Overclock.net › Forums › Intel › Intel CPUs › [MOD] LGA775 Support For LGA771 Xeon CPUs
New Posts  All Forums:Forum Nav:

[MOD] LGA775 Support For LGA771 Xeon CPUs - Page 1196

post #11951 of 12074
Quote:
Originally Posted by lumpen View Post

Test failed after 1 hour - the usual average. But i understand now that the difference lies between what we call a stable OC. When i say OC i mean it: stable stable teaching.gif
Btw i think you overdo with those settings, at the end i could go much lower with the same results.
Thanks for everyone, i'm out.

if blend failed after 1 hour, it can be for this reasons:

1º the most common one: not enough cpu voltage (which would have been discarded if you've done small FTT in prime95)
2º ram needs a bit more voltage (if it was auto, then try a bit more voltage than stated in the stick, like 1,9v if stock is 1,8v)
3º the NB chip is overheating.
4º CPU is overheating.
5º your ram is not capable of doing your timings, and you have to relax them a bit.

I'm 100% sure you can do 4ghz rock solid. But tweaking takes time.
post #11952 of 12074
Quote:
Originally Posted by Piskeante View Post

if blend failed after 1 hour, it can be for this reasons:

1º the most common one: not enough cpu voltage (which would have been discarded if you've done small FTT in prime95)
2º ram needs a bit more voltage (if it was auto, then try a bit more voltage than stated in the stick, like 1,9v if stock is 1,8v)
3º the NB chip is overheating.
4º CPU is overheating.
5º your ram is not capable of doing your timings, and you have to relax them a bit.

I'm 100% sure you can do 4ghz rock solid. But tweaking takes time.
It's rare,but on occasion,turning OFF LLC and setting the voltages in bios manually to make up for the drop, can sometimes make up for an unstable OC. What's happening in those cases is the slight surges to make up for the drops under demand sometimes cause a failure/error. Some with the nvidia sets had better results with it off than on.Maybe the same would apply here in some cases?
post #11953 of 12074
If he's getting an error when stressing the CPU after an hour, I'll put my money under Northbridge getting too hot... Been there before...

Get a fan blowing some cooled air into the Northbridge are and try the run again... The fact that you have 4 ram stick make it even more possible for the NB to be overheated as he is dealing with RAM control (775/771 still don't have memory control being handled by the CPU).

Also, add a little bit more Vcore on the CPU, and if your memory is rated@1.8V put it up as 2.0V in BIOS settings as there is more Vdrop when running 4 sticks....

As you can see, I needed more Vcore after I've added a second RAM kit to my spec (1.325V set at BIOS):

CaveMonkey Rig
(15 items)
 
OldSchool Rig
(13 items)
 
 
CPUMotherboardGraphicsRAM
Ryzen 1700 MSI X370 Gaming Pro Carbon EVGA GTX980Ti Classified  2x8GB G.Skill TridentZ 3600 
Hard DriveHard DriveCoolingOS
Sandisk 256GB M.2 SSD Crucial MX100 256GB Be Quiet Dark Pro 3 Windows 10 Pro x64 
MonitorKeyboardPowerCase
iiyama - ProLite XUB3490WQSU-B1 Ducky Mini Brown Switches Corsair RM850i Fractal Design Define R5 
MouseMouse PadAudio
Logitech G502 Proteus Spectre Qpad CT OldGents.se Creative SoundBlaster Z + Edifier Studio R1600T... 
CPUMotherboardGraphicsRAM
Opteron 175 DFI LanParty NF4 Ultra DR BFG 8800GTS 512MB OC2 (G92) 2x1GB DDR PC4000 GSKILL F1 
Hard DriveCoolingOSKeyboard
Seagate 320GB 7200.11 SP Shythe Katana II WinXP Pro SP3 Duky Zero Red Swithes 
PowerCaseMouseMouse Pad
EVGA 500W Lian Li PC-A05FNB Logitech G300 QPad 
Audio
Onboard 
CPUMotherboardGraphicsRAM
Q6600@3.2Ghz DFI P45 Lanparty Dark T2RS Plus XFX GTX260 2*2GB Crucial Ballistix PC6400 
Hard DriveCoolingOSMonitor
Kingston V300 60GB Zalman CNPS X10 Windows 7 x86 Dell 19'' 
KeyboardPowerCaseMouse
Duky Zero Red Swithes Enermax 625W Pro+ NZXT S340 Black Logitech G300 
Mouse PadAudio
QPAD Creative Audigy Z2 Mod 
  hide details  
Reply
CaveMonkey Rig
(15 items)
 
