Sign in with
Sign up | Sign in
Your question

Upping multiplier affect ram?

Last response: in CPUs
Share
March 5, 2003 1:53:50 AM

Ok heres the deal.

Im using cheap samsung pc2100 ram. SO, when I set my 133/266 2400+xp at 166/333 fsb, the comp wont even start up. people made a good diagnosis, and told me it was most likely beacuse my ram cant run at that speed, as its cheap. Even though I keep the ram speed at 133 (nforce2 board) I was told it will actually run at what the fsb is set at.

BUT, on the flip side, when I keep the FSB at my working OC speed of 148/296 speed, but up the multiplier from simply 7 to 7.5, the same thing happens. No startup. Does this still suggest its the ram?

Something else? Thanks.
March 5, 2003 2:09:01 AM

It suggests that the CPU can't do the speed you are asking for... there are limits, you know.

<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 2:09:03 AM

yeah it's still probably the ram
if you want to run at that speed, make sure your timings are really relaxed, that might do it
i'm thinking.. 2.5-4-4-7-2T
is there an option for a 5:4 ratio by any chance?? probably not, but if so you could run at 166/133 fsb:memory speed
EDIT: after reading Teq's post (he posted at the same time), are you sure you can run at 148 stable? i assumed you were sure that was a stable speed. if you are sure, then it's your ram, otherwise yes it could be your cpu
and no your multiplier does not affect your ram
--------------
I LOVE DANGER DEN WATERCOOLING, they went out of their way to both personalize my kit and change my order when i needed to, i had to change my sig to give them props<P ID="edit"><FONT SIZE=-1><EM>Edited by LtBlue14 on 03/04/03 11:12 PM.</EM></FONT></P>
Related resources
March 5, 2003 2:33:54 AM

Am I missing something here, or are you running WAY underclocked? 148FSB and 7 multiplier = 1036 mhz?
148FSB and 7.5 multiplr = 1110 mhz?
Athlon 2400+ Stock Speed = 1930 mhz.

The processor shouldn't have any trouble with those speeds.

Your multiplier should be 14.5 at 133 FSB (1930 Mhz). A 14.0 multiplier at 148 FSB would give you a clock speed of 2.072 Ghz which would be a mediocre overclock. Multiplier of 7.0 or 7.5 is not what you want.

Get DDR333 or DDR400 RAM dude.



<-----Insert witty sig line here.


<P ID="edit"><FONT SIZE=-1><EM>Edited by Twitch on 03/04/03 11:36 PM.</EM></FONT></P>
March 5, 2003 2:34:39 AM

Yup, there are ratios, though that did not help to much. Hmmm. Seems like FSB too high is hurting...... I dunno. remember my board is the ONLY nforce2 board that does not lock PCImhz, tho Id think 166/333 would be ok.

I guess Ill just wait until Thurs-Fri when my Maze3, thanks to your suggestion, arrives. Pop that in and mess around a little, since the CPU is running to hot to really test anything right now. Then next week hopefully get new ram, and then really try things. Thanks!
March 5, 2003 2:37:52 AM

Right, well see the athlon's are DDR. So the 133 stock means 266. I just say 133 because thats what BIOS says. But the processor is 133/266.
March 5, 2003 2:39:00 AM

I know there are limits. BUt this?

My cpu is made for 133/266fsb..... I would think running 166/333 is not hard to believe. It is a thoroughbred B core.....
March 5, 2003 2:42:37 AM

You were running the FSB at 148/296, right? If that is true and the multiplier you were using was 7 and then 7.5, you were underclocking the processor. You multiply the FSB actual speed (133, or in your case, 148) times the multiplier, and that's how you know your actual speed in megahertz.


<-----Insert witty sig line here.
March 5, 2003 2:44:42 AM

The cpu ran stock at 7. So I had 7X296. I tried increasing it to 7.5X296 but it wouldnt boot. Im back at 7
March 5, 2003 2:51:21 AM

Do this. Set your FSB to 133/266. Set your multiplier at 14. Then sit back and rejoice at how much faster your system suddenly became. And if you want to think that you're running a 3.7 Gig processor, that's up to you. You don't base the multiplier off the 266DDR. You base it off the actual speed of the FSB.



