[SnyperTodd]

waldojim, on 03 November 2010 - 11:02 PM, said:

Well, they run 1600 8-8-8-24 stock at 1.65V bumping up to 1.8 doesn't seem to help anything.7-7-7-18 is as tight as they will go. The system refuses to even BOOT if I try 6-7-7-18 (put the 6 anywhere - it doesn't matter) until I hit 1.9V - and I am not about to leave the system running that way. I might try to lower the clock speed some on the ram - but I am not sure I want to run ~1250Mhz.

WJ, Check out this Everest Cache & Memory benchmark with my Patriots at 1160MHz @ 5-5-4-12 1t.

Broke 10K across the board- no small feat for a dual channel system (of course the i7 9xx will run over 15k all day long). Anyway, that's a perfect example of what you & I have been trying to explain to people- timings can trump clock speed if you can get them tight enough.

Quote

With the X48, the memory controller is still on the motherboard, giving you a bit more flex there depending on the board itself... me - I think I have a processor holding me back.

Exactly right. As you know, AMD's IMC is not known for being very robust. There's a good chance your RAM would perform completely different on an Intel system. I have a feeling you're right about the processor holding you back.

Quote

edit: BTW - for the sake of a random bit of coolness - I played with my Wifes PC a bit - with the extra 2 cores unlocked it will OC to 3.5Ghz, but the impressive part was with only two cores - I was at 3.7 with no effort at all - just set the multiplier, and go. Didn't even change the voltage. That collermaster kept that machine humming along at room temp even! I will toy around more with it later... curious what the dual core can do, and how it fares in game.

Cool! You should run NovaBench on it when you get it tweaked. We've kind of taken this topic off course a little bit, here's your chance to bring it back on track!

This post has been edited by SnyperTodd: 04 November 2010 - 03:09 PM

[waldojim]

SnyperTodd, on 04 November 2010 - 03:08 PM, said:

waldojim, on 03 November 2010 - 11:02 PM, said:

Well, they run 1600 8-8-8-24 stock at 1.65V bumping up to 1.8 doesn't seem to help anything.7-7-7-18 is as tight as they will go. The system refuses to even BOOT if I try 6-7-7-18 (put the 6 anywhere - it doesn't matter) until I hit 1.9V - and I am not about to leave the system running that way. I might try to lower the clock speed some on the ram - but I am not sure I want to run ~1250Mhz.

WJ, Check out this Everest Cache & Memory benchmark with my Patriots at 1160MHz @ 5-5-4-12 1t.

Broke 10K across the board- no small feat for a dual channel system (of course the i7 9xx will run over 15k all day long). Anyway, that's a perfect example of what you & I have been trying to explain to people- timings can trump clock speed if you can get them tight enough.

Quote

With the X48, the memory controller is still on the motherboard, giving you a bit more flex there depending on the board itself... me - I think I have a processor holding me back.

Exactly right. As you know, AMD's IMC is not known for being very robust. There's a good chance your RAM would perform completely different on an Intel system. I have a feeling you're right about the processor holding you back.

Quote

edit: BTW - for the sake of a random bit of coolness - I played with my Wifes PC a bit - with the extra 2 cores unlocked it will OC to 3.5Ghz, but the impressive part was with only two cores - I was at 3.7 with no effort at all - just set the multiplier, and go. Didn't even change the voltage. That collermaster kept that machine humming along at room temp even! I will toy around more with it later... curious what the dual core can do, and how it fares in game.

Cool! You should run NovaBench on it when you get it tweaked. We've kind of taken this topic off course a little bit, here's your chance to bring it back on track!

Sometimes it is just as cool to hear the MEANS to the end, the end results are great for bragging, but I think it is worth it to hear the experiences of other people as well. It helps when understanding WHY something works, HOW the OC comes out so well, what does and does not affect the end result, and most importantly - it is a great learning tool! I am still learning everyday - and yes I have people at this forum to thank for much of that (mostly from thinking in a different direction than I am used to).

The benchies are cool, but knowing how to make the MOST of them - even cooler!

