely gives a hit in the CPU portion of this test. So I wouldn't compare results from XP. @ 3.8 under XP my CPU score is 5865, under Vista x64 it falls to 5384.

Looks good AMD. I might have to build up my 3rd rig again with a DFI mobo and one of these 9850s.

[justapost]

Now what is this.

It validates


In short, tried XEN 4.1 with the 9850 oced to 2.8GHz with an Win 2003 64bit Server Guest os. System runs stable. With an 9600BE it was only stable up to 2.5GHz.

[SocketMan]

Now what is this.

It validates


In short, tried XEN 4.1 with the 9850 oced to 2.8GHz with an Win 2003 64bit Server Guest os. System runs stable. With an 9600BE it was only stable up to 2.5GHz.

What have you done!?!
You've tasted the forbidden fruit of spi
AMD has 2 "new bestsellers" for you:BKDG and a revision guide for 10h Family
both dated April 09,2008. Enjoy
http://www.amd.com/us-en/Processors/..._15343,00.html


@CooperCan you take a picture of the cpu when you change the coolers pls?thx

Hey Wiak are you sure it's the right staff on top of the cpu? Looks a bit like someone got "overexcited" a little bit
I know I would've ,nice pictures thx

[KTE]

I'm waiting for TRUE to arrive with 133CFM fan for some proper action

Guess I can just kill one to try this out

I'll test the MSI board and maybe DFI too tomorrow

Thanks man
Unlike Intel 45nm, you can push high volts into these on air for a bit without trouble if you can keep the temp. below 60C - guide states 1.55v as maximum tolerance by core, so I assume 1.7v for 30 minutes won't be too bad

Trust me, I was at a wall with 1.44v - until I pushed 1.6v into it
Drops to 1.512v load anyway.

SocketMan, thanks for the links! Latest "errata" and other register info such as PVI codes will all be there, updated

[wiak]

no offense taken , was just the amount of notice many of you guys actually take of apps reading voltages amazes me....i keep saying the ONLY way to be sure is measure it properly

get a 790FX board m8...if you must wait for SB700 boards to surface but still go with the new gen and demote the old asus to a home theater rig

dont hold it, if you ask me the SB700 is ONLY 2 more sata + some more usb ports, and meybe 5% faster, anyway i have SB600 and happy with it :P

[wiak]

Wow, that's pretty crazy if a Phenom beats a Kentsfield/Yorkfield clock for clock.

I think the 3dmark run there could have been a glitch... I found this other run...
Check out this Q6600s CPU score @ 3401MHz

i think tony's are non-overclocked graphics card and overclocked phenom