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

[Tony]

BIOS P0J doesn't apply the TLB fix by default for MSI Plat. Tony
Official releases 1.3-1.4 do but the official ones are quite poor for oc but better in stability/compatibility.

EDIT: yeah look to your OS

Send me the bios please

[justapost]

I don't run SP1 on vista, so its not the OS

Boot open AOD, its yellow, bench, take scores, open AOD, set red, bench see increase

Simple way of testing me thinks

Go to bios, set enabled, go to vista..AOD yellow.>
Go to bios, set disabled, go to vista, AOD yellow

DFI

What am I doing wrong here?

What bios version are you using? Have you checked the msr registers c0010015 and c0011023 with crystal cpu's msr editor?

[LIKMARK]

heh, lapping only really makes maybe 2-5 degrees of difference... its not gunna make your 57 degrees drop to 47 degrees load like, say, mine. odd that your W/C is that high. whats your idle temps? im sitting around 29-31 when at a 3.0ghz OC

Idle at about 40c @1,30vcore.

Maybe I need a reseating of the block..

[Tony]

What bios version are you using? Have you checked the msr registers c0010015 and c0011023 with crystal cpu's msr editor?

Official 1.4 on the MSI and 331 on the DFI

I have not checked with Crystal CPU yet..benching winrar on the MSI at this time.

[GunterFalstaff]

Idle at about 40c @1,30vcore.

Maybe I need a reseating of the block..

possibly... im at 3.0ghz with 1.34v core, i know my lap didn't do THAT much...

my temps seem really good when under load and overclocked to 3.0ghz...

this is from the dfi thread.

[justapost]

Official 1.4 on the MSI and 331 on the DFI

I have not checked with Crystal CPU yet..benching winrar on the MSI at this time.

Tried it on the DFI with 307. MSR registers show that TLB caching is enabled no matter what i set in the bios.

At 2.8GHz CPU/2.2GHz NB mem at 533 settings auto i get these winrar results with aod in red or yellow mode, again no matter what tlb caching setting was used in the bios.

Completely unoptimized memory timings here but ~2000MB/s vs. ~650MB/s in green mode.
I guess the bios does the right thing, ignores the bios settings with b3's and always enables tlb caching here.

[wiak]

MSI and DFI have not disabled TLB from bios with B3, you still have to apply a fix. I have reported this to DFI and to AMD, we wait for a fix.

Being real honest the DFI bios has many issues, i have a few end users now suffering from the flashing cursor issues, tRFC non detect etc so we wait on a fix.

The board with the lowest amount of issues (without testing the asus yet) is the Gigabyte DQ6, it just needs separate NB VID and it would be awesome.

The MSI is clocking good on the latest 1.4 Bios apart from the TLB annoyance.

Ambient water temp is between 8 and 14C dependent on how cold it is out side on my water loop.

DS5 to has little bugs to

[wiak]

Can some more people chime in on the Temp situation with their 9850's please?

stock speed, stockfan @ 2800rpm is around 49c idle, load temp is 53c

hopes this helps
guys if you set overdrive to RED button the TLB fix will be disabled, looks like overdrive and bios dosnt detect B3 as bugfree

and temps in everest on mine is kinda off buy a HUGE margin, i can soon frie a egg on this thing

[Tony]

Its looking like yellow or red on AOD is telling US top performance, green is most definitely bad.

I think we need an update on AOD, looks like im blaming the bios when its the tools not working right.

Weird thing is AOD reports red on the Gigabyte where MSI and DFI is yellow

[MonsterMash]

BIOS P0J doesn't apply the TLB fix by default for MSI Plat. Tony
Official releases 1.3-1.4 do but the official ones are quite poor for oc but better in stability/compatibility.

EDIT: yeah look to your OS

Where can i find this bios?
Thank you.

[justapost]

Weird thing is AOD reports red on the Gigabyte where MSI and DFI is yellow

I read in an german forum that in red mode autoxpress features get enabled. The MSI has bios options for autoxpress with at least one bios but KTE could not verify that autoxpress is responsible for the performance increase in red mode.

[Tony]

I read in an german forum that in red mode autoxpress features get enabled. The MSI has bios options for autoxpress with at least one bios but KTE could not verify that autoxpress is responsible for the performance increase in red mode.

I also just ran winrar...in red mode 3.4ghz cpu, 2500mhz NB and 1000MHZ ram im hitting 1.700KB/s on the msi...nowhere near what you are getting there.

Weirdness continues

[wiak]

I also just ran winrar...in red mode 3.4ghz cpu, 2500mhz NB and 1000MHZ ram im hitting 1700KB/s on the msi...nowhere near what you are getting there.

Weirdness continues

am getting 1.411KB/s on 2.7ghz, nb 2000, ram 400 in red mode

[LIKMARK]

I also just ran winrar...in red mode 3.4ghz cpu, 2500mhz NB and 1000MHZ ram im hitting 1.700KB/s on the msi...nowhere near what you are getting there.