[SnyperTodd]

waldojim, on 04 November 2010 - 04:56 PM, said:

Sometimes it is just as cool to hear the MEANS to the end, the end results are great for bragging, but I think it is worth it to hear the experiences of other people as well. It helps when understanding WHY something works, HOW the OC comes out so well, what does and does not affect the end result, and most importantly - it is a great learning tool! I am still learning everyday - and yes I have people at this forum to thank for much of that (mostly from thinking in a different direction than I am used to).

The benchies are cool, but knowing how to make the MOST of them - even cooler!

I was just trying to show you that running the memory at ~1250MHz may not be a bad thing, if you can get the timings tightened up down there. That's all.

It's worth trying, anyway.

[waldojim]

SnyperTodd, on 04 November 2010 - 05:22 PM, said:

waldojim, on 04 November 2010 - 04:56 PM, said:

Sometimes it is just as cool to hear the MEANS to the end, the end results are great for bragging, but I think it is worth it to hear the experiences of other people as well. It helps when understanding WHY something works, HOW the OC comes out so well, what does and does not affect the end result, and most importantly - it is a great learning tool! I am still learning everyday - and yes I have people at this forum to thank for much of that (mostly from thinking in a different direction than I am used to).

The benchies are cool, but knowing how to make the MOST of them - even cooler!

I was just trying to show you that running the memory at ~1250MHz may not be a bad thing, if you can get the timings tightened up down there. That's all.

It's worth trying, anyway.

Please don't misunderstand - I will most certainly take the time to try that! Not tonight I am afraid - I have been working doubles as of late, and need the 4 hours sleep I have coming to me!

That said, on Monday (my Saturday) I will certainly play around some more to see what I can do. The memory controller may not have issues with frequencies down that low - and some nice, tight timings... truth be told, 1333 is probably all that controller can really take advantage of anyways (as that IS the JDEC limit)... So - certainly worth a try!

All I was trying to say, is that while this is a TOUCH OT, I know I am learning a good deal, and I hope others (Rommel! )are learning a bit more as well. I think that is a touch more important than just the bragging rights.

edit:
You know... its funny, you never see Wintard participate in these discussions... I want to see some results from those amazing rigs of his...

This post has been edited by waldojim: 04 November 2010 - 06:32 PM

[waldojim]

Ok, i tried for 5-5-5-16. It took 1.85v @ 1224mhz to even post. Currently running overnight memtest. Then another round of benchies!

[waldojim]

waldojim, on 04 November 2010 - 09:24 PM, said:

Ok, i tried for 5-5-5-16. It took 1.85v @ 1224mhz to even post. Currently running overnight memtest. Then another round of benchies!

as expected - that failed... trying for 1200 now.

[Rommel]

Well its been an interesting time playing with these settings.

First the cpu.
Original start point.
CS 3534
X 15.5
BS 228
HT link 2280
CPU NB 2736

Adjusting CPU NB to 2964 automatically takes me to 3 cores?
Didn't try forcing 4th core, just rolled it back to 2736 then forced 4th core.
These cores have passed intels burn harshes test.

Adjusting HT LINK past 2280 causes boot issues.
2508 booted then crashed 2 minutes later.
Rolled back to 2280.

Adjusting multiplier to 16.5 crashes pc.
Seems happy with X16.

ALL above changes did not effect buss speed???
Buss speed is highlighted but nothings happens in bios when enter is pressed?

Current core speed is now 3648.


Played with DRAM.
Original. 7 7 7 16 @ 608 FSB 3:8 <--- auto 1216
Raised it to 1520 auto----> 9 9 9 20 33 @ 760
Then forced the top 3 to 8s NOVA score 1119

Then went to max option, 1824 that gave 11 11 11 24 40 @ 912 FSB 1:4
Nova scored the same, 3D crashed.

Rolled back to original and went to manual timings.
6 6 6 15 5 20 8 4 The numbers past the 6s are guesses. Just trying to follow what seemed like a pattern from auto settings.
FSB 3:8

All these changes didn't make a major differance in bench scores.