<-----Insert witty sig line here.
March 5, 2003 2:56:34 AM

No, its based off the 266ddr. Thats why Im running 7X148/296 and the listed clock speed in Aida32, BIOS, and Windows Sys information all say around 2070-2074mhz.
March 5, 2003 3:05:48 AM

The thing about overclocking is that you are playing with the tolerances of the chip. Some, where everything went absolutely correct may go higher than 333... some where the chip is stable, but at the edge of it's tolerances may not even tolerate 268.

It's like when they make carbon resistors... they don't make 10k resistors... they make a whole mess of resistors and pick the 10k ones out of the pile. The others will become 33k, 22k, 6.8k and so on... It's all about variations in manufacturing and tolerances.

When AMD makes chips, they will bake and cut a wafer, then mount and test the chips... some will become 2000s, some 2200s, some 2400s, some 2600s etc...

Now if you got a 2600 that's right on the edge of 2600, stable but left will little upward tolerance, it's not going to overclock very well. The next one you buy, may go a lot further... or not as far... it's all in the luck of the draw.



<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 3:09:05 AM

Teq: Have you ever heard of a multiplier multiplying the DDR rating, and not the FSB actual speed? It sounds to me like the multiplier is locked on this cat's motherboard, so he isn't really setting anything with multipliers. I'm not an expert, but I know you obtain speed by multiplying the actual FSB, not the virtual DDR speed.


<-----Insert witty sig line here.
March 5, 2003 3:13:57 AM

No, it's not based off the 266... which is just a system of using both rising and falling edges of a 133mhz clock... The 266fsb claimed by AMD is, at best, a marketing ploy.

It is based off the TRUE fsb speed of 133 (or 148 in your case)... set your FSB to 133 and your multiplier to 14... you will then get the intended speed of the cpu... which I rather suspect will be almost twice as fast as anything you've seen so far.

Oh and, BTW, "DDR" stands for "Double Data Rate" and refers to RAM, not cpu speed.

<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 3:16:45 AM

Ey-yup ... methinks this guy's problem is that he's completely misunderstood how this works.


<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 3:23:14 AM

Are you sure?

I was wrong on my earlier numbers, this is what my CPU summary programs are reporting.

"2200mhz - 7.5X294" Or something around 294. I forget. But that means it doubled what the fsb was set at to get that multiplication.


Xp 2400+'s are rated at 2.0ghz. When I did not oc mine at all and did not touch anything, it reported "1995mhz - 7.5X266" with the normal stock 133 in place?

Im not arguing, just wanting a correct answer so I understand this better.
March 5, 2003 3:29:05 AM

Post deleted by Twitch
March 5, 2003 3:32:02 AM

Actually its not locked... Athlon Tbred Bcores are unlocked, and nforce2 boards are designed to take advantage of this by allowing all settings to be changed in BIOS. This is why I upgraded my cpu and MB to what I did.
March 5, 2003 3:36:45 AM

Did you try setting the FSB to 133/266 and the multiplier to 14 yet?


<-----Insert witty sig line here.
March 5, 2003 3:39:39 AM

No,

Becuse if I do that I will end up with around 1862mhz, less than what I have.

And also, if Im right and it really is 133 multiplied by 2, then Id be diong 14X266 and blow the cpu up. I dont see why my cpu would say 2200mhz "294X7.7.5" if it wasnt right.
March 5, 2003 3:42:07 AM

Well you're right about one thing...the speed would be 1862. The multiplier should be 14.5.

Nothing would blow up. It just wouldn't post.


<-----Insert witty sig line here.
March 5, 2003 3:48:44 AM

Yes, I'm sure... I've built a lot of AMD systems and been around a lot of technical people... You're the first person I've ever seen basing speed on AMD's trumped up 266fsb instead of the true speed of the motherboard: 133.

