• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Can't OC my memory,,,,, WHY ???

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

ZeroAim

Registered
Joined
Sep 7, 2003
Location
Denmark
Hi.

I'm pretty new here at Overclockers.com, but i hope someone here can help me with my "little" problem.

As you can see in my profile i run on a Asus P4P800 Deluxe MB with dual Geil PC 3500 ram. For those who dosen't know the MB, it's locked so that you have to change bouth the CPU and RAM speed simultaniasly. The CPU mutiplier is loocked to.

Now to my problem: When I raise the FSB i gain on my CPU speed but my RAM speed drops ? Neither the CPU nor the RAM overheets at all (CPU = 28c Ram 30 - 35c). At 10% OC RAM proformance drops around 20%... I've tried changing the RAM voltage step by step to the max limit at 2,85v, no change. What do I do ???

I use SiSandra software

I use the Asus bios ver. 1009

Hope someone can help me
Henrik Madsen

N.B. I'm from Denmark, so I hope you understand my english.
 
ZeroAim said:
Hi.

. For those who dosen't know the MB, it's locked so that you have to change bouth the CPU and RAM speed simultaniasly.

not true.



ZeroAim said:

The CPU mutiplier is loocked to.

All non-engineering samples have locked multipliers.

ZeroAim said:
Now to my problem: When I raise the FSB i gain on my CPU speed but my RAM speed drops ?
[/B]

have you tried setting ram speed and raising bus async?


bios_select_memclock.jpg
 
Last edited:
Hi.

Sorry about the untrue statment in my original posting :( but it just shows that I'm a rookie.....

I've tried setting the RAM speed to 266, 320 and 400, but I dont really know what it means :( ... And "rasing bus async" ?????

I'm sure this is some very stupid questions,,,,, sorry !!!
 
First your using Geil mem which is most likely the reason your not able to oc more than 10%. With Geil it is a hit or miss proposition and not all sticks are the same. The spd on their ram can differ between sticks of the same class.

Second, if your getting a 20% decrease in performance of ram after a 10% increase in FSB there is something seriously wrong with your oc or your running them asych without a corresponding increase in FSB to compensate for the loss of the PAT. Even then I would find it hard to believe you are getting a 20% decrease in performance, but I guess its possible.

Try this set your mem to run at 320 instead of 400, then raise your FSB to 235 and see what happens. Also up the vdimm to the highest setting that the board has (Geil mem loves the increase in volts) and set your vcore to 1.6. Mem timings to 2.5 -4-4-7. If it won't boot then set mem timings to 3-4-4-8 and try again. You didn't post your cpu specs, but I am assuming you have a 2.4c or above, so a 235 fsb should be no prob running asych with your mem at that speed.

If you want help your really need to be more specific as to what your set-up is, such as whether the GEil mem is the ultra or dragon version, cpu, and bios settings that are giving you those results.

However, I would bet that its your mem, as you can see from my sig I biased towards Geil as this is my second set of sticks that cannot run at spec or anywhere near there without crashing even at 3-4-4-8 timings.
 
I've found out the problem that caused the big drop in mem speed, it was due to very poor timings that the mem proforms very bad with.

I've made some tests and here's the results from SiSandraMax:

The RAM is Geil PC3500 Ultra

STANDART NO OC !!!
FSB: 200Mhz
Ram set: 400Mhz/auto
Ram speed: 2*200Mhz = 400Mhz... 2-2-2-5
RESULTS:
Buf: 5015/5034
Unbuf: 4343/4349

MAX OC
Ram set: 400Mhz
FSB: 220Mhz
Ram speed: 2*220Mhz = 440Mhz... 2-3-3-6
Volt 2,85 (max on my board)
RESULTS:
Buf: 5016/5041
Unbuf: 4116/4116

(BEST) OC
Ram set: 320Mhz
FSB 240Mhz
Ram speed: 2*192Mhz = 384Mhz... 2-3-3-6
Volt 2,85
RESULTS:
Buf: 5157/5141
Unbuf: 3586/3588

I've triede setting the ram to 266Mhz but the results were poor. (Maybe I'm doing somthing wrong)

I've found that the mem proforms best at timing 2-3-3-6 and when I reach the max OC at this setting there is no change in stability by raising the settings to 2,5-4-4-8, and it want start if I set the cas latency to 3,0 at any speed OC or not.

