differences btwn. X3210/X3220/X3230

Assuming the same stepping (G0), would the only difference between a xeon X3210, X3220, and X3230 just be the multiplier? Or are there manufacturing differences as well that would justify the drastic price increases?
From what I've read here, it seems like going up (or down) a processor line has varying multiplier values, but all other specifications seem the same.

This would lead me to ask, if you were planning on getting a X3220 just to drop the multiplier to 8 (it has a stock multi of 9), would it be better (less money, same performance) to buy a X3210 with a stock multi of 8?

I don't expect lengthy answers, just hoping some people can affirm or disprove what I believe to be true

edit: more info on my current situation and why I am asking this..
I am getting a gigabyte p35 motherboard (ds3r or ds3p I've yet to decide). From various reviews, these motherboards can reach very high FSBs that I wouldnt even bother trying with my quad core.

The memory I have bought are Crucial Ballistix sticks with D9 chips, which easily overclock to 900mhz to increase memory bandwidth, with timings that I am fine with. DDR2 running at 900mhz would mean I would have to be running a 450mhz FSB, which is too much for a 9x multiplier anyways like on a Q6600 (I would prefer 8x450 over 9x500, ASSUMING that a quad-core can even reach that high. In any case, I want to make the most of my memory and will therefore try my best to have a FSB of 400 or higher)

lastly, I am air cooling a quad-core (Q6600, X3220, or X3210) with a thermalright HSF, so no "extreme overclocking" here that would give me any need for a high multiplier,

I tend to think this is true, but the 20 and 30 may be binned slightly higher if Intel discriminates that much in minor differences b/n CPUs.

yeah as far as i can tell its the same thing - since they are both xeons they should be binned really high anyway (although i do not pretend to know all about intel's internal processes)

if you did have the x3220 why would you lower the multiplier though?

if you did have the x3220 why would you lower the multiplier though?

Click to expand...

To keep the memory from clocking so high.

I thought it was to keep the cpu from clocker higher if you have weak cooling.
Also if the chip has stability issues at 3.6ghz, then you can run it at 3.2ghz with a 1:1 mem ratio, with a 8X multi (my goal).

I was thinking of buying a quad core and running it 8 x 400 = 3200mhz
vs running one at 9 x 400 = 3600mhz.

dfonda said:

To keep the memory from clocking so high.

Click to expand...

I think that he was talking about lowering the cpu's multiplier, not the system memory multiplier...lowering the cpu multi and keeping the GHz the same would increase memory speed, not lower it.

With respect to the question about why not just get a cpu with an 8 multi vs one with a 9 multi and dropping the multiplier...in most cases you would be better off going with the 8 multi provided you memory and motherboard are up to the task and you use conventional air or water cooling. The performance boost from the lower multi would be due to the increased memory bandwidth at the same CPU clock speeds.

One other problem with using a lower-than-default multiplier has to do with overclocking of the northbridge. If you are not familiar with this topic, check out this linked article:

http://www.thetechrepository.com/showthread.php?p=32#post32

The main advantage of the higher multi comes to play more with (1) use of an i975 chipset mobo that does not reach high FSB or (2) for the guys that use extreme cooling...in those cases the higher multi is needed in order to reach ultra high CPU GHz.

Reefa_Madness said:

I think that he was talking about lowering the cpu's multiplier, not the system memory multiplier...lowering the cpu multi and keeping the GHz the same would increase memory speed, not lower it.

With respect to the question about why not just get a cpu with an 8 multi vs one with a 9 multi and dropping the multiplier...in most cases you would be better off going with the 8 multi provided you memory and motherboard are up to the task and you use conventional air or water cooling. The performance boost from the lower multi would be due to the increased memory bandwidth at the same CPU clock speeds.

One other problem with using a lower-than-default multiplier has to do with overclocking of the northbridge. If you are not familiar with this topic, check out this linked article:

http://www.thetechrepository.com/showthread.php?p=32#post32

The main advantage of the higher multi comes to play more with (1) use of an i975 chipset mobo that does not reach high FSB or (2) for the guys that use extreme cooling...in those cases the higher multi is needed in order to reach ultra high CPU GHz.

Click to expand...

Thanks for the article, will look to run fsb at 401

Reefa_Madness said:

I think that he was talking about lowering the cpu's multiplier, not the system memory multiplier...lowering the cpu multi and keeping the GHz the same would increase memory speed, not lower it.

With respect to the question about why not just get a cpu with an 8 multi vs one with a 9 multi and dropping the multiplier...in most cases you would be better off going with the 8 multi provided you memory and motherboard are up to the task and you use conventional air or water cooling. The performance boost from the lower multi would be due to the increased memory bandwidth at the same CPU clock speeds.