Not sure how to force the buss speed for the cpu.
PC seems nice and responsive.
Temps are 34 C.
Though I did switch fans to high as I think about. Will lower to middle postition and recheck.

http://valid.canardp....php?id=1467233 11s

http://valid.canardp....php?id=1467275

Still haven't touched voltages.

Rommel

EDIT:
Forgot to mention that I lowered my video card RAM HZ to 1385.
Some 3D test strted crashing.

Original 3D score before any other OCing.
16797

After OCing video card and todays OCing,
17632

http://i18.photobuck...in_/666dram.jpg

This post has been edited by Rommel: 05 November 2010 - 05:25 PM

[SnyperTodd]

Rommel, on 05 November 2010 - 04:58 PM, said:

ALL above changes did not effect buss speed???
Buss speed is highlighted but nothings happens in bios when enter is pressed?

Rommel, try enabling number lock and using the number pad to enter the bus speed directly, or try pressing number pad + / - to adjust the bus speed.

[waldojim]

waldojim, on 05 November 2010 - 04:14 AM, said:

waldojim, on 04 November 2010 - 09:24 PM, said:

Ok, i tried for 5-5-5-16. It took 1.85v @ 1224mhz to even post. Currently running overnight memtest. Then another round of benchies!

as expected - that failed... trying for 1200 now.

1200 was a fail... trying for 1150...

[Rommel]

SnyperTodd, on 05 November 2010 - 07:24 PM, said:

Rommel, on 05 November 2010 - 04:58 PM, said:

ALL above changes did not effect buss speed???
Buss speed is highlighted but nothings happens in bios when enter is pressed?

Rommel, try enabling number lock and using the number pad to enter the bus speed directly, or try pressing number pad + / - to adjust the bus speed.

Hi Snypertodd,

I woke up this morning and started thinking about my OCing yesterday and "boom", that is what I remembered.
So I will spend a brief moment with it then I have things I need to take care of.

Thanks for the reminder.

If you or Waldojim could advise on memory OCing that would be appriciated.
When I forced my timing to 6s, should I let bios auto the other numbers?

I need to google the letters of the other clocks to learn their purpose.

Thanks again,

Rommel

[Rommel]

OK, enough torcher for now.

Raising the BS effects the cpu nb and HT link.
Tried to lower them as raising the BS with a lower multiplier was now good to get past current core speed.

Even disabled unleashing mode to go back to a dual core.

After this round my best is a core speed of 3680 from 3648. Not much.
Later I'll try messing with the multiplier.

As for the RAM.
I kept my forced timings at 6 6 6.
Then turned the other number to auto.

My scores did go up.

NOVA 1120 starting point 1074
3D 17746 starting point from first OC was 16797

http://novabench.com...993568e21428971

3D http://i18.photobuck...Irwin_/3680.jpg

CUPID http://i18.photobuck...rwin_/CPUID.jpg

Any suggestions?

Rommel

This post has been edited by Rommel: 06 November 2010 - 06:37 AM

[Car54]

I like how this benchie thread has turned into a OCing thread

Here are some updated benches and validations:
Default settings before OCing:
http://novabench.com...9d7083a13627939

http://valid.canardp....php?id=1467305

OC settings, that seem stable for now with the tests I've run (including IBT)
http://novabench.com...5c6e0aa71c28970

http://valid.canardp....php?id=1468260

So far the OC CPU temps on air, as I post this are 35c on all 4 cores. On the IBT, they maxed about at 60-65c, Prime95 were @ about 55c

Now I too am at the point of where Rommel is at, and that is what to do and tips on OCing the RAM. Below are the Kingston reference numbers. Do I try for the lower latency config. and bump up my volt to the recommend 1.95v or do in need to go to 2.0v due to my CPU Ocing, or doesn't that make a difference? I know it's trial and error, but the CPU OCing seemed to make a little more sense, this RAM OCing is like "what am I shooting for and why"? Do you have any rules of thumb of why you start at this point or that?



Thanks for your help

This post has been edited by Car54: 06 November 2010 - 06:59 AM

[SnyperTodd]

Rommel, on 06 November 2010 - 04:29 AM, said:

If you or Waldojim could advise on memory OCing that would be appriciated.
When I forced my timing to 6s, should I let bios auto the other numbers?

