so the magic number is 2000? to avoid going over I mean. My old X2 systems didn't like going over 1000 or they acted a lot like your descriptions. Have you tried a lower one?
so the magic number is 2000? to avoid going over I mean. My old X2 systems didn't like going over 1000 or they acted a lot like your descriptions. Have you tried a lower one?
with athlon64 cpus, the dfi 790fx-m2r does well over 2000 nb and approaches 400ht
the 1/3/08 beta bios allowed me up to 2100mhz nb with my phenom (that i attempted), but previously could only approach 2000 (maybe 1975 max)... the november bios supposedly allowed over 2100 nb with a phenom and was erased with the 12/11 final release, but i have not tried the november bios
lowering nb and htt does not help me get passed 234 stable
another tweak i could try at my final 234ht is 9x multiplier for nb (giving me 2106) which would improve memory latency significantly...
the 1/3/08 beta bios allowed me up to 2100mhz nb with my phenom (that i attempted), but previously could only approach 2000 (maybe 1975 max)... the november bios supposedly allowed over 2100 nb with a phenom and was erased with the 12/11 final release, but i have not tried the november bios
lowering nb and htt does not help me get passed 234 stable
another tweak i could try at my final 234ht is 9x multiplier for nb (giving me 2106) which would improve memory latency significantly...
ref clock x cpu multi = cpu speed
ref clock x nb multi = nb speed
nb speed / mem divider = mem speed
ref clock x ht multi = ht speed/l3 cache speed
is this right?
I am seeing ppl talk about 240 as the wall with these DFI boards
ref clock x nb multi = nb speed
nb speed / mem divider = mem speed
ref clock x ht multi = ht speed/l3 cache speed
is this right?
I am seeing ppl talk about 240 as the wall with these DFI boards
that pic is with the gigabyte dq6, which is fairing much better with their bios and phenom overclocking than dfi atm... but im sure dfi will eventually get it under control
My DFI 975x has had quite a few bios upgrades. early version of this board had difficulty getting near 400fsb, but with latest bios ppl have pushed up in the 450's.....
i have a bios setting saved in the board that runs 250ht, 2.75ghz, 2000 nb & htt and is stable enough to run prime95 stability tests for hours without errors... i just cant render a divx clip at all or play a game for more than 10 mins before a freeze, and it has frozen on me just opening a webpage... hence why i made this post...
234ht is my max for utter stability... 240ht let's me render divx clips but it has locked up halfway through and i've played 3d games for as long as an hour before i get a freeze... that may be why they say their limit is 240, or ur mileage just varies between cpus
i've used the 250ht saved settings for running f@h overnight, but it lowers f@h frame times to 11m 18secs from 11m 56secs at 234ht... not worth the hassle of loading it and then worrying about it in my sleep if it has locked up (which it never did), but im just gonna run this 234ht setup while i wait for the AT guide and more dfi bioses
234ht is my max for utter stability... 240ht let's me render divx clips but it has locked up halfway through and i've played 3d games for as long as an hour before i get a freeze... that may be why they say their limit is 240, or ur mileage just varies between cpus
i've used the 250ht saved settings for running f@h overnight, but it lowers f@h frame times to 11m 18secs from 11m 56secs at 234ht... not worth the hassle of loading it and then worrying about it in my sleep if it has locked up (which it never did), but im just gonna run this 234ht setup while i wait for the AT guide and more dfi bioses
1 instance... 2 instances gave between 19 minutes and 21 minutes, for an effective 9m-10m frame time, but closing one of the 2 sometimes causes both to end work prematurely... i've lost a 95% and a 57% from this so im not running 2 clients again until it's supported
this is project 2653 times btw
this is project 2653 times btw
I get 11min51sec frames right now out of both of my instances...I am running 2 for the last 4 days....completing 2 units a day with a +4000ppd....
The only hang up I have had is every once in a while the client stops running..no error...no close down of the program just the .exe are not running in task manager anymore and the pcu is idle....I have to log off and log back on....It has done it twice.
The only hang up I have had is every once in a while the client stops running..no error...no close down of the program just the .exe are not running in task manager anymore and the pcu is idle....I have to log off and log back on....It has done it twice.