OldSchool Rig
(13 items)
 
 
CPUMotherboardGraphicsRAM
Ryzen 1700 MSI X370 Gaming Pro Carbon EVGA GTX980Ti Classified  2x8GB G.Skill TridentZ 3600 
Hard DriveHard DriveCoolingOS
Sandisk 256GB M.2 SSD Crucial MX100 256GB Be Quiet Dark Pro 3 Windows 10 Pro x64 
MonitorKeyboardPowerCase
iiyama - ProLite XUB3490WQSU-B1 Ducky Mini Brown Switches Corsair RM850i Fractal Design Define R5 
MouseMouse PadAudio
Logitech G502 Proteus Spectre Qpad CT OldGents.se Creative SoundBlaster Z + Edifier Studio R1600T... 
CPUMotherboardGraphicsRAM
Opteron 175 DFI LanParty NF4 Ultra DR BFG 8800GTS 512MB OC2 (G92) 2x1GB DDR PC4000 GSKILL F1 
Hard DriveCoolingOSKeyboard
Seagate 320GB 7200.11 SP Shythe Katana II WinXP Pro SP3 Duky Zero Red Swithes 
PowerCaseMouseMouse Pad
EVGA 500W Lian Li PC-A05FNB Logitech G300 QPad 
Audio
Onboard 
CPUMotherboardGraphicsRAM
Q6600@3.2Ghz DFI P45 Lanparty Dark T2RS Plus XFX GTX260 2*2GB Crucial Ballistix PC6400 
Hard DriveCoolingOSMonitor
Kingston V300 60GB Zalman CNPS X10 Windows 7 x86 Dell 19'' 
KeyboardPowerCaseMouse
Duky Zero Red Swithes Enermax 625W Pro+ NZXT S340 Black Logitech G300 
Mouse PadAudio
QPAD Creative Audigy Z2 Mod 
  hide details  
Reply
post #11954 of 12074
Quote:
Originally Posted by CrazyMonkey View Post

If he's getting an error when stressing the CPU after an hour, I'll put my money under Northbridge getting too hot... Been there before...

Get a fan blowing some cooled air into the Northbridge are and try the run again... The fact that you have 4 ram stick make it even more possible for the NB to be overheated as he is dealing with RAM control (775/771 still don't have memory control being handled by the CPU).

Also, add a little bit more Vcore on the CPU, and if your memory is rated@1.8V put it up as 2.0V in BIOS settings as there is more Vdrop when running 4 sticks....

As you can see, I needed more Vcore after I've added a second RAM kit to my spec (1.325V set at BIOS):


i aggree with this, as i had already said that.
post #11955 of 12074
Quote:
Originally Posted by Piskeante View Post

