I don't have the tools to check but when I tested L3 cache with the TLB patch, the L3 cache was not picked up and inaccessible. As if it or a major portion of it had been completely bypassed or.. disabled.
Thus, NB speed should make a good deal of difference. Just run EVEREST L3 cache bench at 1400 vs. 1800 and see the difference.
LIKMARK... can you give me cpu-z dump
If after reboot the NB clock does not stick... that mean your NB speed is locked...
Lol. I can do 7GHz NB too...
Cpuz dump is with tlb patch fix with CrystalCPUID and reboot, still no effect on Winrar bench, and "on the fly" nb multi change with BAR-EDIT(no boot).
You think there could be any way to unlock NB multi?
You have to reboot for the change Likmark.
9500/9600 >9x NB = no way.
9500/9600 <9x NB = yep.
Can you check if AOD shows that change without rebooting too?
EDIT: wait, when did you apply the MSR value change, before or after the reboot?
It should be after. As I said, don't try higher than 9x you're wasting your time as the NB isn't changing from stock at all. Physically it won't regardless of what the register value reads so your performance is just like that of 1800. Just try 7x and 9x and compare performance and you'll see.
Last edited by KTE; 02-26-2008 at 09:02 AM.
I know, I know, when I reboot, it's back to stock speed. IE my nb multi is locked to 9x. That's kinda sad but expected.
AOD shows 7GHz NB too, i'll test rebooting one last time just to check.
Edit: Yes, it resets itself at reboot.
Last edited by LIKMARK; 02-26-2008 at 09:13 AM.
Hmm.. that's the same issue as mine. Although I haven't touched my MSR's and they go higher than that, far higher, and mine were automatic bootup values (BIOS has no option to oc NB). The NB is still at only 9x in my instance though, pretty obvious if you look at performance and I already tested real NB multi change, it needs a lot of volts and the performance gets much better in everything. Plus I already know from many tests real NB speed at 2.4GHz starts getting difficult and 2.6GHz NB (real) is the absolute limit for air on mine even with 0C ambient and 1.58V.
Seems like the registers in Phenoms are basically changing on the fly but the actual speed isn't so software which reads those register values is also reporting wrong values for CPU/NB/HT/RAM.
Oh I totally forgot. You're on an AM2 board which are known to give these problems (8GHz CPU). But I'm on a AM2+ which isn't usually known giving these problems apart from Polygons. Try changing CPU multi too, I bet software shows it changed all the way to >7 or so GHz.
I think the only thing i really can change in my bios is the ht ref clock. I believe this because my system bsods if I push it above 227mhz. I can change cpu multi in AOD, downwards only, naturally, but it should bsod if I put it too low I believe.
EDIT: Or not.
Last edited by LIKMARK; 02-26-2008 at 09:43 AM.
Anyone seen where to do a vcoremod on the Crosshair?
Depends how low really. Should be able to go down to 800MHz max at least with a BE and 1100MHz at least with a 95/96.Originally Posted by LIKMARK
The Crosshair is physically the same board as the P965 Commando, apart from the CPU socket differences and luxury extra's. Check the mods for that, they may very well apply. You can easily find and do a mod yourself though, here's one starter guide: http://www.aoaforums.com/forum/hardw...mod-guide.html
Very easy really just depends if you have the tools and time to waste... or if you have nothing else to do. :p
What are you after though and why?
Tried down to 1000MHz, all stable(only changing multi).
Bought the tools some time ago, just need some potentiometers and resistors.. No, I don't have time for it really (my wife tells me, not happy about amount of time spent in front of the computer..). I'm quite lazy, hence my question
I'm quite sure making good Phenom BIOS'es for Crosshair ain't on top of Asus' list these days, so I might do the voltage regulation manually as there is no voltage regulation control in the current BIOS.
I'm gonna check the commando mods, sure some Intel guys did this a couple of years ago..
Thanks for the link btw
I still don't understand one thing though..
Are you trying to access more CPU voltage? If so, your board won't be the limit there hardware wise, it's your BIOS.
But if you're after lessening Vdroop and Vdrop then yep, you'd need to mod them physically after finding the FB pin.