You would have to ask MarkFW900. I only have a B3 while he has 3 G0 stepping chips. I canhit 3.4ghz max stable for the vcore I wont cross....
I have noticed the F@H program likes system bus speed....
3.3ghz at 11x300 was close to 12min50sec times
3.3ghz at 10x330 is 11min51sec
now the memory was 750ddr2 versus 830ddrr...
I need to test 8x400=3.2ghz with 800ddr2...at 320 with the 4:5 ratio the memory spped would be 800ddr2 and identical to the 8x400. It would let me isolate the bus speed and see for sure.
at 3.2ghz 10x320 I was getting 12min10-15sec.
so u'd be getting around 6-7 minutes for 1 instance at 3.3.... damn sexy
before i did my memory tweak that i stated earlier, i was getting 12m4sec... afterwards (which i only lowered trc/trd from 24t to 16t and a few minor timings) down to 11m56secs... so it's definitely sensitive to memory throughput
EDIT: im gonna try for that 2100 nb setting now... maybe it'll lower to 11m40sec... wish me a safe return
before i did my memory tweak that i stated earlier, i was getting 12m4sec... afterwards (which i only lowered trc/trd from 24t to 16t and a few minor timings) down to 11m56secs... so it's definitely sensitive to memory throughput
EDIT: im gonna try for that 2100 nb setting now... maybe it'll lower to 11m40sec... wish me a safe return
I would like to see if cache size will have an impact when the 6mb cache cores come out. I was one of the first to notice huge gains from the 2mb allendale cores to my 4mb cores when running dual instances of certain WU's....Back then I didn't run smp client. I would load 4 instances of folding at home.
The ram I have hasn't been fully tweaked yet. It is running spd at cas 5-5-5-18, but is rated for 4-4-4-12. I dont think it will do much then 900mhz so I I try for 3.4ghz on this board with 8x412 my memory will be 825ddr2 at best...
The ram I have hasn't been fully tweaked yet. It is running spd at cas 5-5-5-18, but is rated for 4-4-4-12. I dont think it will do much then 900mhz so I I try for 3.4ghz on this board with 8x412 my memory will be 825ddr2 at best...
Markfw
Moderator Emeritus, Elite Member
- May 16, 2002
- 27,309
- 16,143
- 136
The link you posted had to due with stock systems, not overclocked systems. And after you have now insulted me, I leave you to you problems, and wish you all the failure you deserve. When you want real advice that works, let me know.
I on the other hand have no problems, but you want advice, but refuse to listen to it. Also, the link you gave did not refute what I say.
I on the other hand have no problems, but you want advice, but refuse to listen to it. Also, the link you gave did not refute what I say.
Markfw
Moderator Emeritus, Elite Member
- May 16, 2002
- 27,309
- 16,143
- 136
First, if you want advice, you need to stop insulting people. Next you need to stop making assumptions.
At 3550 (my current fastest) with 2 smp clients, running ubuntu, I am getting about 10 minutes 30 seconds for each client while running 2 instances of F@H. You can't extrapolate to 5:15 each... Below is part of my log for one of two clients:
[05:57:15] Completed 175000 out of 500000 steps (35 percent)
[06:07:45] Writing local files
[06:07:45] Completed 180000 out of 500000 steps (36 percent)
[06:18:17] Writing local files
[06:18:17] Completed 185000 out of 500000 steps (37 percent)
[06:28:47] Writing local files
[06:28:47] Completed 190000 out of 500000 steps (38 percent)
[06:39:18] Writing local files
[06:39:18] Completed 195000 out of 500000 steps (39 percent)
[06:49:48] Writing local files
[06:49:48] Completed 200000 out of 500000 steps (40 percent)
[07:00:20] Writing local files
[07:00:20] Completed 205000 out of 500000 steps (41 percent)
[07:10:50] Writing local files
[07:10:50] Completed 210000 out of 500000 steps (42 percent)
noone extrapolated to half... 6-7 mins is 12-14m... did u want me to say 6:15-6:50 instead??
there is something u just dont understand...