Honest... that's how it works... 133 X 15 == 1995

I suggested 133 x 14 as a slight underclock while you got the rest of the system stable enough to begin increasing speed.


<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 3:49:32 AM

Just checked, only goes up to 13. But I think Im right, because all my programs say the same thing, although off by a few mhz here and there, thats its 2200mhz at 7.5X290. Aida32, Windows Sys, BIOS.
March 5, 2003 3:51:05 AM

Like I said BIOS only goes up to 13 on mulitplier. New nforce2 board with bios dated Feb 14 2003 so its new. I dunno...
a b à CPUs
March 5, 2003 3:53:52 AM

d00d, several of us have been through that whole thing before, I even told him that the mutiplier is based on the actual pulses per second of the clock generator (133MHz), and that the 266 comes from the fact that it transfers data twice per pulse, but he wouldn't listen to any of us.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
March 5, 2003 3:58:46 AM

What I'm trying to figure out is this:

If he does indeed have an unlocked T-Bred B core and;
His bios does indeed say the multiplier is X7...
Why would his software say he's running over 2gig?
And why would his multipliers only go up to 13?
It's like the motherboard is disregarding his multiplier and reverting to stock, like on a locked core. Either that, or he's misreading something.

<-----Insert witty sig line here.


<P ID="edit"><FONT SIZE=-1><EM>Edited by Twitch on 03/05/03 01:00 AM.</EM></FONT></P>
March 5, 2003 4:05:37 AM

OK here's the poop on my Gigabyte/via board with an XP1800+ in it...

FSB = 133
CPU SPEED == 1533

Lets do the math... 1533/133 == 11.5

And that's exactly how the CPUID utility detects it.

In any case... it sounds like that stupid board is playing a numbers game... Use AMD's silly trumped up doubled numbers, divide the true multplier (set in the chip's jumper fields) by 2 and get the same answer...

So set the thing to 266 x 7.5 and see what the speed is reported in a CPU ID utility...

In any event, as I said before... you cannot count on overclocking every CPU... I had a P1-200 that I ran at 240mhz for years, my friend had one and couldn't push it past 200 at all. Same with the couple of dozen Durons I've sold recently... some of the 1200s will go 1400 even 1600, some wont go past 1300 ... if you can get 296 (well actually, 148) out of it, be happy and carry on.

And people wonder why I say "Don't overclock, buy a faster chip".


<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 4:10:33 AM

Ok, I understand what youre saying... but...

Lets assume I didnt touch the FSB. It would say 133 also. Like I said, when I first start the comp up, it was FSB 133 multiplier 7.5. Bios reported the MHZ speed as if it were 266X7.5 as well as every other program.

I would try what you are saying by like I said the multiplier only goes up to 13. I clocked my machine with prime95. Although I have no clue whatsoever what the numbers mean, they put me in exactly the spot on prime95's chart with other xp2400's. When I overclocked from 133/266 to 148/296 I beat all 2400's, though I remained behind 2800's etc.

Im happy for now until someone reads this post and can verify yours or my stance, before I try anything. Although I dont know what, since the multi only goes to 13....
March 5, 2003 4:11:47 AM

I know Crash, everything I've ever worked on is that way... but from what he's saying the Bios is doing some number games...

Detect the FSB from the chip's jumper field ... multiply by 2.

Detect the Multiplier from the jumper fiels... divide by 2

It gives the same answer... but it's screwy as hell.

Ever seen that before?


<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 4:14:43 AM

Ok... send us the make and model of that motherboard... I want to check it out on the manufacturer's website, see if they will give a clue...


<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 4:16:12 AM

I don't think it can work that way, TEQ. After all, the 2400+ is a 1930 GHZ processor. Under your system, you'd need a clock multiplier of 7.25 to reach that.

Hen...are there .25 multipliers in your bios?