Weirdness continues

That's Vista right. Sounds like you're running the patch to me

to compare: Xp 32bit with crap memsettings:

[wiak]

That's Vista right. Sounds like you're running the patch to me

to compare: Xp 32bit with crap memsettings:

am on ungaged*

btw, bit-tech tested 9850 and got some nice scores to
http://www.bit-tech.net/hardware/200...9550_b3_cpus/1
and the wreidnessy still continues

[Tony]

That's Vista right. Sounds like you're running the patch to me

to compare: Xp 32bit with crap memsettings:

It gets better:

Yellow mode winrar 1.7ishKB/s like you
Red mode Winrar 1.7Iish KB/s like you
Green winrar 367KB/s

So I can turn the patch on and off, vista is not the issue, something else is speeding up the DFI

Now 3dmark 06 CPU 3.375GHZ, NB 2.5GHZ, ram 5-5-5-15 1000MHZ 4GB

CPU score:

Red 4704
Yellow 4580
Green 4229

Red wins

Cinebench:

Red scores lower than yellow by around 200 points, this is consistent also as I ran it a few times

All I can think of is:

I read in an german forum that in red mode autoxpress features get enabled

The Gigabyte has an option for Autoexpress in bios, the MSI and the DFI do not (as far as I know) BUT I have a feeling the DFI is enabling some tweaks quietly in bios we are not seeing.

My DQ6 is reflashed and in a box winging its way back to me...we will retest this soon also

there is much strangeness with these boards and AOD settings...much strangeness LOL

Edit: had an update which clears some things up.

Yellow mode is default for B3,when its yellow then TLB caching is enabled

If you boot and AOD is showing red then Autoexpress is enabled from bios

If you enable Red in AOD you enable Autoexpress also.

So DFI have some tweaks that leave AOD yellow but speed things up...or their bios is masking what they are doing.

So we need manual settings in bios for Autoexpress now so we have a choice

[massman]

Did anyone notice that core4 is faster than any other core in Cinebench?

[wiak]

Did anyone notice that core4 is faster than any other core in Cinebench?

wreid

[justapost]

win2k 32bit 2.8/2.2 GHz 1.3125/1.27V Mem 5-5-5-16-11 2V ganged vs. unganged

only a marginal difference here 2060 vs. 2100 NB/s.

win2k 32bit 3.0/2.4 GHz 1.4875/1.475V Mem 4-5-5-16-11 2.3V unganged (very hot and unstable setup)

Not really a big gain here ~110MB/s more (5&#37 considering the power the system requires.

@Tony: Thanks for verifying the autoxpress thing on the gbt board.

[wiak]

http://service.futuremark.com/compare?3dm06=6215558
my lastest, phenom in red @ 2.7ghz ganged, 3870 @ 850/1250

[massman]

wreid

Not really, Core1 was faster on the C2Q's as well .

Red/Yellow/Green seems to be influencing the memory bandwidth a lot.

Green: mem read=49xx, mem latency=100ns
Red: mem read=59xx, mem latency=65ns

//edit: need to update bios, switching to red fixes the TLB issue

[wiak]

Not really, Core1 was faster on the C2Q's as well .

Red/Yellow/Green seems to be influencing the memory bandwidth a lot.

Green: mem read=49xx, mem latency=100ns
Red: mem read=59xx, mem latency=65ns

//edit: need to update bios, switching to red fixes the TLB issue

sounds like that TLB still hunts amd since they fixed it

[massman]

Not sure ... I'm using and older bios (1.1) instead of the newest, which has an updated cpu support (1.4). Just read in another thread that only the latest version fixes the TLB issue ...

[KTE]

Where can i find this bios?
Thank you.

Here you go, flash with the correct parameters: http://www.yourfilehost.com/media.ph...376AMS.P0J.zip

I also just ran winrar...in red mode 3.4ghz cpu, 2500mhz NB and 1000MHZ ram im hitting 1.700KB/s on the msi...nowhere near what you are getting there.

Yeah, somethings wrong, YGPM.
The official 1.4 for MSI is bad because many users borked their MB/CPU when they left oc options in. Since then Performance releases are best, and they have AutoXpress in there.
Yeah, we verified a while ago that Red Mode = AutoXpress and it has better perf. for higher end cards especially.

Did anyone notice that core4 is faster than any other core in Cinebench?

Mine is actually a tad slower

Not really, Core1 was faster on the C2Q's as well .

Red/Yellow/Green seems to be influencing the memory bandwidth a lot.

Indeed. Yellow is a good balance [check power with Red ]

[Brother Esau]

I also just ran winrar...in red mode 3.4ghz cpu, 2500mhz NB and 1000MHZ ram im hitting 1.700KB/s on the msi...nowhere near what you are getting there.

Weirdness continues

Yeah no kidding hey? Tony could you please make a suggestion to Oskar to get this darn AMD Overdrive working properly with the DFI Bios? 2.0.17 Is terrible and big issues with Software itself (not Bios) I have installed it twice once with 6400+ BE = NO GO 2 times with Phenom BE = NO GO and it shut my damn machine down just opening the stupid thing just as a note Defaults were set in Bios when Installing and Opening Program

At the bare minimum I would like to see this thing functional on our board and coexist harmoniously without issues I mean just opening the program up from Day one with this board give random 'weirdness" and lock ups let alone trying to use the dam thing!