I need to google the letters of the other clocks to learn their purpose.

Thanks again,

Rommel

Rommel, here's a link that explains RAM timings pretty well and it's fairly easy to understand: http://www.hardwares...om/article/26/1 It's old, but still relevant. What I do for tightening the memory is to set the primary timings and leave the others on 'Auto'. Then I print the complete list of timings from Everest (you can use your favorite freeware system information utility) so I know what the BIOS has done and manually set the later timings to what the BIOS automatically sets them to. From there I tweak each timing individually. It's a lot of messing around, but I don't like having anything on 'Auto' in my BIOS. The BIOS doesn't always get them as tight as they can be, and every little bit of gain helps.

Nice overclocks again! Isn't it a good feeling to spend that time tweaking and rerun that benchmark and see that kind of improvement? A 46 point gain is impressive! Are all of your voltages set to 'Auto'?

WJ- good luck! Running 5-5-5-15 at over 1200MHz is pretty ambitious. I would expect it to pass at 1150MHz. You might even be able to tighten the tRP (3rd number) a notch at that speed. If it's not stable, try loosening the tRCD (2nd number) and see what happens. Some RAM likes that a tad looser than the others if you're really pushing it.

[Car54]

Excellent link Snyper, including the link from there on Memory Overclocking http://www.hardwares...com/article/152

edit: that link should be flagged as a "Reference Work" on RAM/Memory.

This post has been edited by Car54: 06 November 2010 - 07:48 AM

[SnyperTodd]

Car, a 4GHz quad core is something to be proud of! I'm impressed (and jealous )! Your RAM score suffered a little because your timings loosened up. Since you're still running DDR2, I would work tightening those instead of trying to push the speed up anymore. You're basically at 1066MHz right now, so you should be able to run 5-5-5-15 or even 5-5-5-12 if you've got some voltage room to play with. You don't have to run the voltage any higher than the RAM is stable with. The other thing to watch is your FSB strap. I don't have your board in front of me, but it's generally set with the 'System Memory Multiplier' on Gigabyte boards. It should look like a pair of numbers, like this: 1.2:333, where 1.2 is your memory multipler (1.2=5:6, which is what you're running; not all boards display it the same way) and the 333 is your FSB strap. The strap is a set of motherboard timings much like the RAM timings. You want it as low as possible, but it will affect stability and can limit your FSB overclock if it's too low. At your FSB of 445MHz, you might have to run the strap at 400, but try it at 333 just to make sure it won't work. The other motherboard timing to watch is the Performance Level (also called 'Read Delay' or tRD). It's another northbridge timing that can have a drastic effect on overall performance. Again, you want it as low as possible, but it will be dependent on your NB voltage and FSB frequency more than anything you do to your RAM. I would find a program that will show you what timings are set at (like Everest if you have it), take a note of it, and start trying to work it down. Good luck!

[Car54]

SnyperTodd, on 06 November 2010 - 07:23 AM, said:

Car, a 4GHz quad core is something to be proud of! I'm impressed (and jealous )! Your RAM score suffered a little because your timings loosened up. Since you're still running DDR2, I would work tightening those instead of trying to push the speed up anymore. You're basically at 1066MHz right now, so you should be able to run 5-5-5-15 or even 5-5-5-12 if you've got some voltage room to play with. You don't have to run the voltage any higher than the RAM is stable with. The other thing to watch is your FSB strap. I don't have your board in front of me, but it's generally set with the 'System Memory Multiplier' on Gigabyte boards. It should look like a pair of numbers, like this: 1.2:333, where 1.2 is your memory multipler (1.2=5:6, which is what you're running; not all boards display it the same way) and the 333 is your FSB strap. The strap is a set of motherboard timings much like the RAM timings. You want it as low as possible, but it will affect stability and can limit your FSB overclock if it's too low. At your FSB of 445MHz, you might have to run the strap at 400, but try it at 333 just to make sure it won't work. The other motherboard timing to watch is the Performance Level (also called 'Read Delay' or tRD). It's another northbridge timing that can have a drastic effect on overall performance. Again, you want it as low as possible, but it will be dependent on your NB voltage and FSB frequency more than anything you do to your RAM. I would find a program that will show you what timings are set at (like Everest if you have it), take a note of it, and start trying to work it down. Good luck!