<-----Insert witty sig line here.
March 5, 2003 4:18:59 AM

He has an MSI nForce2. It uses Phoenix Award Bios. It has FSB stepping up to 200 mhz. And I do seem to remember THG saying something about how MSI wins a few performance benchmarks by "cheating" on the FSB speed, but didn't say anything about funky multipliers. If you're correct about the bios doubling the multiplier, eliminating the normal .5 multipliers means every processor installed on an MSI motherboard that would normally require a XX.5 multiplier would end up to be either .5 overclocked, or .5 underclocked. Unless, of course, this BIOS has .25 stepping of the multipliers, which I've never heard of. It doesn't make sense.

<-----Insert witty sig line here.


<P ID="edit"><FONT SIZE=-1><EM>Edited by Twitch on 03/05/03 01:25 AM.</EM></FONT></P>
March 5, 2003 4:29:35 AM

Really? I dunno.......

But, Ill do it. Ill let you know how it turns out then..... I might be on the newest bios anyways. Ok ill be back
March 5, 2003 4:31:17 AM

Hineigger...did you by any chance set your FSB/DRAM ratio to 2:1 when you were trying to overclock the FSB without overclocking the RAM?



<-----Insert witty sig line here.
March 5, 2003 4:33:34 AM

Teq: If you want to see exactly what he's looking at, go to this link and click on the thumbnails of the bios...

<A HREF="http://www.hardocp.com/article.html?art=NDM0LDI=" target="_new">http://www.hardocp.com/article.html?art=NDM0LDI=&lt;/A>

I notice the "CPU Ratio" looks like it's on the same screen as the AGP clock and PCI voltage settings? Is that normal?


<-----Insert witty sig line here.


<P ID="edit"><FONT SIZE=-1><EM>Edited by Twitch on 03/05/03 01:36 AM.</EM></FONT></P>
March 5, 2003 4:35:13 AM

Maybe this will give some insite or make a bit of sense......

ive got a T-Bred 'B' 1700+...now in my bios i can select any multipler from 5.0 - 12.5.....i dont get any options higher or lower......now....ive been told if i cut the 4th and 5th L5 bridges(or mayeb its L i cant remember, but its the bridges right below the L1's) that my multipliers will change.....like 5.0 will now be 13.0 and 5.5 will be 13.5 and so on and so forth....

Maybe your reported 7.0X is ACTUALLY 15.0 but all utilities show it as the 7.0...as 7.0 would be the "default" when really its 15.0.....u know what i mean ????

<A HREF="http://www.anandtech.com/mysystemrig.html?id=13597" target="_new">-MeTaL RoCkEr</A>
March 5, 2003 4:35:51 AM

No. You can do ratios but I picked auto, and it said it kept 133..... Tho I dont know if you can trust it.
March 5, 2003 4:38:03 AM

That could be it..... Im not sure.

The thing that gets me is the programs are all reporting 2200 mhz with 266X7.5. If youre right, that means the others are right. 266 = 133 but 7.5 = 15. You know?

By the way to the others, I have the latest BIOS and 13 is still highest ratio. Maybe the guy above is right.
March 5, 2003 4:39:27 AM

Ok, Metalrocker, I think you're on to something. I remember reading about something like that on the overclocking thread. It was a member who couldn't get the max overclock on his 1700+ because he didn't particularly want to cut the bridges, and his bios wouldn't give him the right multipliers. Seemed like there are only so many muliplier slots, and depending an whether or not the chip is modded, those options change...and it did seem like 7 was the default on his MB too...



<-----Insert witty sig line here.
March 5, 2003 4:39:57 AM

That's ok... I downloaded the user manual...

It says FSB can be set from 100 to 200 in 1 mhz steps.
Multipliers can be set up to 16.

It also says the RAM clock is a fucntion of FSB as a ratio between FSB and RAM 1, 1.1, 1.2 etc. The "autodetect" is called "BySPD" which is a special port on the ram where it's parameters should be read...

SO! FSB == 133, Multiplier == 15, Ram == "by SPD" and it should work...


If it doesn't... I give up.