At the 2 max OC results above the system is not 100% stable, but that might be due to CPU overheating, seeing that I use standart BOX cooler.

As the results shows I'm only just able to make the mem run at it's standart speed (433Mhz), but no matter what I do I don't seem to be able to get bette results then with the standart (no OC) setting, but in the link below the exact same RAM is testet to run at 495Mhz and 470Mhz at 2,85v with out any problems. I'm confused.

http://www.extremeoverclocking.com/reviews/memory/GeIL_PC3500_Ultra_Plat_1.html

What do I do when I really want my RAM to run faster ???
 
"What do I do when I really want my RAM to run faster ???"

trust me. running ram faster than the fsb doesn't help performance.

what p4 do you own?

try lowering the memory ratio and seeing how high the processor can go. I'll post back later with a better recommendation.
 
240*14 = 3360.....thats pretty good for stock cooling.

"FSB: 220Mhz
Ram speed: 2*220Mhz = 440Mhz... 2-3-3-6"

I'd go with the 3360 on cpu and 192Mhz on the ram over the 220mhz on the ram and 3080 on the cpu.
 
zabomb4163 said:
240*14 = 3360.....thats pretty good for stock cooling.

Do you thing it's the CPU that limets the OC and not the RAM ??? (that would be great cause I'm going for watercooling soon)

zabomb4163 said:
I'd go with the 3360 on cpu and 192Mhz on the ram over the 220mhz on the ram and 3080 on the cpu.

I agre with you on that one, but I still don't understand that I'm unable to go higher on the RAM, when the link in my last post shows that the exact same RAM OC's to 495Mhz ???
 
I agre with you on that one, but I still don't understand that I'm unable to go higher on the RAM, when the link in my last post shows that the exact same RAM OC's to 495Mhz ???

You are trying to compare apples to oranges here. The test system used in the review was only able to reach that oc because they could oc the processor to hell and back becasue of the phase change cooling AND the epox board had vdimm to 3.2. Your mem is right on target given your oc, and the fact that you cannot oc more is because of the limitations on the processor, not the mem. 2.8's are not the best processors for getting the most out of your mem becasue of the oc unfriendly multiplier:).

One more thing, benching mem is highly dependant on the OS. There is a real good article posted on the home page talking about this very fact. You can increase your sandra marks by 200 to 300 just by going in and eliminating unnecessary services.

Your oc is fine for the cooling you have, but don't expect any spectacular gains out of watercooling with the 2.8, as you will see from my sig below the processor will top out anywhere from 3.45 to 3.55. Even then the increase in FSB will not be that significant for your mem. As I stated above, and as shown by the review, Geil mem needs high vs to really crank.

The only way your going to get this mem to crank is by doing a volt mod if your board does not go above 2.85; get a board that does go to 3.1; or get a 2.4c with water cooling that will allow you to go higher on the fsb at lower ratios.

Your BEST oc seems to be right on target for the processor and mem. I have the same processor and I am water cooling, but I am still only able to get the cpu to a 246 fsb at 5:4 ratios using mushkin mem at the same rated speed. Its not so much the mhz you can get the meme to runa at, but the combination of fsb and timings that get the performance. In other words, your mem seems to be performing as it should, it s the boards vdimm limitation and the processor that are keeping your from getting the most out of them.

I have the same prob with the IS7-G, but I am going to do the vdimm mod on same to overcome the volts prob, and will more than likely get a 2.4c for my main rig to get the most of the system.
 
Back