Has anyone tried this before?

Ok so from what I have read online, the lowest memory multiplier for the i7-920 is 6x, and from what I know the lowest memory multiplier will always give the best performance and is the best to use when possible.

But so far I havent seen anyone online using a 6x memory multiplier on an overclocked i7-920, mostly 8x. (FSBRAM of 2:8)

Before I go testing this, I want to ask if this is logical and if anyone out there has tried this.

Lowering memory multiplier to 6, using DDR3-1333Mhz, CPU base clock would have to be: 1333/6=222 Mhz

Then from there, I would start increasing the CPU multiplier (lowest limit is 12):

222x12=2664 Mhz
222x13=2886 Mhz
222x14=3108 Mhz
222x15=3330 Mhz
etc..

Is there any disadvantage or is it bad to set the BCLK that high but keeping the CPU multiplier low?

And also would 222x12 (2.66Ghz) use the same voltage cpu temp as 133x20 (2.66Ghz)?

And lastly, this is where I am more concerned, applying the same logic if I want to achieve DDR3-1600Mhz, CPU base clock would have to be: 1600/6=266 Mhz

266x12=3192 Mhz
266x13=3458 Mhz
266x14=3724 Mhz
266x15=3990 Mhz
etc..

Is there any problem setting the BCLK to 266Mhz with a reasonable CPU OC to 3458Mhz (266x13)?

Please note that my system is already stable at 3.4Ghz using (166x21) but this gives me an 8x memory multiplier to achieve 1333Mhz on my DDR3 RAM, I want 6x but I didnt want to try (222x15) before asking the experts opinions

Thanks.

My system is:

i7-920 C0 2.66Ghz @3.4Ghz (166x21) on Intel Stock Fan. Max Temp 75 C on Overload.
Asus P6T Motherboard
6x2GB "12 GB" Kingston DDR3-1333Mhz (166x8)
Inno3D Geforce GTX 295 1.7GB DDR3
2x1TB WD10EADS 7200RPM S-ATAII 32MB in RAID0
ThermalTake VH8000BWS Armor+ MX Case
Cooler Master Real Power Pro 1250W PSU

Add a reply

Register or log in to remove.

Why? The FSB vs Multi overclocking has no real life difference. Just leave it at 8 and be happy.

Message edited by Shadow703793 on 10-17-2009 at 08:09:21 PM

------------------------------

http://farm3.static.flickr.com/2617/3815217176_0a5be7955d_o.gif

http://farm4.static.flickr.com/3553/3818083596_1a772f7162_o.gif

Reply to Shadow703793

Well from my expierience the higher the multi the better. Hence why the more expensive chips have unlocked multies. So i think you would get a worse speed.

Also, how will your memory work? Your memory would be at like 3ghz, right?

Message quoted 1 times

------------------------------

http://img191.imageshack.us/img191/885/14y1fuqvybiiwayab.gif

Reply to overshocked

Higher FSB speeds and lower multipliers are a trick from days gone by.
10-15 years ago, you got better performance by raising the buss speed and lowering the mulitiplier, because the bus speeds were so miserably slow. You got much more impact if you could crank up the buss, even if it meant lowering the multiplier to keep the CPU stable.
5x100@500mhz or 8x66@528mhz? The 5x100 would result in a much faster overall system, much faster.
However, with the speeds of todays modern quad pumped buses, hypertransport buses, DDR2 and DDR3 memory, you can throw this right out the Window. It simply don't work that way anymore.

Message edited by jitpublisher on 10-17-2009 at 09:18:05 PM

Reply to jitpublisher

just play round see what works best all overclocking is basically just trail an error

Reply to obsidian86

overshocked wrote :

Well from my expierience the higher the multi the better. Hence why the more expensive chips have unlocked multies. So i think you would get a worse speed.

Also, how will your memory work? Your memory would be at like 3ghz, right?

Wrong. Memory multiplier would be at 6x which is the lowest and best. So if my FSB is 222Mhz, my memory will be locked at exactly 222x6=1333Mhz. After that I can play around with CPU multiplier to achieve watever desired clock speed I want, with the theoretical maximum being 222x21=4662Mhz.

And the more expensive chips have unlocked multipliers so that you can achieve higher CPU clock speeds without affecting your RAM speed everytime you want to speed up your CPU, since the RAM is directly linked to FSB speed.

Anyway thanks guys for your answers, although non of them were correct since I found an article that explained exactly what I was looking for, and disproved every response here.

It says always try to achieve the lowest FSBRAM ratio possible and work your way up from there. Because once you start using higher ratios, 8x and 10x or 12x, performance depreciates.

And as for jitpublisher's response, it still does make a significant difference because at the default i7-920 speed of 133x20, you have to set the memory multiplier at 10x to get 1333Mhz on your RAM. While better performance would be running 1333Mhz at 6x with a higher FSB.

Message quoted 1 times

Reply to rorz

rorz wrote :

Wrong. Memory multiplier would be at 6x which is the lowest and best. So if my FSB is 222Mhz, my memory will be locked at exactly 222x6=1333Mhz. After that I can play around with CPU multiplier to achieve watever desired clock speed I want, with the theoretical maximum being 222x21=4662Mhz.

And the more expensive chips have unlocked multipliers so that you can achieve higher CPU clock speeds without affecting your RAM speed everytime you want to speed up your CPU, since the RAM is directly linked to FSB speed.

Anyway thanks guys for your answers, although non of them were correct since I found an article that explained exactly what I was looking for, and disproved every response here.

It says always try to achieve the lowest FSBRAM ratio possible and work your way up from there. Because once you start using higher ratios, 8x and 10x or 12x, performance depreciates.

And as for jitpublisher's response, it still does make a significant difference because at the default i7-920 speed of 133x20, you have to set the memory multiplier at 10x to get 1333Mhz on your RAM. While better performance would be running 1333Mhz at 6x with a higher FSB.

Oh LOL, i thought you ment having a redonculous bclock like 400.

------------------------------

Reply to overshocked

Quote :

I found an article that explained exactly what I was looking for, and disproved every response here.

Was it this? http://www.anandtech.com/memory/sh [...] i=3589&p=1

Quote :

, common desktop applications such as Excel, Photoshop CS4, iTunes, and others just do not benefit that much from improved memory bandwidth or latencies. That brings us to games. Average frame rate improvements improved up to 7% by increasing bandwidth and reducing latencies but we never noticed the difference when actually playing the game.

And

Quote :

The increase in CPU speed outweighed any potential gains in memory bandwidth or latency improvements in our benchmarks.

Yes, there is a bit of a gain, like 7-8% but that's not really noticeable in the real world.

Message edited by Shadow703793 on 10-18-2009 at 05:49:36 PM

------------------------------

Reply to Shadow703793

Focus On

Forum Overclocking : General Discussions - Has anyone tried this before?