Venice 3000 or 3200

[xer9]

PCPerspective OCd a 3200+ to 2.8Ghz fsb increased, multi at stock 10x... how does that compare to the 3000+ and is it worth $50 more?

 

[imported_llama]

probs worth it

 

[McArra]

3200, you will need the extra multiplier for the OC the Venice can achieve!

 

[deadseasquirrel]

Originally posted by: McArra
3200, you will need the extra multiplier for the OC the Venice can achieve!

naw, it depends on how well your motherboard handles higher FSB speeds. the dfi nf4 handles high FSB well (i'm at 313 * 8x on a 3000+ winnie). 250 * 10x wouldn't have helped me in the least since my ram is value and is pretty much stuck around 200.

 

[McArra]

But Venice can easily do 2800Mhz, let alone a 9x multiplier and.... You better buy 3200 with 10x multiplier this time, it's not a Winnie

 

[sangyup81]

2800 = 312x9

When DFIs can go 330, 340 then why do you need the 10x multiplier again?

 

[Hyperlite]

yeah its a tough call...i'm not sure what i'm going to get yet.

 

[SynthDude2001]

Probably the 3200+.

 

[GML3G0]

depends on if they're speed binned or not.

 

[imported_Kensai]

3200.

 

[Sentential]

Id go with the 3200+ in case they decide to aggresively bin like they did last time :disgust:

 

[McArra]

Originally posted by: sangyup81
2800 = 312x9

When DFIs can go 330, 340 then why do you need the 10x multiplier again?

Even some TCCDs are not easy to get o that speed.... that's why

 

[Solema]

I plan on getting the 3200 for the same reason, that getting my RAM up to 300+ is impossible and I want to get the best chance for a 1:1 OC.

 

[ironique]

3200+ for 10x multiplier and 1:1 ocing!

 

[HeroOfPellinor]

3000+ because it's almost $50 cheaper and you're going to need a divider anyway to hit 2.8 and tests have shown a divider makes hardly any difference.

 

[Solema]

Didn't you read the PC Perspective article? They did NOT need a divider to hit 2.8. The 10X multi is worth it if you have RAM that can run at 1:1 at high speeds.

 

[RichUK]

3200+ 10x multi just to be safe

 

[Duvie]

Originally posted by: McArra
But Venice can easily do 2800Mhz, let alone a 9x multiplier and.... You better buy 3200 with 10x multiplier this time, it's not a Winnie

I agree here for right now......The fact is we have not seen many early results with this chip and the way the CBBID stepping chips were running with high HTT I would be worried about anything too high.....

I will wait and see what the probability from getting each chip may be to 2.8, 2.9 and 3.0ghz and then go from there...I am not against getting a 3500+ if I feel that is what is going to get me to 3ghz. If a lot of these newer 3000+'s crap out in the 2.6 to 2.7ghz range what would eb the point??? My board is tested for 330 but I just dont have the confidence in the chips as of late (post 0448)......

May skip the whole thing if I dont see decent returns above my 2.66ghz now....just put all my efforts in to building the dual core dual opteron system and screw this machine until dual core desktops come out....

 

[deadseasquirrel]

Originally posted by: HeroOfPellinor
3000+ because it's almost $50 cheaper and you're going to need a divider anyway to hit 2.8 and tests have shown a divider makes hardly any difference.

exactly. 10x multi at 280 FSB... most of us out there are still gonna need some kind of divider to hit that. so there won't be any 1:1 overclocking-- and, if you really wanna split hairs about it, i don't think there's any actual 1:1 on the A64s anyway.

the main idea is to get your memory as high as it can handle, while matching it with the right divider to push your CPU speed as high as possible.

so, if your board can handle it, 311 * 9x for 2800mhz. run with a /133 divider if you got cheap stuff that can only overclock to 205mhz or so. or, if you've got the good stuff, run at a higher divider and push your ram more. no matter what, most people won't be achieving any 1:1 overclocks.

