Why would anybody buy a 3200 90nm over a 3000 90nm?

[overst33r]

I really dont understand what the point of spending $50 bucks more for an unnoticable performance gain.

 

[Duvie]

Some ppl feel more secure with the 3200+ and its higher 10x multiplier. The 9x multiplier can really put some stress on the oc as a 2.5ghz oc will require a near 280fsb and a lot of boards can't do that...Maybe only 1 o2 reliably...

 

[Fingers]

Some people but it for the higher multiplier, others are hoping for a better overclock.

Edit: just beaten to the punch.

 

[saimike]

i got the 90nm because the cheapest s939 chip was a 90nm a64 3000+. i wanted s939 for the upgradeability 2yrs down the road.

 

[StriderGT]

Originally posted by: Duvie
Some ppl feel more secure with the 3200+ and its higher 10x multiplier. The 9x multiplier can really put some stress on the oc as a 2.5ghz oc will require a near 280fsb and a lot of boards can't do that...Maybe only 1 o2 reliably...

I doubt that this is the case (280) -but then again its just my board k8n neo2 plat.- I can go as high as 315 FSB prime stable. There might be other factors blocking them mainly wrong HTx selection... and peculiarities with 2/3 divider (doesn't work with odd cpu multipliers)

 

[Avalon]

Originally posted by: Duvie
Some ppl feel more secure with the 3200+ and its higher 10x multiplier. The 9x multiplier can really put some stress on the oc as a 2.5ghz oc will require a near 280fsb and a lot of boards can't do that...Maybe only 1 o2 reliably...

Pretty much. I'd buy the 3200+ over the 3000+ as well. The 10x multiplier is a sweet spot. You only need 260FSB to hit 2.6ghz, which won't be too hard to achieve with most decent overclocking boards.

 

[Zebo]

Horrible multi on the 3000. Not all boards will be even able to let you max this chip out. AFAIK only the Epox or MSI NF3 is even capable of uber FSB.

 

[Duvie]

Originally posted by: StriderGT

Originally posted by: Duvie
Some ppl feel more secure with the 3200+ and its higher 10x multiplier. The 9x multiplier can really put some stress on the oc as a 2.5ghz oc will require a near 280fsb and a lot of boards can't do that...Maybe only 1 o2 reliably...

I doubt that this is the case (280) -but then again its just my board k8n neo2 plat.- I can go as high as 315 FSB prime stable. There might be other factors blocking them mainly wrong HTx selection... and peculiarities with 2/3 divider (doesn't work with odd cpu multipliers)

I have left mine at the odd multi 9 and booted at 311fsb with the 3:2 divder (133mhz).....But as the others have said after me not many have done this 290-300 range. Pastorjay had bleedinedge.com article did 335fsb with their board buy taking a cpu with a higher multiplier and just droping it to like 7 or something just to test the fsb on these boards...

 

[Speedo]

Interesting thread, since I'm also trying to decide wether to get the 3000+ or 3200+. Btw, it seems as atleast some of you here should be able to answer some of the questions in my thread here (which I haven't go any answers to yet): http://forums.anandtech.com/me...7831&enterthread=y

 

[StriderGT]

Factors that should be considered and ruled out before blaming the FSB:

1. Real Memory Operating Freq and timings (MSI BIOS misreports mem freq and Auto mem settings should be never used when going async)
2. SATA NF3 problems (only Slot3/4 working stable etc), I am using a PATA disk.
3. 1T/2T stabillity issues. Use 2T at the beginning, then switch to 1T and relax your mem timings if things become unstable.
4. Never use Spread Spectrum: Its unstable even when NOT Oced...
5. Non-existant multipliers: There is no 1x,2x,2.5x HT multiiplier at least in the MSI K8N Neo2 Plat. I can post at max 3x353FSB, 2.5x353FSB, 2x273FSB, 1x278FSB, you get the picture: anything below 3x is pointless...
6. Keep Hypertransport frequency below 1050, even though I ve heard reports of reaching 1200, so experiment.
7. Some cpu multiplier/mem divider combinations are peculiar. Eg post crash when I use odd cpu multiplier, mem divider 2/3 and FSB beyond 210. Propably BIOS bug because I can boot at 200FSB and up it within Windows with Clockgen without any further problems... (I could use your feedback on this issue Duvie: Is it 3000+ only related? Is it the official 1.3 MSI K8N Neo2 plat. BIOS to blame?)
8. Stay within ALL these boudaries simultaniously...

You can find a more detailed analysis of the various FSB experiments in the 1T BIOS Test worksheet within the calc below.