ur 1st post is vdroop
i reply it's there by design
u ask for proof
thor86 says my benchmarks mean nothing on an unstable system, even tho i clearly write "these specs are my maximum stable settings, meaning the system can be fully multitasked as though u were running stock settings with 100% peace of mind"
i reply to him asking why it isnt stable
U reply with "you did", and duvie follows u with agreement and calls my vdroop remark lame
at this point how can i not feel that choosing this site to post was a mistake since noone can understood what i was writing and people with THOUSANDS of posts are throwing opinions out that clearly contradict articles written on this very same site... not to mention i have not insulted anyone, only received it
now here at the end u wish me bad luck that i deserve... thank you so much... as a follow-up, what makes u so egotistical that i am hoping to get more "help" from u after reading what u wrote in this thread?? obviously im not saying u dont have much to share as we dont know each other at all, but seriously how can i be wishing for u to make another post in this thread atm?
there is something u just dont understand...
ur 1st post is vdroop
i reply it's there by design
u ask for proof
thor86 says my benchmarks mean nothing on an unstable system, even tho i clearly write "these specs are my maximum stable settings, meaning the system can be fully multitasked as though u were running stock settings with 100% peace of mind"
i reply to him asking why it isnt stable
U reply with "you did", and duvie follows u with agreement and calls my vdroop remark lame
at this point how can i not feel that choosing this site to post was a mistake since noone can understood what i was writing and people with THOUSANDS of posts are throwing opinions out that clearly contradict articles written on this very same site... not to mention i have not insulted anyone, only received it
now here at the end u wish me bad luck that i deserve... thank you so much... as a follow-up, what makes u so egotistical that i am hoping to get more "help" from u after reading what u wrote in this thread?? obviously im not saying u dont have much to share as we dont know each other at all, but seriously how can i be wishing for u to make another post in this thread atm?
nb 9x multi and 2000+mhz is not possible yet, only with the 8x multi... there are half multi's in the bios for nb but it currently does not work either
Markfw
Moderator Emeritus, Elite Member
- May 16, 2002
- 27,309
- 16,143
- 136
The whole point he was trying to make was that the motherboard would do that, so at idle it didn;t over vcore the cpu. When you OC, you almost always up your vcore. And its still a lame argument, when many of the best reviewers use the X38-DQ6 or P35-DQ6 to do OC'ed reviews, since it is such a good motherboard, and has no vdroop. So he contends it is a feature to have a power delivery system that can't regulate well ? but the best motherboards do ? I stand by my statements, vdroop is bad for stability. In his chart, if the vcore was a straight line, it would have been great.
im not here to prove i have a stable 234 ref dfi 790fx on phenom... that's not even anything to brag about... im here to describe my type of instability i have past 234 ref cause it's so damn weird i have managed 250ref prime95 stable (blend and in-place) for 16+ hours but the system locks up 5 - 15 seconds after i close prime95, or i cant play any games or render divx videos in particular
noone is trying to brag about any system here... if anything im damn jealous i didnt wait to save the money for q6600+mobo to buy at the same time instead of getting phenom for my m2n-e, and later getting this dfi
that's all this is about!! stop with the egos, im not trying to prove anything
edit: Now that I have read over the posts again, I can see that most of the comments are made in jest.
It is nice to see how a Phenom rig works in a non-review type situation.
It is nice to see how a Phenom rig works in a non-review type situation.
Thor86
Diamond Member
- May 3, 2001
- 7,888
- 7
- 81
I'm not complaining about your bragging rights, I just want to see if you are using all 4 cores fully loaded when doing Prime95 blend tests.
You never mention this, that is why I suggested what I posted before. Sheesh, people take things waay too seriously these days. You should be having fun, and not bitch at people who are trying to help you out. :frown:
ok so u think im stable in prime because only 1 thread is running... now i understand where u wanted to go... im using prime95 25.5, it makes 4 threads automatically and sets their affinity
i have SM Bus Controller with yellow exclamation in system devices.... and i am finding nothing on the web for a driver... from what im reading, this could be huge for stability
i have SM Bus Controller with yellow exclamation in system devices.... and i am finding nothing on the web for a driver... from what im reading, this could be huge for stability
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter