Yeah, found the same here but not at my fully stable daily clocks, only when i started pushing the cpu.Quote:
Originally Posted by Oj101
At 22x204 in bios it fluctuated between 204 and 206 bus speed. :shrug:
Printable View
Yeah, found the same here but not at my fully stable daily clocks, only when i started pushing the cpu.Quote:
Originally Posted by Oj101
At 22x204 in bios it fluctuated between 204 and 206 bus speed. :shrug:
Same issue with the 2 HT mhz jumping up and down, will check later today if i only have it when pushing.
Already mailed asus, but getting some standard story back that i should use factory settings for the bios.
factory settings ? :shocked:
yeah, dont really know what he means with that either. thought 6ghz+ was factory? :D
But it's bloody annoying, cant push properly with this flaw :(
In general my CH IV issues : the HT jumps indeed are too big, Plus the final HT speed is far from stable either. On top of that I got CPU/NB voltage issues on my sample board since 0707 bios (mostly only a cold boot gives me correct values again)... Efficiency at high ram speeds is horrible compared to 790FX boards... was about 5-8secs difference in pi32mb with a 790FX and 890FX at +1900mhzC7. I'm thinking this high ram speed stuff is overrated...
Use the Turbo V tool Charged, that one still works fine for me... sadly I can't select the cores via the ROG connect tool on my lappie, using K10 stats instead...
Hi my friend:)Quote:
Originally Posted by Leeghoofd
There is definitely a ways to go yet with bandwidth efficiency with the CH4 bios. In saying that, it still holds true that to take advantage of high ram frequency, we still need the capability to run the NB at twice the ram frequency. As with earlier Phenom II's, it is next to impossible unless under LN2. With my current Thuban, a little over 3200Mhz is max even under a 404A Prommy, which at 2 to 1 makes 1600Mhz ram speeds with tight timings the most efficient theoretically.
The inefficiency of the mem subsystem archictecture harkens back to the original Athlon days, when the L2 was off die. When it was moved on die, and finally at full processor speed we finally had something. Then the A64 architecture took it a step further and we were eating P4's for breakfast:) I have not gotten too excited with the 2000Mhz+ speeds on this platform yet because of the inability to scale the NB higher. Without that ability, high ram frequency, it seems, or the ability to run it, is only eye candy and this platform is not fully capable of utilizing the bandwidth that it can produce.
Best regards,
Randi:D
I though using 70% nb vs core speed was best ?
I see everyone with thuban is using 75%+..
now the other problem with that is not having enough core speed to take advantage of the NB speed. seem like the cores start to bottleneck it then.
with 4000mhz NB and 2000Mhz ram you need like 5.7ghz cores ? lol better off for Ln2 users
Agree with you there Randi, i'd like to add that on top of high CPU-NB speed you will need very high speed on the cores as well to max things out. In other words rams at DDR 2000, CPU-NB at 3.2G will show less bandwidth at 20x200 than at 30x200 for example.
For most ppl over here who are using air or watercooling, DDR1800 w/ tight timings will be more than enough as the CPU core limit won't exceed 4.5-4.6G -- or will it? :D
Zeus:)Quote:
Originally Posted by Zeus
Could you imagine if the design incorprated the ability to have CPU Mhz and NB Mhz 1-1. At stock speeds you could use 1333 ram at tight timings with excellent bandwidth, over 4Ghz CPU step up to 1600Mhz tight timings and on up the ladder. Performance would be, I believe phenomenal(No pun intended:)) Might be a bit closer to Intel in that regard.
Best regards,
Randi:D
Thanks for being my friend, you are my first friend here :) (besides Zeus but he wants my memory kit)Quote:
Originally Posted by MrIcee
I know Randi, but everybody links 890FX and Thuban with 2000mhz ram speeds... my testing showed 1) to get the rams at 2000mhz you have to cripple them a bit ( subtiming tweaking) and then 2) efficiency is far fetched at that moment... for daily users indeed go with 1600-1800freq range and with tight timings.... even for the LN2 guys I wonder what difference a Pi32mb run at eg 6Ghz, +4Ghz NB speed between 1800-1900C6 and 2000C7 will be... Will have to find the latter out for my team mate's compo next week...
i've tried fast, gotten up to 1030mhz CL7-7-7-20-24-1T @ 90ns with pretty normal subtimings. this is @ 1.7V with dominator GT 2000CL8 on a 0801 bios CH4 with a 1090T.
pi32m ? 1m is 2100mhz stable, Pi32mb is another story...