 

[hamidjunior]

I'm planing on geting the new MSI RS480M2-IL that has 1000 FSB. Also choosing between 3000+ and 3200+ Winchesters. Cooling will be Zalman 7000 Alcu with Arctic Cermique. U think getting OCZ Plat Rev 2 or Ballistix is unnecessary for geting system 2.6 prime stable, if yes then what mem? And which of those two procs would be better for this setup?

Granted hard to answer this one coz this board didn't come out yet and on top of that it has completely new chipset (Xpress 200). But everybody including AT raving about great OC potential of this chipset.

 

[StriderGT]

Originally posted by: hamidjunior
I'm planing on geting the new MSI RS480M2-IL that has 1000 FSB. Also choosing between 3000+ and 3200+ Winchesters. Cooling will be Zalman 7000 Alcu with Arctic Cermique. U think getting OCZ Plat Rev 2 or Ballistix is unnecessary for geting system 2.6 prime stable, if yes then what mem? And which of those two procs would be better for this setup?

Granted hard to answer this one coz this board didn't come out yet and on top of that it has completely new chipset (Xpress 200). But everybody including AT raving about great OC potential of this chipset.

To get to 2.6 you do not need special memory, A64 runs perfecctly async, so what you will need is extra volt, extra cooling and luck with your specific CPU chip...

 

[Shimmishim]

i'd say mostly for the multiplier

lets say your chip tops out at 2.7 ghz

since the 3000+ is top locked at 9x, you'll need to run a HTT of 300 to get there.

the 3200+ is top locked at 10x, so you need to run a HTT of 270 to get there.

how high of a HTT you can run depends on the board of course and the mem controller...

 

[nealh]

will the 3200 do 2.5ghz reliably??? I thought the 3200 were not reaching the same speeds as 3000....I thought on 3000..2.4-2.6 was a very good chance but necessarily with the 3200

I would like to hit 2500 with 3000/3200 but do not want to spend $300 on ram...I have some valuram from kingston that is running at cas2.5/3/3/7 (Abit IT7 Max 2 rev2)2.7v @ 200 fsb ...is there a way with multipler ratios to get the ram at 200...and fsb at 250 with a 3200

 

[WuGahCha]

agree with Duvie

for overclocking, its like either spend $40 more on a the mobo or spending $40 more on the 3200+ for the 10X multiplier

but even with the very top OC mobo, u r not garanteed to get up to 300HTT or wutever, so buying a mediocore mobo and clocking the HTT lower, and multiplier higher is a much safer bet

 

[Bateluer]

Some people don't OC.

 

[StriderGT]

Originally posted by: nealh
will the 3200 do 2.5ghz reliably??? I thought the 3200 were not reaching the same speeds as 3000....I thought on 3000..2.4-2.6 was a very good chance but necessarily with the 3200

I would like to hit 2500 with 3000/3200 but do not want to spend $300 on ram...I have some valuram from kingston that is running at cas2.5/3/3/7 (Abit IT7 Max 2 rev2)2.7v @ 200 fsb ...is there a way with multipler ratios to get the ram at 200...and fsb at 250 with a 3200

check the calc!

 

[slash196]

Originally posted by: Bateluer
Some people don't OC.

 

[superkdogg]

Losers!

They're the same people who drive the speed limit, aren't they?

J/K

 

[WuGahCha]

so people should either buy the 3000+ or the FX55 rite?? cuz small increments are not noticable

 

[nealh]

StriderGT..very nice OC calc but I am too stupid to understand it....

 

[StriderGT]

give it a second try
green items are the ones you wish your system to operate with and can be altered.
red items are the ones you should set in your bios to achieve the green ones.
and blue items are the actual clock freqs of your system if you have selected the corresponding red ones in BIOS

 

[Markfw]

Or because your mobo sucks at OC'ing like mine. I can only get 2.2 ghz, and still have a problem once in a while with F@H. (Abit AV8) and thats on an Antec true power 380, and Kingston Hyperx PC3500 CAS2 (the good stuff) only running at 189.

 

[PVD]

check my thread on my 3500+ overclock. I am at 2.607 on stock voltages and an FSB of only 237 (x 11). I am going to go higher by dropping the multiplier to 10, but you get the point. This chip and memory kick ass. I would spend the extra hundred to get fx-55 + speeds. I bought mine for $260 OEM from Monarch.

 

[nealh]

so does the 3200 do as well in o/c as the 3000...not in % but max speed...

so if a 3000 can hit 2.5-2.7(if really lucky) will a 3200 also hit 2.5-2.7 reliably...

Also how good is the retail HSF.....