Page 1 of 1
PC8500 Memory Timings
Posted: Tue Feb 19, 2008 11:40 pm
by sygnus21
I have 2gig of Corsair Dominator PC8500 memory coupled with a Q6600 (2.4gig) on an Intel 975XBX2 MB. Vista Ultimate 32bit is the OS. The RAM is running at 800MHz.
I was running the CPU @ 2.8gig, and had my memory timings at 4-4-4-12.
I decided to run the CPU @ 3.06gig but found that Vista won’t boot unless I have the timings set @ 5-5-5-15.
The default timing of this memory is 5-5-5-18, voltage is 2.2v.
Thus my question: What the better performer? 2.8gigs with memory timings of 4-4-4-12, or 3.06gigs with 5-5-5-15?
I’m a little fuzzy on the memory timing thing so don’t shoot me.
Thanks
edited to add BIOS settings:
BIOS Settings: FSB = 334, Frequency Reference = 266, Memory Frequency = 800
BIOS Settings: FSB = 334*9 = 3006
FSB/Mem Ratio = 266/800/2 = 0.666 = 2/3
Memory clock = 334 fsb/0.666 memory freq = 502 (502*2 = 1004)
DDR2 Equivalent = 502*2 = 1002Mhz (I’m running DDR2 PC8500 which will run up to 1066MHz)
The above calculations were based on this model here:
http://www.peakin.com/xbx2/, or you can see a calculator for the Intel 975XBX2 board here:
http://www.peakin.com/xbx2/calculator.html
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 1:32 am
by ibleet
Thats a very good question and I'm sure you will hear arguments going each way.
I personally would choose the quad at 3.0Ghz. I ran benchmarks with different cpu speeds and ram timings, and I always noticed the biggest difference with the cpu changes.
I overclocked a q6600 to 3.2Ghz and ran 3DMark06 with 6-6-6-18 ram timings...I then changed the timings to 4-4-4-15 and the benchmark didn't increase more than 100.
Now I realize that benchmark is GPU and CPU intensive, but ram does come into play.
Your actual mileage may vary.

Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 6:22 am
by DMB2000uk
What MHz was the RAM being overclocked by (or what divider was it at)?
If you have a good chip and board, you might be able to get the PC8500@4-4-4-12 if you lower the RAM divider and push the CPU further.
Dan
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 10:19 am
by sygnus21
DMB2000uk wrote:What MHz was the RAM being overclocked by (or what divider was it at)?
If you have a good chip and board, you might be able to get the PC8500@4-4-4-12 if you lower the RAM divider and push the CPU further.
Dan
"I have 2gig of Corsair Dominator PC8500 memory coupled with a Q6600 (2.4gig) on an Intel 975XBX2 MB."
The RAM is running at 800MHz.
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 11:03 am
by sygnus21
Guess I should include this info:
BIOS Settings: FSB = 334, Frequency Reference = 266, Memory Frequency = 800
BIOS Settings: FSB = 334*9 = 3006
FSB/Mem Ratio = 266/800/2 = 0.666 = 2/3
Memory clock = 334 fsb/0.666 memory freq = 502 (502*2 = 1004)
DDR2 Equivalent = 502*2 = 1002Mhz (I’m running DDR2 PC8500 which will run up to 1066MHz)
The above calculations were based on this model here:
http://www.peakin.com/xbx2/, or you can see a calculator for the Intel 975XBX2 board here:
http://www.peakin.com/xbx2/calculator.html
Thanks.
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 11:30 am
by DMB2000uk
The second post was the info I was after, thanks.
Have you tried starting at 2.8Ghz, and upping the FSB MHz a little at a time to see at what speed your RAM fails at with the tighter timings?
Have you also tried ruling out the NB as not being able to cope with those timings while overclocked. (To eliminate it try increasing the NB/MCH voltage a little bit to see if it will then boot).
I take it that you are unhappy with increasing voltage to the RAM to see if it will work at the tighter timings with more voltage.
Let us know how you get on.
Dan
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 12:39 pm
by sygnus21
DMB2000uk wrote:The second post was the info I was after, thanks.
Have you tried starting at 2.8Ghz, and upping the FSB MHz a little at a time to see at what speed your RAM fails at with the tighter timings?
As my post stated, I was at 2.8gig then jumped to 3gig; However I did notice that I can run at the low end of 2.9Ghz (323*9 = 2907), with the memory running at 978MHz, timings 4-4-4-12. I guess that means anything past 1000MHz the timings get loose. Hmmm I have to run memtest to see if this is stable, if so, I'll prpbably go with 2.9gig at 4-4-4-12
Have you also tried ruling out the NB as not being able to cope with those timings while overclocked. (To eliminate it try increasing the NB/MCH voltage a little bit to see if it will then boot).
I've bumped the MCH by .50 which is recommended by Corsair.
I take it that you are unhappy with increasing voltage to the RAM to see if it will work at the tighter timings with more voltage.
Corsair's recommended voltage is 2.2v - In fact they guarantee the RAM to run at 1066MHz at 5-5-5-18, 2.2v. So no, I havent bumped the voltage pass 2.2.
Let us know how you get on.