i aggree with this, as i had already said that.
lol,several people told him similar,he seems to think that that all hardware of same configurations should OC the same. Doesn't want to hear that cpu's OC differently because they've been used harder for years before this,memory by DIFFERENT makers has different results even if the settings are set the same, and boards run different in general depending how they've been used. 2 p5qpro's with 2 x5470's with 8 sticks of the same ram may have very different results with the same settings.
#1 one or the other cpu may be less capable then the other so will fail/not OC as well.
#2 Any 1 stick of the 8 could be older/more abused & less likely to hold up under an extreme OC /heat load. Very hard to find the 1 when always testing multiple sticks.
#3 one systems PSU may be giving low voltages or voltage spikes causeing test failures.
#4 one board or the other may have been abused by a previous owner
#5 local temps may be significantly different to cause overheating for air cooled systems on 1 board while the other is in a controled temp enviroment.
And THAT'S using the SAME settings on the "same" systems. @ least if 1 person does THIS,you know they are using the same definition of "stable" for their testing. What We do is similar,but without a set definition of STABLE. What's stable for member # 1 who is only using the rig for gaming,may be a disaster for member #2 who's using the rig for professional video work (where it's encoding the video)and storing it.(data corruption issue). Seems safe to say that unless the person is willing to PUT THE TIME INTO THE OC, that extreme OC's (4.5 & up) are only really viable for gaming. While base clock to 3.6 should be able to be achieved easier and on a more stable basis much more easily. between 3.6 and 4.5 seems to be where some work can give most of us a "stable for our needs" setting without driving us to the point of throwing the system off the balcony. rolleyes.gif
My point is that OC'ing is a work in progress. It takes time and effort to get your rig where you want it.If you're not willing,or unable to put the time in,then the OC won't happen. Just using another person's settings isn't likely to be the perfect answer.It may give you a place to start,but you'll most like still need to tweak your own.
Everything on AUTO doesn't count BTW. as tweaking. thumb.gif
post #11956 of 12074
Quote:
Originally Posted by schuck6566 View Post

lol,several people told him similar,he seems to think that that all hardware of same configurations should OC the same. Doesn't want to hear that cpu's OC differently because they've been used harder for years before this,memory by DIFFERENT makers has different results even if the settings are set the same, and boards run different in general depending how they've been used. 2 p5qpro's with 2 x5470's with 8 sticks of the same ram may have very different results with the same settings.
#1 one or the other cpu may be less capable then the other so will fail/not OC as well.
#2 Any 1 stick of the 8 could be older/more abused & less likely to hold up under an extreme OC /heat load. Very hard to find the 1 when always testing multiple sticks.
#3 one systems PSU may be giving low voltages or voltage spikes causeing test failures.
#4 one board or the other may have been abused by a previous owner
#5 local temps may be significantly different to cause overheating for air cooled systems on 1 board while the other is in a controled temp enviroment.
And THAT'S using the SAME settings on the "same" systems. @ least if 1 person does THIS,you know they are using the same definition of "stable" for their testing. What We do is similar,but without a set definition of STABLE. What's stable for member # 1 who is only using the rig for gaming,may be a disaster for member #2 who's using the rig for professional video work (where it's encoding the video)and storing it.(data corruption issue). Seems safe to say that unless the person is willing to PUT THE TIME INTO THE OC, that extreme OC's (4.5 & up) are only really viable for gaming. While base clock to 3.6 should be able to be achieved easier and on a more stable basis much more easily. between 3.6 and 4.5 seems to be where some work can give most of us a "stable for our needs" setting without driving us to the point of throwing the system off the balcony. rolleyes.gif
My point is that OC'ing is a work in progress. It takes time and effort to get your rig where you want it.If you're not willing,or unable to put the time in,then the OC won't happen. Just using another person's settings isn't likely to be the perfect answer.It may give you a place to start,but you'll most like still need to tweak your own.
Everything on AUTO doesn't count BTW. as tweaking. thumb.gif

should i consider your post as a personal attack?? for me it definetely is. ok so, i get the point. redface.gif

Just to clarify somethings.

1º when telling people a place to start, you are not considering any of the aspects you said. Am i supposed to know if a component has been treated good or badly??Does even the owner know that?? Does any of your tips consider the same arguments as you are throwing at my face?? I'll have to see some of your posts, in case you did try "la bola de la bruja lola", to figure out if someone can or cannot do overclocking considering the real state of the hardware. When you advice someone you know in advantage if he/she can or cannot do that oc??? Or you just give them an idea??
I don't know if what bothered you was my statement saying " if they work for me, should work for you". Well, those options are standard. they are all around internet. They are even adviced in many guides of 775 overclocking. They will work in a big % with all people in his config.

2º his mobo is almost exactly the same as mine, except for very minor improvements, and not getting ddr3 support. My board will do very easily 4ghz. Almost any X5470 will do 4 ghz, and that board should handle that 100% sure.