Change VCore and CPU VID like this in Windows, try it (it's real-time): http://www.xtremesystems.org/forums/...&postcount=554
Well, I thought to do both things actually, but as you say accessing more vcore is (of course) bios limited..
I'll test doing it realtime in windows.
Sorry, 9500 won't allow more than 1.25VID (VID locked, which means no more voltage than what BIOS allows).
Check this post out and add this link/pic to the first post please: http://www.xtremesystems.org/forums/...postcount=1448
It's very valuable, AMD/MB MFGs are hearing us. Actually, thank Sami for initiating this, MB MFGs can be very hard to work with.
Second post updated.
Check the link out in my linked post, and it'd be useful to post that as it contains the description of the fields in the table (they're TLB Fix enable/disable through the BIOS fields, whether AMD tested it to work or MB MFG, or both).
LIKMARK: Add these to the first post please, all AM2 which officially fully support Phenom X4 CPUs.
ASRock:
ALiveDual-eSATA2
ALiveN570SLI-eSATA2
ALiveNF5-eSATA2+ R3.0
ALiveNF5SLI-1394
ALiveNF5-VSTA
ALiveNF5-VSTA R2.0
ALiveNF6G-GLAN
ALiveNF6G-VSTA
ALiveNF7G-HD720p R1.0
ALiveNF7G-HD720p R3.0
ALiveNF7G-HD720p R5.0
ALiveXFire-eSATA2 R3.0
AM2NF3-VSTA
ALiveNF7G-FullHD R1.0
ALiveNF7G-HDready
Even a 480X in there
new crosshair bios here
any chance that the MSI K9A platinum will support a phenom 9x50?
not yet: http://global.msi.com.tw/index.php?f...&cat3_no=#menu
I don't know if they are updating bioses for Phenom on that board
My buddies ECS A770M just died with a 9850.
"Angels on the sideline, Puzzled and amused.
Why did Father give these humans free will?
Now they’re all confused."
9850(JAAFB AA 0810BPAW), DFI 790FX M2RS(3/05 Bios), G.SKILL 2 x 2GB 1T-5-5-5-15, OCZ 600, MINI NINJA, WD SE16 AAKS 320, HIS3850, bluegears b-Enspirer, Vista 64
Yes. If it's a board with less than five pwms the board have to have quality pwms (and probably good cooling) to handle a 9850.
He added cooling but I just thought I'd post so you could note.
"Angels on the sideline, Puzzled and amused.
Why did Father give these humans free will?
Now they’re all confused."
9850(JAAFB AA 0810BPAW), DFI 790FX M2RS(3/05 Bios), G.SKILL 2 x 2GB 1T-5-5-5-15, OCZ 600, MINI NINJA, WD SE16 AAKS 320, HIS3850, bluegears b-Enspirer, Vista 64
Thanks for the heads up, I recon it was the mosfets that popped?
VRMs are rated for the amperage, frequency and efficiency they work at under specified ambient temperatures. The motherboard PCB heat capacity is a large limiting factor for their performance, being limited to 105C quite often - at max the components themselves have to stay below 125C to perform within the accepted amperage and temperatures. Their biggest impedance as with any IC is temperature.
I'm not aware of the VRMs used on many of these MBs but a typical 3-phase has maximum efficiency at 50A whilst a typical 4-phase PWM has a maximum efficency at 60A at 250-400kHz at 25C ambients with 200 LFM air flow. After that the power loss increases rapidly and hence the efficiency and heat of the IC and PCB both increase in the negative way. The 6-phase VRMs on the other hand only start to become highly efficient from 60-70A output but their efficiency decline, hence power leakage is much slower and gradual. It's the typical problem found in SMPS really.
If you have a good enough LFM over the PWMs, more than the conventional 200 LFM found in desktop systems, they will increase in efficiency, power delivery and lifetime - 3-phase circuitry won't do too badly as long as the current is not out of spec for the components or of the inductors being used.
they need to its one of the top end AM2 boards from them it would be stupid not to...
Posting Permissions
Reply With Quote
Bookmarks