I agree, mine is can do 367 on air.
Printable View
I agree, mine is can do 367 on air.
My Crosshair IV and 1055T combo are acting up...
I have the following settings applied:
In the first boot and ONLY in the first boot, after Windows pops up, it works for a while and then, inevitably after a few seconds or minutes i get a BSOD. I then shut down the power supply in the power button, let the motherboard "discharge" and power on the PSU once again. I press the power button and it boots happily to never crash again until i leave it off for a good 6-7 hours time.Code:CPU Ratio Setting 13.5x
Turbo Boost Disabled
CPU Bus Speed 300
PCIE Clock 100
DRAM Clock 1600
CPU/NB Auto
HT Link Auto
(DRAM SETTINGS) 8-8-8-24-1T (all the rest is Auto)
CPU Voltage 1.4750000
CPU/NB Voltage 1.2000000
CPU VDDA Auto
DRAM Voltage 1.6500000
HT Voltage 1.3250000
NB Voltage 1.3250000
NB 1.8v Voltage 1.8285000
SB Voltage Auto
VDDR Auto
VDDPCIE Auto
DRAM Ctrl Ref Voltage Auto
DRAM Data Ref Voltage Auto
Load Line Calibration Enabled
CPU/NB Load Line Cal Enabled
CPU Spread Spectrum Disabled
PCIE Spread Spectrum Disabled
Also, RAM is in "UNGANGED" mode and INTERLEAVE in Auto
If i turn the PC off and back on 3-4 hours later, this doesn't happen. It does happen if i turn it off now, go to bed and turn it back on - BSOD after windows loads as i described above.
The RAM i'm using is GSKILL Trident 2x2GB PC3-12800 8-8-8-24 at 1.60v.
Any ideas why this is happening? :shrug:
make a bootable USB key via the HP tool, copy the dos files to it and you can flash firmwares biosses etc... eg linkie here : http://www.lowfps.com/creating-a-boo...sb-thumb-drive
Brian:)
Quite right on the ability to run 350Mhz FSB. Up to that point it is very easy utilizing the right ram divider.
On the FSB/ram ratio's, wouldn't 1-1 be almost impossible? You would need an 800Mhz HT and 1600Mhz ram(800Mhz CPUz) to be at 1-1. 400HT and 1600Mhz ram would be 1-2 ratio, which would be about the best you could attain on this architecture.
If AMD could have only refined the NB even further would have been a major performance advance on this architecture. Having the cache running at 1-1 at CPU speed in itself would take us several steps up, with the memory at 1-2.
I too am perplexed at the Asus boards inability to scale HT/FSB above 350-360Mhz. I figured it had to be the board and/or bios. When I tested for DFI back in the NF4 era alot of the NF4 boards had trouble pushing high HT. Abit, MSI, ECS...I had a bunch of them and none would do much better than mid 350's. The early DFI NF4 would not boot up at 400Mhz HT or higher, but we could push it up booting at a lessor HT and then using SetFSB to take it where we wanted it.
Subsequent boards had registers in the bios adjusted based on our feedback data and could boot at 500Mhz HT and higher.
Are you working on the Asus or Gigabyte Brian, or both? Can you boot at 400+ or are you setting it after boot into Windows?? Thanks in advance my friend:)
Best regards,
Randi:D
Randi, I work on everything and anything, just stating thuban itself is capable.
420 HT i'm referring to is a boot stable/benchable. I hate that software crap.......it's misleading.
400 HT will give users 1600 ram IIRC with the 800 ram divider. More importantly that equals to a nice 4000 usable NB for the cold guys.
Here's a shot for you that I was booting at.
Everything was eliminated from equasion partly due to using a stock x4 cooler.
http://chew.ln2cooling.com/AMD%20build1/420ht2.jpg
Excellent my man:up:
Thats what I wanted to hear re: that you are booting into Windows at 400+HT. 400HT+ and 1600Mhz Memory+ and 4000Mhz NB will give you guys running LN2 a decided boost at 6000Mhz + CPU speed. With the improved NB where alot of the guys here are at 3200Mhz or a little better, the 400HT and 1600 Mem will also give them a nice boost in bandwidth at moderate overclocks.