<b>(</b>It ain't better if it don't work.<b>)</b>
March 5, 2003 4:44:05 AM

Okay. I give up too. I still think that the bottom line is that dude needs PC2700 or, better yet, PC3200 RAM. He's overclocking with PC2100.



<-----Insert witty sig line here.
March 5, 2003 4:56:00 AM

That may have been my thread =)

I wanted to use multipliers above 12.5.... =)

and the only way to do this on my board is to cut the 4th and 5th L5 (or L3) bridge....which will than make my board default to 15x Multiplier....and i believe the multipleri r than interpreted as follows..

Here is what <A HREF="http://forumz.tomshardware.com/modules.php?name=Forums&..." target="_new">phsstpok</A> wrote about what i stated about the interpreting of 7.0 as 15 etc......
Quote:
Without BIOS multipliers of 13.5x-24x there is only one thing that can be done, mod the chip. By cutting the 5th L3 bridge you gain access to the higher multipliers but it's not as simple as you would hope. (I'll explain this in a bit).

After you boot you run into a remapping problem. If you set your multipliers in BIOS and FSB to 100 Mhz you'll get the following remappings.

7x ----> 15x
7.5x --> 22.5x
8x ----> 16x
8.5x --> 16.5x
9x ----> 17x
9.5x --> 18x
10x ---> 23x
10.5x -> 24x
11.5x -> 19x
12.5x/13x -> 20x

That's the good news the bad news is it only works with low bus speeds 100+ mhz. At 133 Mhz you are stuck at the default multiplier at which you mod'd your chip. Don't know why this is true.

This fact makes it important to choose your default multiplier wisely. I liked 15x because 15 x 133 is 2000 Mhz a decent overclock yet still within the capabilities of Tbred B 1700+. If you want to use a higher bus speed you'll want a lower default multiplier but remember you won't be able to change the multiplier from BIOS any longer so too low a multiplier may limit your overclock. 15 x 133 gives a good starting point, 2000 mhz (XP 2400+). Overclocking the bus from there lets you get into the typical Tbred B 1700+ overclock levels. 150 Mhz gets 2250 mhz. 160 mhz gets 2400 mhz which you probably can't get much higher with a 1700+.

2250 is my max with a rather high 2.15 VCore but I've settled on 2100 mhz, 15 x 140. I've only tested with KT133 and KT133A mobos but I've read that people have run into the same issues with KT133E, KT266, KT266A, and KT333 mobos.

If you want to try the same multiplier (15x) then the L3 bridges should look like this C--C--C--O--O. For an XP1700+ you need to cut the 4th and 5th L3 bridges.

So far I've only read about people using Tbred B 2100+ and 2400+ and XP 1700+ or XP 1800+ being mod'd to look like one of the first two.

So now u see what im saying ??

I think its just a wierd thing the Motherboard is reportign to all software ID Programs....its saying its usign as Multtiplier of 7x when in ACTUAL its 15x......get it ???

<A HREF="http://www.anandtech.com/mysystemrig.html?id=13597" target="_new">-MeTaL RoCkEr</A>
March 5, 2003 5:00:24 AM

Well, you may be right, but weren't you using a KT133A motherboard? Because this guy's got a new MSI nForce2, and according to the manual, should have settings up to 16X. But he says his bios is only giving him up to 13X.

Did you look at the link I gave above? What is the CPU Ratio? Is that the multiplier, or does it have something to do with the AGP / PCI Bus voltages or something? I've only ever heard of it referred to as Multipliers, and I'm wondering if he's setting the wrong thing.


<-----Insert witty sig line here.
March 5, 2003 5:06:27 AM

the Chipset is not the thing...i was saying my T-Bred 'A' 1800+ is on a KT133A and my T-Bred 'B' is on an SiS 745 platform.....i cant choose multipliers above 12.5 cuz these chipsets dont support it UNLESS i cut the 4th and 5th L3 bridge......