Dan
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 1:05 pm
by ibleet
Edited because I could not delete...sorry.
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 1:36 pm
by sygnus21
I have a B3 stepping.
I just tested 2.9gig @ 4-4-4-12 timings with 3DMark06 and got a score of 12051, whereas 3gig gave me a score of 12113. So I lost a whole 62 points. Basically all I get out of that is that there really isn't any difference between to the two.
I'm going to go back to 3+ gigs though with the looser timings of 5-5-5-15. That ran stable for me. As far as cooling, I have a Zalman CNPS7700-AlCu seen here:
http://www.pcbuzz.com/_e/Zalman/product ... 0_AICu.htm. It's big, though low profile, quiet (fan set to high with fan mate ll), and it gets the job done. My temps dropped by 12 degress celsius.
But getting back to my question, what I really wanted to know was - is
2.8 - 2.9gig at 4-4-4-12 better then 3gig at 5-5-5-15? I guess that's not easily answered as It depends on programs ran??? 3DMark says no, but what would others say?
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 3:05 pm
by DMB2000uk
Like you say, it's dependent on what you want to run. But the vast majority of (If not all) programs will like the faster CPU speed than tighter RAM timings.
Also there is something psychological about 3Ghz that makes it sound better to run it at. It just sounds nicer than 2-point-something GHz.
Dan
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 3:46 pm
by sygnus21
Well I just tried to overclock to 3.2gigs...system wouldn't post - had to remove & reinstall CMOS battery and reset CPU & Memory settings to get system to post. Tried it at 3.1...same results, same remedy

So now I'm running at 3gig (334*9 = 3006), memory timings at 5-5-5-15. Seems my B3 won't do more then 3gigs.
So now my question becomes: 2.9gigs with 4-4-4-12 timing or 3.0gigs with 5-5-5-15 timing?
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 4:59 pm
by DMB2000uk
I trust that you took the RAM down a divider to the 667Mhz one, and tried increasing the CPU core voltage?
Dan
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 6:14 pm
by sygnus21
DMB2000uk wrote:I trust that you took the RAM down a divider to the 667Mhz one, and tried increasing the CPU core voltage?
Dan
Actually, I didn't...the whole point of buying the PC8500 was so I could run the divider at 800Mhz. I don't want to bottleneck the RAM. As far as upping the CPU voltage, I was under the impression that you didn't have to up the voltage as long as you stayed below 3.2gigs. Considering the heat this thing puts out, the last thing I want to do is add more voltage! However I do have "Enhanced Power Slope" enabled.
At any rate, seems like I'll be running at 2.9gig anyway. After bumping my CPU to 3gig, I ran Memtest and it crashed/failed. Dropped the CPU to 2.9 (323*9) and memtest passed. So now I'm running 2.9gigs with mem timings of 4-4-4-12, RAM divider at 800Mhz.
If I'm missing something please let me know as it's been awhile since I've overclocked anything.
Thanks
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 6:31 pm
by DMB2000uk
sygnus21 wrote:Actually, I didn't...the whole point of buying the PC8500 was so I could run the divider at 800Mhz. I don't want to bottleneck the RAM. As far as upping the CPU voltage, I was under the impression that you didn't have to up the voltage as long as you stayed below 3.2gigs. Considering the heat this thing puts out, the last thing I want to do is add more voltage! However I do have "Enhanced Power Slope" enabled.
At any rate, seems like I'll be running at 2.9gig anyway. After bumping my CPU to 3gig, I ran Memtest and it crashed/failed. Dropped the CPU to 2.9 (323*9) and memtest passed. So now I'm running 2.9gigs with mem timings of 4-4-4-12, RAM divider at 800Mhz.
If I'm missing something please let me know as it's been awhile since I've overclocked anything.
Thanks
If you want to keep the RAM running on that divider then fair enough.
It depends on how good the core's in your Q6600 are as to how far you can go before more voltage is needed. Some are better than others. It's also understandable that you don't want to add more heat to the furnace by increasing volts.
I don't know of any other tweaks that you can do, So it looks like the answer to your question has been forced by default.
Dan
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 7:07 pm
by sygnus21
Thanks Dan, indeed the answer to my question has been forced on me. That's cool though.
Thanks for the help and suggestions.
Re: PC8500 Memory Timings
Posted: Wed Feb 20, 2008 11:19 pm
by ibleet
The Go stepping model will typically go to 3.2Ghz with no voltage increase. Im not surprised that the B3 will not.
Re: PC8500 Memory Timings
Posted: Sun May 04, 2008 6:10 pm
by IRQ Conflict
Not necessarily. I just bought a Q6600 G0 stepping. The newer batch#'s (mine) L805A320 for example required a volt bump to 1.350 from 1.325 to be prime stable @3.2Ghz and 57c.
If you can find a G0 with an earlier batch# "L7" it should give better results. I think with the "L8" and later we are starting to scrape the bottom of the barrel.
Re: PC8500 Memory Timings
Posted: Wed May 07, 2008 9:34 pm
by ibleet
On average I would say that you just got unlucky. You roll the dice even with Go stepping as all chips have different die sizes. Bottom of the barrel is a good analogy.
