Printable View
I guess you just run two instances of a single threaded application like Prime95 (see how hot you get!) or SuperPi. I think two instances of Prime95 will tell you whether your setup has sufficient cooling. My guess is that air cooling temperatures on overclocked processors will run through the roof.Quote:
How do you engage each core separately so you know what is running what??
Two instances of Prime....Quote:
Originally Posted by joe2004
Isn't there a little trick----make a second shortcut and then add something at the end of the executable on that shortcut-----something like that....
You also have to set something in the prefs if I remember....
Am not looking for massive overclock-----just a goal of 2.8 and am hoping for voltage around 1.55-----I use tccd and will do a low FSB/High Multi overclock so my VDIMM should be around 2.75. Looking for around 235 FSB with a 12 multi. I've been getting VERY stable results form this apparoach from San Diegos and Winnies in comparison to higher HTTs and lower multi overclocks----and while the benches are just a teeny bit slower because of RAM bandwidth-----the CPU power is the same----and the gains in stability are tremendous
My 120 x 38 Panaflo and an XP 120 HS ain't bad at all for air.....
Should be fun....
Thanks
John
Copy prime folder to somewhere... make short cut that makes it 2 prime 95 you can run.
You can set processor affinity from preference or go to task manager and set it.
As for aircooling, I am using Thermaltake Big typhoon with stock fan atm and it is easy to see 60c dual priming 1.57v @ 2.65ghz... with ambient temp of 25-27c. So, it is hotter than venice once voltage is there. (on stock though it runs even cooler than venice of same speed simply because stock vid is at 1.3v instead of 1.35 or 1.4v on venice) :P
Just download prime2004, you can select which core you want each instance to operate from or it will share if you have "no afinity" selected. I also noticed you can bring up taskmanager and select the core you want a particular app to run from if you like.Quote:
Originally Posted by johnrr6
Got my 4400+ on Saturday and I've definately got 1 duff core and one good one.
Core 0 hit 2.6GHz at 1.4v 12 hours Prime95 small FFT stable
Core 1 needed 1.57v to hit the same speed and be Prime95 stable
Core 0 is happy all the way up to 2.8GHz at 1.57v but Core 1 bombs at anything over 2.6GHz no matter what the voltage.
I'm running 2 instances of Prime95 small FFT over 24 hours (one bound to each core) to see if that will sort it.
Rilot,
I'm guessing you must be Rilot from OCuk :)
What's the stepping and week info on yours?
I've yet to test my 4400+.
Yup. That's me.
Same stepping and week as NickK so: CCBWE 0517MPMW
How do you do that exactly? I don't see that option? :confused:Quote:
Originally Posted by highoctane
Right mouse on a process, select "affinity"Quote:
Originally Posted by xgman
Anyone notice a drop in Sandra mem scores when changing form single to dual core? Seems like Sandra memory benches aren't allocatig the memory as it was on single core, loosing quite a bit of bandwitch on the test. Is there a better memory benchmark test that is appropriate for dual cores?
hmm. never saw that option. I'll have to give it a try. John asked a good question though. What if you have 6 apps open/running at one time. What happens between the cores automatically?Quote:
Originally Posted by rilot
It's cuz you went from an FX55's true 128bit memory controller to the x2's 64 bit controller ;) The FX55 will ALWAYS be faster in terms of bandwidth at the same speeds/settings.Quote:
Originally Posted by xgman
Hmm, I did not know there was that difference. I guess I've been on the FX boat for so long it didn't occur to me. I wonder how much real world difference that makes?Quote:
Originally Posted by s7e9h3n
So it's like this quote:Quote:
Originally Posted by s7e9h3n
"The one limitation that both AMD and Intel have is bandwidth. In order to maintain compatibility with present day Socket-940 and Socket-939 motherboards, AMD could not increase the pincount of their dual core processors. The benefit is that AMD's dual core CPUs will work in almost all Socket-940 and Socket-939 motherboards (more on this later), but the downside is that the memory bus remains unchanged at 128-bits wide and supports a maximum memory speed of DDR400. So, while single core Athlon 64 and Opteron CPUs get a full 6.4GB/s of memory bandwidth, today's dual core CPUs are given the same memory bandwidth to share among two cores instead of one."
However that's like saying a 3700 isn't as fast as an FX55/57 - completely ignoring the fact that the chip is dual core and the way it does things suits a different purpose better.Quote:
Originally Posted by s7e9h3n
In the end bandwidth means how fast you can get data/code to/from the memory when requested. If the FX55 is running at a multi that allows the cache to work to ensure every possible capacity of that memory bandwidth can be used then your right.
However that's dependent on the code instruction distribution.
X2 with it's limited multi doesn't use it's cache in the same way - instead the second core has the chance to get in and use the memory bandwidth. So between the cores the X2 should use more of the bandwidth - if the application is processing the same work but in a multithreaded fashion it's quite capable that the X2 will equal the FX55's use of memory bandwidth.
Single core vs. one of the X2's cores - your correct. However, as weve seen from multithreaded apps the X2 does a pretty good job of matching all but the extreme overclocked FX55/57.
Am I missing something? DC on all s939 MB's list as 128-bit.
He's taking the pish :D :pQuote:
Originally Posted by ewitte
Basically as the dual cores have a single memory bus to fight over, a simplistic statement is that it's equivilent to a pseudo 64bit memory bus.
A dual core has 128 bit access to memory. However it may have to wait for the other core's memory bus dependant operation to complete first.
Just a quick sidestep from all this ocing talk for anyone else with an X2 CPU...
Device Manager -> (your pc name) -> Computer -> xxxxxxxxx
What computer type have you got in the xxxxxx area above?
Have you got ACPI Multiprocessor or MPS Multiprocessor?
:)
I forced "ACPI Multiprocessor" so thats what shows up ;)Quote:
Originally Posted by [timko]
ACPI Multiprocessor PCQuote:
Originally Posted by [timko]
and use these
http://www.amd.com/us-en/Processors/..._13118,00.html
ACPI Multiprocessor too - using AMD's X2 driver too.
4800+ X2 in and running
Got my 4800+ X2 in my DFI rig and she fired right up....... Bios 623-3
Made 2.8 ridiculously easy at 1.42 vcore-----HTT 234 x 12 Multi
Ran two superPi's simultaneously----1 meg did 30 seconds each
Ran two Primes----Core 1 Keeps chugging along----Core 2 drops after about an hour so I'll have to tweak a bit. Temps get as high as 50C under stress and I'm on air.....I have a whole bunch more stability testing but think I'm in the ballpark with a goal of a 2.8 overclock on air....
Wasn't there a way to set memory for prime95 so they didn't share when running two instances??
This dual core stuff is a whole different ballgame.....
Ran 7 apps simultaneously and alt tabbed between them------it was instantaneous----no lag at all....
Need to test it while one core is REALLY busy with something demanding----say rendering a DVD----see how fast it switches and responds then.
Can someone explain to me how the cores respond???
When does Core 2 kick in??? At a certain percent of CPU load??? When a new application requiring a certain amount of resources starts up???
Whats the rhyme or reason a certain core would do a certain app???
In other words----can someone explain to me how the pie is divided when there are say 7 apps running????
John
John, glad you like the 4800+. If you really want to see you proc perform run the new PC Mark05. At one point it runs 5 or 6 different apps at once. PCMark is a good test for the multi-processor system.
What was stepping and date of your processor? :up:
Congrats john. I have the same questions. I'm not sure anyone other than Bill Gates knows for sure. ;)Quote:
Originally Posted by johnrr6
I don't think you can get 2x prime not to share memory. I'm also not so sure we should be all that hung up on why that 2nd instance of prime fails so much sooner than the other as long as what you really do on the computer is stable.
What stepping codes did you get?
I see, therefore the CPU where both cores are good ones will be sold under 4800 while those that have one core "faulty" will go with lesser numbers.Quote:
Originally Posted by johnrr6
Meaning you are really getting what you paid for.
AMD has this yield problems for years. Intel manufacturing of separate core makes it for the better (read lower) price but it suffers in performance.
NOT . . my 2nd core on a 4800 from the same place is way under the first core.Quote:
Originally Posted by joe2004
CCBWE 0518 TPMWQuote:
Originally Posted by xgman
My 2nd core will do about an hour of Prime 95----then poops out.
I will play with it some more-----but I'm kinda in agreement with you about the 2nd Core------I'm not sure the quicker failure rate is a big deal----especially since the 2nd core seems to act as a "booster" for the first core.
Am going to raise the vcore a tad more----but since I'm on air----can't really go much higher and be safe----at least in my mind.
What are good scores for the new PCMark 05???
I just downloded it to try out. EDIT: Got 1750 as a score----seems paltry but maybe this is a REALLY tough test.
Anybody got any comparisons----or is there something amiss with my rig....
John
Here's what Beilyaal said to me over on the DFI forum about how the X2 allocates processing power. Acts kinda like a "booster" if I understand his logic.
"Windows will allocate the cores equally among threads with the same priority. If one app has a higher process priority class and is running two threads at full load that app will take up most of the time and other apps will not get much processing time.
If two apps are running with the same priority both apps will get equal time on the cores.
If one app is running 1 thread another app is running 1 thread and a third app is running 2 threads, the third app will get 50% of the processing power while the other two will have to split the other 50% of power.
Generally calculation work should be done on a low priority thread so the gui will get more processing time and keep the computer experience fast, but most apps will calculate demanding things like video encoding with normal priority, and can thus make the experince laggy.
Two cores help with this problem, because there is less chance that two demanding caclulations are done at the same time, but if you start a lot of programs that does anything mayor (like rendering a movie or batching a job) your computer will become slow like a single core again.
Generally if you don't do anything with the programs you have running they don't take up any processing time, so just having 7 programs started will not slow you down."
There is a bug with X2 in PCMark05. Going by the charts an X2 @ 2.7Ghz should get about 5900. I can't get a good run either. It seems all the problems are on HDD related tests. At least 3 other people on the futuremark board also have the problem.
Eric
well so much for that test then. Is there anything left in this world withoujt "a bug" :rolleyes:
:banana::banana::banana::banana::banana: happens doesn't it.......... When you jump on the new stuff early to "Blaze New Trails", this is often the result. In time it will all get resolved! ;) ;)
Regards-
My score was 910 w/ my X2@2700. :p: Oh, yeah, there is definately a bug.Quote:
Originally Posted by ewitte
Funny thing with "burn-in" which I never really benefited from on CPU at least.
2.65ghz that was stable with 1.57v is now stable @ 1.46v... Going to be long week slowly burning in and ramping up my 4200+... Maybe 2.75ghz stable hopefully. (oh this is regarding both core, if it can't run prime 95 for more than 8 hours on dual prime setup, I don't even bother)
since yesterday i also have a 4200+ in my board
my first OC result after few hours was 2620Mhz at 1,5Vcore. But the secound core is holding me back :( the first can easily run 50Mhz higher at this Vcore
but my water wasn't very cold, today i had to turn on the fans :banana:
i just tested it with Prime. mayby it is a prime-only problem?
in next time i test the stability and ignore prime, but with more load on the secound core
Quote:
Originally Posted by ShoNuff
are you using the nvidia sata on your drives?
This will also show up in super pi and systool overclocking utility and maybe other places. I am beginning to think that there is something else going on here. It seems like 90% of these duals have a slow poke 2nd core, although not always #2 core. maybe there is some other interaction in play here that is causing this. It is difficult to pin down, because in normal use it is hard to tell what core is doing what amd when. You would think that two cores off the same die would be closer in top end at least most of the time, and we are seeing the exact opposite.Quote:
Originally Posted by centaur
Yep, I am. Are you aware of any problems/conflicts with the nVidia SATA drivers?Quote:
Originally Posted by xgman
I ran it again and got a whopping 2780. :YIPPIE:
That has been suggested. That's why I wnat to hear from anyone with an X2 not on Nvidia sata connectors/drivers.
I'm not running raid but are using the nvidia sata drivers without issue........
Quote:
Originally Posted by pcdoc1
without issue on PCMark 05? If so what score did you get.
Okay, at 270x10, I am perfectly stable with 1.52v with my 4400+, in both Prime95 Large/Small FFTs and SuperPi32M. At 9x300, I have instability in both. Weird because I can Memtest just fine at 300MHz 1:1 at 2.5-4-4-8 with my Patriot XBL.
Score......Quote:
Originally Posted by xgman
Obviously I have a little more tweaking to do.......
Hmm, I wonder what is different on your X2 setup whereas your score looks normal.
Is everyone having the problem using a DFI board? Don't know why that would be a variable but there must be something in common.
Edit: It's also weird that I think I was the only one that couldn't successfully run PCMark04. Couldn't get a hardrive score higher than(1).
Isn't ShoNuff up a few posts on an Asus same as you?
Maybe it's the brand of drive? :confused:Quote:
Originally Posted by pcdoc1
pcdoc and shonuff both show raptors, although shonuffs shows additional tk750 "storage" drive.
PCDoc isn't using RAID, how about everyone else?Quote:
Originally Posted by xgman
Actually, it DOES look like it may be a prob with RAID #1 x2 score ise IDE.
All the top scores are single drives....
wonder why it is primarily showing up on x2's though?
Probably some reason similar to why the x2's can't complete the grammar check on '04 :p:Quote:
Originally Posted by xgman
Probably because the BIOS is not mature. They did X2 testing but there is no way it was a great deal of testing. Even now with Retail/OEM shipping its probably not even .005% of the users.Quote:
Originally Posted by xgman
Eric
Its probably messed up. Version 1.3 was to fix grammer failing in another AMD cpu.Quote:
Originally Posted by s7e9h3n
ERic
Yup, it fixed in the fx53 iircQuote:
Originally Posted by ewitte
Guess this is the best it's gonna do........... :(
My first PCMark score was only 949.
My raptors are not set up in a raid array.
pcdoc how are your temps with the Koolance? I think that case is uber sexy.
My X2 seems to be doing very well as far as clocking is concerned. But, for some reason i get like 50sec PI times, complete crap 3D results. I have it running at 2.8GHz so i know this is not normal. Done a full format/reinstall, installed the AMD driver, 1T, latest NF4 drivers, everything i can think of. WTF is going on here. Some benches seem fairly normal, while 3D, PI and sandra memory bandwidth are WAAAAAY low. Any help would be nice.
What version of the DFI BIOS are you running?Quote:
Originally Posted by Sucka
If it's 6xx series, there is a performance fix for the X2 in a later version.
35 idle, 43 under load…… That’s with 120mm radiator fans turned down to the point of being inaudible…….Quote:
Originally Posted by ShoNuff
6026
http://service.futuremark.com/compare?pcm05=9997
Install the AMD drivers using the .EXE not the .ZIP.
Eric
good find!
here is the link:
http://www.amd.com/us-en/assets/cont...ies/amdcpu.exe
We think the only difference is it modifies the boot.ini file with the /usepmtimer switch.Quote:
Originally Posted by xgman
Check if you're running "ACPI MULTI-PROCESSOR" or "ACPI UNI-PROCESSOR"Quote:
Originally Posted by Sucka
Just flashed to 6/23, fixed the problem. That is to weird..
Well my DFI NF4 SLI DR up and died before I could get any real stability testing in.
Here's what I did before the board stopped booting. I'm really thinking the power draw at 3Ghz may have killed it. It was running i turned it off to attach another fan blowing over the DIMM area and it just never posted again. It's a shame because I was really ramping up. Burn in helped me with this I did burn in at 2.4 for a cpl hours then 2.6. By the time I was done with 2.6 I could do both of them at default volts. Next came 2.8 at near default volts but anything after that was really sketchy. I didn't really stability test I'd call 2.8Ghz benchmark stable though no problem. I ran 3DMark 01, 03, and 05 w/out a reboot. I ran every Sandra CPU and Mem test twice as well. I didn't notice higher than normal evap temps until I gave this more than 1.475v then it really wan't a linear increase in heat output. Honestly I think that's a bit too much voltage for the processor considering the current draw at that voltage and stpeed. This processor feels FAST in windows. I've never booted that fast in my life. I've just ordered a new DFI and had it overnighted so hopefully I can get this system back up and running tomorrow for weekend benches.
What clocks are everyone getting when using phase, don't seem to be many poeple posting results?
http://premium1.uploadit.org/jokester//toledo.JPG
I was eventually able to get both cores 8MB Superpi stable at just over 3GHz. Not sure what speed Prime stable is as I haven't been able to test it for prolonged periods yet.
Jokester
I reinstalled those suggested AMD CPU drivers...
fixed PCMark 05
My score went to 6421 from 1750
Ran two instances of Prime Blend for 24 hours and passed both
Ran two instances of Stress Prime 2004 Blend for another 24 hours and passed both
Ran two instances of small FFT for an hour and passed both----but that one DOES heat things up!
Highest CPU temp was 62
Highest PWM Temp was 59
Am using 1.350 Vcore With a special Vid of 110% ----I wind up with a vcore of 1.48 in Smart Guardian
I'm on air and at 2.8 megahertz----seems totally stable and my Mushkins are at 2,3,3,6
Will test some more----but so far----I'm liking my 4800+ X2 experience
Seems snappy with a lot of apps open......
Just the way I like it.......
I'm slowly pushing mine up, been Folding 24/7 for about 48hrs error free at this speed. :woot:
Just got mine and already in the hall of fame ;)
2.8GHz @ 1.55v. Seems that burnin had a huge impact since at start it was only stable at 2700Mhz with 1.73v.
http://service.futuremark.com/compare?pcm04=3013048
Burn In seems good for me too, i could not get this speed even with 1.55 volts to start with either.
Day by day i get a little faster as long as F@H work units are error free. :)
However - I thought folding didn't check your results unlike Prime?Quote:
Originally Posted by daffy67
No not nearly as good indication. There are 2 school of people... if it feels stable than it is group and tard like me who insist everything must be prime 95 stable :P (and everything else of course). It usually takes me a while to get there too.
If the work units are erroring out then there's a problem, just check the log file. I run Prime 95 first for a few hours, then switch to Folding where CPU cycles are going to something useful.
Quote:
Originally Posted by daffy67
However surely only exceptions on the flow of instructions will be picked up but not mathematic inaccuracies that it won't detect and report in the folding log.
Submitting a incorrect result isn't good, if they check multiple run results of the same work unit against each other your threatening the validity of a 3 run check. If they're comparing 5 runs you're just using up their bandwidth, drive space fo nothing.
As i said, i run Prime 95 for a few hours first, then switch over. I check the WU regularly for problems. :)
I noticed this as well, in nearly every review the X2 gets a slightly lower BW than an equally clocked Venice.Quote:
Originally Posted by xgman
I'm guessing they only allow each core to use 95% (or something) of the total bandwidth, so there's always some room for the other core to communicate with the ram at low latencies so things stay snappy at all times.
Well latest update... I haven't got anything more out per say after the initial burn in... for each steps it did work wonderfully by increasing additional few mhz which I thought would add up (started from 1.2v or so) but now that I am @ 2.65ghz... I only reduced another 0.01v to be stable for spending last 2 days or 3 days of constant burning in... now to 1.45v... This could just be the TIM settling down to be honest though.Quote:
Originally Posted by jinu117
Maybe the old days (I am talking decade ago) practice of computer manufacturers making system and going through 48 hours of burn in is as much stability there need to be as is. (out of those 48 hours, actual testing is usually 24 hours of running through stuff and rest was used on configuring, setting up, let it idle eating lunch, etc)
Maybe there will be more marginal gain over time as I keep using it (which I could since it IS dual core so I can switch back and forth cores as needed) but I do think I did max out what I can get out of this for now. Oh... now that I have another .1v to go through before feeling unsafe about temp or voltage on this chip for 24x7 usage... I hope to achieve another nice 100mhz... Gonna take another few days... lol going for that dual prime 2.75ghz setup on air :)
im sp2004 teting at 2880 on a x24800... using 1.48 vcore cpu is 46c on water but my pwm temp is currently 84c and rising slowly. what temps is everyone else getting on the pwm? and what are you using to cool it. also hammering my 12v rail to set at 12.48.. under 100% load its at 11.96 lol
At full load on water im seeing no more then 40C using CPU Burn... Im using a G4 Storm. My PMW sits between 32C and 44C depending on what time of day it is and how much load im pushing. FYI, its a good idea to have a couple fans blowing across your DFI's mb. It helps quite a bit, esp for stable oc's.
yea ive jammed a 120mm fan in there blowing on the pwm and ram area got it down to around 35 idle 45 load now on the pwm.