280 * 10x with a divider for your ram is going to give you the same performance as 311 * 9x with a divider for your ram (provided your ram is running the same actual speed on both). if your board can handle it, why not save the $50? granted, this is coming from a guy who blew an extra $375 for SLI, so i do agree it is up to everyone to make up their own mind.

 

[sxr7171]

I think the 10x mulitplier is invaluable. Yes, I use a RAM divider with my Winchester 3200+ and I find that get the best performance running 276MHz at 9.5 mulitplier. If I go down to 9x mulitplier then I have to use a lower RAM divider ratio and that affects performance as the RAM runs below 200MHz in that case. When you consider the new Venice cores you might get upto 3GHz and 330 x 9x could be pushing things pretty hard. It depends a lot on your memory and how flexible it is. You still want to try to run the memory as high as possible. In other words you don't want to take your memory down to a lower divider setting and run it at 179 MHz or something because it won't do the approx. 235MHz or so at the next higher divider setting. If we had more choice in divider ratios then we could get away with the 3000+. Still running 330MHz bus might require a little extra NB cooling.

 

[Solema]

sxr, that is EXACTLY why I sold my 3000+ and got a 3200+. With the divider I could not get my ram to run past 187MHz or the next step up was too fast for it to handle. Of course I shot myself in the foot when the 3200+ I got was a CBBID.

 

[sxr7171]

Originally posted by: Solema
sxr, that is EXACTLY why I sold my 3000+ and got a 3200+. With the divider I could not get my ram to run past 187MHz or the next step up was too fast for it to handle. Of course I shot myself in the foot when the 3200+ I got was a CBBID.

Oh man, bad luck. I guess that Venice will change things for the better.

 

[Zebo]

The bad news: Id hate to get a 3000Mhz capable chip and be limited by mobo... Even if you get 333HTT off mobo, this also puts hella stress on Mem controller according to OCZ guy tony in his testing, some simply can't do it.

The good news: 3000 is da bomb due to it's excellent 9x and capable speed when combined with mem it beats the 10x in most configs.

If you get the 3000 w/ 9x and clock to 2800Mhz...That means HTT is 311

You have 3 choices on mem speed with most boards:

1. 200 = 311 mhz, very possible with TCCD. I don't recommed due to sloppy timings, least you get major bandwidth. And will have no problem utilizing expensive TCCD's full potential. This is the worst choice but still there as an option.

2. 166 = 255 Mhz excellent speed and awesome timings w/ OCZ VX for 2-2-2ish or ballistix 2.5-3-2 ish or new BH-5 1.5-2-2. BEST choice.

3. 133 = 205 Mhz, every mem can run this. A boat load of mem can run this at tight timings. The value choice.



With the winny, the 3200 was the way to go because it was only ~2600mhz capable and made for some weird mem combos when using 3000, this has changed with veince. Buy the cheapest, make the best. :thumbsup:

 

[Zebo]

Continuation of my reply above:

say you can swallow that 33% price premium for the 3200 and still clock to the same 2800Mhz. Was it worth it?

with 10x multiplier Htt = 280

1. Memsetting @ 200 = 280Mhz. Only TCCD can do this or ballistix at horrendous timings. And the Ballistix must run 2T.:thumbsdown: TCCD is being robbed of 20 Mhz at least.

2. Mem Setitng @ 166 = 230 Mhz. Sure you could get VX, Ballistix, BH-5 all running at this speed trivially, too trivially. You're robbing yourself of 20-30Mhz you paid for with this ram and really underclcoking it by my standards. Some budget ram will clock here but few.

3, Memsetting @ 133 = 185 Mhz = you're not serious about performance.

Less performance for more $ = Not worth it.

 

[stelleg151]

Yeah, but lets not judge the entire future of the venice on these few oc's that we have right now to work with. Zebo I see your point, but it hasnt been ruled out that a venice might be able to hit up to 3GHZ +, and I would hate to be mobo limited if I had a chip like that. Also not everyone has the Lanparty, my chaintech will only go around 305.