3º every X5470, even the worst ones, will do 4ghz stable. Even degraded cpus will clock a bit less than expected but 4ghz?? come on..... Only those oced to the extreme and degraded by years of overvoltage will not do that, but it's not a thing we can know. When we give "our config" we just consider all the hardware is in good condition. If you have a way to see if a hardware is in good or bad condition, just with your imagination, well, just let us know.

4º in relation to ram settings, timings , voltage, and other things, may differ from one to another pc, so that's why i told him it could be NB overheating, Ram needing more voltage or even relaxing times? Where do you see a neglecting attitude or a bad advice in my post??

5º stable has to be put in context. stable for me may not be the same as stable for you, though we can share a minimum. For me the minimum is being able to pass one of this : prime 95 1 hour , IBT 20 pass, or linx 20 pass. If you are doing content creation, then you may want rock solid rig. If not, you can be going around at very high speed without having any issues, though knowing it's not stable. I told him that, so i don't know here where do you happen to see another neglectable attitude on my side.

6º it happens to be that the settings on auto i told him do not vary at all the system stability, since NB, SB, PCIE, GTL, are options that do not matter while only doing 4ghz. As he was going 800mhz ram, the mobo is perfectly capable of running auto timings (specially if ram training option is enable).They can be left perfecttly on auto. So, what you are trying to tell me, is that if anyone does not know his timings, but will run his memory stock, he should not use auto settings? you must be kidding.

Auto will work in almost all conditions, except if you are overclocking, WHICH HE WASN'T. You do not need to mess with that. Also consider the fact that he was "in his last try" so there is no point in telling him: well, you can try upping the voltage on only single option, changing every single timing, to see if you go spot on. So, when someone comes "in his last try" your option is "go for a lot of tries with all options???", that's nonsense, mate.

i should tell you i didn't expect such a personal attack. this is a bit disgusting. Well, weren't you the one that told me that "personal attacks" were not to be allowed??? (to refresh your memory, was between Tnlgg and me talking about stability). btw, i did apologize to Tnlgg in case my words had been misunderstood at that moment.
post #11957 of 12074
Quote:
Originally Posted by Piskeante View Post

should i consider your post as a personal attack?? for me it definetely is. ok so, i get the point. redface.gif

Just to clarify somethings.

1º when telling people a place to start, you are not considering any of the aspects you said. Am i supposed to know if a component has been treated good or badly??Does even the owner know that?? Does any of your tips consider the same arguments as you are throwing at my face?? I'll have to see some of your posts, in case you did try "la bola de la bruja lola", to figure out if someone can or cannot do overclocking considering the real state of the hardware. When you advice someone you know in advantage if he/she can or cannot do that oc??? Or you just give them an idea??
I don't know if what bothered you was my statement saying " if they work for me, should work for you". Well, those options are standard. they are all around internet. They are even adviced in many guides of 775 overclocking. They will work in a big % with all people in his config.

2º his mobo is almost exactly the same as mine, except for very minor improvements, and not getting ddr3 support. My board will do very easily 4ghz. Almost any X5470 will do 4 ghz, and that board should handle that 100% sure.

3º every X5470, even the worst ones, will do 4ghz stable. Even degraded cpus will clock a bit less than expected but 4ghz?? come on..... Only those oced to the extreme and degraded by years of overvoltage will not do that, but it's not a thing we can know. When we give "our config" we just consider all the hardware is in good condition. If you have a way to see if a hardware is in good or bad condition, just with your imagination, well, just let us know.

4º in relation to ram settings, timings , voltage, and other things, may differ from one to another pc, so that's why i told him it could be NB overheating, Ram needing more voltage or even relaxing times? Where do you see a neglecting attitude or a bad advice in my post??

5º stable has to be put in context. stable for me may not be the same as stable for you, though we can share a minimum. For me the minimum is being able to pass one of this : prime 95 1 hour , IBT 20 pass, or linx 20 pass. If you are doing content creation, then you may want rock solid rig. If not, you can be going around at very high speed without having any issues, though knowing it's not stable. I told him that, so i don't know here where do you happen to see another neglectable attitude on my side.