This probably isnt any help,I am not doing any burn in, but here are the results so far.
X2 4400 @ 2640mhz mushkin redline 4000 at 240fsb.
watercooled cpu and gpu.
temps
cpu=30 idle - 37 load
Nf4 =44 idle - 47 load
Pwmic =42 idle - 48 load
core volts are 1.5
havent pushed the cpu further yet, i have been too busy gaming with the new system, which is rock solid.
do you think 1.6-1.65 is safe for wc? tempes havent hit 40 yet even under load.
I'm in that range under watre after burning in at 1.71 for a few hours. temps in 30's seems fine even at 2900 mhz.Quote:
Originally Posted by Usurper
Quote:
Originally Posted by pancake
84 deg>???????????
Something isn't right. Mines well under 50. Check your ldt voltage also.
I have my LDT at 1.37 is that ok?
Xgman what sort of volts in that section of the bios should i be running?
i got ldt at 1.4 (1.37 in windows) chipset at 1.7 (1.66 in windows). waht water cooling setup you got usuper
Danger Den TDX , Danger Den Maze 4 acetal , 50Z pump , BIX2 rad.
i seem to be crappin out at high 2.6-low 2.7. maybe up my chipset and core then try again.
Just got the A64 4200+ and 3800+ venice. Dont know which one to keep ?Quote:
Originally Posted by Eto
Any ideas
buit less than venice is me guess just due to heat... :PQuote:
Originally Posted by irfs
Maybe 50-100mhz less in general? not bad really though as both do seem to clock very high anyway.
First test on the X2. 32Million Digit calculation @ 2.4ghz 1.26vCore while watching TV :). Currently testing 2.5ghz @ 1.26vCore - 32M Digit calculation...
http://homepage.ntlworld.com/najmudd...26vCore_X2.jpg
jus tested 1 core at 32M Pi @ 2.8ghz 1.53vCore
http://homepage.ntlworld.com/najmudd..._1.53vCore.jpg
I'm hoping to hit 2.9ghz and i think i'm going to need plenty of burnin
I just preordered my 4400+ from Monarch, can't wait!
my X2 4400+ toledo is running smooth, quiet & cool on silent water :D
i do love this baby
http://mitglied.lycos.de/xxmartin99/...2255@1.280.png
http://mitglied.lycos.de/xxmartin99/...2299@1.280.png
http://mitglied.lycos.de/xxmartin99/...2354@1.280.png
http://mitglied.lycos.de/xxmartin99/...2409@1.280.png
http://mitglied.lycos.de/xxmartin99/...2453@1.280.png
http://mitglied.lycos.de/xxmartin99/...2508@1.280.png
http://mitglied.lycos.de/xxmartin99/...2508@1.280.png
http://mitglied.lycos.de/xxmartin99/...4RPMW/4400.jpg
http://mitglied.lycos.de/xxmartin99/...24RPMW/box.jpg
still not maxed out
btw.
check out this one from IcyRocket
http://www.forumdeluxx.de/forum/showthread.php?t=137667
for some high 4800+ scores :)
Quote:
Originally Posted by Usurper
I've turned all my volts down a bit with the X2. I run LDT and chipset 1 setting over the lowest and it did seem to help pmw temps.
2x 2.7 GHz @ 1.424V still running smooth on water
http://mitglied.lycos.de/xxmartin99/...2706@1.424.png
Is that the max? You got some spare volts left !!!Quote:
Originally Posted by xxmartin
Temps's already at 45° .. not that much spare volts.. if at all...
45C isn't really hot ...........
I had lots of AMD cpus that had BSODs above 40° while overclocking. At least 50° is the limit.
240x11, stock cpu voltage, air. Still haven't even reached its max yet.
Well, I just went for a SLIGHT overclock, since I am on air cooling at the moment.... Waiting a little longer before I get the water cooling.... I have it up to 2.4 GHz...
A question, what do I use to check my core temps. when in Windows?
Yan