One other problem with using a lower-than-default multiplier has to do with overclocking of the northbridge. If you are not familiar with this topic, check out this linked article:

http://www.thetechrepository.com/showthread.php?p=32#post32

The main advantage of the higher multi comes to play more with (1) use of an i975 chipset mobo that does not reach high FSB or (2) for the guys that use extreme cooling...in those cases the higher multi is needed in order to reach ultra high CPU GHz.

Click to expand...

I remember that article and it turns out it may have been an error. The strap and mem bandwidth does change but the NB itself isn't oc'd internally iirc. Also the strap changes are not an issue with P35 chipset.

Reefa_Madness said:

I think that he was talking about lowering the cpu's multiplier, not the system memory multiplier...lowering the cpu multi and keeping the GHz the same would increase memory speed, not lower it.

With respect to the question about why not just get a cpu with an 8 multi vs one with a 9 multi and dropping the multiplier...in most cases you would be better off going with the 8 multi provided you memory and motherboard are up to the task and you use conventional air or water cooling. The performance boost from the lower multi would be due to the increased memory bandwidth at the same CPU clock speeds.

One other problem with using a lower-than-default multiplier has to do with overclocking of the northbridge. If you are not familiar with this topic, check out this linked article:

http://www.thetechrepository.com/showthread.php?p=32#post32

The main advantage of the higher multi comes to play more with (1) use of an i975 chipset mobo that does not reach high FSB or (2) for the guys that use extreme cooling...in those cases the higher multi is needed in order to reach ultra high CPU GHz.

Click to expand...

Yep!

MadMan007 said:

I remember that article and it turns out it may have been an error. The strap and mem bandwidth does change but the NB itself isn't oc'd internally iirc. Also the strap changes are not an issue with P35 chipset.

Click to expand...

Do you happen to have any links to discussions on this topic...if this has been debunked, I sure would like to read up on it. Certainly don't want to be spreading incorrect info, but at the same time, I haven't seen any "Corrections" posted on any of the links that I have on the topic.

also...no mention of P35 chipset in that first post, at least none that I could find.

Reefa_Madness said:

also...no mention of P35 chipset in that first post, at least none that I could find.

Click to expand...

well there is now ...

shirker said:

edit: more info on my current situation and why I am asking this..
I am getting a gigabyte p35 motherboard (ds3r or ds3p I've yet to decide). From various reviews, these motherboards can reach very high FSBs that I wouldnt even bother trying with my quad core.

The memory I have bought are Crucial Ballistix sticks with D9 chips, which easily overclock to 900mhz to increase memory bandwidth, with timings that I am fine with. DDR2 running at 900mhz would mean I would have to be running a 450mhz FSB, which is too much for a 9x multiplier anyways like on a Q6600 (I would prefer 8x450 over 9x500, ASSUMING that a quad-core can even reach that high. In any case, I want to make the most of my memory and will therefore try my best to have a FSB of 400 or higher)

lastly, I am air cooling a quad-core (Q6600, X3220, or X3210) with a thermalright HSF, so no "extreme overclocking" here that would give me any need for a high multiplier

Click to expand...

that said, if a G0 x3210 can OC to the same speeds as a x3220 or q6600 equivalent, there is no question that the x3210 would be a better choice for me, since I currently intend on running a G0 Q6600 at 8x450 (3.6ghz) or 8x425(3.4ghz), unless someone could kindly point out any obstacles I might run into by trying to use a high fsb like 450mhz with a quad-core

Also, any input on the original X3210 vs X3220 vx X3230 question?
thanks to everyone so far

Your original question was answered right away: the max multi is different, and they may be binned differently if you took large samples but the chip-to-chip variance on a given luck-of-the-draw purchase could be equivalent. As for the FSB I know P35 boards can do upwards of 450 FSB with quads, you may want to research your board choice in particular to make sure though.

Reefa - I can't find any solid links now, but I think what happened was the strap changes, which changed internal chipset latencies thus affecting bandwidth which is what people were seeing as performance changes, was tricking some of the system-reading software like SiSoft Sandra into thinking the NB was running faster. Sort of the way that Coretemp will often display wrong final speed when you alter a CPU multiplier in BIOS. I think Gautam might remember more about this..

MadMan007 said:

Your original question was answered right away: the max multi is different, and they may be binned differently if you took large samples but the chip-to-chip variance on a given luck-of-the-draw purchase could be equivalent.

Click to expand...

right, but the answers were far from absolute and firm. the third time's the charm though, it seems like there really isn't any difference besides the multiplier. Would it be safe to assume that a X3210 with 8x multi can overclock to the same speeds as a X3220 dropped down to an 8x multi?
All signs are pointing to yes, but sometimes the most obvious questions can overlook some critical bit of information that might affect the answer..

shirker said:

I currently intend on running a G0 Q6600 at 8x450 (3.6ghz) or 8x425(3.4ghz)

Click to expand...

i doubt u will be able to they seem to be getting around 3.2Ghz with the Non-ES versions