My mere point to my last post was to show that hsi board is just displaying 7x instead of 15x but is ACTUALLY using 15x....its just a read error or misinterpretation....

as 7x = 15x when the Multipliers r unlocked....i think hsi motherboard is just getting confused......

<A HREF="http://www.anandtech.com/mysystemrig.html?id=13597" target="_new">-MeTaL RoCkEr</A>
a b à CPUs
March 5, 2003 6:06:53 PM

I'm sure you'll find (if you look hard enough) that some multipliers are repeated. 7x for example is also the setting for a higher multiplier, but I forgot what it is. I'm sure an AMD Expert will be here shortly to clarify.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
a b à CPUs
March 5, 2003 6:14:45 PM

It sounds like the "7x" is actually being used as one of the higher multipliers. The hardware setting for each multiplier only goes so high, then repeats. I don't remember at which point it repeats or which setting it repeats from...but if the 13x setting is a repetition of the 6x setting, the 14x setting would be a repetition of the 7x setting. I know the two points I speak of are near 13x and 6x, but I could be as far as 1x off on either or both, since I haven't really looked into it.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
a b à CPUs
March 5, 2003 6:14:45 PM

It sounds like the "7x" is actually being used as one of the higher multipliers. The hardware setting for each multiplier only goes so high, then repeats. I don't remember at which point it repeats or which setting it repeats from...but if the 13x setting is a repetition of the 6x setting, the 14x setting would be a repetition of the 7x setting. I know the two points I speak of are near 13x and 6x, but I could be as far as 1x off on either or both, since I haven't really looked into it.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
March 5, 2003 8:19:32 PM

MeTaL RoCkEr, please stop quoting me.

The remapping only applies to chipsets that can't manipulate the higher multipliers, like KT133 thru KT333 and only if the 5th L3 bridge is closed. Nforce2 is not included in these motherboards. Nforce2 can control the higher multipliers.

If the 5th L3 bridge is open (as it is on any Tbreds lower than XP2100+) then multipliers, 5x-12.5x, all work normally on all motherboards.

It becomes interesting <b>when the 5th L3 bridge is closed</b>. This is true on Tbreds XP2100+ and higher when they are stock and also other Tbreds when they are <b>modified</b>.

If the motherboard is unable to manipulate the higher multipliers then when you adjust multipliers in BIOS you get a remapping, 7x becomes 15x, 10x becomes 23x, like that.

Some mobos have problems with this with higher bus speeds (133 mhz and higher) but this is not part of this discussion.


Now lets move on to the nForce2 mobos. These boards <b>can manipulate the higher mulitpliers</b>. However, it works in a funny way.

If you have processor lower than XP2100+ then you only see 5x-12.5x in BIOS. If you have an XP2100+ and higher (or a modified processor) then you can see only higher multipliers. (This may have been fixed with a BIOS revision).

To manipulate the multipliers, at all, the processor must be unlocked. The L1 bridges must be closed.

All Tbred B's are unlocked. Tbred A, not sure which ones are unlocked but I think XP2200+ and higher. The lower Tbred A's are locked. Palominos are all locked.

<b>How do you get to the higher multipliers when you are using a Tbred B XP1700+ or XP1800+?</b>

You cut the 5th L3 bridge.

When you are using an nForce2 mobo cutting the 5th L3 bridge will add 8x to the default multiplier. 11x will become 19x and 12x will become 20x. <b>This is not a remapping</b>. You are simply adding the high-order multiplier bit, e.g. 11x + <b>8x</b> = 19x. The problem is that now your PC will want to POST with a very high multiplier, 19x for the mod'd XP1700+. This is not good because 19 x 133 is 2533 mhz which is too much of an overclock.

To fix the problem you mod some more L3 bridges to get a lower default multiplier. If you have a good HSF I suggest 15x. If not choose a lower multiplier. With an nForce2 it doesn't really matter what you choose as long as you can get it to POST. After it POSTs you'll be able to change the higher multipliers in BIOS just as if you had true XP2100+ or higher.

<b>99% is great, unless you are talking about system stability</b>
!