Had to 1st type:Quote:
Originally Posted by Forsaken1
atiflash -unlockrom 0
Then normal flash code.
Good to go now.
Oh nah, sorry. just 1M indeed :) was fiddling about for a bit.Quote:
Originally Posted by Leeghoofd
Oh and, i get the 2mhz jump all the time, i was at like 4ghz or so and i still had the issue. it's starting to get on my nerves now actually :(
Same here, but with M4A89GTD. I hope it will be fixed in the next bios release.Quote:
Originally Posted by thijs
I think it was stable with some of the previous versions, maybe I should check it,
but a manual reflashing is necessary, because the new version keeps telling me that my board is newer than the bios I want to update with. It's obvious that this is not true :).
EZ Flash was working well with the previous 3 releases while flashing back to older version.
"Childish" problems that occur with Crosshair IV keep me waiting for newer bios version, or the Extreme variant. Will wait some more time and see...
I have had a couple mhz swings on about every board I own.. why is this so wierd?
Corrupted bios on CH IV, now always "new cpu installed" @ post:(
Can't set oc settings on memory, multi and nb from bios.
Waiting for bios chip from Asus.
Ouch Dumo, you don't seem to have much luck with the Crosshair IV, do you? Did you try bios recovery method? Like putting the bios on a thumbdrive or a cd, put it in and just boot? I think it's called Asus crashfree bios option or something.
Thank god Asus decided to put the bios chip in a socket again insted of soldering it straight onto the pcb.
Try using the board in a eg 4ghz limited competition, it's near impossible to keep the speeds under control... my board varies about +/-2-3 Ht clock, imagine with a 30 multi... that's more than 60-80Mhz difference, same goes for ya rams max clocks... makes the board pretty hard to stabilise....Quote:
Originally Posted by charged3800z24
Exactly, if you are pushing it, 50mhz up and down kind of prevents you from getting the max out of it.
I will try a few bios versions later tonight / tomorrow and see if i can get some improvement. really hope they will fix it fast :)
Oh, on the HT clock, I thouht you ment core clocks. Well, since I don't get to compete, doesn't bother me as much:p:Quote:
Originally Posted by Leeghoofd
Of course I cannot get my board to do it, you must live in a very unique place that has some sort of time/space distortion going on. :rofl: However, we were able to replicate it elsewhere so we are looking into it now for the next BIOS release. ;)Quote:
Originally Posted by Leeghoofd
I will try it, thanks. I am so use to AOD any more:p:Quote:
Originally Posted by Leeghoofd
On another note, has anyone had any xtreme undervolting going on when OCing? I mean I just started getting 1.5 volt droops last night when running WCG @ 4.05ghz. I went into the bios and up'd the bolts .5 and not it is over volting under load.... Not sure what changed.
Give me your board, i'll give you mine. :DQuote:
Originally Posted by bingo13
Yesterday when i pushed it some it was fluctuating again, about 2MHz ht.
At my daily 20x205 the board seems stable though.
Nice car in your avatar, is that a Mustang Shelby?
Not sure why you guys are crippling sub's. I'm not.Quote:
Originally Posted by Leeghoofd
Goona do some testing soon at high ram verus tight soon on ln2. ty to make heads or tails of it but I do know for a fact that at normal user speeds 2000 ram is a waste.
I did a test on d9 before with 5-5-5 versus 6-6-6 and on air 5-5-5 kicked 6-6-6 ass.......when cpu was at 6 gig it was a totally diff ballgame, 5-5-5 got it's ass handed to it.
1m is 2200 stable :PQuote:
Originally Posted by Leeghoofd