6º it happens to be that the settings on auto i told him do not vary at all the system stability, since NB, SB, PCIE, GTL, are options that do not matter while only doing 4ghz. As he was going 800mhz ram, the mobo is perfectly capable of running auto timings (specially if ram training option is enable).They can be left perfecttly on auto. So, what you are trying to tell me, is that if anyone does not know his timings, but will run his memory stock, he should not use auto settings? you must be kidding.

Auto will work in almost all conditions, except if you are overclocking, WHICH HE WASN'T. You do not need to mess with that. Also consider the fact that he was "in his last try" so there is no point in telling him: well, you can try upping the voltage on only single option, changing every single timing, to see if you go spot on. So, when someone comes "in his last try" your option is "go for a lot of tries with all options???", that's nonsense, mate.

i should tell you i didn't expect such a personal attack. this is a bit disgusting. Well, weren't you the one that told me that "personal attacks" were not to be allowed??? (to refresh your memory, was between Tnlgg and me talking about stability). btw, i did apologize to Tnlgg in case my words had been misunderstood at that moment.
NOOOOOOOO! I'm sorry if you took me wrong! If you look back @ his post, you'll see I tried helping him in the beginning and he wanted a quick fix! THAT was all I was talking about. If I came across differently then I apologize to you and anyone else who took me wrong.redface.gif Edit: By him I'm refering to lumpen.
Edited by schuck6566 - 4/21/17 at 10:05am
post #11958 of 12074
To anyone and EVERYONE who may have taken my prior post about Overclocking the wrong way. ALL that was meant was that different hardware can present different issues and that we needed to be willing to work for an acceptable OC. NO digs,cuts,or disparagements were meant towards people offering advice. I was simply stating my OWN opinion on the vagaries of the hardware we're dealing with.
Edited by schuck6566 - 4/21/17 at 10:06am
post #11959 of 12074
Quote:
Originally Posted by Piskeante View Post

i aggree with this, as i had already said that.

one thing i noticed in his link to his settings is northbridge voltage is set to 1.2v. when raising FSB to the 400 or better range with 4 sticks of memory i've had to use 1.35v and put a fan on the heatsink of every board i've ever used. maybe thats just my experience though.
Edited by ried16 - 4/21/17 at 2:23pm
post #11960 of 12074
Quote:
Originally Posted by ried16 View Post

one thing i noticed in his link to his settings is northbridge voltage is set to 1.2v. when raising FSB to the 400 or better range with 4 sticks of memory i've had to use 1.35v and put a fan on the heatsink of every board i've ever used. maybe thats just my experience though.

Nice info.

When overclocking memory, yes you need to increase your CPU NB voltage and ram voltage. it's very well known that if you have 4 pairs of ram, the ram voltage will have to be increased a little, this is not the case of Lumpen, but sometimes increasing the ram voltage helps, from 1,8 to 1,9 or even 2.0 as Crazymonkey said (depending of the type of ram and voltage obviously). it's not that that ram needs more voltage to work at stock but the fact that the chip is not able to supply all 4 sticks the same amount of voltage, so you have to increase it a bit.

i would say if you have 4 sticks you have to increase the voltage, even if running stock fsb clock. What you are saying actually makes sense, but only with 4 sticks, as with two there seems to be no voltage issue.
i run ddr3 1510mhz 9.9.9.24 2T and my system can pass memtest completely with auto NB voltage and auto ram voltage. But i have two sticks.

and for the heatsink, it depends. I've read many times that something over 1,4 on the NB voltage will need active cooling. Heatsinks will handle till 1,4 as general, (many boards won't so take this a particular statement). i dont have almost any experience on NB voltage overheating, since for me NB voltage does nothing in providing stability. If i'm unstable at auto, i'll also be unstable at, 1,3 or 1,4. For me it changes nothing. This is only my case, and may vary from one rig to another.
Edited by Piskeante - 4/21/17 at 2:48pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › [MOD] LGA775 Support For LGA771 Xeon CPUs