I was working with mine this morning at low CPU mhz, but on the Asus CF IV w/ 801 bios anything over 365ish is a no go. I've got the bios here on my laptop to look at right now and tinker with. 400HTT and the 800 divider are important. I tried maintaining 1-2 ratio with 333HT/1333 Mem with tight timings but even with a 3500Mhz NB the system bandwidth was not as high as I last posted in my own thread.
Great work Brian, looking forward to great things:)
Randi:D
Do me a favour and next time it crashes , reboot and go into the bios and read out the CPU NB voltage plz... if it's lower then the voltage you applied report back... though I must say 1.2CPU/NB voltage is already low... Try upping to 1.25-1.3ish
I made a small homevideo of my NB voltage issue I've been experiencing since 0707 bios... passed it on to the big shots at Asus...
Thank you Leeghoofd, i'll do that in the morning as it is the next time it will exhibit that behaviour :up:
I'll try upping the CPU/NB as you mentioned.
Seeing this is turning in to a general CH4 topic, isnt it smart to change the thread name and such?
Well I've been busy trying to persuade Jack C for weeks about it, he still says it's safe mode all the time Brian... some people never listen... so I won't rest till it gets solved or it pops up in the review next week...
I need people to back me up... to get them moving !
Not sure if you remember but m4a79_T deluxe ddr III had the same exact issue but with vcore. Work around was to not set vcore to ignore so it would peg the overvolt warning every boot.
When it stopped warning you that you were over volted you knew your vcore had dropped. I think that issue and this issue may be a form of OCP kicking in which i also found to be the case on the m4a89TD Pro.
Sure they will get it nailed down but i've already moved on and am working on other things to better end users experience with locked chips now :D
Another nice one. I was just pushing HT ref clock to see how high it would go. At 330 (stock ht and NB volts) it refused to boot and i had to press reset twice. After that it booted at default settings but all settings including OC profiles were gone!
The only thing that remained were clock and date settings. :( Has anyone else had this before?
I had a slimilar thing once .
all settings went back to auto , and lost my first 4 saved oc profiles in the bios . the last ones were still there .
this was with 0801 bios .
Also i need to powercycle my psu every time an overclock locks up . :down:
Yah thats not uncommon. I have run into that issue on asus and gigabyte.
When pushing the limits things like that are to be expected. Luckily im a bios setting crack fiend. I know all my settings by heart for every board i have :ROTF:
It's certainly an issue; quite funny how some people refuse to acknowledge there's a problem :D.
This whole problem makes me feel like there's a VID setting issue. Have you tried to replicate this behavior using the offset voltage mode (remember: I assigned you this as homework!)? If there's no problem with the offsets, it's definitly a VID table screw-up ... something which has been on some AMD mainboards in the past.
Well, I just got my first message on the subject, but I assure you it will be addressed tomorrow night with engineering. ;) In the meantime, I must have a perfect setup on four boards here as I started testing it and have not been able to replicate it yet. Albrecht is going to send me his OC profile so I will start from that point. Of course that means I have to roll back to a crappy CPU stepping to match his system. :rofl:
What is HT link speed set at? If it is auto, go in and back it off to 2000MHz as you clock HT ref up. Also, cooling and CPU quality makes a difference in this area but with decent water cooling or really good air, I can clock my good 1055T up to around 410 HT ref clock. I have to really fine tune the voltages (HT, CPU VID, CPU-NB VID, and NB/SB VIDs) around 370~380 to get that last bit out of the board with the Thubans.
Any chance for us with 1055T to be able to use a multiplier of 16.5 if we disable turbo? Would love that feature!
Just read this.
I'm not sure it's even possible to do it using the existing DDR2 memory dividers. Waaaaaay back when the first DDR3 boards were launched, I already asked some manufacturers to look into this, but most of them reported it's not as easy as just opening up the 'locked' dividers.
Check out page 185 of the BKDG of AMD's 10h. In short, the type of DDR memory detected determines what memory dividers are available. When DDR3 is used, the Ddr3Mode register is set to 1 thus enabling 400/533/667/800 dividers.
Do note that there are unused (reserved) divider registry keys available, but as far as I understand they are locked and not useable.