Very nice Snyper, very nice, thank you for all that valuable info.
I will print this off, and go into the BIOS and see where I'm at.
As always, thank you for your time and effort

P.S. I agree, the 4GHz is where I'm going to stay, that's good enough for me on that front. (if not maybe even backing it down a bit, it's nice that I can get there, but not necessary to stay there)

This post has been edited by Car54: 06 November 2010 - 08:20 AM

[Car54]

Just an update, I did get an error on this round of Prime95 on the 4th core shutting down ....and so it begins, the tweaking and adjusting, and to think I almost thought I was lucky enough to get by on my first couple of tries, yeah, right..lol

Correction on my Prime temps listed in my previous post, around 60-65c

This post has been edited by Car54: 06 November 2010 - 10:43 AM

[Rommel]

Car54, on 06 November 2010 - 06:48 AM, said:

I like how this benchie thread has turned into a OCing thread

I agree.
Snypertodd called it right too.
You need to backup your claims.

Very nice CPU speed. I bet that made for a noticable differance in response time.
I am stuck at 3680 but thats higher than the last time I tried.
Interesting you also experiance a core shutting down on some of your adjustments.

Snypertodd, I haven't touched voltages yet. Not sure what my current values are. Bios says auto but highlighting it does give an option.
Must be typed in. Need to researched my componets and see their rated ranges.

Played with my timings based on your link.

TCL, TRCD, TRP do not like anything below 6 so that is where they are.
TRAS is 15
TRT P in auto.
TRC is 22
TWR TRRD, both in auto.

If the article covered these I missed it. Will go over it again.

6 6 6 15 22
613.4 at 1T.
Feels better having an understanding of these things now.

Also happy to report no smoke detectors have been activated.

As Car54 said and I agree, once I feel I found my max I'll back off some for a safty buffer.

Hopefully Waldojim found his sweet spot.

Rommel

EDIT:
Had a "wake up from sleep" misque.
Overclock failed.
Reset TRC to auto.
Defaulted to 27, tested, wokeup.
??

This post has been edited by Rommel: 06 November 2010 - 02:30 PM

[Car54]

Rommel, on 06 November 2010 - 02:16 PM, said:

Also happy to report no smoke detectors have been activated.

lol

It's been about 4 hours, and Prime is liking 420 MHz (3.78GHz..420x9) allot better. If that stays solid, I may just stay there, as I would've backed it down from 4 anyway. Then I'll make some RAM adjustments.

[waldojim]

Car54, on 06 November 2010 - 06:48 AM, said:

I like how this benchie thread has turned into a OCing thread

Here are some updated benches and validations:
Default settings before OCing:
http://novabench.com...9d7083a13627939

http://valid.canardp....php?id=1467305

OC settings, that seem stable for now with the tests I've run (including IBT)
http://novabench.com...5c6e0aa71c28970

http://valid.canardp....php?id=1468260

So far the OC CPU temps on air, as I post this are 35c on all 4 cores. On the IBT, they maxed about at 60-65c, Prime95 were @ about 55c

Now I too am at the point of where Rommel is at, and that is what to do and tips on OCing the RAM. Below are the Kingston reference numbers. Do I try for the lower latency config. and bump up my volt to the recommend 1.95v or do in need to go to 2.0v due to my CPU Ocing, or doesn't that make a difference? I know it's trial and error, but the CPU OCing seemed to make a little more sense, this RAM OCing is like "what am I shooting for and why"? Do you have any rules of thumb of why you start at this point or that?



Thanks for your help

The ram can generally take a bit extra, especially to drop timings, I do NOT remember what the upper LIMITS for DDR2 is, but I DO know, that I ran 2.1V for almost 2 years. I would suggest trying to stay at or under that if you can.